P. 1
0199609837 Constructivism

0199609837 Constructivism

|Views: 29|Likes:
Published by transgresivac

More info:

Published by: transgresivac on Jul 15, 2013
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





Constructivism in Practical Philosophy

Constructivism in
Practical Philosophy
James Lenman and
Y onatan Shemmer
Great Clarendon Street, Oxford, OX2 6DP,
United King4om
Oxford University Press is a department of the University of Oxford.
It furthers the University's objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxrord University Press in the UK and in certain other countries
© the several contributors 2012
The moral rights of the authors have been asserted
First Edition published in 2012
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloging in Publication Data
Data available
ISBN 978-{)....19-960983-3
Printed in Great Britain by
MPG Books Group, Bodmin and King's Lynn
Links to third party websites are provided by Oxford in good f.Uth and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work
In 2009 the editors of this volume were supported by a grant from the Arts and
Humanities Research Council's Research Networks and Workshops Scheme for the
purpose of holding a series of workshops and a major conference on the subject of
Constructivism in Practical Philosophy at the University of Sheffield. This project
would not have been possible without this generous funding support.
The project went ahead, beginning with a workshop on 7 February on Construc-
tivism iri Political Philosophy with papers from Kerstin Budde, Aaron James, Andrew
Williams, and Miriam Ronzoni. This was followed on 28 March with a workshop on
Epistemological Issues in Constructivism with papers from Simon Blackburn, Valerie
Tiberius, Matthew Chrisman, and James Lenman. A final workshop on 20 June looked
at Constructivism about Practical Reason with papers from Jussi Suikkanen, Michael
Ridge, Carla Bagnoli, and T. M. Scanlon. Finally, our conference on 14-16 August
comprised papers by Jay Wallace, Robert Stem, Nadeem Hussain, Yonatan Shemmer,
Sharon Street, Dale Dorsey, and Michael Bratrnan as well as further contributions from
James, Ridge, Tiberius, Lenman, and Williams. In this book we have mainly sought to
publish all (with the one exception ofWilliams's) and only (with the one exception of
Scanlon's) papers from the concluding conference, but following the great success of
the project over all four stimulating meetings we would like to express our thanks to all
those who contributed.
We are especially indebted to Michael Ridge and Valerie Tiberius for their wise
counsel over various issues in the running of the project.
We owe special thanks to Heather Arnold, our administrative assistant on the
project, whose energetic support relieved us of much of the tedious minutiae of
organizing such events and made a huge difference to the success of the project.
Finally, we would like to thank Peter Momtchiloff of Oxford University Press for
his support and advice throughout.
List of Contributors
james Lenman and Yonatan Shemmer
Constructivism about Normativity: Some Pitfalls
R. Jay Wallace
Coming to Terms with Contingency: Humean Constructivism
about Practical Reason
Sharon Street
Constructing Protagorean Objectivity
Aaron james
Constructivism, Agency, and the Problem of Alignment
Michael E. Bratman
A Puzzle for Constructivism and How to Solve It
Dale Dorsey
Constructivism and the Argwnent from Autonomy
Robert Stem
Kantian Constructivism: Something Old, Something New
Michael Ridge
Constructing Coherence
Yonatan Shemmer
A Problem for Ambitious Metanormative Constructivism
Nadeem]. Z. Hussain
Constructivism and Wise Judgment
Valerie Tiberius
Expressivism and Constructivism
james Lenman
The Appeal and Limits of Constructivism
T. M. Scanlon
List of Contributors
MicHAEL E. BRATMAN is Durfee Professor in the School of Humanities and Sciences, and
Professor of Philosophy at Stanford University.
DALE DoRSEY is Assistant Professor of Philosophy at the University of Kansas.
NADEEM). Z. HussAIN is Associate Professor ofPhilosophy at Stanford University.
AARoN JAMEs is Associate Professor of Philosophy at the University of Calfornia, Irvine.
JAMES LENMAN is Professor of Philosophy at the University of Sheffield.
MICHAEL RIDGE is Professor ofPhilosophy at the University of Edinburgh.
T. M. SCANLON is Alford Professor of Natural Religion, Moral Philosophy, and Civil Polity at
Harvard University.
YONATAN SHEMMER is a Lecturer in Philosophy at the University of Sheffield.
RoBERT STERN is Professor of Philosophy at the University of Sheffield.
SHARON STREET is Associate Professor ofPhilosophy at New York University.
VALERIE TmERIUS is Professor of Philosophy at the University of Minnesota.
R.}AY WALLACE is Professor ofPhilosophy at the University of California, Berkeley.
James Lenman and Yonatan Shemmer
1. Constructivism
Constructivism has come to be a major theme in contemporary practical philosophy.
But it is not always clear how it is best understood. What exacdy is constructivism?
Its roots might be traced back at least as far as the writings of Hobbes and Rousseau
which give classical expression to the central ideas of the contractarian tradition in
moral philosophy that views certain moral and political truths as determined by what
deliberative agents, variously characterized, may be expected to agree to in particular
conditions, variously characterized. A still more important influence is Kant, who took
moral truths to be determined by the categorical imperative procedure whereby
maxims governing action are to be tested by ascertaining whether their universal
application is something we are able to conceive as governing a moral community in
which we would be willing to live.
The constructivist dimension to Kant's moral philosophy was emphasized by John
Rawls in his Lectures on the History of Moral Philosophy! and inspired the development of the
constructivist aspect ofhis own work on political philosophy, which in turn has been the
main inspiration for the continuing interest in constructivism in contemporary practical
philosophy more generally. Rawls contrasted constructivism with what he called rational
intuitionism which understood facts about, say, justice (Rawls's own central concern) as
constituted by "self-evident truths about good reasons" where these reasons are "fixed by
a moral order that is prior to and independent of our conception of the person and the
social role ofmorality".
Rawls's constructivism, by contrast, took principles ofjustice to
be determined by the outcome of a procedure (in Rawls's case, the choice of principles of
justice by the parties to the Original Position) designed to express certain moral ideals he
took to be implicit in our public political culture.
Other moral philosophers in recent
times, most notably David Gauthier,• have developed constructivist understandings of
Rawls 2000, esp. Chapter 6 of the section on Kant.
• Rawls 1999, p. 344.
Rawls 1980, 1993.
• See especially Gauthier 1986.
morality that have drawn their primary inspiration from Hobbes, but the term has
primarily come to be associated with the work of Rawls and those closely influenced
by it, notably the inB.uential group comprising his colleagues (f. M. Scanlon, Christine
Korsgaard) and students (notably Korsgaard again, Onora O'Neill, Thomas E. Hill) at
Harvard, and some of their students in tum (notably Valerie Tiberius (Hill), Tamar
Schapiro, and Sharon Street (both Korsgaard) and Aaron James (Scanlon)).
These observations, however, do not go far in characterizing what distinctively
unites these and other constructivist philosophers beyond the contingencies of aca-
demic geography and pedagogical relations. Many of Rawls's own extensive remarks
on constructivism are too heavily loaded with the details of his own particular
theoretical position in political philosophy to be very readily generalizable beyond
these. We would, however, suggest that one feature ofRawls's constructivism that has
appealed to many who have since styled themselves constructivists is his attempt to
make good sense of moral justification in a way that is maximally free of any
controversial metaphysical suppositions. In the version of constructivism Rawls can-
vasses in Political liberalism, what he calls political constructivism, this metaphysical mod-
esty has a rather particular significance. Here, rather than competing with rational
intuitionism and other metaethical theories, political constructivism seeks to be neutral
between them.
This with a view to offering an understanding of moral justification
that can operate in a society that is pluralist about the various views that divide
adherents of competing "comprehensive doctrines", meta ethical views included.
Nonetheless, if we find ourselves persuaded that Rawls has succeeded in adumbrating
an understanding of moral justification that relies on only the most minimal of
metaphysical commitments, it might be thought that such success could hardly fail to
count as a dialectical strike both against such relatively unparsimonious rival under-
standings as rational intuitionism and against forms of moral scepticism and error theory
that might view the possibility of any such justification with philosophical suspicion.
Other more positive common features unite all or most moral philosophers who
class themselves, or who are helpfully classed, as constructivists. Here we note two
central such features. First, of course, there is the issue of procedures. A constructivist
account of some domain of normative facts is often characterized as one that under-
stands them to be determined by the outcome of some procedure.
This might be
understood in terms of agreement among contracting parties characterized as con-
forming to certain standards of rationality or reasonableness (e.g. Rawls, Scanlon,
• The bibliography included in this volume lists relevant work by all these philosophers.
• Rawls 1993, p. 95.
Ibid., pp. 97-8.
• Street 2010, p. 365 quotes Darwall, Gibbard, and Railton's defmition (1997, p. 13):
[T]he constru(:tivist is a hypothetical proceduralist. He endorses some hypothetical procedure as determining
which principles constitute valid standards of morality .... (He) maintains that there are no moral facts
independent of the fimling that a certain hypothetical prO(·edure would have such and such an upshot.
--along with Enoch's (2009) characterization of it as "canonical".
Gauthier) or in tenns of some kind of reflective scrutiny (e.g. Korsgaard, Street). In
some cases this procedure may be the same as the procedure we employ in order to fmd
such facts out, but it is not taken to be something that tracks the facts in question, these
being conceived as constituted prior to and independendy of it. Rather, the procedure
having the outcomes it does is what determines or constitutes the facts. The procedure
may well be central to a correct epistemological understanding of how we can know
about the domain in question but the central, distinctively constructivist claim is the
metaphysical claim about the order of determination moving from procedure to facts
and not vice versa. It is here that the parallel with constructivism in mathematics is most
clear. For mathematical constructivists, mathematical proofS are conceived not as
devices whereby we discover an independent domain of mathematical facts. Rather,
provability just is truth: that we can construct a suitable proof of some theorem is what
makes the theorem true.
Another central commonality is a concern with the grounding of normative truths
in what is sometimes called the practical point of view.
What is meant by this is simply the
point of view characteristic of a deliberating agent, subject to all the motivating states
agents are subject to: desires, plans, intentions, and, perhaps in particular where
constructivists are concerned, normative and evaluative judgements. Constructivists
take this stuff, or some of it, and make it the raw material for their constructive
endeavours. Thus we may read Rawls as using as raw material certain normative
judgements implicit in our political culture which he then seeks to express in the
original position from which his two principles of justice then flow. In this way, for
many constructivists, it is distinctive of constructivism that the raw materials on which
constructivism goes to work are already normative. This might be thought the clearest way
to distinguish constructivism from more reductionist versions of response-dependent
Contemporary constructivists promote a variety of different views and often disagree
about the goals and the shape of the constructivist project. Rawls's political construc-
tivism is a local constructivism, concerned with questions of political justice. So too is
Scanlon's contractualist account of morality.
The more local fonns of constructivism
may be understood as seeking to construct some of the judgements constitutive of
some proper subset of the normative domain, such as justice or morality, using
materials found elsewhere in the practical point of view. Some contemporary con-
structivists about practical reason have sought to defend a global form of constructivism
applicable to the whole normative domain.tt Such ambitiously thoroughgoing
This is central to the characterization of constructivism favoured in Street 2010, a paper that has helped
us immensely in writing this Introduction.
Scanlon 1998.
" E.g. Korsgaard 1996b, 2003, O'Neill 1996, Street 2008, Lenman 2010.
,. "Thoroughgoing" is Street's favoured terminology in her 2008 and 2010 where she contrasts resaicted
and thoroughgoing versions of constructivism.
constructivism cannot set out, like more local constructivism, to ground one domain of
nonnative thinking in materials supplied by another. The concern is rather to offer a
constructivist account of the whole thing where the grounding considerations appeal
to abstract and general features attaching to the practical point of view as such,
appealing, for example, to allegedly constitutive features of the practical point of view
or of normative judgements in general to ground a constructive account.
Particularly influential here is the work ofKorsgaard who has articulated in a series of
books and articles over recent years a form of Kantian constructivism about practical
A second and perhaps not as obvious source of the contemporary construc-
tivist position has been Harry Frankfurt who has done much to develop our under-
standing and thinking about the idea of the practical point of view.
Frankfurt's own
recent nonnative writings might indeed be best characterized as a form of Humean
constructivism, though he has never adopted this tide. We did not brief the contribu-
tors to the present volume to focus their attention on such ambitious global forms of
normative constructivism but many have done so and they have come to form a central
theme of the present volume.
Constructivism differs from rational intuitionism in its metaphysical modesty. If it is
true, the intuitionist's realm of independent facts need not be supposed to exist for us to
make good sense of normative justification. How it differs from standard response-
dependence theories is perhaps most naturally specified by reference to constructivist
understandings of the practical point of view as characterized above, whereby con-
structivist characterizations of the various domains of normative judgements typically
do not seek to free themselves of normative concepts. Constructivism's relation with
expressivism is a rather delicate matter. Sometimes the two are taken to be in competition.
Many writers take constructivism to be a form of cognitivism and hence in competition
with expressivism. But this concern is no big deal: contemporary expressivists would
distinguish their view from noncognitivism,
happily recognizing the truth-aptitude of
all and the truth of some normative judgements if only in some suitably deflationary
"quasi-realist" spirit.
At deeper levels there might seem to be important things in
common. Both aim to be metaphysically parsimonious, consistent with a f.Urly sparse
naturalistic metaphysics. More positively, for both expressivists and the more thorough-
going, global kind of constructivist, normativity is taken to be a feature of distinctive ways
in which what from a third-person perspective are simply features of our psychological
u Korsgaard 1996a, 1996b, 2008, 2009.
•• See Frankfurt 1988, 1998, 2006.
" See Blackburn 1996, pp. 82-8. The point is emphasized by Wallace, this volume, pp. 22-3.
" In Samuel Freeman's generally excellent book on Rawls we find the following rather unhappy
;entence: "As opposed to moral skepticism, which denies moral truths or objective standards of correctness
:e.g. 'expressivism' says that moral statements are expressions offeelings), constructivism affinns that there are
:ruth-conditions for moral statements" (Freeman 2007, p. 291). Expressivism emphaticaUy is t1ot a species of
nora) sn:pticism.
dispositions represent themselves to us from a first-person standpoint. Thus Korsgaard
If constructivism is true, then nonnative concepts may after all be taken to refer to certain
complex facts about the solutions to practical problems faced by self-conscious rational beings.
Of course it is only viewed from the perspective of those who actually face those problems in
question that these truths will appear normative. Viewed from outside of that perspective, those
who utter these truths will appear to be simply expressing their values.
This is almost exactly what expressivists believe. Given which, it might come as no
surprise that Allan Gibbard in his critical study of Korsgaard's Sources of Normativity
took her position to be one that was best made sense of in distinctly more expressivist
terms than Korsgaard herself has professed herself willing to embrace. Moreover,
constructivism is sometimes criticized for being unhelpfully unforthcoming when it
comes to central metaethical issues about the semantics and the logic of normative
concepts about which expressivists have had a lot to say. Thus we find e.g. Wallace
proposing that constructivism and expressivism are mutually consistent answers to
different questions,
and Lenman
urging that not only can versions of these views
be combined but that such combination might enhance them both.
Further controversies abound even about how constructivism is best understood.
Some constructivists have offered their constructivist views with a clear view to
illuminating the central issues of metaethics.
Others have professed themselves
unconvinced that these views shed any light on such central metaethical concerns as
the semantics of normative discourse.
Constructivism is ordinarily understood, in the
words of David Enoch, as "a metaphysical thesis about the relations of truth-making or
correctness-priority between substantive results and the procedures leading to them" .
But not everyone fmds the characterization in terms of procedures helpful or clear.
Some have thought constructivism, or something very like it, might help us to
understand features of moral or normative epistemology.
And it has been proposed
as an objection to constructivism that it fails to make clear and adequate sense of the
possibility of ethical knowledge.
Finally, the practical point of view may be understood in conflicting ways. As noted
above, many constructivists take it as a distinctive feature of constructivist views that
Korsgaard 2008, p. 325. She goes on in a footnote to this passage to say that this way of seeing things
makes expressivism "ttue ... but in a way that makes it boring". For discussion of this rather odd remark, see
Ridge, this volume, footnote 2.
Gibbard 1999.
This volume.
Lenman 2010 and this volume.
Notably Korsgaard 1996b, 2003, 2009, Street 2008.
Notably Hussain and Shah 2006. Cf. Ridge, this volume, Dorsey, this volume, Lenman, this volume.
23 2009, p. 322.
See e.g. Street 2010, p. 265, Ridge, this volume.
e.g. Lenman, this volume, Tiberius, this volume.
See Chrisman 2010.
the raw materials that furnish the construction of normativity, morality or whatever is
at issue, are already normative in character.
For others this is too narrow. Thus Ridge
thinks some readings ofKant that view the relevant procedures not as normative but as
grounded in our conception of ourselves as autonomous are properly classed as
If the raw materials are taken to include ordinary desires, it may
become hard to represent constructivism as a position genuinely distinct from more
general and familiar forms of subjectivism, with the character and role of the practical
point of view no longer interestingly different from that accorded in Bernard Wil-
liams's existence-intemalist about normative reasons to the agent's "subjective moti-
vational set".
Construing these too narrowly, on the other hand, might risk
disqualifying positions thought to be illuminatingly understood as contructivist in
But it would be beyond the scope of this brief introduction fully to document and
discuss the innumerable loci of philosophical disagreement both within the construc-
tivist camp and between it and the adherents of rival views. Rather, we tum to offering
some brief summaries of the articles that follow, articles which will, we trust, speak for
2. The chapters
R. jay Wallace
Constructivism is primarily a view about the nature of normativity. Quite a few authors
in this collection have identified the constructive procedure as the heart of that view.
According to these authors a view is constructivist if it sees norms as the output of a
(specified) procedure of construction. Jay Wallace, on the other hand, sees the consti-
tutive relation as the central element of a constructivist position. On this understand-
ing, fundamental principles of normativity are normative in virtue ofbeing constitutive
of our deliberative activity (activity aimed at solving practical problems). In other
words, fundamental normative principles are normative if in the very activity of
deliberation we are committing ourselves to be guided by them. While constructivism
shares with expressivism its naturalist aspirations, it adds to expressivism this under-
standing of the nature of normativity, and differs from both expressivism and realism in
its ability to thereby answer sceptical questions. A deliberating sceptic will always have
available to her an answer to her scepticism since in the very activity of deliberating she
is committing herself to the normative principles inherent to this activity.
r> E.g. Street 2010, Lenman, this volume .
.. Th1s volume.
•• Williams 1981. Concern about distinctiveness is to the fore in Enoch 2009, pp. 228-30. And cf the
taken by Ridge's critique ofStreet, this volume.
"' The inue i• one that divides the editors.
In his chapter, "Constructivism about Normativity: Some Pitfalls", Wallace con-
siders three pitfalls that a constructivist should beware o£
The first is psychologism. It is easy-but not advisable-to understand claims about
constitutive relations as claims about psychological necessities: the necessity to act in
accordance with some purported normative principles. But then it is open for one to
ask: By what right do these principles govern the will?
The second is bootstrapping. Constructivists claim that our activity of deliberation
commits us to comply with certain principles. Being committed in this way involves
having an attitude of endorsement towards the relevant principle; an attitude that is at
once both a disposition to comply with it and a normative endorsement of the
principle. But a normative endorsement might be mistaken; to insist otherwise is to
bootstrap our potentially fallible normative endorsements into normativity. The con-
structivist has a ready reply for at least one version of the bootstrapping objection. She
can explain local error in normative endorsement by appeal to the possible gap
between our actual endorsing attitudes and the endorsing attitudes we would have
had if we were fully rational. Actual endorsing attitudes that conflict with those we
would have if we were fully rational are mistaken.
But this reply leads us to the third pitfall: the gap between actual endorsing attitudes
and fully rational ones is generated by some normative principle of rationality. The
only way to justify this principle is to claim that it is a principle we are already actually
committed to. But then any other particular principle that the agent is actually
committed to would thereby be normative as well and the constructivist would lose
her ability to explain the possibility of local error.
Sharon Street
Sharon Street's "Coming to Terms with Contingency: Humean Constructivism about
Practical Reason" develops ideas she sketched in her 2008 paper "Constructivism
about Reasons". Here, as in the earlier paper, she defends a form of metaethical
constrnctivism: "[t]he truth of a normative claim consists in that claim's following, as a
logical or instrumental matter, from within the practical point of view". In the 2008
paper, Street distinguishes between two views. The first, mainly associated with
Christine Korsgaard, Street calls substantive metaethical constructivism. According to
that view, there are certain substantive normative claims, notably moral claims, to
which any rational agent, as such, is committed. The second, formal metaethical
constructivism, is Street's favoured view. According to this view, "there is nothing in
particular that one must value if one values anything at all" (Street 2008, p. 244). In the
present chapter she continues her discussion of these positions, now rather more
snappily relabelled Kantian and Humean constructivism respectively, and develops a
detailed critique of the former view and defence of the latter.
The central section of her chapter comprises a reconstruction and a critique of
Korsgaard's regress argument from Sources to the conclusion that we are rationally
required to value our humanity. At a crucial stage of this argument, Korsgaard raises the
question whether one has reason to take anything to be a reason at all, a question Street
thinks illegitimate, making no sense without an illicit assumption of a kind of realism
Korsgaard rejects. The way Street thinks the regress of normative reasons comes to an
end is in recognizing that, once one's web of normative judgements has been rendered
fully coherent by suitably rigorous mutual scrutiny, any demand to provide them with
some further, deeper justification simply fails to make sense.
But ifKorsgaard's argument fails, can moral requirement be categorical? According to
Street, we should allow that moral requirements are categorical with respect to "some
important parts of our evaluative as desire, or what is appealing or pleasant-
but it makes no sense to demand that they be categorical "with respect to any evaluative
nature an agent might have". We should accept contingency: "if one lacks moral concerns
altogether, then morality doesn't bind one". However, this should not undermine our
ability to take morality seriously. Here Street makes two points. The ftrst is that there is a
sense in which, for many of us, it is not contingent that we are moral agents. For many of
us, our moral commitments are so fundamental a part of our evaluative natures that losing
them would be a kind of death. The second is an analogy with love. It is a contingent
matter whom we love, but appreciation of that does not mean we love them any less.
Aaron james
In "Constructing Protagorean Objectivity", Aaron James defends an objectivist form
of constructivism.
The first two elements of this view-the rejection of the possibility of brute error
and the rejection of scepticism about practical reason--are assumed, according to
James, by the constructivist methodology and therefore do not call for a uniquely
constructivistjustiftcation. The third and the fourth elements of this view are in need of
detailed defence: the third is the claim that ideal reasoners will converge on the content
of substantive norms; the fourth consists of three claims: a. that constructivists are
capable of accounting for the legitimacy of a demand to justify the norms of rationality;
b. that by doing so they will not fall prey to sceptical arguments; and c. that con-
structivists are capable of satisfying the demand to justify the norms of rationality.
According to James, constructivists cannot assure us of the truth of the third claim a
priori-whether it is true or not in part depends on the persistence of disagreement
between reasoners about substantive norms. However, the truth of the fourth claim
can be established by constructivists by appeal to constitutive arguments. A constitutive
argument is an argument to the effect that there are norms internal to a certain
activity-norms constitutive of that activity--and that those who fail to comply with
these norms thereby fail to engage in the activity. Korsgaard has famously claimed that
we can ftnd norms constitutive of the activity of choice. James worries, however, that
an appeal to the activity of choice will tum constructivism into a form of voluntarism
and suggests we should focus instead on the activity of making reasoned judgements
about how we ought to act and on the norms constitutive of that activity. Once we
focus on the activity of making judgements we realize that we cannot
undentand ounelves and othen as reasoning unless we comply with certain norms of
reasoning. It is this realization that provides us with an answer--intellectualist and
objectivist-to the demand to justify our norms of reasoning.
A wony of particular importance for James is the possibility of creatures whose
reasoning activity diverges significantly from our own. How can we determine which
one of us reasons correctly and can we answer such a question without appeal to facts
that are independent of our reasoning practices? This wony echoes the central concern
of Hussain's paper. According to James, we answer such a worry by noting that
whatever these other creatures are doing, their activity is not intelligible to us as a
form of reasoning and so ipso facto cannot be a form of valid reasoning.
Michael E. Bratman
Constructivists believe that our judgements about reasons are justified if they are the
output of a (specified) process of reasoning. In particular, a few constructivists see the
process of reasoning as a process of submitting the judgements we wish to justify to
scrutiny from the standpoint of the agent's other judgements about reasons. Call the
judgement whose justification is in question the output judgement and call the
judgements from whose penpective it is scrutinized the input judgements. In his
chapter, "Constructivism, Agency, and the Problem of Alignment", Michael Bratrnan
reminds us of the pressures we must take into account when we come to undentand
the nature of both the input and the output judgements. He then raises the worry that
given these pressures and the nature of the process of scrutiny the constructivist might
find herself with judgements of incompatible nature.
Consider fmt the output judgements. According to Bratman, these, being norma-
tive judgements, must be seen as subject to correction by relevantly situated thinken,
or in other words, to standards of intenubjective convergence.
Consider now the input judgements. Constructivists appeal to these judgements
because they represent the agent's point of view. The nature of that point of view has
been debated at length in the literature about identification and one central lesson of
that debate is that often the point of view of the agent is constituted by attitudes that go
beyond the agent's own normative judgements; it is constituted by conative attitudes
such as love or caring or intending. And these attitudes are not necessarily subject to
pressures of intersubjective convergence.
Finally, consider the nature of the process of scrutiny. In order to characterize that
process in a way that does not assume any substantive normative standard, constructi-
vists such as Street appeal to formal standards that are constitutive of the attitude of
judgement. According to Street, to withstand scrutiny from the point of view of the
agent's judgement is to withstand, for example and among other things, the application
of a formal standard of consistency.
But now we face the problem of alignment. If we allow the input judgements to be
constituted by attitudes such as love and caring, that are not necessarily responsive to
intersubjective pressures, and if we merely apply to them formal constraints like the
constraint of consistency, it is not clear how we are going to get output judgements that
will be susceptible to standards of intersubjective convergence.
Dale Dorsey
In his chapter, "A Puzzle for Constructivism and How to Solve It", Dale Dorsey is
interested in constructivist views such as Street's which understand the truth conditions
of nonnative judgements as "bearing a favored relation to other normative judge-
ments". This relation, in Street's case, is given in terms of their withstanding scrutiny
from the standpoint of other normative judgements. But what, Dorsey asks, do
normative judgements mean? A natural thought is to take their meaning to be given
by their truth conditions, but that appears to lead to the uncomfortable upshot that the
semantic content of some normative judgement nj can be analysed in terms of nj's
bearing the favoured relation to other normative judgements. This is circular and
unilluminating. It also tells us so little about the meaning of normative judgements
that it makes their scrutiny-withstandingness from the standpoint of each other a rather
uninteresting accomplishment. '
The way out of the difficulty Dorsey favours is "to break the link between the
meaning of normative judgments and their truth conditions". In the normative
domain, though not across the board, truth should be understood in terms of coher-
ence. That leaves us free to give an account of the meaning of normative judgements in
terms of their ascription of nonnative properties, such as right and rational, to actions.
But constructivists typically take themselves to be opposed to normative realism, and
thus committed to denying that normative properties exist, so an error theory seems
unavoidable. Not so, according to Dorsey, precisely because we have cut meaning
loose from truth conditions. Normative judgements may refer to normative properties
and normative properties may not exist but that does not suffice to make normative
judgements false. What makes a normative judgement false is rather a failure of
coherence. So we combine a realist semantics for normative judgements with an
anti-realist understanding of their truth conditions.
But how should we understand the truth of claims about coherence? Since, accord-
ing to Dorsey, it is only in the normative domain that truth is understood in terms of
coherence and since whether a set of judgements are coherent is not a normative issue,
the truth of coherence claims is not itself a matter of coherence. This "alethic
pluralism" might seem vulnerable to serious objections regarding "mixed" contexts:
it is not clear how logical connectives might combine normative and non-normative
sentences or how arguments can draw conclusions from a combination of normative
and non-normative premises. Dorsey thinks this can be handled by resorting to a more
general notion of truth* such that some judgement is true* if it is "true as evaluated by
the trnth predicate appropriate for its domain". We now give an account of the meaning of
connectives and of what valid inferences preserve in terms not of truth but of truth*.
Robert Stem
One central objection to value realism that has motivated the development of con-
structivist views is the objection from autonomy. According to this objection, an order
of independent values threatens the autonomy of agents since such values bind them to
act in certain ways and diminish their domain of control. In his chapter, "Constructiv-
ism and the Argument from Autonomy", Robert Stem sets out to evaluate the details
of this argument. He identifies three versions of the argument: first, the claim, inspired
by Kant, that in following moral principles we are determined by our inclination (e.g.
our inclination to pursue perfection); second, Rorty's position that any order of
independent facts-moral or not-is a threat to human autonomy; and third, the
view that the obligatoriness of values is a threat to autonomy. Stern believes that the
third version of the argument is the one to which realists will have the hardest time
responding. Realists might try to claim that the constructivist notion of obligatoriness is
incoherent-either because self-legislation undermines itself or because it is not dear
what is the source of its authority-and therefore that the argument does not pose a
special threat to realism. Stem thinks, however, that constructivists may appeal to
constitutive arguments to explain why self-legislation is not self-undennining and to
the authority of reason over the passions to explain the source of authority of self-
Failing to establish a challenge for constructivists, realists will try to defend the claim
that the obligatoriness imposed by an independent order of values does not threaten
autonomy. Following Clarke, realists might claim that obligatoriness does not require a
legislator since it may be one of the objective properties of values, thus alleviating the
threat to autonomy posed by a legislating agent. The problem with Clarke's position is
that it tells us nothing of how obligatoriness comes about nor of what it consists. It is
here that Stem suggests realists can help themselves to a Kantian solution: obligatori-
ness, on this Kantian understanding, is not imposed by any obligating entity that limits
our autonomy, nor is it an objective property of values. Rather, obligation is the name
we give to the tension that imperfect, passion-driven agents like ourselves experience
when confronted with objective moral reality.
Michael Ridge
Michael Ridge, m "Kantian Constructivism: Something Old, Something New",
addresses the question of whether constructivism can be credibly regarded as a position
in metaethics which is "at once thoroughgoing, novel, and plausible". His defence of a
negative answer proceeds by examining two key texts in recent constructivist litera-
ture, Christine Korsgaard's "Realism and Constructivism in Twentieth-Century Moral
Philosophy" and Sharon Street's "Constructivism About Reasons". In the former
paper Korsgaard proposes that normative concepts do not serve the function of
describing reality but rather of helping us solve certain distinctively practical problems.
Ridge is doubtful of her claim that this constitutes a genuine alternative to cognitivism
and non-cognitivism as this account appears, on the contrary, to be consistent with
either. He goes on to object that Korsgaard's version of constructivism, in seeking to
understand normative concepts as referring to solutions to practical problems, fails to
provide a full and satisfactory account of such concepts. This talk of problems and their
solutions "is itself normative talk" so all Korsgaard has really done is move the
metaethical bump to another part of the carpet.
Ridge now turns to Street, according to whom a normative judgement of some
agent is true if and only if it is able to withstand scrutiny from the standpoint of that
agent's other normative judgement. Ridge's concern is that this leaves it unclear what a
normative judgement is. If normative judgements are representational states like beliefs
or perceptions with word-to-world direction of fit, what is their content? A normative
judgement's content better not be just that it withstands scrutiny from the standpoint of
other normative judgements or the account will be circular. But if it has some other
representational content surely that will serve by itself to determine a truth condition
for normative judgements, making the constructivist machinery redundant. Street
takes judgements about reasons to be judgements about what counts in favour of
what, where this notion is taken to be irreducibly normative. But if we take them to be
belief states that represent irreducible normative properties, this would seem to push
Street, given her naturalistic rejection of such properties, in the direction of an error
theory. These difficulties might be avoided by taking normative judgements, or at least
the primitive normative judgements that furnish inputs to Street's constructive proce-
dure, as pure feelings with neither word-to-world nor world-to-word direction of fit.
But Ridge objects that this is implausible for many reasons and threatens anyway to
collapse into a form of cognitivism. That leaves the most attractive way of understand-
ing primitive normative judgements as a kind of desire. But on this reading, Ridge
doubts if Street's view constitutes a "genuinely new and thoroughgoing approach to
metaethics" as opposed to "a form of sophisticated subjectivism" of the sort defended
by Bernard Williams and others.
Yonatan Shemmer
According to constructivists, the process of norm construction is governed by the
principle of practical consistency. The principle of consistency is a thin principle of
rationality that prohibits agents from both adopting and rejecting the same goal at the
same time. However, according to Yonatan Shemmer, the principle of consistency is
too thin to account for the kind of structural restrictions that agents impose on the
dynamic process of goal management. To account for these restrictions, constructivists
must add another principle of rationality--a broad principle of coherence. The broad
principle of coherence demands integrity or unity among our goals that goes beyond
!trict consistency. In his chapter, "Constructing Coherence", Shemmer explains why
he thinks the principle of consistency is insufficient to account for our habits of goals
management, offers an account of the broad principle of coherence that together with
the principle of consistency governs our goal-adoption process, and explains how
constructivists can account for the normativity of this broad principle by showing that it
also can be the object of construction.
Shemmer considers two challenges to his account of the construction of the broad
principle of coherence:
1. Since on constructivist views the broad principle of coherence governs the
construction of norms, it may be hard to see how to provide a non-circular
constructivist account of its nonnativity. Shemmer argues that the broad princi-
ple of coherence can itself be the output of the process of construction and thus
receive a constructivist justification as long as we do not think of it as a necessary
or universal principle.
2. Since the broad principle of coherence governs the construction of norms it must
have nonnative priority over particular newly constructed norms. Shemmer
explores ways of justifYing that priority without conceding a despotism of our
past selves over our future selves. We can explain such a priority by appeal to an
analogy with the normative status of constitutions in the political domain. There
is, however, a price to that explanation: we are bound to admit a certain inherent
fragility in our normative constructions.
Nadeem]. Z. Hussain
In his chapter, "A Problem for Ambitious Metanormative Constructivism", Nadeem
Hussain attempts to show that any view developed along constructivist lines which
aims to offer a distinct metanormative position will face a version of Russell's Bishop
Stubbs objection. Metanormative positions, according to Hussain, are committed to
providing us with an account of normative facts that does not merely tell us when a
putative normative fact obtains but that also offers an explanation of the nature of these
facts. One natural way to develop such an account leads back to reductive naturalism
and is thus not metanormatively ambitious, since it does not give us a distinct
metanonnative view. The only alternative is to specify the constructive procedure
(the procedure whose output constitutes normative facts) in nonnative terms, that is,
say that the procedure, when properly carried out, ought to yield a certain output. And
that means that the fact that the procedure yields a certain output will itself be a
normative fact. The constructivist will then have to say that this further normative fact
must be understood as constituted by the fact that the procedure of construction yields
yet a further output. The repeated application of the procedure will produce an infinite
series of elements each one of which constitutes the obtainment of the previous
normative fact in that series. Imagine that the constructivist has produced such a series
in order to explain the obtainment of the fact that Bishop Stubbs is rational. A parallel
series can now be constructed in order to explain the obtainment of the fact that Bishop
Stubbs is irrational. And, says Hussain, there are no facts we could appeal to in order to
determine which of these two series is such that all its elements obtain and which one
is not. If we appeal to nonnative facts whose obtainment is explained by other
metanonnative positions we lose the distinctly constructivist nature of the view,
whereas if we try to appeal to facts whose obtainment is explained by the constructivist
story we face the prospect of two new parallel series of construction.
Valerie Tiberius
In "Constructivism and Wise Judgment", Valerie Tiberius presents in outline a first-
order constructivist view that sees all-in reasons as constituted by the judgements of a
wise person. Tiberius hopes that her discussion will help to shift the focus in the
constructivist debate away from semantic and psychological questions and towards
epistemological issues.
The chapter defends three central claims:
1. Practical questions about what we have all-in reason to do are complex and are
therefore best addressed by a theory that is both flexible and capable of incorpor-
ating a multiplicity of fundamental values without committing in advance to the
content of these values. A theory of wisdom promises to satisfy this demand.
2. We can develop a theory of wisdom that would be responsive to theoretical
constraints central to the constructivist approach: that the theory be authoritative
for all agents, that it be empirically sound, and that it be action guiding.
3. A theory that views the output of the deliberation of a wise judge as constitutive
of solutions to practical problems qualifies as a constructivist theory.
james Lenman
James Lenman's chapter, after some taxonomical preliminaries, turns to a critical
examination ofStreet's views. He notes the obvious affinity with expressivism in her
strategy for understanding normative language by seeking to make sense of facts about
reasons in terms of facts about what we judge to be reasons. He goes on to argue that
shifting her view in the direction of expressivism would promise to avoid various
difficulties that beset it, notably the following.
(a) A lack of clarity about what kind of mental state normative judgements are
supposed to be is resolved on an expressivist account which takes them to be a
distinctive subspecies of desire, broadly understood.
(b) For Street, what makes a normative judgement true is just its surviving scrutiny
in reflective equilibrium, giving the idea of reflective equilibrium constitutive as
well as epistemological significance. Scanlon has objected to this claim, raising a
worry about circularity, that the process of arriving at reflective equilibrium
cannot be understood as a method for constructing reasons since it itself involves
making judgements about what is or is not a reason. By moving in an expressivist
direction, Lenman suggests, a constructivist can tame this objection.
(c) On a robustly cognitivist reading of Street where we take normative judgements
to be straightforwardly b e l i e f ~ . their content is left looking mysterious.
Their truth conditions cannot credibly be supposed exhaustively specified in
terms of their coherence with other such states. An expressivist semantics for
such judgements that understands them as a species of desire again offers to fix
Lenman goes on to note Hussain and Shah's objection to Korsgaard's constructivism
that it fails to offer a distinctive position in metaethics. Certainly it is true, he notes, that
Korsgaard offers little illumination of normative semantics. However, an expressivist
understanding of the latter might fruitfully be married to constructivist ideas in ways
that might promise to supply the deficiencies ofboth views.
Finally, Len man proposes an account of moral reasons. He offers a constructivist and
contractualist understanding of the justification of moral judgements, or, more pre-
cisely, of what such justification might amount to given the extreme metaphysical
sparseness of the expressivism he favours. Such an understanding, he argues, can be
combined with an expressivist understanding of the semantics of such judgements to
shed light on the role and status of intuitions in moral argument.
T. M. Scanlon
In characterizing a position in moral philosophy as constructivist, one thing we do is
invite comparison between it and constructivist views about mathematics and set
theory. Such comparison forms a major focal point for T. M. Scanlon's chapter,
"The Appeal and Limits of Constructivism". In both morality and mathematics we
have a subject matter that raises philosophical difficulties. In both cases it is not clear
what grounds our judgement that some claims are true and others false. In both we
need a way of appraising to what extent such truths are independent of us. In both cases
we need an epistemology that explains how we can discover facts about the subject
matter simply by thinking and reasoning about it. And, in the case of morality, we need
a way of accounting for the practical significance we take moral truth to have.
In the case of arithmetic and, more tentatively, in the more complex case of set
theory, Scanlon urges that we can make significant headway with such questions
without engaging in second-order metaphysical theorizing but simply by characterising
the subject matter in first-order terms, that is, with concepts internal to these domains.
Such an enterprise might be constructivist if it characterizes its target domain in terms
of some constructive procedure. But this constructive procedure will eventually be
grounded in foundational judgements whose validity cannot be established by the
constructive procedure, but are rather justified by reflective equilibrium. Thus Rawls's
constructivist account of justice takes justice to be determined by judgements we
would expect to be made by agents, in the Original Position, about what would be
rational for them to choose, and not their judgements about what would be just. The
rationale for accepting this constructive account will employ the method of reflective
equilibrium but this method is not itself constructive. Likewise Scanlon's own con-
tractualist account of right and wrong is constructivist in a similar way.
Both Rawls's constructivist account of justice and Scanlon's of morality have the
virtues of illuminating how these subject matters can be independent of our particular
judgements and choices without being independent of what human beings are like.
The fmal pages of his chapter are devoted to airing doubts as to the possibility of a
constructivist account of reasons quite generally. Scanlon is doubtful about the pro-
spects of deriving such an account from an understanding of what is constitutive of
agency in the manner favoured by Korsgaard and other constructively minded neo-
Kantians. Nor does he see how a constructive characterization of reasons could usefully
be derived from an understanding of rationality. Lasdy, he considers and rejects the
view that the idea of reflective equilibrium could itself be the basis of a constructivism
about reasons. After all, not any old way of arriving at reflective equilibrium will do.
Something is a reason for me if the judgement that it is so is among my judgements in
reflective equilibrium only if the judgements I make in arriving at this equilibrium are
themselves sound. But then the process of arriving at reflective equilibrium cannot be
understood as a method for constructing reasons since it itself involves making judge-
ments about what is or is not a reason.
Chrisman, Matthew. 2010. "Constructivism, Expressivism and Ethical Knowledge," in Intema-
tional]oumal of Philosophical Studies 18, pp. 331-53.
Darwall, Stephen, Allan Gibbard, and Peter Railton. 1992. "Toward Fin de siecle Ethics: Some
Trends," in Philosophical Review 101, pp. 115-89, reprinted in Darwall, Gibbard, and Railton
(eds). 1997. Moral Discourse and Practice (Oxford: Oxford University Press), pp. 3-47.
Enoch, David. 2009. "Can There Be a Global, Interesting, Coherent Constructivism about
Practical Reason?" in Philosophical Explorations 12, pp. 319-39.
Frankfurt, Harry. 1988. The Importance of What We Care About: Philosophical Essays (Cambridge:
Cambridge University Press).
Frankfutt, Harry. 1998. Necessity, Volition and Love (Cambridge: Cambridge University Press).
Frankfurt, Harry. 2006. Taking Ourselves Seriously and Getting it Right (Stanford: Stanford
University Press).
Freeman, Samuel. 2007. Rawls (London: Routledge).
Gauthier, David. 1986. Morals by Agreement (Oxford: Clarendon Press).
Gibbard, Allan. 1999. "Morality as Consistency in Living: Korsgaard's Kantian Lectures," in
Ethics 110, pp. 140-64.
Hussain, NadeemJ. Z. and Nishi Shah. 2006. "Misunderstanding Metaethics," in Oxford Studies
in Metaethics 1, pp. 265-94.
Korsgaard, Christine M. 1996a. Creating the Kingdom of Ends (Cambridge: Cambridge University
Korsgaard, Christine M. 1996b. The Sources oJNormativity (Cambridge: Cambridge University Press).
Korsgaard, Christine M. 2003. "Realism and Constructivism in Twentieth-Century Moral
Philosophy," in journal of Philosophical Research APA Centennial Supplement, pp. 99-122 and
in her The Constitution of Agency, pp. 302-26.
Korsgaard, Christine M. 2008. The Constitution of Agency (Oxford: Clarendon Press).
Korsgaard, Christine M. 2009. Self-Constitution: Agency, Identity and Integrity (Oxford: Clarendon
Lenman, James. 2010. "Humean Constructivism in Moral Theory," in Oxford Studies in MeUI·
ethics 5, pp. 175-93.
O'Neill, Onora. 1996. Towards Justice and Virtue: A Constructive Account of Practical Reasoning
(Cambridge: Cambridge University Press).
Rawls, John. 1980. "Kantian Constructivism in Moral Theory: The Dewey Lectures 1980," in
Journal of Philosophy 77, pp. 515-72 and Collected Papers (Cambridge, MA: Harvard University
Press, 1999, pp. 303-58 (page references to latter).
Rawls, John. 1993. Political Liberalism (New York: Columbia University Press).
Rawls, John. 2000. Lectures on the History of Moral Philosophy (Cambridge, MA: Harvard Univer-
sity Press) .
Ridge, Michael. "Kantian Constructivism: Something Old, Something New," this volume.
Scanlon, T. M. 1998. 1-Vhat We Owe to Each Other (Cambridge, MA: Harvard University Press).
Street, Sharon. 2008. "Constructivism About Reasons," in Oxford Studies in MeUiethics 3,
pp. 207-45.
Street, Sharon. 2010. "What is Constructivism in Ethics and Metaethics?" in Philosophy Cornpas$
5, pp. 363-84.
Wallace, R. Jay. "Constructivism about Normativity: Some Pitfalls," this volume.
Williams, Bernard. 1981. "Internal and External Reasons," in his Moral Luck (Cambridge:
Cambridge University Press), pp. 101-13.
Constructivism about Normativity:
Some Pitfalls
R. jay Wallace
The most ambitious version of constructivism in contemporary philosophy is construc-
tivism about normativity or practical reason. My agenda in this chapter is to offer a
limited assessment of this view. I begin by sketching the main features of the construc-
tivist position, situating it in relation to other approaches, and highlighting its most
important claims. My implicit touchstone in this part of the chapter will be the work
on normativity by Christine Korsgaard, who has done more than anyone else to
advance a kind of Kantian constructivism as a systematic approach to these issues.
I then discuss some potential objections to the constructivist approach to normativ-
ity. There are three potential pitfalls that I shall focus on, which may be referred to with
the slogans "psychologism," "bootstrapping," and "theory of error." I believe that
satisfactory responses are available to the first two objections, and my discussion of
them will help to bring into focus the appeal of the constructivist view. But I shall argue
that the last issue poses serious problems.
1. Normativity and normative skepticism
It is common these days to understand normativity at least partly in terms of reasons.
A typical view of this kind is that of Joseph Raz, who has written that "(t]he
normativity of all that is normative consists in the way it is, or provides, or is otherwise
related to reasons" (Raz 1999, p. 67). Reasons, in tum, are commonly taken to be
considerations that recommend or speak for (and against) prospective attitudes or
responses on the part of a person or agent (cf Scanlon 1998, Chapter 1). Thus reasons
for belief are considerations that speak in favor of affirming or revising one's judgments
about what is the case, while reasons for action recommend the affirmation or revision
of one's intentions.
Some philosophen hold that there are, in addition to reasons, also nonnative requirements. These are
stru<.tural or "wide-scope" principles that regulate combinatiom of t t i t u d e ~ rather than attitudl's considert·d
Considerations ofboth kinds have an important, indeed constitutive connection to
deliberation about belief and action. Deliberation may be understood as reflection that
aims to resolve the question of what one should think or do. So understood, it is an
activity that can be engaged in either individually or collectively; thus a group of
people might engage in deliberation together, with an eye to resolving the question of
what they should undertake to do as a group. Reasons and rational requirements, then,
are considerations that are constitutively suited to figure in deliberative reflection of
this kind. We resolve the question of what we are to think or to do by complying with
rational requirements and reflecting on the considerations that count in favor of
holding certain belie& or acting in certain ways. Normative considerations are thus
considerations that those who are favorably situated will take into account and respond
to, insofar as they are deliberating correcdy.
We might put this point by saying that
normative considerations have a certain intrinsic authority for deliberative reflection.
An important question, to which I shall return, is how we are to make philosophical
sense of the authority of the normative for deliberation.
If normativity has to do with reasons and requirements in this way, then norms may
be understood as principles that state, in substantive terms, what people have reason to
think and to do, and that formulate the structural requirements that govern rational
reflection. Epistemic norms will be principles that specify, for instance, what counts as
evidence in favor of the truth of propositions of various kinds. Practical norms, on the
other hand, will tell us in general terms what agents who are situated in various ways
have reason to do. In what follows I shall focus primarily on practical rather than
epistemic norms, though it is worth bearing in mind that there are analogues to the
issues I discuss that pertain to theoretical as opposed to practical reason.
An important aspiration for much philosophical work on normativity has been to
place this phenomenon in relation to the world of natural objects and processes.
Consider, in this connection, norms of instrumental rationality. These norms instruct
agents to take those means that are necessary in relation to their given ends. In the
modem era, instrumental rationality has widely been viewed as the single unproblem-
atic requirement of practical reason. The instrumental principle makes no assumptions
about the rationality of people's ends, taken one by one. Rational assessment of
people's aims in acting apparendy presupposes that there are objective reasons and
values, providing standards for assessment of ends that are independent from psycho-
logical facts about what people happen to be motivated to pursue. But concerns about
naturalism have led many to doubt whether such independent standards can be
reconciled with the metaphysical commitments of contemporary scientific practice.
singly or in isolation; see e.g. Broome 1999. There are interesting questions, which I shall bracket for purposes
of this discussion, about whether there really are any normative requirements in this sense, and if so how their
normativity is to be explained (e.g. whether it ultimately reduces to the nonnativity of reasons).
Correcdy, that is, in a substantive rather than a merely procedural sense. In this substantive sense, to
deliberate correcdy just is to reach true conclusions about what one has reason to do.
A world that is shorn of objective values or norms leaves no room for rational criticism
of people's ends, but only for Weberian Zweckrationalitiit: the rational detennination of
means to the realization of ends that are taken to be given, as a matter of human
psychological fact.
The naturalist credentials of this position are specious, however. The instrumental
principle says that we stand under a rational requirement of some kind to take the means
that are necessary to achieve our ends; if the principle represents a binding norm of
practical reason, then we are open to rational criticism to the extent we fail to exhibit
this kind of instrumental consistency, regardless of whether we want to comply with
the principle or not. If naturalism really entails that there can be no objective or mind-
independent norms or values, it seems there is no room for the instrumental require-
ment, as an objective principle of practical reason. A more consistendy naturalistic
position would be to reject even Zweckrationalitiit in favor of a skeptical attitude towards
normativity in all its forms.
Complete normative skepticism is, however, a radical philosophical position. Taken
literally, it holds that there are no correct norms of practical reason, and correspond-
ingly no facts about what we as thinkers and agents have reason to believe and to do.
Deliberation, on this view, is an intellectual activity without a subject matter, an
attempt to get clear about questions that have no real answers.
One familiar naturalistic strategy for avoiding this extreme conclusion is to develop a
defiationist account of the activity of deliberation. The aim of such accounts is to
reconstruct the main features of deliberative thought, showing how such thought can
be a sensible or advantageous activity to engage in, and one that has roughly the
features we observe deliberative thought to possess, even if it is not ultimately
answerable to facts or truths about the normative domain. The basic idea in these
expressivist approaches is to explain deliberative thought as the activity of mobilizing
and imposing discipline on a set of distinctively practical attitudes. The attitudes in
question are practical, insofar as they are at bottom detenninations of the will rather
than representations of the way things are independendy of the will. Practical thought,
understood in these terms, is not an illusion, because it is not in the business of making
claims in the ftrst place; hence the fact that there are no mind-independent facts or
truths about the normative does nothing to show that practical thought trades in false
representations about what is or is not the case.
It will be helpful to introduce constructivism about normativity by comparing it to
expressivism in this form. Constructivists, to begin with, apparendy share the natural-
istic metaphysical commitments of expressivists. In particular, both constructivists and
expressivists agree in rejecting a certain kind of realism about the normative realm: the
position, namely, that there are facts or truths about normativity that are prior to and
independent of those who engage in deliberation about what to think and to do.
Constructivists agree with expressivists, furthermore, in holding that practical thought
is a matter of solving practical problems rather than identifying the solutions to
theoretical questions or describing a reality that is taken to have independent existence.
They apparently differ from expressivists, however, in postulating that there are
normative facts or truths, and that these facts are dependent on the will of those
whom they regulate. The idea that is central to constructivism is that we construct
or bring about through our agency the normative realm. Thus, whereas the normative
realist holds that there are normative truths that are prior to and independent of the
will, a constructivist position will say that such truths are distinctively the result of our
volitional activity.
I shall come back to the comparison between expressivism and constructivism
below. The contrast I have drawn suggests that proponents of these views disagree
about whether there are normative facts or truths; but their disagreement on this point
turns out to be more elusive than this initial crude contrast might suggest. Before
turning to this issue, however, we need to get clearer about the central commitment of
the constructivist position, which is the claim that normativity is somehow the creation
or product of our activity as agents. This is a metaphor of a kind; the question is
whether it can be unpacked in terms that are plausible and perspicuous.
There is one class of positions in moral and political philosophy that offers a
comparatively clear account of the idea of constructing a normative realm, namely
contractualist positions. These accounts represent the basic normative principles of
justice or morality as the result of the choices of hypothetical agents under carefully
circumscribed conditions. Thus John Rawls defines justice by reference to principles
defming the basic structure of society that a self-interested chooser would select from
behind a veil of ignorance (Rawls 2005). Similarly, T. M. Scanlon takes morality to be
specified by principles for the general regulation of conduct that could not reasonably
be rejected by persons who are informed about their situation, and motivated by the
concern to arrive at a set of principles that no one could reasonably reject (Scanlon
1998). But it is striking that these theories attempt to construct only a part of the larger
realm of normative concepts: in Rawls's case, the realm ofjustice, and in Scanlon's the
morality of right and wrong? The contractualist apparatus that gives a content to the
metaphor of construction is in each case embedded within a larger context of norma-
tive concepts that are not themselves explicated in terms of a hypothetical choice
situation, including reasons for action that are grounded in agents' projects, life plans,
relationships, aspirations, conceptions of the good, and so on.
Constructivism about normativity tout court takes a considerably more ambitious line.
It holds that all normative facts and principles are ultimately constructed through the
volitional activity of the agents to whom they apply. To make good on this position,
however, it seems that we will need to unpack the metaphor of construction in terms
that are different from the terms that contractualism offers. For I doubt that we can
Thus as I mentioned above, Scanlon's position is a fomt of realism about nomtativity in general
(including what he would call morality in the broader sense), within which his constructivism about the
"morality of right and wrong" (or morality in the narrower sense) is embedded. I discuss this aspect of his
position in Wallace 2006, Section 4; see also Street 200!!, Sections 3 and 4.
really make sense of the idea that all nonnative principles are the result of a hypothetical
choice under specified conditions. To render this idea coherent, we would need to
imagine agents making choices under parameters that can be represented without
recourse to any normative concepts whatsoever, and our ideas of agency and choice
would arguably give way under those conditions. Thus, the adoption of intentions and
plans by agents ordinarily takes place within the context of a set of views about what
there is reason for them to do; remove that context, and it is doubtful that we would
see them as genuine choosers (as opposed to creatures who make intentional move-
ments in response to stimuli of various kinds). As Thomas Nagel has said, some
unchosen normative constraints on human choice would seem to be among the
conditions of its possibility (Nagel 1970, p. 23).
If I understand the constructivist position, it would seem to move in a rather
different direction from these familiar contractualist accounts. The generalized con-
structivist account of normativity rests ultimately on the idea that the normative
principles that govern the will are the very same principles that are constitutive of
the will in the first place. This is a theme that has become especially prominent in
Christine Korsgaard's recent work, which treats normative principles as standards that
are constitutive of agency itself (see Korsgaard 2009, Chapter 2). Thus, in a sympathetic
discussion of Kant's views on freedom and morality, she writes that "the categorical
imperative is a principle of the logic of practical deliberation, a principle that is
constitutive of deliberation, not a theoretical premise applied in practical thought"
(Korsgaard 2008, p. 321). The idea seems to be that the normativity of practical
principles is to be accounted for by showing that we commit ourselves to complying
with them when we undertake to act. Normative principles are universally binding on
all agents, in these terms, just in case a commitment to such principles is constitutive of
willing anything at all. Candidates for universal principles of willing in this sense
include the instrumental principle discussed above, telling us to take the necessary
means to our ends, and (in Kantian variants, such as Korsgaard's) the categorical
imperative or the moral law.
The element of construction in this approach is to be understood not in terms of the
hypothetical choice of principles (as in contractualist theories). It refers, more basically,
to the fundamental role of practical activity in grounding the principles that regulate
that very activity. Principles of practical reason are not principles whose normative
force is prior to and independent of the will. Rather they are principles that are
rendered normative for us by our act of willing itself, insofar as they describe constraints
that we constitutively aspire to comply with in willing as we do (compare Street 2008).
2. Three pitfalls
With this oudine in place, I would like now to return to the comparison between
expressivism and constructivism, which will help to bring into focus the challenges for
constructivism that I wish to address. Constructivists and expressivists, I said above,
agree in accepting an essentially practical conception of deliberative thought, while
disagreeing about the postulation of normative facts or truths. But this is not quite
Note, for one thing, that many contemporary expressivists are happy to agree that
normative thought involves cognitive states whose contents are assessable as true or false
and subject to the standards of truth-functional logic.
The expressivist point is not that
deliberative thought cannot trade in such attitudes, but that they are to be explained in a
distinctive way. In particular, deliberative thought does not exhibit these features
because it is answerable to an independent domain of normative facts or truths. Rather,
these features are applicable to practical thought because the motivational states that it
involves are structured in such a way as to undergird the ascription of truth to proposi-
tional contents associated with them. There are, fundamentally, just the practical
orientations of the will that are implicated in agency, such as intentions, plans, and
commitments to action; these give rise to thoughts that are truth-assessable and subject
to logical requirements because of the distinctively practical structure that is latent in
them. The expressivist thus explains normative truth and belief practically, showing
how we "earn the right" to talk of truth in this domain when and because our agential
commitments are appropriately patterned.
But this comes very close to the construc-
tivist claim that normative truths are not prior to and independent of the will, but
created by our volitional activity.
The constructivist might respond that different kinds of dependence are at issue in
the two positions. The expressivist is interested in characteristically metaethical issues,
about the semantic and logical properties of the language we use to talk about
normative and evaluative issues. Constructivism, by contrast, appears to be making a
high-level claim about the truth-makers of normative assertions, to the effect that
normative claims are true in virtue of our commitment to the principles that determine
them. Does this contrast help us to isolate the commitment that is distinctive of the
constructivist position? This question brings me to the fl.rst of the pitfalls for construc-
tivism that I wish to discuss.
(a) Psychologism
Expressivists and others are likely to object to the proposal just made about how we
should understand constructivism. The proposal gives us a way of distinguishing
constructivism from expressivism, but at the expense of committing the constructivist
to an implausible position, one moreover that is a position of the wrong kind.
• See, e.g., Blackburn 1998, Gibbard 2003. For a critical discussion of this line of thought, see also Broome
2008, pp. 102-9.
' If expressivism is developed in this way, the question arises as to how it can avoid saying that normative
thought involves an illusion, since the nonnative beliefs that it trades in &il, by hypothesis, to represent an
independent domain of normative facts or relations. One response would be to deny that the beliefs that are
at issue are descriptive or representational states; see e.g. Horgan and Tinunons 2006. I return to this question
Constructivism is presumably intended to be a general philosophical account of the
nature ofnormativity. Like expressivism and other metaethical theories, it abstracts in
the first instance from the content of our normative views, and asks what it is for our
normative views to be correct. The thesis that it advances about the dependence of
normative truth on agential commitments is a general philosophical position of this
Consider, now, the suggestion just made about how constructivism differs from
expressivism: the constructivist holds, unlike the expressivist, that "normative claims
are true in virtue of our commitment to the principles that determine them." This
proposal is naturally read as a first-order normative view, to the effect that we have
reason to do something just in case, and just because, we are committed to principles
that entail that this is the case. On this substantive reading, the constructivist is saying
that our agential commitments have a certain kind of normative significance. Either
they constitute in themselves reasons for action, or they make it the case that we have
reason to do what they commit us to. But if we interpret it in this way, then
constructivism appears to consist in a set of first-order claims about the kinds of
considerations that ground or constitute our reasons for action. Furthermore, the
first-order claims that it endorses, on this interpretation, are highly controversial
ones. It certainly does not seem to be the case that facts about our practical commit-
ments are the only considerations that count in deliberation as reasons for action. Nor
does it strike us that psychological facts of this kind are the only considerations that
make it true that we have reasons for action of various kinds. Constructivism, on the
reading that has been suggested, appears to offer an implausibly psychologistic account
of the considerations that constitute or ground our reasons, one that is at odds with
many of our considered deliberative convictions.
I assume that this psychologistic interpretation of constructivism should be rejected;
the question is whether an alternative can be formulated that preserves an interesting
distinction between constructivism and expressivism. As a first step toward an im-
proved interpretation, we should note the emphasis that constructivism places on the
perspectival character of deliberative reflection. We engage in such reflection, Kors-
gaard contends, from the first-personal standpoint of an agent who is confronted with a
practical predicament of some kind (Korsgaard 1996, Lecture 3). Deliberation from
that point of view reaches a successful conclusion when the practical predicament has
been resolved through a determination to act one way or another in response to the
considerations that present themselves. First-order normative conclusions, we are to
suppose, are conclusions that are endorsed from this deliberative perspective and that
play a significant role in determining how the deliberative predicament is thus brought
to resolution.
Now constructivism, as a philosophical theory about the nature of nonnativity, is
presumably not meant to be a set of claims that are themselves to be endorsed within
this kind of perspectival deliberation. It is better understood as a theory about such
deliberation, offering an account of the nature of reasons for action that explains why
they are suited by their nature to figure within deliberation in the way they clearly do.
The theory says that reasons are detennined by the principles that are constitutive of
the deliberative point of view, insofar as we commit ourselves to complying with them
when we engage in reflection from that point of view. These principles include both
contingent commitments (connected e.g. to practical identities that differ from one
agent to another), as well as commitments-if there are such-that are inescapably
built into the nature of practical reflection itsel£
In its abstraction from first-order practical deliberation, constructivism resembles
expressivism in certain important respects. Both of them are meant to be theories that
are about normative thinking rather than first-order normative positions in their own
right. Both of them, in addition, attribute a certain kind of significance to distinctively
practical commitments to action in their accounts of what is involved in deliberative
thought. But there are noteworthy differences as well. Expressivists offer an interpre-
tation of practical reflection as the formation and elaboration of such states of mind as
intentions, plans, and practical commitments. The idea is to explain in terms of such
states how the language that is deployed in normative reflection could come to exhibit
the semantic and logical features that are proper to it; an explanation that is couched in
these terms, if successful, will enable us to make sense of normative thought and
discourse without needing to postulate a realm of non-natural normative objects or
properties or relations.
Constructivism, as I understand it, is addressed to a different set of questions. It is not
intended as an account of the semantic and logical features of normative discourse, but
as a philosophical explanation of the nature ofnormativity, offering a general account
of the conditions that make true claims about what a given agent has reason to do. The
proposal, as we have seen, is that normativity is a product of practical activity, insofar as
considerations are made normative for us through the commitments that constitute our
standpoints as agents. A theory of this kind is not, at least in the first instance, in
competition with expressivism, because it is attempting to account for a different set of
phenomena. Thus, for all I can see, expressivism and constructivism may be compatible
with each other. The constructivist could endorse the expressivist explanation of the
semantic and logical properties of normative discourse, at least as it figures in contexts
of actual deliberation (a possibility that I shall come back to below). And it is open to
expressivists for their part to accept what the constructivist says about the nature of
It will therefore be possible to distinguish between philosophical and deliberative accounts of the "truth-
makers"' for particular nonnative claims. Within deliberation, the considerations that make it true that I ought
to do X will be f.nt-order considerations that are determined by the particular nonnative principles I accept.
So if I am conunitted to the principle that people ought to keep their promises, then the fact that I have
promised to do X is something that I might cite, within deliberation reflection, as "making it true"' that
I ought to do X. The philosophical story about what makes nonnative claims true in general is not intended
to play the same deliberative role. It should be possible, however, for deliberating agents to acknowledge the
truth of a philosophical theory such as constructivism, even if it is not meant as a direct answer to deliberative
questions. What effect this acknowledgment might have on agential deliberation is an interesting question
that I cannot go into here.
Both theories begin by stepping back from the deliberative point of view
that they theorize about, but they are attempts to account for different sets of phenomena;
if this makes it difficult to situate them in relation to each other, it also opens the possibility
that the two theories might be combined.
The project in which the constructivist is engaged is to make philosophical sense of
normativity, providing a kind of explanation ofhow normativity is so much as possible.
But what kind of explanation is the constructivist after? To shed light on this question,
it might be fruitful to compare constructivism with normative realism. Korgaard herself
famously accuses the position she refers to as "substantive realism" of a kind of
dogmatism (Korsgaard 1996, pp. 28-48). Though sympathetic to normative realism
myself, I think I understand the point she is making. The normative realist posits that
there are mind-independent facts or truths about what people have reason to do,
arguing that this assumption is not metaphysically objectionable, and that it can be
combined with a plausible account of the epistemology and motivational output of
deliberation (Scanlon 1998, Chapter 2). But these arguments, even if successful on
their own terms, do not amount to a comprehensive philosophical explanation of the
phenomenon that they attempt to account for.
Thus, there is nothing in the realist position that tells us, in perfectly general terms,
what it is for something to be a reason, in a way that would illuminate what I referred
to above as the authority of normativity for deliberation: the idea that deliberative
thought and motivation are answerable to normative considerations, in virtue of the
nature of those considerations. On the contrary, there are features of the realist position
that make normativity seem even more puzzling than we might otherwise have
thought it to be. On the realist approach, deliberative activity is said to be answerable
to principles that are prior to and independent of it. But it might intelligibly be asked
why we should care, when we engage in deliberation, about principles that are by
hypothesis independent of the activities they are said to govern.
The structure of
the realist position invites skeptical questions of this kind, while depriving us of the
resources to answer them.
Constructivism, by contrast, offers a general philosophical account of normativity,
appealing to the idea that considerations become normative for us through the
Compare Korsgaard 2008, p. 325, where she remarks that expressivism is true (albeit "boring"). For their
part, expressivists who accept constructivism might well have to give up claims that actual expressivists
sometimes espouse; for discussion, see Street 2010, Section 7. My point is not that the position that results
fiom marrying expressivism and constructivism would not differ from the position that many expressivists
actually accept in various respects, but merely that the core commitments of these positions could coherendy
be combined.
This is, of course, a question that is couched in normative terms. It trades on the distinction between two
kinds of reasons: reasons for action, on the one hand, and reasons for caring, on the other. Constructivism
coUapses this distinction, treating reasons for action as considerations that agents are already conunitted to
thinking that they have reason to care about; this is what ensures that there will always be a dialectically
effective answer to the skeptical question when it is posed of considerations that in fact qualify as reasons, by
the theory's lights.
conunitments that are constitutive of agency in the first place. This general account,
moreover, seems suited by its nature to illuminate the authority of normative con-
siderations to govern the deliberative reflections of the agents to whom they apply. If
principles become normative for us through our volitional and deliberative activities,
then in engaging in those activities we have already committed ourselves to complying
with the principles to which the activities are answerable. Skeptics will therefore find
that they cannot consistently challenge the authority of nonnative principles while
occupying the standpoint of agency (K.orsgaard 2009, pp. 32-3). The constructivist's
philosophical account of normativity promises to shed light on the fundamental idea
that there are normative truths to which we are responsible when we deliberate about
what to do. It does this by interpreting those truths as created by the very activities they
are to govern.
To this it may perhaps be responded that the anti-skeptical strategy latent in
constructivism is psychologistic in a different way. The virtues of constructivism over
realism, I have suggested, consist in the constructivist's ability to illuminate the
authority of normative considerations for the rational will. The constructivist otfen a
general account of what it is for something to be normative, from which it follows that
normative considerations have to be taken seriously by the agent to whom they apply.
In his illuminating recent discussion of Korsgaard's position, however, Derek Parfit
contends that its account of authority trades on a subde equivocation between norma-
tive and psychological force (Parfit 2006, pp. 351-80). The constructivist maneuver of
anchoring normative facts in the will, Parfit argues, would establish at most that such
facts represent psychological necessities for the agents whom they purport to govern.
The skeptical question about authority, however, should be understood in irreducibly
normative terms. It is not the question of whether normative principles inevitably
determine the will, but the different question: By what right do purported normative
principles govern the will?
Parfit is, of course, correct that a psychologistic response to the question of authority
should be rejected. But I do not believe that the account of authority implicit in
Korsgaard's constructivism needs to be understood in these psychologistic terms.
A better label for her strategy might be dialectical. Skeptical questions in this general
area, including the question about the authority of basic normative requirements, are
posed from the perspective of deliberative agency. Skeptics are imagined to be asking,
of some candidate principle or norm, why they ought to care about complying with it,
and this is itself a deliberative question. On the constructivist account, what makes a
principle normative for a given agent is the fact that the agent is committed to
accepting it from the deliberative point of view that they necessarily adopt when
they undertake to act. If this is right, however, then it will follow from the construc-
tivist account that for any question of authority that might be raised about a genuinely
normative principle, there must be an answer to the question that is available to the
skeptic, and couched in terms that the skeptic is already committed to endorsing. To be
dear: it is not the fact that the skeptic is committed to a given principle that itself
answers the question of authority. That quasi-psychological fact is not a consideration
that is to be presented to the skeptic in deliberation, as a reason for accepting the
authority of the principle to which the skeptic is said to be committed. Rather, the fact
about the agent's commitments enters into a philosophical account that explains why
there must be a deliberative answer to any skeptical question that agents might try to
raise about the principles that are normative for them. The role of normative principles
in constituting the standpoint of agency makes it the case that the authority of those
principles carmot be questioned at a fundamental level, from the only standpoint that is
available to agents for raising normative questions in the first place.
I said above that the constructivist theorizes about the standpoint of practical reason
rather than engaging in first-order reflection about what there is reason to do from
within that first-person point of view. This is what enables the position to evade the
charge of psychologism, on either of the two versions of it that I have considered. It
should be added, however, that it is open to the constructivist to contend that the
nature of agency and normativity is such as to determine certain answers to first-order
normative questions. Thus, Kantian constructivists such as Korsgaard hold that the
principles constitutive of agency ultimately include the moral law, a principle enjoining
us to acknowledge the value of humanity as a constraint on our actions and choices.
This is a first-order normative conclusion, and in endorsing it the constructivist might
appear to be ignoring the distinction I have been at pains to emphasize in this section,
between first-personal deliberation on the one hand and detached philosophical
reflection about the nature of such deliberation on the other (compare Smith 1999).
The worry is misplaced, however. Though I do not believe that Korsgaard's Kantian
argument works (see Wallace 2004), it is perfectly coherent to suppose that philosoph-
ical reflection on the nature of agency might reveal that a commitment to comply with
certain substantive principles is constitutive of it, including e.g. a commitment to the
moral law.
(b) Bootstrapping
The constructivist idea is that our activity as agents commits us to complying with the
principles that define what we ought to do. But what is it to be committed in this way
to complying with a given principle of action? For help in answering this question, we
might consider the instrumental principle, which tells us to take the means that are
necessary relative to the ends we have chosen to pursue. A constructivist account of this
principle says that the attitude or stance of intending a given end carries with it a
commitment to comply with the instrumental principle, so construed (Korsgaard 1997,
p. 245). The idea seems to be that we would not really count as intending the end if we
were not committed to taking the means we regard as necessary for its realization. The
instrumental principle can be described as a constitutive principle of the will, insofar as a
certain level of compliance with it is necessary if we are to will an end in the first place,
and this is supposed to explain why it is binding on all agents who will anything at all.
Here, the stance of commitment seems to involve. at the least. a kind of active
disposition or striving. To intend an end is to be actively engaged in bringing the end
about, and one can be said to have this attitude toward the end only if one is similarly
committed to taking the necessary means.
An active disposition to comply with a principle, however, would not alone seem
sufficient to make the principle normative for an agent, on the constructivist view. To
see this, it will help to recall the perspectival character of deliberation, as the construc-
tivist conceives of it. Deliberation is a response to a practical predicament, one that is
constituted in part by our capacities for reflection. We find ourselves confronted with
multiple options for choice, and are able to step back in reflection from the desires and
inclinations that initially make some of those options appear more attractive than others
(Korsgaard 1996, pp. 92-100). We do this by posing to ourselves a question that is
couched in explicidy normative terms: "Granted, I find that I am attracted to option
X. But is this an option that I really ought to choose or go in for?" Deliberation is
brought to a successful conclusion, Korsgaard suggests, when it results in a state of mind
that answers the question from which reflection began. One commits oneself to a law
or principle of some kind, and thereby resolves the predicament to which deliberation
is a response. If the act of commitment is to play this role, however, then it cannot be
understood as a brute disposition or causal tendency to action. A motivational state of
that kind would simply be another mental item of the kind that one initially stepped
back from in thought, not a state that is suited by its nature to answer the normative
question from which deliberation begins.
For these reasons, it seems that something else must be involved in the stance of
commitment, beyond a brute disposition to action. But what might the additional
element be? On the most plausible interpretation, I would suggest, we should under-
stand commitment to incorporate an attitude of normative endorsement. To be com-
mitted to complying with a principle, in the way that renders it normative, is not
merely to be disposed to comply with it, but also to regard the behavior to which one is
thus disposed as justified, something that one has reason to engage in.
It is to have a
kind of normative belief, accepting the principle as a normative "law" from which it
follows that one ought to act in a certain way. Only an attitude of this kind, it seems,
Compare Street 2008, pp. 230--1. Street holds that the commitments that ground reasons are like desires
in being essentially motivating attitudes, but unlike desires in involving beliefS with nonnative content. But
she also holds that the motivations at issue need not be commensurate with the normative contents to which
they are can fail to do what one believes one has most reason to do (see Street 2008,
pp. 228-9, n. 37, where situations of this kind are said to involve a form of irrationality). This raises two
questions. First, what role in the constructivist position is played by the stipulation that normative commit-
ments are motivating, if we concede from the start that the relevant motivations might diverge in causal
strength from the judgments with which they are connected? That is, would it make any difference to the
consmiCtivism of the position if we allowed that nonnative commitments might sometimes not be motivating
at all? Second, how do we make sense of the irrationality of cases in which agents are not motivated to a
degree commensurate with their normative convictions? (lf there is a principle that stipulates that it is
irrational to fail to intend to do what one believes one has most reason to do, then that principle had better
admit of a constructivist explanation.) Korsgaard, for her part, does not seem to countenance this form of
irrationality; for discussion, see Wallace 2001, Section 1.
could resolve the deliberative predicament, since it alone involves acceptance of a
proposition that answers the normative question to which practical reflection is a
It remains somewhat obscure, however, how this further dimension of volitional
commitment helps to account for the normativity of the principles to which agents are
committed. If the story I have been telling is on the right lines, then to be committed to
a principle is (inter alia) to regard compliance with the principle as something one has
significant reason to go in for. In these terms, the instrumental principle would be
binding on us, for instance, insofar as we already regard it as binding in willing
something as an end. You cannot set yourself an end in action without having a
tendency both to comply with the principle, and also to judge that you should adopt
the means that are necessary relative to the end you have chosen.
To this it will be objected that the constructivist story involves an implausible form
of bootstrapping. We are looking for a philosophical account of the normativity of
practical requirements such as the instrumental principle; the constructivist approach, as
I have been reconstructing it, explains normativity by appeal to motivating attitudes of
normative endorsement. From the mere fact that I regard a principle as normative,
however, justifying the actions that comply with it, it does not appear to follow that the
principle is normative in fact. There is a potential gap between the subjective attitude
of normative endorsement and the objective normativity of the principle thus en-
dorsed. In light of this gap, it is still not clear how the constructivist explanation I have
sketched could be made to work. Even if there are principles that agents are committed
to complying with, in the sense that involves normative endorsement, it would still
seem to be an open question whether those principles really are normative for human
agency. Thinking something is the case--even when it is true that one cannot help
thinking that it is the case--does not make it so.
It will help at this point to distinguish between local and global versions of the
bootstrapping objection. It would clearly be implausible to offer a theory of norma-
tivity that had the consequence that we are infallible about what we ought to do; there
has to be some room for at least local errors in normative reasoning, cases e.g. in which
agents mistakenly believe that they ought or ought not to do something. To make
room for this possibility, constructivism requires a theory of error, an account that
explains how there can be mistakes in normative reflection despite the fact that
normative principles are anchored in the attitudes of those they govern. The general
shape of the constructivist response to this problem is to appeal to procedures of ideal
reflection, contending that normative facts are determined not by an agent's actual
normative attitudes, but by the attitudes that the agent would arrive at by following the
specified ideal procedures. It is those attitudes that represent the agent's true volitional
commitments, and that therefore determine what the agent genuinely ought to do.
I shall raise some questions about the constructivist theory of error below. Even if we
accept that it accounts for local errors in normative thought, however, the resulting
theory seems to have the consequence that there is a kind of global bootstrapping of
normativity that remains potentially problematic. The theory says that the totality of
reflectively stable attitudes that a given agent has about normativity determines what it
is in fact normative for the agent to do. But this again erases the important distinction
between subjective views about what is normative and objective facts about norma-
tivity. It has to be at least conceivable that an agent's procedurally corrected normative
attitudes might be mistaken, and this is a possibility that the constructivist appears,
following the tradition of a kind of idealism, to deny.
In response to this concern, constructivists should, I believe, simply bite the bullet. The
global bootstrapping worry, in a nutshell, is that normative thought might ultimately fail
to correspond to the reality it attempts to represent. As I noted above, however,
constructivism seems to rest on a qualified skepticism about the very idea of a completely
mind-independent reality that is authoritative for deliberative thought and activity. The
anti-realist idea here is that it is not really plausible to suppose that an activity of this kind
should be answerable to a realm of facts that is genuinely prior to and independent of it. If
this supposition is suspect, however, then the gap that I just noted, between an agent's
(corrected) normative attitudes and the objective facts about normativity that those
attitudes attempt to represent, cannot really open up after all, because there are no
mind-independent facts of the relevant kind. The constructivist simply rejects the con-
ception of objectivity that the bootstrapping objection presupposes (Street 2008).
Another way to put this point is to note that normativity, at least on Korsgaard's
constructivist account of it, is an artifact of the deliberative point of view (compare
Street 2009, sees 2-3). Normative facts come into existence through our deliberative
and volitional activity, and they can be grasped or made sense of only from the
deliberative point of view. When we step back from the normative commitments
that are constitutive of that standpoint, and ask whether those commitments do or do
not correspond to putatively objective normative facts, then normativity seems to
disappear from the scene. We are left with a conception of the world that includes the
attitudes of individual agents-including attitudes with a normative content-and a
world that is devoid of independent normative constraint.
This brings us back to the comparison between constructivism and expressivism.
Proponents ofboth of these positions seem to agree that there is no normative reality
that is ultimately mind-independent. When one abstracts from the deliberative point of
view and engages in philosophical reflection about normative thought, all that will
come into focus are the practical attitudes of the agents who are engaged in such
thought, agents who are by hypothesis operating in a world that is devoid of objective
normativity. Expressivists and constructivists agree on that much. Unlike most ex-
pressivists, however, constructivists do not conclude from this that there are no
normative facts. The correct conclusion, they maintain, is that normative facts are
not independent of the deliberative activities that they regulate. We make principles
binding on ourselves through our (normative) commitment to comply with them in
deliberation, and we grasp the normative conclusions to which we are thus committed
from the perspective that is constituted by that deliberative activity.
Consider, in this connection, the role of beliefS in deliberative reflection. I noted
above that many contemporary expressivists are prepared to acknowledge that norma-
tive thought involves attitudes ofbelief, construed as cognitive states that are open to
assessment as true or false. But expressivists typically keep these cognitive deliberative
states figuratively at arm's length, treating them as bits of deliberative phenomenology
that are to be explained theoretically, but not really taken seriously in their own right.
Witness the tendency to downplay their representational or descriptive function; the
concern seems to be that to ascribe to normative beliefs a robusdy representational
character, we would need to give up the naturalistic metaphysical picture that ex-
pressivists take as their starting point. Constructivism, by contrast, gives us a way to take
more seriously the normative beliefS that deliberation apparendy involves. On the
constructivist view, there are after all normative facts that are available to be repre-
sented in normative thought; it is just that they are created by the deliberative activity
that is set in motion when we confront the "normative problem" in practice and apply
our minds to resolving it. When we reach normative conclusions in deliberation, then,
there is a subject matter for those conclusions genuinely to be about.
There is a residual puzzle about Korsgaard's emphasis on the perspectival character of
normative thought, however. I said above that normativity is, on her interpretation of
constructivism, an artifact of the practical point of view; normative facts are both
constituted by our practical commitments and come into focus only from the perspec-
tive of those practical commitments. But it is unclear why we should accept the second
part of this position. If there is indeed a fact about what I ought to do that is available
for me to grasp in deliberative reflection, why isn't that fact also available to be grasped
by a third party who engages in detached reflection on my deliberative situation
(a philosopher, say, or an impartial observer, or someone who wishes to give me
advice)? Once we allow for genuine normative representations, it seems that these have
to be available to be grasped in reflection from different points of view, including
points of view that are not themselves constituted by deliberative predicaments. In
rejecting this possibility, Korsgaard reveals her apparent agreement with expressivists
about the essentially motivating or practical function of normative thought. Given this
conunitment-which takes on an existentialist character in Korsgaard's development
of her position
-it follows that normative conclusions can be embraced only from a
perspective that is essentially deliberative (where deliberation can extend to the
formation of plans or intentions for hypothetical situations that one does not actually
find oneself in). But it is not clear that a constructivist about normativity has to agree
with Korsgaard's existentialism in this matter. If constructivism is compatible with a
qualified version of the expressivist metaethic, it also seems at least possible to marry it
to a more conventionally cognitivist account of moral discourse.
The dimension is the idea that practical reasoning bottoms out in orientations of the will
that are themselves prior to reason and justification, insofar as they bring reasons into existence in the first
place. See Korsgaard 19lJ6, Lecture 4; also the remarks about "radical choice" in Street 200H. Section 8.
(c) Theory of error
Constructivism traces nonnative facts to the deliberative activities of agents, in partic-
ular to their nonnative beliefS and commitments. But it is a commonplace that beliefS
and commitments of this kind can be mistaken. Not everything that a given individual
or society might think about what there is good reason to do is correct. How are we to
think about the possibility of error in normative conviction if we follow the construc-
tivist in grounding nonnative facts on such convictions?
The general constructivist approach to nonnative error, as I noted above, makes use
of the notion of idealization. Our actual nonnative attitudes may reflect numerous
errors of nonnative fact; specifically, we might endorse mistaken conclusions about
what we ought to do, while failing to acknowledge true conclusions of the same kind.
But such attitudes do not necessarily reflect our true nonnative commitments. To
identify those, it is necessary to subject our actual attitudes to procedures of rational
reflection, procedures that have the result of eliminating views that are incorrect (the
false positives, as it were), and adding to our inventory attitudes corresponding to the
previously unacknowledged truths. Our real nonnative commitments, on the con-
structivist conception, are the attitudes we would arrive at if we subjected our actual
normative views to this kind of procedural correction.
In an earlier paper I distinguished between two different notions of commitment
that seem called for on the constructivist picture (Wallace 2004). There are, first, the
motivating attitudes of normative endorsement that figure saliently in Korsgaard's
account of deliberative thought. And there is, second, a notion of commitment that
is given content by the specified procedures of reflection, procedures that mediate
between the principles an agent actually endorses-their actual normative commit-
ments, in the first sense of this term-and the idealized principles that determine what
they really ought to do. It is this second notion of commitment that leaves space in the
constructivist picture for substantive error in normative thinking.
But the irnpottance
that is ascribed to this notion by the constructivist account also renders it vulnerable to
serious objection-indeed, to different and more powerful versions of the two lines of
objection sketched above.
The question that immediately suggests itself about the role of rational procedures in
the constructivist account is: What accounts for their authority over the agent? The
I here disagree with Street, who at one point denies that the constructivist notion of commitment
involves a notion of"rational entailment" (Street 2008, p. 232). To take one ofStreet's own examples (Street
2008, p. 230, n. 38): people who believe that agents shouldn't be blamed for things that are beyond their
control, but who blame their own children for catching the flu, are making a mistake precisely because they
are not drawing conclusions that are rationally entailed by the general normative judgment they accept. The
conclusion that one should not blame one's child in this case is not something that is "constitutively entailed"
by the general judgment (in the way that Street believes the thought that one has reason to take the means is
constitutive ofjudging that one has reason to achieve the end). Rather, it follows rationally from that general
judgment, insofar as universal generalizations entail their instances. I therefore prefer the characterization in
Street 2010, p. 371, where constructivism is represented as appealing to facts about "whatfollouJS (as a logical
and instrumental matter)" from a given agent's evaluative commitments.
procedures are naturally described as standards of correctness in reflection, and it is
tempting to appeal to this aspect of them in accounting for their authority.
The procedures are authoritative for agents, according to this line of thought, because
they tell us what it is to engage properly in the activity of deliberative reflection.
But the constructivist cannot accept this answer as it stands. It appears simply to
presuppose the normativity of the specified rational procedures in accounting for
their authority over the agent, and this is at odds with the basic philosophical aspiration
that we have seen to animate the constructivist position: to develop a global theory of
normativity that explains why normative considerations are authoritative for the
attitudes of those they govern (Compare Hussain and Shah 2006, pp. 290-2).
It does not follow from this, however, that the rational procedures must be
understood by the constructivist in non-normative terms. For that would seem to
entail that the specified procedures are not authoritative at all, thereby undermining in
a different way the whole constructivist account. Consider an agent whose thinking
allegedly incorporates normative errors, and who is told in support of this conten-
tion that the application of the critical procedures to their actual normative oudook
would result in crucial modifications (including the abandonment of some views
and the adoption of others). Our agent could, and mosdy likely would, respond by
posing a variant ofKorsgaard's normative question, asking why they should care about
the modifications that would counterfactually be induced in their actual attitudes
by submission to the specified procedures. There had better be some positive answer
that can be returned to this question on the constructivist's behalf if the account
of error is to be at all convincing. Moreover, the answer that is provided had better
be one that is in the spirit of the constructivist's general theoretical approach to
That approach, as we are now in a position to see, would lead us to seek a dialectical
account of the normativity of the rational procedures. What makes those procedures
normative for us as agents-procedures that define genuine standards of rationality or
correctness-is the fact that they are firmly anchored in our actual commitments. If
something like this could be established, it would follow that a convincing answer to
the normative question about the procedures would be available to any agent who
might happen to pose it. In reflecting about why you should care whether you comply
with the specified procedures, you would find that you are already committed to
accepting them and to caring about whether you have satisfied them. This would be a
distinctively constructivist explanation of the normativity of the rational standards that
figure in the constructivist's theory of error. Note, however, that the commitments
that are appealed to at this point in the constructivist story have to be of the first of the
two kinds distinguished above: they are actual attitudes of norrnative endorsement, not
attitudes that the agent would arrive at if they followed the specified rational proce-
dures. (It would be viciously circular to appeal to commitments in the latter sense in an
account of the normativity of the rational procedures, because the second notion of
commitment itself presupposes the authority of those very procedures.)
We see, then, that the procedural standards that are invoked in the constructivist
theory of error have to be standards that are always already endorsed, explicitly or
implicitly, by the agents who are allegedly mistaken in their nonnative attitudes. Only
if this very stringent condition is satisfied can the procedural standards possibly do the
work in the constructivist story that they are meant to perform. It is at this point,
I believe, that genuine difficulties begin to arise for the constructivist position. I shall
conclude my discussion by making three critical observations about the constructivist
theory of error, focusing in particular on the nature of the agent's commitment to the
procedural standards that the theory relies on.
The first point to note is that the constructivist account of the authority of rational
procedures suggests a fairly rigorous restriction on the content of those procedures. The
procedures have to be ones that are already actually acknowledged by the agents whose
attitudes they regulate, and it would seem that only the most obvious and fundamental
principles of rationality are likely to satisfy this stringent condition. Plausible candidates
for procedural requirements of the right kind are requirements to comply with the
principle of non-contradiction and with the instrumental principle, as it was intro-
duced above. This outcome is, of course, not necessarily problematic in itself; perhaps
there are good reasons for favoring a narrow and restrictive interpretation of the
requirements of procedural rationality. An interpretation of this kind, however, will
make correspondingly difficult the task confronting proponents of more ambitious
Kantian versions of the constructivist program. Positions of this kind hold that funda-
mental moral requirements are universally binding on rational agents, insofar as a
commitment to comply with those requirements is built into the structure of agency.
This means, in the terms that I have proposed, that application of the procedural
standards to any agent's actual set of nonnative views will result in a modified set of
views that includes acknowledgement of something like the Kantian moral law. But it
seems unlikely in the extreme that narrow requirements of consistency and instrumen-
tal rationality would alone suffice to push any agent-whatever normative assumptions
they might start from-to this interesting and controversial conclusion (compare Street
A second issue concerns the significance attached to an agent's actual normative
views by the constructivist theory of error. I have said that the procedural standards can
do the work the constructivist allots to them only because they are standards that agents
can be assumed already to acknowledge, implicitly or explicitly. Their authority, on
this account, can be traced to the fact that they are standards that the agent actually
accepts. But if actual acceptance grounds the authority of the standards that
are appealed to in criticizing an agent's normative views (the standards, that is, that are
invoked to explain the errors in those views), then it seems that it should have a similar
significance in other cases as well. That is, for any candidate normative principle, the fact
that a given agent actually accepts the principle ought to confer at least defeasible
authority on it over their thought, so long, at any rate, as the attitude of acceptance
remains unrevised.
This outcome has important consequences for the theory of error, however.
Consider what I referred to above as "false positives": attitudes of normative endorse-
ment that one would abandon if one applied the procedures of rational correction in
normative thought. The constructivist wants to say, on the one hand, that these
attitudes are mistaken, insofar as they would not survive reflective scrutiny of the
appropriate kind. But on the other hand, as long as they continue to be sustained by the
agent, they have to be accorded default authority, so that the agent who holds them
really does have some reason to act in accordance with them. This is, in fact, precisely
what Korsgaard herself says about an example of G. A. Cohen's: the Mafioso who
reflectively endorses the requirements of a code of strength and honor, taking himself
to be obligated to comply with those immoral requirements even when he feels
tempted to defect from them (Korsgaard 1996, pp. 256-8; c£ pp. 183-4). About this
figure, Korsgaard writes: "There is a sense in which these obligations are real-not just
psychologically but normatively. And this is because it is the endorsement, not the
explanations and arguments that provide the material for the endorsement, that does
the normative work" (Korsgaard 1996, p. 257).
For a long time I was puzzled by this remark, wondering why Korsgaard did not
follow the apparendy simpler route of denying the normative significance of volitional
commitments that would not survive the process of reflective scrutiny. But I now see
that her position is the right one for a constructivist to adopt. If the thing that confers
authority on procedural standards is the fact that one actually endorses them, then one
has to grant provisional authority to other principles that are actually endorsed by the
agent; the logic of the larger constructivist account forces this conclusion. Having said
that, however, I would add that the conclusion does not seem to me to be a very
plausible one. It gives up too much that we should want a theory of normative error to
accomplish, reintroducing the problem of local bootstrapping that the theory was
originally meant to address. On the picture that results, normative thought becomes
peculiarly self-validating-thinking that you ought to do X really does make it the case
that you ought to do X -and an account that has this result should be rejected (unless,
of course, the alternatives are even more problematic, something I cannot go into
Consider, fmally, a different kind of situation, involving what we might refer to as
recalcitrant irrationality: I am thinking of an agent who acknowledges that the
procedural standards they accept require a revision in their actual attitudes, but who
refuses to make the revision that they see to be thus required. The Mafioso just
described, for instance, might see that their endorsement of a code of dan honor
would not survive the consistent application of the procedural standards, but care more
about their narrow conception of honor than about consistency (compare Williams
1986, Chapter 2). We want the theory of error to deliver the result that normativity is
ultimately determined by the attitudes of endorsement that would emerge if the
procedural standards were applied to the agent's actual attitudes. In cases of recalcitrant
irrationality, however, the agent actually acknowledges the truth of the relevant
counterfactualjudgment, but fails in fact to revise their attitudes accordingly. A stance
of this kind might in a certain sense be reflectively unstable, insofar as it would not (by
hypothesis) survive a complete course of procedural reflection. But it might be very
stable indeed in practice, grounded in fierce emotional attachments to the ends and
values that reflection would require the agent to abandon.
A plausible theory of error should say about cases of this kind that the agent's actual
attachments, however fiercely they may be held, are mistaken; they are, after all, cases
of recalcitrant irrationality. But what does constructivism say about these cases? It can
agree, I think, that the label "irrationality" is correctly applied to them. After all, the
agents we are considering acknowledge that their own attitudes are criticizable by the
light of procedural principles that they themselves accept; there is, accordingly, some-
thing internally problematic about their overall point of view. But it is also true that the
agents in question do not fundamentally care about this aspect of their own position, or
at any rate they do not care about it as much as they care about the first-order ends that
a more consistent outlook would require them to give up.
Proponents of a cognitivist version of constructivism about normativity might not
be too troubled by this possibility. According to this kind of position, we reach
normative bedrock when we establish that the attitudes of recalcitrant agents are
criticizable by their own lights; to the extent this is the case, such agents are not able
fully to endorse the ends that they actually pursue, and this is enough to establish that
those ends are not really normative for them. Once it takes this form, however,
constructivism seems to sacrifice some of its apparent advantages vis-a-vis normative
realism. In particular, it agrees with realism in holding that a significant gap can open up
between the acknowledgment of normative fact by an agent and the agent's actual
volitional commitments. Once this concession is made, it may be wondered what the
real advantages might be of constructivism over realism, especially in light of some of
the other difficulties I have touched on for the constructivist account.
Korsgaard, for her part, favors a more existentialist interpretation of the constructivist
approach. This position represents a starker alternative to normative realism. But the
cost is an implausibly psychologistic account of what I have called recalcitrant irratio-
nality. If a given group of agents fundamentally do not care enough about their own
irrationality to be moved to revise the offending attitudes, then the standards
of rationality themselves lose their authority for those agents; we may call them
irrational if we like, and the agents might even agree, but this form of criticism loses
its critical animus.
In its attempt to anchor norrnativity in the agent's actual volitional
The constructivist might at this point appeal to facts about identity, arguing e.g. that the agent's actual
motivations are less authentic somehow (or less "central" or "strongly held" or "truly their own") than the
nonnative attitudes with which they conflict; compare Street 2008, pp. 234-6, who appeals to factS of this
kind to draw distinctions within the class of an agent's normative attitudes. The challenge here will be to
develop the notion of identity in a way that doesn't simply presuppose what is to be established, namely the
authority of an agent's normative judgments. Highly relevant here is the extensive debate about identification
set in motion by Frankfurt 1971 and Watson 1975.
commitments, constructivism in its purest form ends up collapsing the important
distinction between norm and psychological fact, ought and is, in ways that fmally
call into question its credentials as a theory of normativity .U
Blackburn, Simon. 1998. Ruling Passions. A Theory of Practical Reason (Oxford: Oxford University
Broome, John. 2008. "Comments on Allan Gibbard's Tanner Lectures," in Allan Gibbard et al.,
Reconciling Our Aims. In Search of Bases for Ethics (New York: Oxford University Press),
pp. 102-19.
Frankfun, Harry. 1971. "Freedom of the Will and the Concept of a Person," in journal of
Philosophy 68, pp. 5-20.
Gibbard, Allan. 2003. Thinking How to Uve (Cambridge, MA: Harvard University Press).
Horgan, Terry and Mark Timmons. 2006. "Morality without Moral Facts," in James Dreier
(ed.), Contemporary Debates in Moral Theory (Malden, MA: Blackwell Publishing), pp. 220-38.
Hussain, Nadeem and Nishi Shah. 2006. "Misunderstanding Metaethics: Korsgaard's Rejection
ofRealism," in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics 1 (Oxford: Clarendon
Press), pp. 265-94.
Korsgaard, Christine. 1996. The Sources ofNormativity (Cambridge: Cambridge University Press).
Korsgaard, Christine. 1997. "The Normativity oflnstrumental Reason," in Garrett Cullity and
Berys Gaut (ed.), Ethics and Practical Reason (Oxford: Oxford University Press), pp. 215-54.
Korsgaard, Christine. 2008. "Realism and Constructivism in Twentieth-Century Moral Philos-
ophy," as reprinted in her The Constitution of Agency. Essays on Practical Reason and Moral
Psychology (Oxford: Oxford University Press), pp. 302-26.
Korsgaard, Christine. 2009. Self-Constitution. Agency, Identity, and Integrity (Oxford: Oxford
University Press).
Nagel, Thomas. 1970. The Possibility of Altruism (Princeton: Princeton University Press).
Parfit, Derek. 2006. "Normativity," in Russ Shafer-Landau, (ed.), Oxford Studies in Metaethics 1
(Oxford: Clarendon Press), pp. 325-80.
Rawls, John. 2005. A Theory of Justice, original edition (Cambridge, MA: Harvard University
Raz,Joseph. 1999. "Explaining Norrnativity: On Rationality and the Justification ofReason," in
his Engaging Reason. On the Theory of Value and Action (Oxford: Oxford University Press),
pp. 67-89.
Scanlon, T. M. 1998. What We Owe to Each Other (Cambridge, MA: Harvard University Press).
Smith, Michael. 1999. "Search for the Source," in Philosophical Quarterly 49, pp. 384-94.
Street, Sharon. 2008. "Constructivism about Reasons," in Russ Shafer-Landau (ed.), Oxford
Studies in Metaethics 3 (Oxford: Clarendon Press), pp. 207-46.
I have received very helpful feedback on this chapter from Anne Burkard, Benjamin Kiesewetter,
Jinuny Lenrnan, Kaaien Schaubroeck, Y onatan Shemmer, Marcus Willaschek, participants in the Confer-
ence on Constructivism in Practical Philosophy at the University of Sheffield in August 2009, and members
of discussion groups in Oxford (on nonnativiry) and Berlin (the Thomas Schrrudt colloquium).
Street, Sharon. 2010. "What is Constructivism in Ethics and Metaethics?" in Philosophy Compass
5, pp. 363--84.
Street, Sharon. 2012. "Coming to Tenns with Contingency: Humean Constructivism about
Practical Reason," this volume.
Wallace, R.Jay. 2001. "Normativity, Commitment, and Instrumental Reason," in Philosophm'
Imprint 1, no. 3, http:/ /hdl.handle.net/2027 /spo.3521354.0001.004.
Wallace, R.Jay. 2004. "Constructing Normativity," in Philosophical Topics 32, pp. 451-76.
Wallace, R. Jay. 2006. "Scanlon's Contractualism," as reprinted in R. Jay Wallace, Normativity
and the Wrll. Selected Essays on Moral Psychology and Practical Reason (Oxford: Clarendon Press),
pp. 263--99.
Watson, Gary. 1975. "Free Agency," injoumal of Philosophy 72, pp. 205-20.
Williams, Bernard. 1986. Ethics and the Umits '?{Philosophy (Cambridge, MA: Harvard University
Coming to Terms with Contingency:
Humean Constructivism about
Practical Reason
Sharon Street
1. Introduction
What is value, and how did it enter the world? Metaethical constructivism gives the
following answer: Value is something conferred upon the world by valuing creatures,
and it enters and exits the world with them. Causal forces, evolutionary and otherwise,
gave rise to conscious living things capable of valuing things-whether it be food or
their bodily integrity or each other--and with this state of mind of "valuing" came
truths about what is valuable--truths which hold from within the point of view of
creatures who are already in that state.
Call the standpoint occupied by any creature who is in the state of mind of"valuing"
the practical point of view. I argue elsewhere that metaethical constructivism is best
understood as claiming the following: The truth of a normative claim consists in that
claim's following, as a logical or instrumental matter, from within the practical point of
view, where the practical point of view is given a formal characterization. To give the
practical point of view aformal characterization is to give an account of the standpoint of
valuing as such, where this involves giving an account of the attitude of valuing that
does not itself presuppose any substantive values but rather merely explicates what is
involved in valuing anything at all. Logical and instrumental "requirements," as these
govern practical reasoning, are explained not as substantive values, but rather as features
constitutive of the attitude of valuing. To ignore these "requirements" in full con-
sciousness of what one is doing is not to make a mistake about a normative matter; it is
merely to fail to value. Apart from the attitude of valuing, there is no such thing as a
mistake. Either one values things or one does not, and whether there exist any
normative reasons for one to do anything depends upon it.
' I explain and develop this view in Street 2001la and 2010.
As this rough characterization suggests, if one accepts metaethical constructivism,
then one is an anti-realist about value in an important and traditional sense, holding
that value is an attitude-dependent property. One regards the attitude of valuing as the
more fundamental explanatory notion, and understands value itself as a "construction"
of that attitude. Things are valuable ultimately because we value them, not the other
way around.
The realism/anti-realism debate, understood in this "Euthyphronic" fashion, is a topic
I explore elsewhere.
In this essay, I wish to set that debate entirely to one side and
focus on an intramural debate among metaethical constructivists-what I'll call the
debate between Kantian versus Humean constructivists.
As I'll draw the distinction,
Kantian and Humean constructivists agree that the truth of a normative claim consists
in its following from within the practical point of view, where the practical point of
view is given a formal characterization. What separates them is their position on
whether moral conclusions follow from within the practical point of view given a
formal characterization. According to Kantian versions of metaethical constructivism,
moral conclusions do follow. That is, we may start with a purely formal understanding
of the attitude of valuing, and demonstrate that recognizably moral values follow from
within the standpoint of any valuer as such. The most prominent contemporary
defender of such a view is Christine Korsgaard.
Humean versions of metaethical constructivism, in contrast, deny that substantive
moral conclusions are entailed from within the standpoint of valuing as such. Accord-
ing to the Humean, the substantive content of an agent's normative reasons is a
function of his or her particular, contingently given, evaluative starting points. On
this view, "pure practical reason"-understood as the standpoint of "valuing," or
normative judgment, as such--commits one to no specific substantive values. Instead,
the substance of one's normative reasons must ultimately be supplied by the particular,
contingent set of values with which one finds oneself alive as an agent. The Humean
constructivist accepts the idea that if one had entered the world with a radically
different set of values, or were merely causal forces to e1fect a radical change in one's
existing set of values, then one's normative reasons would have been, or would
become, radically di1ferent in a corresponding way.
The goal of this chapter is to weigh in on the Humean side of this debate, and to
sketch the broader picture of practical reason that it involves. The name Humean
constructivism seems appropriate for a view that draws with equal prominence on both
the Kantian tradition, especially as articulated by Rawls,
and the Humean tradi-
tion. Roughly speaking, from Kant come the ideas of "giving laws to oneself," the
Street 2006, 2008b, 2009, 2011, and MS.
• Others have drawn similar distinctions between Kantian and Humean versions of constructivism. See
Bagnoli 2002, p. 131; Dorsey MS; Lenman 2010; and Tiberius 2008, p. 189. I am introducing the terms
.K.mtian constructivism and Humean constructivism in my own sense, but there are important ideas common to all
of these discussions.
• Korsgaard 1996b.
See Rawls 1999.
"practical point of view," and the associated emphasis on the attitude of valuing, as
opposed to mere desiring, as the key to understanding the phenomenon of normativity
and the possibility of truth and falsehood in the normative domain. From Hume, on
the other hand, come the skepticism about the ability of "pure" practical reason to
tell us how to live and the associated emphasis on the need to look to contingendy
given, more "passionate" states of mind for the source of the substantive content of
our reasons.
While the debate between Kantian and Humean versions of metaethical construc-
tivism is stricdy intramural, the interest and importance of the debate are not. First, the
outcome of this debate is important for the assessment of the plausibility of metaethical
constructivism across the board, both in itself and as it compares with other metaeth-
ical views. For example, were a Kantian constructivist argument to succeed, metaethical
constructivism could boast to have captured and vindicated the intuitions of robust
moral objectivity that otherwise might tempt many theorists to go realist rather than
constructivist. A different set of theorists, for their part, might be tempted by a veruon of
constructivism--Humean constructivism-that is upfront about what many people regard
as the fairly obvious fuilure of ambitious Kantian arguments for morality. Historically,
constructivism in ethics has been so closely associated with the Kantian tradition that one
mistakenly might have thought they were inseparable, and a Humean version might
win some new converts. More important than any of this, however, is the fact that the
debate between Kantian and Humean constructivism is ultimately just a debate about
how to live and why, and about the fundamental nature of our relationship with moral
requirements. Is morality something we are bound to by the mere fact that we are valuers
at all, or is our relationship with morality more contingent than that, though perhaps no
less dear for that fact? I will be arguing for the latter view.
2. The attitude of valuing
Let's start with a brieflook at the attitude of valuing. In my view, much of the resistance
to attitude-dependent conceptions of normative reasons in general (of which metaeth-
ical constructivism is one
) is due to a failure, on the part of the view's supporters and
opponents alike, to locate accurately enough the type of attitude on which normative
reasons are said to depend. It is widely assumed that if an agent's normative reasons are
always ultimately a function of his or her attitudes, then the attitudes in question must
be the agent's desires. While not stricdy speaking inaccurate, due to the philosophically
cultivated latitude in what may plausibly be labeled a "desire," the language of desire is
exceptionally misleading in this context due to its meaning in ordinary speech. I think
it is hard to overstate the extent to which the emphasis on "desire" distorts the debate
• Examples of other attitude-dependent v i e w ~ include Williamsi9K1; Lewis 1989; and Schroeder 2007.
regarding the attitude-dependence of normativity, making the attitude-dependent view
seem less plausible than it is.
To see the full attractions of an attitude-dependent account, it is essential to
distinguish between what I will call, quite stipulatively, the attitude of valuing and
the attitude of mere desire, and to recognize that if value is indeed conferred upon the
world by our attitudes, as the anti-realist claims, then it's the attitude of valuing, and not
the attitude of mere desire, that does the conferring. A full development of the contrast is
impossible here.
Here I wish merely to note three key differences.
First, more is constitutively involved in the attitude of valuing than is constitutively
involved in the attitude of mere desiring. As we might also put it: The attitude of
valuing is characterized by a "discipline" that the attitude of mere desiring lacks.
Consider Korsgaard's example of a Civil War soldier who, if he is to live, must have his
leg sawed off without the benefit of anesthetic.
Imagine first that the soldier says:
1. I desire to live, and I know that I must have my leg sawed off without the benefit
of anesthetic to do so, but I have no desire to have my leg sawed off.
While the language is of course far too controlled for the circumstances, claim
(1) would nevertheless be an entirely coherent and comprehensible thing to say. We
all know perfectly well what it's like to desire the end but not the means; far too much
of life has exactly this structure, and the case of a life-saving amputation is a vivid
example. But now imagine the solider says:
2. I have most reason to live, and I know that I must have my leg sawed off without
the benefit of anesthetic to do so, but I have no reason whatsoever to have my leg
sawed off.
In contrast to claim (1), it's hard to make sense of this. The speaker appears to be
making some kind of conceptual mistake. It's not that the soldier understands perfectly
well the concept of a "reason," and then is making a substantive mistake about the
reasons he has; it's that he seems not to know what he's saying.
The lesson, I would argue:
The state of mind of desiring the end does not
constitutively involve desiring what one is fully aware is the necessary means to that
end. This is so, anyway, in a perfectly familiar sense of the word desire-the sense that
involves, roughly, a feeling ofbeing pleasantly attracted. This is the state of mind I'll
refer to as mere desire. In contrast, the state of mind of taking oneself to have reason to
pursue the end constitutively involves taking oneself to have reason to take what one is
fully aware is the necessary means to that end. This is the state of mind I'll refer to,
A full development would draw upon the rich literature on these topics in the philosophy of agency,
espedally the debate between Frankfurt 1982 and Watson 1982b.
• I borrow the language of"discipline" from Wright 1992.
Kongaard 1997.
'" For more detailed discussion, see Street 2008a and 2010.
interchangeably, as valuing, or taking something to be a reason, or normative judgment.
Valuing an end, in contrast to merely desiring it, constitutively involves valuing what
one is fully aware is the necessary means to that end.
Second, as the idea that one can "value" having one's leg sawed off suggests, the state
of mind of valuing is characterized by a much broader array of conscious experience
than is the attitude of mere desiring. Unlike the state we naturally think of when we
hear the word desire, which calls to mind paradigmatic examples such as sexual
attraction, thirst, a craving for chocolate, and so forth, the state of mind I am stipula-
tively calling valuing is characterized by all the range, nuance, and depth of human
emotion and feeling. Consider getting up at 4 a.m. to finish grading papers; breaking
bad news to a friend; or risking death to fight an authoritarian regime. We have
affective, sometimes intensely emotional, experiences of such things as "demanded"
or "called for," but they do not call out to us in a way that is pleasant, as the term desire
suggests, but rather in a manner that involves something more like anxiety or sickness
at the thought of not doing them, feelings of gritty determination to preserve what one
cares about most in spite of everything, and so on. The language of desire is completely
inadequate to capture the kind of emotional and phenomenological complexity
involved in the kinds of attitudes that confer value on the world, if any do. The
relevant attitude can be directed at things we simultaneously fmd unpleasant, terrible,
anguish-inspiring, and so forth.
Third, the attitude of valuing is characterized by greater structural complexity than
the attitude of mere desiring. We tend to think of "desiring" as directed at a single
object or state of affairs: I desire a donut, for example, or to be rich or to be liked.
Evaluative experience of the kind that confers value if anything does, however, is
structurally a great deal more complicated than that. It often involves experiencing very
specific features of the world as "calling for" or" demanding" or "counting in favor of"
other very specific things. For example, I experience the fact that a friend lent me her
car two months ago as counting in favor of saying "yes" to the favor she's asking me
now; I experience someone's youth and inexperience as ruling out a harsh reply; and so
on. Such states of mind are very different from simply wanting a donut. While no
doubt we can speak of desiring things in virtue of specific features they have, there is
nevertheless a tendency to think of desiring as involving a simple focus on one object or
state of affairs as something to be possessed or brought about. The attitude of valuing
involves much more complex attitudes toward the world and one's own potential
responses to it.
Attitude-dependent conceptions of normative reasons owe much of their inspiration to
the Humean tradition. If we are to see the full promise and plausibility of these views,
however, we need to abandon another, much less helpful, element that tends to be
associated with the Humean tradition-namely the constant use of the language of desire,
which fails to draw our attention to the kinds of attitudes we really need to be focusing
on, which are characterized by a discipline, a range and subdety of emotional experience,
and a structural complexity that are foreign to the everyday concept of desire.
3. A reconstruction ofKorsgaard's argument
for Kantian constructivism
Korsgaard's position in The Sources of Normativity is the leading contemporary example
of a Kantian version of metaethical constructivism. I will assume for the sake of
argument that Korsgaard is a metaethical constructivist in the sense I have outlined,
holding that normative truth consists in what follows from within the standpoint of a
valuing creature, where that standpoint is given a formal characterization. As a specifi-
cally Kantian metaethical constructivist, Korsgaard holds in addition to this that there
are certain substantive values, and in particular moral values, that follow from within the
standpoint of any valuing creature. In this section I offer a reconstruction ofKorsgaard's
argument; in subsequent sections I explain where I think the argument goes wrong.
The motivations for this critical discussion are threefold. First, examining Kors-
gaard's argument and seeing why (in my view) it fails is a useful means of developing
the more basic, underlying position-metaethical constructivism-that does not stand
or fall with the Kantian argument for morality. Second, even ifKorsgaard's argument
does fail, it is an important example of the general strategy that in my view must be
pursued by anyone who wants to try to combine an anti-realist, in the sense of attitude-
dependent, understanding of normativity with a robust form of objectivity about
morality or any other substantive value. Third and finally, the specific position I wish
to defend-Humean constructivism-is defmed negatively in contrast to Kantian
constructivism: it is characterized by a general skepticism about arguments which
seek to establish moral values as following from the standpoint of "agency" as such.
Therefore, the obvious way, and perhaps the only way, to argue for Humean as
opposed to Kantian constructivism is negatively-by examining the best attempts to
establish such substantive values and showing why they fail.
Korsgaard's argument, as I reconstruct it, consists of six steps. We view the argument
as starting with a purely formal understanding of the attitude of valuing. We imagine
someone who values things, but we assume nothing in particular about what he or she
values: the individual merely has to be recognizable as valuing something or other. We
then seek to show that moral commitments are entailed from within that individual's
own evaluative point of view. If we succeed at this, then we will also have shown that
morality is entailed from within our point of view, since we are valuing creatures too.
Step 1. Suppose, then, that you take something or other to be valuable; it doesn't
matter what. To put it in the language of practical identities: Suppose that you take
some practical identity or other (whether it be your identity as a mother or Mafioso
or something else) to be normative for you. Since we are assuming nothing about
the substantive content of the values in question, we are arguing from within, or
addressing, the practical point of view "as such"-the standpoint of anyone who values
anything-the standpoint of an agent, any agent.
Step 2. If you do value anything, or if you do embrace any practical identity as
normative for you, then you must take your endorsement of that value (or your
embrace of that practical identity) to be supported by further reasons. (In Korsgaard's
view, these further reasons are given by a further practical identity, also embraced as
Discussion of step 2. Why, if one accepts a value or a practical identity (any practical
identity), must one take one's acceptance of it to be supported by further reasons?
I think this is best read as a point about what is involved, as a constitutive matter, in the
attitude of valuing. In particular, we may view it as the same point that Kant is making
when he says "we cannot conceive of a reason which consciously responds to a bidding
from the outside with respect to its judgments. "
The idea seems to be: If you value
something, then you cannot--simultaneously, in full, conscious awareness-also think
that there is no reason whatsoever to value it. The force of the "cannot" here is
conceptual, not rational. Kant's suggestion seems to be that a faculty isn't recognizable
as the faculty of reason at all if that faculty consciously takes itself to have absolutely no
reason whatsoever for what it is doing or thinking at that moment.
Step 3. From step 2 it follows that in order to be regarded as normative, every
practical identity must be regarded as having a further practical identity-itself also
embraced as normative--lying behind it and supplying a reason for accepting it. But
such a regressing chain of supporting practical identities cannot extend forever. There-
fore, if the original practical identity supposed in step 1 (call it P) is to be regarded as
normative, then it must-when one thinks about it-be regarded as ultimately
supported by a practical identity that brings the regress of identities to a satisfactory
end. If, contrary to this, one thinks the regressing chain bottoms out with no reason,
then (in accordance with step 2), one will think one ultimately has no reason for P
either, and cease to regard it as normative. At the bottom of the chain, then, we need a
practical identity with the following two characteristics: (a) it must supply a reason for
accepting other practical identities; and (b) it must not itself require any further
normative practical identity to lie behind it. Call a practical identity that has these
two characteristics a source practical identity (my term).
Step 4. Our practical identity as human beings, accepted as normative, has these two
Step 4a. Our practical identity as human beings, accepted as normative, fulfills the
first role of a source identity: it supplies a reason for embracing our other practical
identities as normative. What is this reason? According to Korsgaard, one's identity as a
human being is one's identity as a creature who is able to distance herself from her
impulses and desires, and ask "Should I act?" If (and only if) we embrace this identity as
Kant 1785/1959. Kongaard quotes Kant's remark in her 1996b, p. 94.
This claim is not fal,ified by cases in which, as we might describe it, we decide "It's time to do
something for no reason at aU." Such cases are not examples of regarding oneself as "having no reason
whatsoever" in the leJlle relevant to Kant's point. In such cases as they ordinarily occur, one would at least
implicidy be thinking !IOmething like "It's time to loosen up," "let myself go," "stop thinking so much and
just do wmething on impulse," "foUow my whims," etc., and merely in having this rationale in the
background one would, in the relevant sense, stiU be oneself as for a reason.
normative, then this sets us up with a problem, according to Korsgaard: we need
reasons to act. But we can have reasons to act only if we embrace some particular
practical identity or other. Why? As I read it, the idea here is the central metaethical
constructivist point that one has reasons to act only if one Ulkes something or other to
be a reason; this is because "practical identities," or the attitude of "valuing" itself, is
what sets the standards determining a creature's reasons. (At work in the background
here is the rejection of realism.) Since we need reasons, and we can have reasons only if
we endorse some particular practical identity or other, it follows that we have a reason
to endorse particular practical identities. Thus, if we take our practical identity as a
human being to be normative, then we may say of the original practical identity P: one
reason that we have to accept this particular identity is that as human beings, we need
some reasons or other, and identity P supplies us with such reasons if we go ahead and
accept it as normative.
Step 4b. Our practical identity as human beings, taken as normative, also fulfills the
second role of a source identity: it does not require any further practical identity to lie
behind it and impart it with its own normativity in tum. One sees this, according to
Korsgaard, when one attempts to continue the regress of questions and asks why one
should take one's practical identity as a human being to be normative. Korsgaard
[N]ow that you see that your need to have a normative conception of yourself comes from your
human identity, you can query the importance of that identity. Your humanity requires you to
conform to some of your practical identities, and you can question this requirement as you do
any other. Does it really matter whether we act as our humanity requires, whether we find some
ways of identifying ourselves and stand by them? But in this case you have no option but to say
yes. Since you are human you must take something to be normative, that is, some conception of
practical identity must be normative for you. If you had no normative conception of your
identity, you could have no reasons for action, and because your consciousness is reflective, you
could then not act at all. Since you cannot act without reasons and your humanity is the source of
your reasons, you must value your humanity if you are to act at all.
Notice what is going on here: I am asking what reason I have to treat my practical
identity as a human being as normative, and the answer is that if I do not, then none of
my other identities will be normative, and I need some normative conception of my
identity. But this reason for treating my practical identity as a human being as
normative just springs from the same practical identity that is under question-namely,
one's identity as a human being, who needs reasons and who therefore needs some
normative conception or other of her identity. In this way, Korsgaard suggests, our
practical identity as human beings is self-supporting and so brings our regress of
questions to an end: the very reason for embracing it as normative springs from itself
" Korsgaard 1996b, p. 123.
Step 5. The next step is to claim that our practical identity as human beings, endorsed
as normative, is the only practical identity capable of serving as a source practical
identity. This step involves rejecting all other candidates for the role of bringing the
regress of questions to a satisfactory end. Leading contenders for this role include states
such as desires or pain, or independent facts about reasons of the kind posited by realists.
Korsgaard's arguments against such alternative contenders appear in various places
throughout The Sources of Normativity .
Step 6. It follows from all of this that if we are to embrace practical identity P
(whatever it might be) as normative, then we must accept our practical identity as
human beings as normative. Why? From steps 2 and 3, it follows that if one is to regard
any given practical identity as normative, then one must regard it as ultimately
supported by a source practical identity. From steps 4 and 5, we see that one's practical
identity as a human being, if embraced as normative, is not only capable of playing this
role but also the only thing capable of playing this role. Therefore, if one is to regard any
given practical identity as normative, one must also take one's practical identity as a
human being to be normative. In other words, if one is conunitted to the value of
anything, then one is conunitted to the value of one's own humanity .
4. The problem with Korsgaard's argument
While suggestive, this argument fails, in my view. My criticisms of Korsgaard's
argument will be internal ones, in the sense that my objections will be raised entirely
from within the larger metaetbical constructivist framework. I will argue that Kors-
gaard's argument does not ultimately remain true enough to the central constructivist
point that there are no facts about normative reasons apart from the standpoint of an
agent who is already taking things to be reasons. The rejection of realism is merely
being assumed, without argument, as a given throughout the following discussion.
Let's begin with a look at Step 4b. The first sign that something is amiss with this part
of the argument is the apparent circularity of the answer that is given to the question of
why we should embrace our practical identity as human beings as normative. The
question at hand is: Why should I take my identity as a creature who needs reasons (and
hence normative conceptions of my identity) to be normative? And the question-
begging reply essentially seems to be: You need reasons (and hence normative con-
ceptions of your identity). This makes it unclear exactly how one's practical identity as
•• Realism is comidered at length in Lecture I; desires are discussed in section 2 of Lecture 3 and section 3
of the Reply; and pain is discussed in section 4.3. See also Korsgaard's interpretation of Kant's argument for
the Formula of Humanity in her 1996a, especially pp. 12(}-3.
" !u Korsgaard poinll out, one might think that t h i ~ argument "shows only (or at most) that you must
place a value on your own humanity, but not yet that you therefore have obligations to other human beings"
(1996b, p. 130). Kongaard therefore olfen a supplemental argument in Lecture 4 to show how this latter
conclusion is entailed as well. Since I believe the af!(Ument fails already in the stage we have just considered.
I will not consider the Lecture 4 argument. Examining the Lecture 3 argument is enough for my larger
purpose of illustroting the basic idea behind a Kantian version of metaethkal constru<·tivism.
a human being is supposed to fulfill the second role of a source identity and bring the
regress of questions to a satisfactory end; after all, one could stop at any point in the
regress of questions and merely insist upon whatever value is being questioned.
But I don't think circularity or lack thereof is the real problem here. The real
problem lies with the question at hand, and with the suggestion that there could be
any coherent answer to it. As Korsgaard phrases it, the question is: "Does it really
matter whether we act as our humanity requires, whether we find some ways of
identifying ourselves and stand by them?" And her reply is: "But in this case you have
no option but to say yes." But I think that as a matter of fact we do have an option here,
and that the proper answer is not to say yes, but rather to reject the question at hand as
ill-formulated. Translated into the language of"mattering" and normative reasons, the
question at hand is: Does it really matter whether I take anything at all to matter? Or: Is
there any reason why I should take something or other to be a reason? According to
metaethical constructivism, however, the standards that determine whether something
"really matters" are set by one's own judgments about what matters; the standards that
determine one's reasons are ultimately set by one's own judgments about what count as
reasons; and there are no standards apart from this: this is the rejection of realism. It is
therefore illegitimate to stand apart from every last one of one's judgments about what
matters, and then to ask whether something further matters; it is illegitimate to stand
apart from all of one'sjudgrnents about what count as reasons, and then to ask whether
one has some further reason. Such questions are ill-formulated in the way that "Is the
Empire State Building taller?" is ill-formulated: one has failed to supply the standard
that would make the question make sense.
This is precisely the mistake that is involved in the question about the value of
humanity, in my view. In asking whether one has reason to take anything at all to be a
reason, one is posing a normative question; and yet at the very same time one is
stepping back from and suspending one's endorsement of all values, thereby robbing
the question of the standards that could make the question make sense. In effect, one is
asking: Does it matter (as judged apart from the standards that determine what matters)
whether I take anything to matter? Or: Do I have a reason (as judged apart from the
standards that determine what counts as a reason) to take anything to be a reason? The
proper reply to such questions is not to say yes, but rather to reject the question and say:
Either you take something or other to matter or you don't; either you take something
Recall that here I am taking it as a given--agreed upon between the Kantian and Humean constructi-
vists-that realism is &lse. Since there are, on the views of both parties to the discussion, no independent
normative facts, with standards of correctness for normative judgments existing only from within the
standpoint of a valuing creature, constructivists can agree that there's a sense in which the question being
asked here is ill-formulated. In a broader metaethical context, however-where realism, for example, was still
on the table as a live option-it would not be appropriate to put things this way. In other words, the question
whether there are independent normative facts makes sense; it's just that I think the answer is no. Once
one decides that, though, certain questions--especially when asked by a fellow constructivist-will seem to
be ill-formulated. Compare: "No detergent is better than any other. Which detergent is better, Tide or Arm
& Hammer?"
or other to be a reason or you don't. If you do, then something matters for you;
then you have reasons. If you don't, then nothing matters for you; then you have no
If this is right, then what Korsgaard calls the "value of humanity" is not a coherent
value. For the value of humanity, expressed in various ways, is this:
• I have a reason to have some normative conception or other of my identity.
• I have a reason to take something or other to be a reason.
• It matters that I take something or other to matter.
• It is valuable that I take something or other to be valuable.
In an important sense, such claims do not have meaningful content, any more than
"The Empire State Building is taller," for these claims are asserted without reference to
any standards which could determine their correctness or incorrectness (again, taking it
for granted that realism is false). They are not asserted from any standpoint, but rather
from nowhere, and from nowhere there are no normative standards, according to
In answer to this criticism, the Kantian constructivist might reply that these claims are
asserted from somewhere-namely, the standpoint of agency itself. All we have done,
the Kantian might say, is to abstract from all particular normative commitments; but this
still leaves us with an evaluative standpoint that can set a substantive normative
standard. What we have arrived at, on this view, is the standpoint of an agent as
such-i.e. the standpoint of a creature who is able to distance itself from its unreflective
evaluative tendencies, and who needs an answer to the question "What should I do?".
Such a creature, according to Korsgaard, "needs reasons," and therefore needs some
normative conception or other of its identity to supply those reasons. In Kantian
language, what we've arrived at is the standpoint of the spontaneous will, which has
a problem: it needs some law or other in order to be a will, but if it is to be a free will,
this law cannot be supplied by any alien cause.
What makes this idea of the standpoint of"agency as such" sound plausible is the use
of words such as need and problem. Such language makes it sound as though there is
indeed a certain universal standpoint that one can identify with, a standpoint that is
identified not by any particular normative commitment but rather by the general
commitment to having some normative commitment or other. But the language of
need and problem ultimately misleads here by trading on an ambiguity. Consider the
basic claim at issue (put in a few different ways):
• An agent needs to take something or other to be a reason.
• A will needs some law or other.
• An animal with a reflective structure to her consciousness needs some normative
conception of her identity or other.
This is Korsgaard'< interpretation of Kant in her I 'N6a.
There are two different ways to understand such claims about "needs." On one
interpretation, such claims state conceptual, "constitutive" truths, such as "To be a
parent, one needs to have children." On this interpretation, these claims merely state
what is involved in being an agent, a will, or an animal with a reflective structure to her
consciousness. On a second interpretation, however, such claims state substantive
normative truths, according to which an agent has a reason to take something or
other to be a reason, or a will has a reason to have some law or other, or an animal
with a reflective structure to her consciousness has a reason to have some normative
conception of her identity or other.
Only the first interpretation of the above claims is proper. On a constructivist
picture, an agent is simply defined by the fact that it is a creature that takes something
or other to be a reason. Similarly, a will is defmed by the fact that it has some law or
other (a principle, accepted as normative); and having a reflective structure to one's
consciousness is defined by the possession of some normative conception of one's
identity or other (i.e. the taking of something or other to be a reason). To advance the
second, normative interpretation of these claims is to say something akin to "A parent
has a reason to have children." But this is confused: a "parent" who does not have
children does not have a reason to have children; rather, he or she is not a parent at all.
Similarly, an "agent" who doesn't take anything at all to be a reason does not have a
reason to take something or other to be a reason; rather, "he" or "she" is not an agent
at all. A will cannot have the problem of not having a law. If there is no law, this does not
mean there is a will with a need; it means there is no will. If this is right, then the alleged
value of "humanity" -i.e. the taking of oneself to have reason to take something or
other to be a reason-is not a coherent value, and hence cannot bring the regress of
questions to a satisfactory end.
5. How the Humean constructivist thinks
the regress ends
How then does the regress of questions end? According to the Humean constructivist,
eventually (at least in theory, if we pursue our reflections far enough) we get to a point
where we have arrived at a coherent web of interlocking values. At that point, one can
ask: "But why should I endorse this entire set of normative judgments? What reason do
I have to endorse this set as opposed to some other set, or as opposed to no set at all?"
The proper answer at this point, according to the Humean constructivist, is that the
question is ill-formulated. One cannot coherently step back from the entire set of one's
interlocking normative judgments at once, and ask, from nowhere, whether this set is
correct or incorrect, for on a constructivist view there are no independent standards to
fix an answer to this question; this is the rejection of realism. It is important to be clear
here, lest it sound like the Humean constructivist is ruling out perfectly acceptable
questions such as "Do I have a reason to reject the values of the Taliban in favor of my
own?" or "Are the nonnative conunitments of Albert Schweitzer superior to my
own?" Such questions are entirely in order, according to the Humean constructivist-
just so long as one is, at least implicidy, posing them from the standpoint of some further
set of values (however vague or inchoate) concerning what makes one set of nonnative
conunitments more worthy of endorsement than another.
The one thing that one
cannot do, coherendy, is to step back from every last one of one's nonnative judgments
at once and try to pose such questions from nowhere-asking, while suspending one's
acceptance of any value that might be capable of setding the matter, whether one
should endorse one's own set of values, or some other set, or none at all. If one tries
this, then one has stepped, for the moment, outside the standpoint of agency, into a
realm where there are no normative facts, and one's question is ill-formulated. On the
Humean constructivist view I am proposing, then, the regress of normative questions
comes to an end not with any substantive value, but with an understanding of the exact
moment at which normative questions cease to make sense-namely, the moment one
divorces oneself from the practical point of view altogether, refusing, for that moment,
to take any value for granted.
6. But if I think valuing is the source of value,
mustn't I think valuers are valuable?
But the debate is not over yet. The Kantian constructivist might think that what the
Humean constructivist has said so far is largely correct, but that there is a final
implication that the Humean is neglecting. In particular, the Kantian might point
out that if you accept metaethical constructivism, then you think, roughly speaking,
that the fact that you take something to matter is ultimately what makes it matter. And
this conunits you to a further thought, the Kantian may suggest: namely that you, as a
source ofvalue, are valuable.
To argue in this way is to restate the most basic and
intuitively appealing point that lies behind Kant's argument for the Formula of
and the "fancy new model" of this argument offered by Korsgaard. As
Korsgaard puts it: "Kant saw that we take things to be important because they are
important to us-and he concluded that we must therefore take ourselves to be
important. "
While this thought has some intuitive appeal, I don't think it is correct. One may
coherendy think that one's attitudes of "valuing" confer value without also thinking
Here I dnw on Taylor 1982.
•• In order to derive morality, the Kantian would need to go on to argue that it furthemmre follows that all
valuen (not just you) are valuable. Here again, though (cf. note 15 above), I think we need only dwell on the
first step of the argument to see why it fails; intuitively. the step to the value of one's own self seems as though
it should be the easiest of all, so the prospects for deriving the value of all valueB seem especially dim if we
can't even get to the value of our own selves.
A ~ interpreted by Kongaard in her 1'J96a and 1'J%b, section 3.4.1!.
Kongaard 1996b, p. 122.
that oneself has any value; one need only think that this is what value is-namely,
something that exists from the standpoint of a creature who takes something or other
to be valuable, but who need not take herself to be valuable. One thing that makes the
Kantian suggestion seem plausible is that as a contingent matter, most actual valuing
creatures do as a matter of fact take themselves to be valuable. And it is clear, from a
biological point of view, why this would be: roughly speaking, creatures who do not
value themselves do not look out for themselves, and so they die off. To make clear
why the Kantian point doesn't go though, then, it is useful to illustrate with a science
fiction example of a creature that does not take itself to be valuable.
Consider, then, the case of a reflective social insect, living on another planet in
another galaxy. She is a worker in a colony of highly intelligent ant-like creatures,
headed by a queen (the worker's mother, and the mother of all members of the
colony), who is the only one able to reproduce. The colony is made up of millions
upon millions of sisters, all of whom share three-quarters of their genetic material in
The worker is like us in that she values many things, but she is very unlike
us in that she takes her own self to be of only the most trivial and purely instrumental
importance. She takes the well-being and survival of her queen and colony to be the
most important things in the world, feeling happy and content so long as their future is
secure, and feeling horror and anxiety at the very thought of any harm coming to
them. When it comes to her own personal welfare and survival, however, she is mosdy
heedless of these-except as a purely instrumental matter. Though it requires effort and
concentration (it doesn't come naturally), she does her best to look out for herself so
long as she knows her continued existence is important to her mother and sisters. But
the moment she notices that she could help her family more by giving her life (for
example, in the service of a construction project), she does so cheerfully and with
eagerness-glad to be of some use, and delighted in the knowledge that the colony is
Suppose that this worker also takes the pursuit of science and philosophy to be
valuable (after all, the colony's success over the generations owes much to its scholarly
achievements, and she also views these subjects as worthy in themselves). So in her
spare time she studies biology and metaethics, and eventually arrives at the metaethical
constructivist conclusion that normative truth consists in what follows from within the
practical point of view, given a formal characterization. As part of this, she concludes,
It is this high degree of genetic relatedness that explains how a creature with the set of values I am about
to describe could arise via natural selection. Because this creature is more closely related to her sisters than she
would be to her own of!Spring (with whom she would share only one-half of her genetic material), she can
actually do "better" (as judged from an "evolutionary point of view") by forgoing reproduction and devoting
herself utterly to the promotion of the welfare of her sisters and queen. For a summary of the relevant
biology, see Trivers 1985, pp. 169-79. Of course, the origins of the creature in my example are really beside
the point here--ultimately it does not matter whether such a creature could actually evolve or not. The
imponant point is that such a creature--whether it emerged via natural selection or the handiwork of a god
or magician-1:ould value something or other without being rationally required to value its own self in
for example, that what it is for her queen's life to be valuable (for her, the worker) is for
this conclusion to follow from within the standpoint of her (the worker's) own set of
values. And this it does, she is certain: at every level of her emotional being, she is
inclined to regard her mother's life as terribly valuable, and when she subjects her
normative judgment to reflective scrutiny, she finds that she is glad to be like this (even
when fully aware of the evolutionary origins of this unreflective tendency in herself).
As judged from the standpoint of all her other values--such as her love of her sisters,
philosophy, science, and so on--she has no reason whatsoever to reject the judgment
that her mother's life is valuable and no reason whatsoever to start doing battle with her
overwhelming unreflective filial devotion.
The question at issue between the Kantian and Humean constructivists is this.
The worker now sees that what it is for her queen's life to be important is that she
(the worker) takes it to be important, and that this judgment follows from within her
(the worker's) own set of values. Must the worker now, in light of this realization-
that she is a source of value, so to speak-take herself to be important? Must she now
start treating herself as an end and no longer merely as a means to the welfare of her
mother and sisters? I think the answer is clearly no, if we understand "self" in any
ordinary sense of the word. The worker has recognized that her own values set the
standards determining what is valuable for her, but she need not conclude from this
that she herself, as an individual, is of any non-instrumental value. She may quite
properly decide to go on as before, treating her own survival and welfare as important
only insofar as they tend to promote the survival and welfare of the colony as a whole.
To the Kantian theorist who tries to convince her that she must take herself to be non-
instrumentally valuable, she might say: "The fact that it is my own values that
determine what is valuable (for me) does not imply that I am valuable. Rather, this
fact implies nothing more than what it says-namely that this is what value is-
something determined by the normative judgments of valuing creatures like myself.
Indeed, I am not valuable as legislated by my own normative judgments (except
instrumentally). You may have strong unreflective tendencies to value yourself non-
instrumentally (most animals do)-but I don't: I honestly don't care a hoot for my own
survival unless it's clear that it is benefiting these other things (my mother and sisters
and the cause of science and so on) that I do value so deeply in themselves. And I see no
reason to change that about myself, even though now I see that what it is for these
other things to be valuable is for my own set of values to say so."
I think that the reflective social insect would be right to give such a reply, and that
this provides further illustration of the Humean point that nothing substantive-here,
nothing substantive about the value of one's self-is implied by the mere fact that one
takes something or other to be valuable. At this point, the Kantian might grant that the
reflective social insect need not take herself to be valuable in the ordinary sense of this
expression (a sense we normally take to involve valuing one's own survival, bodily
integrity, and so forth), but insist that what the creature must do is to take herself to be
valuable in the sense of valuing her own valuing. So far as I can see, however, this
understanding of the "value of self" is empty in the sense that it directs nothing
substantive: it just says to value what you value. An agent can embrace any set of values
whatsoever and satisfy this "requirement." We have derived no substantive constraints
on how the reflective social insect must treat herself, much less all other valuing creatures.
7. Morality and contingency
If Humean metaethical constructivism is correct, then morality does not follow from
pure practical reason, understood as the standpoint of a valuer as such. Instead, we must
conceive of our relationship with morality as more contingent than that. Moral feeling
is something that an agent ultimately either has or does not have as part of his nature,
and if an agent has no trace of it anywhere in his evaluative makeup, or if that feeling
does not run deep enough when push comes to shove, then this is not a mistake in any
genuine sense, on the Humean constructivist view; in that case, the agent simply is not,
at bottom, a moral agent.
From a Kantian point of view, this thought might seem especially distressing. For on
a characteristically Kantian way of thinking about morality, it's part of the very idea of
morality that its requirements are categorical-not something whose "bindingness"
one may escape merely by failing to care about it. From this point of view, the
conclusion that morality is not dictated by pure practical reason could seem tantamount
to the conclusion that there is no such thing as morality. Certainly Korsgaard in The
Sources of Nonnativity takes her task to be nothing less than showing how morality
follows from within the standpoint of any valuer, and it is a plausible interpretation of
Kant that he thought morality had to be categorically binding in this way if it was not
to be a "phantom."
But we should question this conception of morality. It seems right that it is part of
the very idea of morality that its requirements are categorical with respect to some
important parts of our evaluative nature--for example, that it is categorical with respect to
what we desire to do in an ordinary sense or what we fmd most appealing or pleasant.
But it seems to me that we have gone too far if we think that it is part of the very idea of
morality that its requirements are categorical with respect to any evaluative nature an
agent might have. Instead, on the view I am suggesting, it is constitutive ofbeing a moral
agent that one take certain requirements (of a certain characteristic content concerning
the equal treatment of others) to be binding even if carrying them out goes against
certain large aspects of one's evaluative nature--including what feels easiest, what is
pleasant, fun, what one finds most naturally appealing, and so on. But the part of
The agent might, of course, still be making an instrumental mistake as dictated by the tenns of his other
values. For all the usual Hobbesian reasons, for example, the agent might have self-interested reason to act
morally or perhaps even to cultivate moral feeling in himself. The H umean meta ethical constructivist can
accommodate such thoughts as well as anyone. But the Kantian constructivist, like many theorists, is looking
for more.
oneself that takes the requirements to be binding in this way is just another part of one's
evaluative nature, and it goes too far to think that morality is not morality unless it
binds us independendy of that part of our evaluative nature as well. Properly under-
stood, in other words, the requirement to tell the truth, for example, binds us
independendy of whether we feel like telling the truth, or whether we would find it
fun or pleasant or easy or to our personal advantage narrowly understood, and so forth.
But it is a mistake to think that the requirement must also bind us independendy of
whether we accept that very requirement.
On this view, the Kantian constructivist is making a mistake a bit analogous to what
the Kanti.an and Humean constructivists both agree is the realist's mistake. The realist's
mistake is the mistake of thinking that requirements must bind us independendy of the
evaluative standpoint itsel£ The Kanti.an and the Humean constructivists are united on
this point against the realist. But the Humean constructivist will think the Kantian
constructivist is committing another mistake-less severe than the realist's, but still a
mistake, and in the same ballpark. The Kantian constructivist's mistake is the mistake of
thinking that moral requirements must bind us independendy of the particular evaluative
nature with which we fmd ourselves-and in particular, independendy of whether we
already have moral concerns as a deep part of our nature. But the right view, according
to the Humean constructivist, is that moral requirements do not bind us irrespective of
our particular evaluative nature. In particular, if one lacks moral concerns altogether,
then morality does not bind one. But if one is a moral agent, as opposed to just an
agent, then part of what that involves is taking oneself to be bound categorically (in
certain cases) with respect to what one feels like doing, what one finds pleasant and
attractive, and so forth. Part of the work of standard normative ethics, on this view, is
the precise mapping of the contours of the relevant part of our evaluative nature-our
moral nature. Depending on one's view, that part of our nature could be characterized
as a commitment to living in ways others could not reasonably reject, or to weighing
the happiness of all impartially, and so on-and doing so even when it is unpleasant, or
indeed much worse. It is important to map out the nature of these commitments as best
we can. In so doing, we are mapping out the values that are constitutive of moral
agency. But it is a mistake to think that we have to show how these values follow from
the standpoint of agency as such, on pain of their not being categorical in the right way.
The "categoricity" of moral requirements is with respect to one important part of our
contingent evaluative nature, but not the whole of our contingent evaluative nature.
This line of thought raises the worry, though: Can one sustain one's commitment to
morality, when (as Korsgaard emphasizes) what it demands is sometimes so hard, and
when one comes to view it as a contingent matter that one is a moral agent in the first
place? This question raises large issues, and here I will merely offer two brief thoughts.
First, it is and it is not a contingent matter that "I am a moral agent." It is a
contingent matter in the sense that morality is not required by practical reason as
such. It is also a contingent matter in the sense that there is a perfectly intelligible
conception of personal identity on which I can imagine myself without moral
commitments. But there is another sense in which it is not at all a contingent matter
that I am a moral agent. In a different, deeply intuitive sense of personal identity,
commitments like this are constitutive of who I am, such that I would regard myself as
having vanished or died were I to lose them. Here the close connection with
Korsgaard's own thinking about practical identities is clear. For those of us who have
moral commitments as a fundamental part of our evaluative nature, to lose those
commitments entirely would be to perish in a deeply intuitive sense: these are the
commitments that define ourselves in large part in our own eyes, and so, when we look
at ourselves, we cannot just think "It's a contingent matter that I am a moral agent."
For if I imagine that, perhaps due to brain injury or illness or just the wear and tear of
life, I were to lose my moral commitments entirely, then my view of that hypothetical
case is that I, as I think of myself, would be gone from the world-in roughly the same
way I think I would be gone from the world were I to develop a severe case of
Alzheimer's. Someone l s ~ person with the same personal identity in one sense, but
not the sense that intuitively matters-would have taken my place.
Second, to the extent that our relationship with morality does strike us as contin-
gent, an analogy with love suggests itsel£2
Unless one is a hopeless romantic, one will
think that there are any number of human beings running around on the planet today
whom one might have, in some very real sense of "might have," met, fallen in love
with, married, and built a happy life with, had things gone just a bit differendy. But
many of us also think that it is possible to acknowledge this contingency fully without
having it undermine one's lifelong love and commitment to the person whom one did,
as it so happened, actually meet, fall in love with, marry, and build a happy life with.
On the contrary, there is a way in which the very contingency of the relationship--the
extreme fragility of the whole way things happened, the ease with which they could
have gone differendy-can be seen as making the relationship all the more dear: this is
the one, whom, among all the possible ways things could have gone, I did meet and
become bound to. Our relationship with morality might be seen as similar.
A cynic might at this moment point to the high divorce rates and object that this last
point is more disturbing than comforting if the goal is to reassure us about our
relationship with morality. And I am the last one to suggest that these questions
about morality are easy. The above remarks are meant only as preliminary thoughts,
and in the meantime I might answer the cynic with the cynical remark that there's
always the threat of prison.
8. Conclusion
I will close by discussing a puzzle that might seem to be raised by a Humean, as opposed
to a Kantian, version of constructivism. If one accepts Humean constructivism, then
•• See Lenman 1999. p. 167, for a similar point.
one accepts that contingencies-social, historical, biological, and otherwise-have
played a crucial role in determining what nonnative reasons one has.
One might
wonder how this recognition squares with a point we considered earlier, in the context
of the second premise ofKorsgaard's argument, according to which (in Kant's words)
"we cannot conceive of a reason which consciously responds to a bidding from the
outside with respect to its judgments." It might seem that to accept Humean construc-
tivism is to try to do exactly what Kant says is inconceivable, namely consciously to let
one's reason be determined "from the outside." Practical reason is being asked to look
to a source outside itself for the ultimate substantive content of its reasons.
We may clear up the apparent puzzle by distinguishing between the application of
Kant's point at the level of individual values, and the application of the point (or rather
the lack thereof) to one's entire set of values taken at once. It is entirely correct,
according to the Humean constructivist, that in the case of any individual value, and
indeed in the case of any subset of your total set of values, you must (when you think
about it) regard yourself as having some reason or other for endorsing them. Such a
reason, if one has it, will be supplied by the standards set by one's other values. When it
comes to the level of one's entire set of values taken at once, however, the Kantian
point no longer holds. One can, and indeed one must, consciously accept that one's
starting set of values was given to one "from the outside." This does not mean that one
is forced to accept any one of them; by no means. Any one of them may be called into
question and subsequendy rejected, if one takes oneself to have reason to do so. What it
does mean is that there is no escaping the fact that you have to start somewhere as an
agent-with the acceptance of some values or other--and that this starting point
cannot itself be chosen for a reason, since there is no standpoint prior to agency from
which one could do this. To put it another way, it is only causes, and not reasons, that
can catapult one into agency. Of course, your parents and other agents can have reasons
for bringing about your birth, and for nurturing you on into reflective agency. The
point is that you cannot; you must simply come alive. And this is the sense in which you
can and must accept direction from the outside with regard to your judgments: you
must be born as a reflective creature with some values or other. After that, however,
things are very much up to you (for now there is a you). When it comes to the case of
any individual value or subset of your values, you cannot consciously let your judg-
ment be directed from the outside. But this is just to say: you cannot consciously let
your judgment be directed from outside the further values that are you.
Even on a Kantian version of metaethical constructivism, contingency plays a large role in shaping one's
reasoru. On this point, see Korsgaard 1996b, pp. 241-2. But the point obviously holds even more strongly for
Humean versions of metaethical constructivism.
For helpful feedback on earlier versions of this material, I am indebted to Melissa Barry, Michael
Bratman, Nadeem Hussain, Aaron james, Christine Korsgaard,James Lenman, Mike rudge, T. M. Scanlon,
Yonatan Shemmer, Robert Stem, Valerie Tiberius, David VeUeman, and Jay Wallace. I am also indebted to
audiences at Brown University, the Ohio State University, the University of California at Los Angeles, the
University of Connecticut at Stom, the University of Michigan at Ann Arbor, the University of Pittsburgh,
the University ofSheffaeld, and the University of Vermont at Burlinl!ton.
Bagnoli, C. 2002. "Moral Constructivism: A Phenomenological Argument," Topoi 21,
pp. 125-38.
Dorsey, D. Unpublished ms. "Relativism and Constructivism: A Humean Response."
Frankfurt, H. 1982. "Freedom of the Will and the Concept of a Person," in Watson 1982a.
Kant, I. 1785/1959. Foundations of the Met4physics of Morals, translated by L. W. Beck (New York:
Macmillan Library of Liberal Arts).
Korsgaard, C. 1996a. Creating the Kingdom of Ends (Cambridge: Cambridge University Press).
Korsgaard, C. 1996b. The Sources of Normativity (Cambridge: Cambridge University Press).
Korsgaard, C. 1997. "The Normativity oflnstrumental Reason," in Ethics and Practical Reason,
eds G. Cullity and B. Gaut {Oxford: Clarendon Press).
Lenman,J. 1999. "Michael Smith and the Daleks: Reason, Morality, and Contingency," Utilitas
11, pp. 164-77.
Lenman, J. 2010. "Humean Constructivism in Moral Theory," Oxford Studies in Metaethics
vol. 5, ed. R. Shafer-Landau (Oxford: Clarendon Press), pp. 175-93.
Lewis, D. 1989. "Dispositional Theories ofValue," Proceedings of the Aristotelian Society, suppl. 63,
pp. 113-37.
Rawls, J. 1999. "Kantian Constructivism in Moral Theory," in Collected Papers, ed. S. Freeman
(Cambridge, MA: Harvard University Press).
Schroeder, M. 2007. Slaves of the Passions (Oxford: Oxford University Press).
Street, S. 2006. "A Darwinian Dilemma for Realist Theories ofValue," Philosophicol Studies 127,
no. 1, pp. 109-66.
Street, S. 2008a. "Constructivism about Reasons," Oxford Studies in Metaethics, vol. 3, ed.
R. Shafer-Landau (Oxford: Clarendon Press), pp. 207--45.
Street, S. 2008b. "Reply to Copp: Naturalism, Normativity, and the Varieties ofRealism Worth
Worrying About," Philosophical Issues (a supplement to Noas), vol. 18 on "Interdisciplinary
Core Philosophy," ed. W. Sinnott-Armstrong, pp. 207-28.
Street, S. 2009. "In Defense of Future Tuesday Indifference: Ideally Coherent Eccentrics and the
Contingency of What Matters," Philosophical Issues (a supplement to Noas), vol. 19 on
"Metaethics," ed. E. Sosa, pp. 273-98.
Street, S. 2010. "What is Constructivism in Ethics and Metaethics?" Philosophy Compass 5,
pp. 363-84.
Street, S. 2011. "Mind-Independence Without the Mystery: Why Quasi-Realists Can't Have It
Both Ways," Oxford Studies in Metaethics, voi. 6, ed. R. Shafer-Landau (Oxford: Clarendon Press).
Street, S. Unpublished rns. "Objectivity and Truth: You'd Better Rethink It."
Taylor, C. 1982. "Responsibility for Self," in Watson 1982a.
Tiberius, V. 2008. The Rejleaive Lift: Living WISely with our Limits (New York: Oxford University
Trivers, R. 1985. Social Evolution (Menlo Park, CA: Benjamin/Cummings Publishing).
Watson, G. 1982a. Free WiU (Oxford: Oxford University Press).
Watson, G. 1982b. "Free Agency," in Watson 1982a.
Williams, B. 1981. "Internal and External Reasons," in Moral Luck (Cambridge: Cambridge
University Press).
Wright, C. 1992. Truth and Objectivity (Cambridge, MA: Harvard University Press).
Constructing Protagorean
Aaron james
Values have protagorean objectivity, let us say, just in case they are (i) dependent on us by
their very nature, and yet (ii) fully objective as values. Especially since the late Early
Modern period, a central aim of philosophical ethics, if not its Holy Grail, has been to
explain how features (i) and (ii) might be combined-to explain, that is, how depen-
dence on us, of the right kind, does not preclude the possibility of fully objective values.
Constructivist theories, perhaps more than any other approach, purport to characterize
"dependency on us" of the needed sort: values are (in a sense specified momentarily) a
construction of practical reason. They are thus independent of what individuals or groups
simply happen to think or prefer, but no more or less real, or problematic, than the forms
of reasoning which express our practical reasoning capacities. Accordingly, the idea goes,
we can reject subjectivism, conventionalism, cultural relativism and any other view that
immunizes individuals or groups from error, but without thereby accepting Platonism,
theological voluntarism, or any other view that allows values to be radically esoteric, and
so not values in the ordinary sense. Instead of choosing between standard "realist" and
"anti-realist" views, we instead take an attractive middle way.
Unfortunately, however, even the best-developed constructivisms have not clearly
marked the hoped-for distinctive position. According toT. M. Scanlon, constructivism
is applicable to morality but unfit as a general foundational theory.
John Rawls and
Christine Korsgaard do favor constructivism "all the way down" (or without assuming
any other view, on the way down).
But both affirm the autonomy of ethics in a way
that seems to reject the very questions that standard realist or irrealist views seek to
For comments or relevant discussion of this material I am grateful to Michael Bratman, William Bristow.
ElllUnno Bencivenga, Nadeem Hussein, Christine Korsgaard, James Lenman, Peter de Mameffe, James
Nickel, David Plunkett, James Pryor, Andrews Reath, Michael Ridge, Faviola Rivera-Castro,
T. M. Scanlon, Tamar Schapiro, Yonatan Shemmer, David W. Smith, Gary Watson, Nicholas White, and
audiences at UC Riverside, Universidad Nacional Autonoma de Mexico, Sheftield University, Arizona State
University, and the UC Irvine seminar which I co-taught with David W. Smith.
• Scanlon (this volume). ' Rawls 19'16, 1999d; 1996b, 2003.
answer. Constructivism is not then a distinctive foundational answer to those questions,
at least not as conventionally understood.
Our aim will be to set out a version of constructivism which explains the possibility
of protagorean objectivity in a distinctive way. With no pretense of providing an
adequate defense, or of defining anything righdy called "constructivism," our aim is
merely to oudine the main elements of one possible and unduly neglected account.
While perhaps not a "metaethical" proposal in every sense in which that troubled
name is used, our constructivism seeks a particularly deep kind of foundational
explanation, which we presendy explain.
Each age struggles to shake off its own set of misguided questions. So we should not
take any set of"metaethical" questions wholly for granted, however often they have been
asked by respectable people. The metaethical questions we ask, as well as the answers we
give, need to be justified. The proper rationalist methodology, I take it, is what might
be called default quietism: any "meta-question" is presumed misguided, as the default
position, until we see that it is well motivated and admits of illuminating answers. For
the constructivist, some received metaethical questions will not pass muster and fall by the
wayside. But other foundational concerns are so central to the constructivist project that
they cannot be readily disavowed. In particular, constructivism suggests that it is a deep
mistake to conclude from the rejection of subjectivism or conventionalism that objective
values must be located somewhere within the impersonal world, seen from the "view
from nowhere," in Nagel's in Platonic heaven, the will of God, or
otherwise in the metaphysical fabric of reality. What is this deep mistake? Constructivism
should answer this question. But the standard constructivists do not, at least not in
unambiguous terms. Our version of constructivism is distinctive in part because of the
way it answers this profound question about the nature of the ethical domain.
1. Constructivism: four theses
The version of constructivism we elaborate can be stated as a family of four main theses.
The first thesis, as applied to a class of ethical propositions-whether about morality,
prudence, or reasons for action-is the Basic Thesis:
Necessarily, a proposition of the kind in question is true, or a fact, if and only if (and
because) anyone capable of following the norms of practical reasoning would affirm
it (as true) on the basis of valid, fauldess reasoning, in conditions optimal for practical
• Darwall, Gibbard, and Railton, 1992, pp. 137-44, puzzle about this. Gibbard 1999 assimilates Kors-
gaard's constructivism to his own expressivism Shafer-Landau 2003, Chapter 2 argues that constructivism
ultimately collapses into either relativism or realism. Hussain and Shah 2006 argue that Korsgaard's construc-
tivism does not provide an alternative to familiar metaethical views.
' The Basic Thesis can also be expressed as a property identity, e.g. as "An ethical property P == the
property of being judged as having P under the relevant conditions." Note that I bypass sensitive questions
This is to deny realism, the view that well-reasoned judgment, however idealized,
involves tracking truths which are independent of idealized practical reasoning. It is also
to deny views that regard ethical truth and well-reasoned judgment as interdependent,
which would be to say that ethical truth is not simply a function or construction of
well-reasoned judgment.
The Basic Thesis assumes the idea of validity in practical reasoning. But in what does
validity consist? The second thesis, the Constitutional Conception of Validity, answers as
The property of validity in practical reasoning just is (identical to) the property of being required
of a reasoner, on a particular occasion of judgment, if he or she is to fully understand the activity
of practical reasoning.
In other words, to reason validly in making an ethical judgment (e.g. whether slavery is
wrong, or pain is bad) is nothing more, and nothing less, than to manifest a full
understanding of the activity of reasoning, much as speaking well reflects one's full
understanding of a language, or playing well reflects one's full understanding of a game.
For all the Constitutional Conception (of Validity) says, good practical reasoning
may be wholly esoteric or unavailable to us. The third thesis, the Protagorean Constraint,
rules this out:
Validity in practical reasoning cannot outstrip the patterns of thought exhibited in the human
practice of questioning, attending, considering, judging, and deciding in matters of practical
Thus, if, for example, humankind generally fails to reason as homo economicus, then the
norms of orthodox rational choice theory thereby fail to qualify as norms of practical
reasoning. The difference between performance and the putative standard of compe-
tence need only be sufficiently great.
The above theses do not themselves answer various forms of skepticism. The fourth
thesis, Rationalist Anti-skepticism, has this role. It has two parts:
(i) Good practical reasoning favors a range of particular ethical conclusions asso-
ciated with commonsense ethics (e.g. "Slavery is wrong," "Pain is bad," or "We
have reason not to irreparably degrade the planet"}.
(ii) We cannot adequately understand ourselves except as creatures endowed with
practical reason, understood to include reasoned (non-instrumental) practical
Thus, given (i), constructivism offers a non-revisionistic account of what values are
objective. According to (ii), this account is not founded upon a necessary fiction but
about where exacdy to place the modal operator, "necessarily." The intent is simply to rule out what I call
"brute error" below.
• See e.g. McDowell 1985; WiAAios 1987; Johnston 1lJI!9.
rather the best conception of ourselves, of what we ultimately are. We cannot plausibly
accept conceptions of ourselves that efface our capacities of practical reasoning. And so
skepticism that regards practical reasoning as just so much confabulation and rationali-
zation, a by-product of psychic forces and social conditioning-in short, as pure
(though perhaps practically unavoidable) fiction-mischaracterizes human nature.
2. Precluding brute error
Consider first the role of Rational Anti-Skepticism. Its presence in the constructivist
package implies that the other theses are not intended to vindicate the objectivity of
values against skeptical challenge by themselves. Constructivism assumes, and is not
meant to wholly replace, an independent anti-skeptical critique. In defending a version
of sub-claim (i), for example, Dworkin and Nagel argue, in Moorean style, that basic
practical and moral claims-that pain is bad, that exploitation is unjust-are manifesdy
more plausible than any philosophical theory that would reject them.' Korsgaard
takes a similar negative tack, rejecting Humean skepticism about practical reasons as
inconsistent with the Humean's own afflilllation of instrumental reason.
Versions of
sub-claim (ii) may be similarly defended, in purely negative terms, against the likes of
Thrasymacus, Nietzsche, Hume, Darwin, and Foucault. Upon close critical scrutiny,
the argument will go, we cannot recognize ourselves in the proposed terms.
In other words, constructivism assumes that sub-claims (i) and (ii) can each be
negatively justified, by rejecting any version of their denial on their own terms,
without assumption of constructivism's other positive theses. Indeed, Rationalist
Anti-skepticism is by itself fully consistent with unqualified quietism, the rejection of
all positive philosophical theses about ethics as misguided or confused. On the unqual-
ified quietist's reading, claims (i) and (ii) are afflitlled only as high-level substantive
ethical theses, for which no positive philosophical explanation is possible or required.
Unqualified quietism rejects any and all metaethical questions, including standard
questions such as: Are any ethical expressions apt for truth? If so, are any true rather
than false? If so, why? How then would we come by ethical knowledge? And why
might such knowledge motivate action in the way it seems to? For present purposes,
we need not pause over familiar ways of motivating these questions. For even if we
should say something in reply, that is not yet to say why the positive thesis in the
constructivist package should be necessary, even from a broadly rationalist point of
view. For minimal realism afflitllS the metaethical theses that particular ethical claims
are true, objectively true, and known, but stops there.
If those claims can be defended
Nagel 1997, Dworkin 1996.
• Korsgaard 1997.
• This is one reading of Nagel 1997 and Korsgaard 1996b, 1997.
Nagel 1'197, Dworkin 1996, Raz 2002, Shafer-Landau 2003, Padit 2011. Scanlon 1998, Chapter 1 is
ambiguous berween realist and constructivist readings.
(e.g. in Nagel's or Dworkin's Moorean way), the question becomes: Why think some
further form ofmetaethical explanation is needed?
Our constructivism agrees that there are known objective ethical truths (and perhaps
objective truths we as yet know not of). It can even agree that rational vindication of
this thesis requires no further explanation. It also suggests that a further kind of
explanation is nevertheless possible and highly desirable: we should explain how ethical
truths can count as objective and yet "depend on us." Our constructivism does this by
addressing the possibility of brute error, of error despite the best of human reasoning.
What is distinctive about constructivism, in contrast with minimal realism, is that the
possibility of brute error is explicitly ruled out.
Let us assume for the sake of argument that constructivism meets relevant require-
ments of non-circularity.U The Basic Thesis proposes a kind of explanation, and so
assumes, for example, that the norms governing practical reasoning will count as valid,
for the activity of reasoning, independently of the truth of the judgments to which they
The constructivist characterization of "optimal conditions" will be elaborated
only in terms of such truth-independent norms, which are in tum seen as imperatives of
thought andjudgment. Ethical propositions (e.g. "We have reason to protect the environ-
ment for its own sake"), which admit ofbeing true or false, are thus explained in terms of
imperatives (''In considering the value of environmental protection, judge that it is to be
protected, independently of human benefit!"), which can be applicable or inapplicable in a
given situation, or valid or invalid for an activity, but do not admit of truth or falsehood.
One kind of thing is thus explained in terms of a different kind of thing. And this is all it
takes to raise Euthyphro's question of order of detennination (or, if you like, of"consti-
tutive grounding," rather than mere modal co-variation). Is the validity of imperatives of
reasoning to be explained in terms of some relation to independent propositional truth, or
is propositional ethical truth to be explained in terms of independently valid norms? The
realist takes the former, Socratic, view. Constructivism takes Euthyphro's side.
The question, then, is what is at stake in this dispute. Our answer is that minimal
realism does not provide an explicitly protagorean account of objective values. To see
why, consider an ethical disagreement. One party (an individual or community} holds
that the United States has reason to protect the environment for its own sake. Another
party denies this, holding that the United States should protect the environment only
to protect humankind. Assuming these views are logically inconsistent, and leaving
aside who is in the right and why, we may ask: What, if anything, explains how one of
these could be the truth of the matter?
According to minimal realism, there just is an
I take up the issue in detail in James 2007.
But other kinds of circularity hannlessly remain. For instance, because the Basic Thesis is not conceptual
analysis, it can mention the concept being explained in its explanans (Wright 1992).
" I develop this point inJames2fX17, in part to chzllenge the idea that constn1ctivist views must specify the
nonns of reasoning as a method or procedure of judgment (as e.g. according to Scanlon, this volume).
•• Uke minimal realism, constructivism has no reason not to assume a robust cognitivism. By contrast,
expressivisu must earn the right to speak of truth aptness, as in Blackburn 1993 or Gibbard 1990, 21103.
objective truth or fact of the matter, which we are able to know. The question,
however, is what, if anything, explains how this is possible, beyond simply asserting
that it is so. Initially, one wants to say that "One party has the better arguments." But
this requires interpretation, since it can be read as a version of constructivism: roughly,
best arguments determine truth. If it is not read this way, however, it is hard to see how
objective truth is ultimately protagorean. Assume, then, that ethical truth is indepen-
dendy ftxed. What is its relation to good reasoning, to having the better arguments?
The relation is either contingent, or it is necessary. On either reading, a form of brute
error is allowed.
If the relation is contingent-if having the best arguments is just a reliable way of
coming to know what is independently true-then it just turns out that best arguments
take us to truth. At least in theory, we could have the best arguments and yet fall into
global ethical error, because the ethical facts are other than they could ever seem to us.
And if the relation is instead necessary-because "best" arguments are defined as just
whatever leads to independent truth, or because our ideas ofbest arguments and truth
are thought to be interdependent-this merely pushes the problem further back. We
can now be in brute error about "best reasoning." For all we have said, the best
reasoning humans could ideally do about what is true, and what arguments are best,
could still come up radically short.
Compare, in this light, Platonism and theological voluntarism. Platonism would
allow that values might have been wholly inaccessible to us; we might have lacked our
powers to "recollect" or intuit the Forms. Similarly, according to theological volunta-
rism, God's will, and so the objective values, might remain esoteric in the absence of
revelation. These views do not aspire to protagorean objectivity, and indeed invite
concerns of radical error by characterizing values as radically independent of us. While
minimal realism hardly invites the concern, neither does it explicidy rule it out. Given
that most modem rationalists, including minimal realists, adopt a broadly protagorean
orientation and methodology anyway,
it is presumably desirable, if possible, to have a
metaethical view that articulates this fundamental and distinctively modern position.
Constructivism does just this. The Basic Thesis precludes brute error at the level of
basic ethical judgment. The Constitutional Conception of Validity, along with the
Protagorean Constraint, precludes brute error at the level of reasoning.
To illustrate, return to a case of disagreement. If the Basic Thesis is correct, then all it
takes for one party to a disagreement to have the truth of the matter is for the norms of
practical reasoning, as followed in optimal conditions for reasoning, to favor his or her
" Debates over naturalism, the relation between reason and passion, instrumental and other human
rationality, and the "normativity" of ethical concepts usually remain close to actual human practice. Even
minimal realists eschew any metaphysical conception of values that could put great distance between
objective values and human valuation.
•• The "availability" of morality is central to Schneewind's account of specifically modem conceptions of
morality (Schneewind 1998}. While Korsgaard 1996 portrays this as an issue of "normativity," I treat it as a
specifically rpistemo/ogical constraint in the first instance.
view. His or her view thereby counts as true, rather than false; there is no possibility of
error despite best reasoning, and so no reason to posit a further "truth-maker,"
metaphysical or otherwise. Moreover, because the nonns of practical reasoning cannot
outstrip human reasoning practice, given the Protagorean Constraint, there is no
possibility that valid reasoning is unavailable to us by way of the fonns of thought,
reasoning, and reflection that we ordinarily use in ethical judgment. There is no need
for special powers of rational insight, as according to Platonism, or for revelation, as
according to theological voluntarism. If or when we do go wrong in our ethical views,
there is an explanation why within reason-that we fail to reason, that we are incapable
of reasoning well, that we reason poorly, or that we face suboptimal conditions for
Of course, this will fail on its own tenns if it precludes the objectivity of values. But it
is not clear why it should. We may still firmly reject any view that relativizes ethical
truth to the actual views or dispositions of any particular agent or community.
According to Bernard Williams's neo-Humean subjectivism, for example, if Caligula
has no disposition, grounded in his contingent "motivational set," not to kill at whim,
then he has no reason not to kill at whim." We can regard this as compromising the
objectivity of values as follows. In general, we may say, ethical truths count as
objective, or "true independently of us," just in case their truth value is invariant
with respect to certain possible differences in our individual or collective attitudes.
That is, an objective truth is "true anyway"; we can't change the ethical truth simply by
changing our individual or collective minds. Yet objectivity, at least in the case of
values, does not require any explanation in particular for why the required relations of
attitude invariance hold. It is enough that they do hold, or that they have some
explanation or other; objectivity doesn't care what the explanation is.
ln the present
case, then, we can affirm that the ethical truth is invariant with respect to any given
agent's contingent motivations and dispositions of evaluation. We need only say that
the norms of practical reasoning are not so constrained: the correct judgment, accord-
ing to those norms, may be that Caligula is ignoring considerations relevant to his
reasons, such as needless suffering or loss of life.
We can reject conventionalism, cultural relativism, and any such relativization to our
actual attitudes in the same general way. Such views might agree that Caligula's reasons
are not determined by his own attitudes, because they are instead determined by some
conventional or cultural practice. So if Caligula's culture approves of his killing at
whim, he has no reason to do otherwise. Here, again, in order to affirm that ethical
truth does not necessarily vary with such community practices, we need only say that
the norms of practical reasoning always allow us to open such practices to critical
scrutiny and in some cases come down against them. That itself suffices to deny a
Williams 191!1, 1995. On Calligula and objectivity, !lee Gibbard 1990.
•• 1 defend this view in James 2006. For similar proposals, see Blackburn 1 993, p. 1 53, Gibbard I '190,
pp. 164--6, 2003, p. 11!3, and Dworkin 1996.
general relation of detennination by community attitudes. More generally, for any
such subjective condition of an agent, or intersubjective condition of a group, C, as
long as we can, within reason, ask critical questions about C's relevance, and in some
cases side against it, as irrelevant or not detenninative in the case at hand, then ethical
truth does not vary with C, and is in that respect objective.
Does the same reasoning undercut the constructivist's deep identification of ethical
truth with the output of idealized human reasoning practice? Might that output be a
subjective condition C, which is open to question and counter-judgment, by the best
standards of good judgment, in which case constructivism fails to capture full objectiv-
ity? It does not seem so. Suppose we have on hand the most well-reasoned, fully
informed judgment that any person could ever make about what one has reason to do.
Could that judgment really involve a radical, fundamental mistake? We cannot engage
in the sort of critical evaluation which seems open when subjectivism and convention-
alism are our concern. We cannot say, that is, that some relevant humanly available
information is missing, or that some defect in humanly possible reasoning has been
made (e.g. some relevant possibility is ignored). For these diagnoses of error are already
built into the constructivist idealization. Instead, the supposed fundamental mistake
would have to exist for no reason, other than its failure to represent a supposed fact to
the contrary. The error must be brute. But if we have no idea even of how we might
evaluate the present idealized judgments as mistaken, it does not clearly make sense to
suppose there is nevertheless a real possibility of going wrong. The natural doubts we
may have about conventionalism and subjectivism, within reason, do not arise for
idealized human reasoning in general.
3. The autonomy of the ethical from the outside
Even if the Basic Thesis proposes a kind of non-circular explanation, the question
immediately arises what the explanation might accomplish.
It may seem to simply
rephrase the explanatory task, in terms of the status of the supposed "norms of practical
reasoning." It at least seems fair to ask certain general questions about validity, such as:
What does validity in practical reasoning consist in? or What does it mean to say that
reasoning is "valid"? But in that case, potential answers include:
Reasoning Platonism: a pattern of thought is valid as reasoning just in case it conforms
to a Norm ofReasoning, known by Intuition.
Reasoning Subjectivism: a pattern of thought is valid as reasoning, just in case it
constitutes what Bernard Williams calls a "sound deliberative route" to a motivation
to action from an agent's "dispositions of evaluation. "
Reasoning Conventionalism: a pattern of thought is valid as reasoningjust in case it is
conventionally recognized as valid for reasoning in a particular community.
•• Enoch 2011.
Williams 1981. 1995.
Reasoning Expressivism: to take a pattern of thought as valid reasoning is to express
one's acceptance of a norm that pennits that pattern of thought in reasoning (or on
Gibbard's more recent version, to express a plan to so reason).
Constructivism is hardly of deep foundational interest if it must assume some such
familiar view. The question, then, is whether there any alternative.
One alternative is quietism about the questions of validity which these views seek to
answer. It is fine, on this reply, to ask specific, substantive questions about which fonns
of reasoning and judgments are valid. But any general question-about what validity
generally consists of, or about the meaning of "valid" as applied to reasoning--is ill
formed or misguided. And since the views above then answer one or
another misguided question, constructivism need not be distinguished from them. It
counts as distinctive because it transcends rather than enters the traditional metaethical
debate. Rawls hints after such a view.
Korsgaard all but avows it. She claims that
norms grounded in rational agency only arise from the "practical point of view" and
otherwise rejects the "theoretical" or "explanatory" aspiration of much traditional
metaethical inquiry.
For our purposes, such quietism is at best premature. To see why, consider that it
leaves Allan Gibbard room to grant Korsgaard's position from within the "practical
point of view" and still treat ethical judgment as a mere expression of certain attitudes,
when viewed from an external perspective.
This displays one of expressivism's chief
virtues, its account of what Gibbard calls the "transcendental turn": if an expressivist
conception of meaning is correct, this provides a general, "external" rationale, of what
might be called philosophical anthropology, for the autonomy of the ethical domain.
A question of ethics can only be an "internal" question, an occasion to express one's
The quietist might of course simply dig in, rejecting any and all external
questions, about what validity generally is or consists in, or of what "validity" means.
Yet the result is precisely what we are after, an account of what we called the "deep
mistake." The better reply, in that case, is not quietism but rather to accept general
questions about the nature of validity in reasoning and provide a constructivist answer
which explains the deep mistaken in a different, distinctive way.
Indeed, Korsgaard's project of grounding norms of practical reason in rational
agency, as "internal standards" of action, can be understood in just this way. Certain
norms count as standards because particular acts can be evaluated with reference to
them. They also count as internal to action, however, in the sense that approximate
Gibbard 1990, 2003.
Rawls 1999a and 1996, p. 101, where Rawls writes: "(for Kant! the affirmation of reason is rooted in
the thought and practice of ordinary (sound) human from which philosophical reflection must begin."
Korsgaard 1996, pp. x-xii, 116, and 1997. Kongaard 2009, has a more metaphysical hue.
Gibbard 1999. In a passing comment, Kolligaard 2003, n. 9, seems to concede this ground. For a similar
expressivilt constructivism, see James Lenman (this volume).
See Gibbard 2003.
compliance with them is constitutive of action, or part and parcel of what performance
of an action, by its very nature, is. Here Korsgaard can grant the question, What does
the validity of the relevant standards consist in? Her answer is that their validity for
action just is their status as constitutive of action: they are valid for action because part
of what action is. Moreover, if validity is nothing more, and nothing less, than the
requirements of full-fledged action, whatever they may be, then any further external
question about validity ("But what is validity, really?") will be confused. It will count as
confused, not because the initial general question of validity was misguided or unin-
teresting, but because the answer to that question has not been registered. For once we
invoke the relation of constitution, any further questions about validity can only be
internal questions, about what full-fledged action substantively involves. (Korsgaard's
claim that the Categorical and Hypothetical Imperatives are "internal standards" is thus
an internal claim, defended by internal argument.) We thus have the desired account of
the "transcendental turn," but without assuming a (potentially problematic) expressi-
vist theory of meaning. The autonomy of the ethical then reflects a deep truth of
philosophical anthropology, about the very nature of agency, which explains, from an
external perspective, why ethical questions must begin from within the practical point
of view.
4. An intellectualist approach
Korsgard's appeal to constitution may be read as a voluntaristic version of the Constitu-
tional Conception of Validity as formulated above. The activity of "practical
reasoning" is understood as the activity of choice. While this is progress for our
purposes, it also threatens the objectivity of values (of which more below). We
therefore now elaborate a less encumbered intellectualist version of the Constitutional
Conception. It sees "practical reasoning" not as action per se, but as a species of
judgment, guided by imperatives of thought, which may or may not call for intention,
choice, or action.
On this view, the deep mistake in supposing that objective values
must be located in the impersonal world is not to miss the truth of expressivism, or a
deep feature of agency per se, but rather to miss the constitutive relation between
validity and human reasoning practice-the practice, as I will call it, of reasoned practical
The activity or practice of reasoned practical judgment is not all of "practical
reason," but one of its central elements: a familiar process of (non-instrumental)
reasoned consideration. The process begins when one poses a question about what is
just or right or to be done given the (non-evaluative) facts of some situation. It ends
when one reaches a judgment on the matter, whether or not action is also required for
Rawls l999b, p. 72, n. 23, hints after intellectualism in writing: "A man who rejects the conditions
imposed on a considered judgment of a competent person could not longer profess to judge at all." See also
Rawls 1999c.
further reasons of practical rationality. (One will not e.g. be required to act or choose in
judging a situation one could not possibly occupy.)
Within a process of consideration, reasoning is psychologically guided by nonns,
understood as imperatives of thought.
Imperatives of the form, "Consider this!"
"Ignore that!" "Conclude thus and so!" "Reconsider!" tell us such things as: which
of the innumerable &cts of the situation to attend to as presenting "relevant" con-
siderations, and which to simply ignore; what more general principles or conclusions to
take as potentially explaining the various relevant features; which possibilities to
entertain or disregard; how one is to weight various features; how long one is to
deliberate; and so on. Being so guided may involve consciously following a general
procedure, but the guiding role of nonns is normally transparent to us. We normally
just think about what is to be done, given the situation.
The same guiding nonns provide a basis for evaluating a process of reflection as
defective, in light ofhow one attends to the situation being judged. Depending on the
case, a person's process of consideration may have underestimated the force or weight
of certain factors, given weaker considerations priority over stronger ones, counted
&cts in favor of a response they count against, ignored relevant features of a situation,
overlooked a crucial distinction, drawn distinctions arbitrarily, been distracted by or
fixated on irrelevant details, been insincere, ill motivated, biased, unimaginative,
unsympathetic, woodenheaded, and so on.
This suggests an initial, very general constitution thesis:
The activity of reasoned practical judgment, as undertaken by someone on a particular
occasion, is constitutively open, by its very nature, to assessment as well or badly done.
The idea that reasoning can be "correct" or "mistaken," "valid" or "invalid," "excel-
lent" or "defective," as according to some norms of reasoning, simply comes along
with the kind of activity that reasoned practical judgment is. Were one to instead
proceed at whim, posing questions or making assertions without accountability to
some (perhaps implicit) expectations about how to proceed, one would not be engaged
in reasoned practical judgment, but doing something else-perhaps deference to au-
thority or just make-believe. Reasoned practical judgment is a norm-structured activi-
ty by its very nature.
If that is to say that reasoning assumes some standards of appraisal or other, it is not
yet to say why particular norms or expectations of judgment should themselves count as
valid or invalid in a process of thought. To answer, a further, more specific kind of
constitution thesis posits specific conditions of participation in an activity, such as the
"ground rules" of chess:
77 Because practical reasoning is an activity. it cannot be represented as a series of propo!itions, according
to the traditional "practical syllogism." See Ham1an 191!6.
• Por a similar list, !lee ibid, p. 7. For further d!!I<:U,sion, ,ee James 2007.
For a relevant kind of activity, to violate certain specific requirements is to thereby
not be engaged in the activity at all.
To violate the ground rules of chess is not to play chess. Similarly, to violate certain
requirements of reasoning (perhaps certain "formal constraints," such as consistency),
means that one is not engaged in the activity of reasoning, but doing nothing or
something else.
It is commonly objected that such "constitutional" requirements cannot function as
norms. A failure of performance disqualifies one from participation in the activity and so
cannot also amount to defective engagement. If you are not playing chess, you are not
playing chess badly, and if your thought is not consistent enough to be reasoning, you
are not reasoning defectively. This problem is easily avoided: we merely say that only
suffidently gross failure of "constitutional norms" disqualifies participation. Provided a
background of sufficient approximate compliance, failing any particular imperative of
thought or action means only that one does not fully understand the activity one is
engaged in, not that one is not at all engaged. As Korsgaard would put it, we have the
"same activity, badly done."
This issue is especially important for Korsgaard, who sees the Categorical and
Hypothetical Imperatives as something like ground rules for action. It is less important
for intellectualism, which is more concerned about the status of specific patterns of
attention and inference in a process of reasoned consideration. A still different kind of
constitution thesis applies to such specific patterns of thought.
Return to the disagreement over environmental protection, and suppose both
parties to the dispute are fully capable of and well placed for making a reasoned
practical judgment. Even so, it may be natural to say that one party is mistaken in his
or her reasoning because of how he or she attends to the situation at hand; he or she
might not be considering the environment aside from its use value, perhaps showing
lack of imagination or an unquestioned belief that all value is use value. That, we may
say, amounts to competence failure, not in the sense that the party is not engaged in
reasoning at all, but in the sense that he or she fails to fully understand the activity of
reasoned practical judgment in the case at hand. Much as a chess master or poker pro
might observe an amateur compete and conclude that he knows how to play but does
not fully understand the game, we might say of our mistaken disputant that he or she
does not fully understand the activity he or she is engaged in.
Simply by virtue of
overlooking a certain distinction or possibility, or being fixated on certain irrelevant
details, the mistaken disputant-but perhaps not the party in the right-fails to fully
understand their common activity, of practical reasoning.
Our intellectualist version of the Constitutional Conception generalizes from that
idea. It starts from the idea that fully understanding what is involved in making a
.. Kongaard 2009, Chapter 2.
Burge 2005, appeals to the notion of full undentanding in another a priori context.
reasoned practical judgment, in general or in the specific case at hand, means knowing
how to think and reason in various ways. As in the case just suggested, if one doesn't
respond in those ways in reasoning, one doesn't fully understand the distinctive kind of
activity that reasoned practical judgment is. But that, according to the Constitutional
Conception, is all validity in reasoning is: fully competent performance, in the activity
of reasoned practical judgment, is nothing more than meeting the range of conditions,
whatever they are, which are required for one to fully understand that activity. When
one's aim is to make a reasoned practical judgment, validity in reasoning, in one's
process of consideration, just is one and the same thing as the process of thought one
must carry out in order to fully understand the distinctive activity one is engaged in, the
distinctive activity that reasoned practical judgment is.
Now, without offering further clarification and defense of this thesis, the crucial
point for our purposes is that, if correct, it would explain the deep mistake. The fact of
constitution would provide a general, external rationale why further questions about the
general nature or fact of validity are in a deep way confused. Further questions of
validity could only be internal questions.
To elaborate: we can regard the Constitutional Conception as a thesis external to
practical reason, but within philosophical anthropology, in the following sense. It is (i)
about the relevant domain (validity in reasoning) in general, and (ii) neutral as regards
its substantive content (what reasoning is valid). (An "internal" claim, by contrast,
would be either not general (e.g. "Promise breaking is wrong") or general but not
neutral ("All right actions maximize non-moral good.").) But the external thesis is that
validity in reasoning really is nothing more than competent performance within a certain
framework of thought. Pulling a leaf from Camap's philosophy of mathematics, then,
we can treat this very relation of constitution itself as a rationale why there is no further
apt external question of what validity is, really, in a framework-independent sense.
Questions of validity will be apt only as internal questions, if not entirely for Carnap's
reasons, then for reasons flowing from the kind of constitution relation involved. On
our intellectualist account, a judgment of validity is an exercise in understanding,
where understanding, we may say, is invariably first personal and so not a purely
affair. That is, for broadly Davidsonian reasons, any classification of a
requirement as necessary for someone to fully or partly understand a type of activity
cannot but rely on one's understanding of the activity in question, including one's best
understanding of what good practice of the relevant kind involves.
So, for example, in chess there is no way to classify a rule as a ground rule without
consulting those who play and, ultimately, relying on one's own understanding of the
game. Without a first-personal grasp or sense of the game, an anthropologist would be
at a loss for what people are doing, beyond playing some game or other. The case of
practical reasoning is similar. We can only make a judgment about what reasoning is
•• Yablo 1\198. pp. 234-5, discusses a similar Camapian thought, founded on an idea of
valid by drawing on our own best understanding of what the activity involves. We can
only answer by carrying out practical reasoning in some range of cases, explicating what
we come up with, and subjecting that explication to further critical scrutiny, based on
our understanding of the activity of practical reasoning itsel£
5. The intelligibility argument
Of course, if validity in reasoning is fixed by practice, the natural question is why
constructivism is not ultimately a form of conventionalism. For all we have said so far,
what counts as an objective value just depends on a conventionally established
reasoning framework, of which there may be many and indeed inconsistent forms.
We might ourselves have used a radically different reasoning framework, in which case
the ethical truth would be radically different than it is. But constructivism is then just a
kind of "framework-relativism"-unless, that is, it can explain why we would then
have been in the wrong, and how we in fact get things right. Here, as always, Platonic
Norms of Reasoning loom.
A constructivism seeking objectivity will indeed reject conventionalism about
validity in reasoning, for the same reasons it rejects conventionalism about straight
ethical truth. If we can critically evaluate any conventional practice in behavior, as
suggested above, then we can equally do so in conventional practice in reasoning.
Indeed, any truly reasoned practical judgment would be open to such evaluation in both
cases. As above, this undercuts variance with subjective conditions, which is inconsis-
tent with the objectivity of values. Does it assume Platonism? Not given the Protagor-
ean Constraint: norms cannot then count as valid, constitutive norms of practical
reasoning if they radically diverge from, and outstrip, the patterns of thought found
in human reasoning practice. The putative possibility of radical divergence is in effect
blocked. But there need not be an independent fact of validity that makes the radical
alternative framework wrong. The point is precisely that such a framework could not
constitute the activity of practical reasoning in the first place.
In defense of this position, constructivists may offer a final kind of constitutive
argument, an Intelligibility Argument. Imagine that we have discovered creatures who,
despite lifelike activity, exhibit patterns of thought radically different than any we
could recognize as valid reasoning. According to the Intelligibility Argument, such
beings would not intelligibly count as engaged in the activity of reasoned practical
judgment. We do not then have to explain how the patterns of thought displayed
somehow count as invalid reasoning. For they do not qualify as engaged in practical
reasoning to begin with. The same will be true even ofbeings which can intelligibly be
said to have something like a mental process which issues in decision, perhaps along
with some rudiments of practical rationality. If they do not sufficiendy approximate
I hope to elaborate more fully elsewhere. For related discussion of "constructive interpretation," see
James 2005.
forms of thought and judgment that we could recognize as valid, we cannot intelligibly
count them as beings that engage in reasoned practical judgment. They could only be
doing nothing, or something else.
Constitutive failures also arise in more ordinary cases. Consider someone who
decides it is better to be a lawyer than a doctor for no other reason than that she had
a ham and cheese sandwich for lunch. Or consider Parfit' s case of someone who judges
that some painful experience will not be bad simply because it will occur on a Tuesday.
These strange ways of attending to a situation are not bad practical reasoning; they are
not practical reasoning at all--at least not unless we imagine sufficient compliance in
the background.
Other cases are still more ordinary. In diagnosing an ethical disagreement (e.g.
over environmental protection), it is often f.Ur to say that one party has not really
thought much about the issue, or that he or she has thought about it in the wrong
way; the view reached may be a sort of reflex, not really a reasoned practical
judgment. Indeed, someone who at first seems to be taking the question seriously
may be rationalizing a prior view which is not open to question, with no intention
of reasoning anew about any true merits of environment protection with an open
mind. Perhaps he or she sees no possibility of questioning social policies of environ-
mental exploitation which serve the political party in power. Indeed, the same may
be true of someone sincerely seeking to affirm a judgment. The person may not
seem engaged in reasoned judgment if it becomes clear that he or she is really just
sincerely parroting the dictates of an ideology, religious authority, trusted pundit,
political party, or tradition. Despite what may be good intentions, he or she ends up
doing something else.
If the Intelligibility Argument addresses radical divergence, are not less radical
alternative frameworks equally a problem? If some frameworks within the bounds of
intelligibility are to count as invalid, or not equally valid, what is supposed to explain
this? Here the Intelligibility Argument has the limited role of sorting such cases in
either of two ways. On the one hand, in many cases it will tum out, upon closer
scrutiny, that there is insufficient compliance with what we recognize as practical
reasoning. The imagined framework can then be disqualified as unintelligible, as in
cases of radical divergence. On the other hand, there may be cases in which the
imagined framework cannot be wholly disqualified. Then, however, the argument
will shift to internal argument, and make either of two claims. Initially dissimilar
patterns might be internally validated as based in deeper agreement with familiar
reasoning norms. Or the imagined reasoning might be internally invalidated, based
on internal claims about what full understanding of practical reasoning involves.
Constructivism thus also rejects what might be called "protagorean relativism," the
view that ethical truth is relative to human reasoning practice, whatever it happens to
" Stroud 1965 makes a vell>ion of th1s point as regards l o g ~ c a l necessity.
be like. Constructivism is not the view that the actual ethical truths are "true for" us,
given actual human reasoning practice, while other (perhaps inconsistent) propositions
would have been "true for" a different group of human beings, with a different
reasoning practice. All such cases of possible divergence in framework are instead to
be sorted in either of the two ways outlined above. We either cannot make out the
imagined form oflife as including the activity of reasoned practical judgment, in which
case there is nothing to relativize truth to, or we appraise the imagined framework in
wholly internal terms, as valid or invalid, according to our understanding of what
reasoned practical judgment involves.
The main role of the Protagorean Constraint, as motivated by the Intelligibility
Argument, is to support our way of explaining the autonomy of the ethical from the
outside. Given the Constitutional Conception, so long as we have practical reasoning
intelligibly on hand, the only remaining questions of validity are internal questions, for
that very activity. And if this is correct, we can say what the deep mistake is in
supposing that objective values must be, or even can be, located within the wholly
impersonal world: it is to miss the deep constitutive relation between validity in
reasoning and human reasoning practice.
If this is right, it follows that vindication of the objectivity of values (to the extent
vindication is required at all) cannot depend on the plausibility of situating objective
values within the impersonal world. Even so, objectivity (and perhaps its vindica-
tion) will arguably depend on interpersonal reality. To engage in practical reasoning
about what is right or prudent or to be done in some situation is, by very nature of
the activity undertaken, to speak for humanity about the case at hand. One in effect
aims to say "what anyone would say." Correspondingly, one incurs a responsibility
of due deference in the face of disagreement--to count reasoned disagreement as
pro tanto grounds that one has oneself fallen into at least partial error in one's own
reasoning, or, if one retains one's own view, to fairly diagnose the error in reasoning
made by those who disagree. In this way constructivism fiirts with error theory: the
truth of any ethical proposition depends on the possibility of explaining the spread
of ethical disagreement we actually fmd, as manifesting one or another form of
unreason. If that position cannot be plausibly maintained, there is no truth of the
matter in the proposition at hand.
So although Rationalist Anti-skepticism claims
that reason is on the side of specific conclusions, a thorough defense of this position,
across a range of ethical cases, must plausibly take the reality of disagreement into
.. See e.g. Joyce 2002.
This is one grand vindicatory aim of Rawlsian political philosophy. The task is harder or easier
depending on how much we should trust our own judgment in the face of disagreement. See Fddman
and Warfield 2010.
6. Further concerns
We close by considering--all too briefly-several further questions or objections.
The Authority of Reason. Constructivists start from practical reason. So it is fair to ask
why practical reason should be the kind of thing which could explain objective truth-
why, as one might put it, it should have a special kind of authority. I am not likely to pay
fifty dollars to someone who says that norms of shmeasoning tell us to conclude that
I should pay up. Why are norms of practical reasoning any more authoritative about
how I should spend my money?
Our general if familiar answer is: Reason is the only game in town. We do accord
reason authority. It is insane say to say that so many coin tosses, as such, could be
authoritative about whether or not I should take a certain job. But ifl believe, having
thought about the matter a little, that I should tum down a certain job, and later change
my mind after giving the matter careful attention, I will rightly trust my reasoned
judgment more. Moreover, it is very plausible to hold that, insofar as other forms of
authority-law, practice, tradition-are indeed authoritative, they gain this status by
having a relation to reason of the right kind.
In that case, the question is simply
whether "reason" can be characterized in a way that suits constructivist explanation.
I submit that our intellectualist account does the job.
J.Vhy not voluntarism? But would it not be more plausible to locate the source of
authority within a person's own agency, as according to Korsgaard? Voluntarism is
appealing for this reason. But it comes at the high cost of failing to capture the
objectivity of values, much as with Williams-style subjectivism. Whether in a Kantian
or Humean form,
as long as practical reasoning is governed only by formal norms of
choice which have no substantive implications, an agent's reasons will vary with
whatever subjective conditions (practical identities, intentions, motives, reason judg-
ments, according to the version in question) do generate the reasons an agent has. So,
on Humean constructivism, if Caligula could not come to see reason not to kill at
whim, within the relevant elements ofhis practical standpoint, he has no reason not to.
And even ifKantian voluntarism can capture the objectivity ofbasic moral and perhaps
prudential reasons, because formal Kantian standards do bar Caligula from killing at
whim, the same problem arises where formal constraints have no say-reasons to value
friendship, higher rather than lower pleasures, environmental conservation, and so
forth. To the extent we commonly do assign these and other ethical values objectivity,
voluntarism seems committed to subjectivism or error theory. It gives up the quest for
full objectivity.
•• As Pufendorfexplained, even views of religious authority which urge that "fallen" human judgment is
not to be trusted are still subject to our own internal sense of what is reasonable and right. Tite grand question
of "faith venus reason" is at most a que5tion of degree.
" For a Humean version, ~ e e Sharon Street, this volume.
Normativity? Voluntarism might well be necessary for certain requirements of mo-
rality, prudence, or rationality, depending on what their "normativity" requires. If so,
our intellectualist account will require supplementation. I take no position on the
matter here.
The possibility of agnosticism suffices to show that a constructivist
account of protagorean objectivity need not give normativity center stage.
The normativity of reasoning? Still, must not the norms that guide reasoning have some
hook, as it were, in the agent doing the reasoning? For voluntarism, the trigger for their
application to the agent is his or her making a choice. For the intellectualist, the trigger is
engagement in judgment. What would be defective thought or attention, in a process of
reasoned practical judgment about a situation S, would not necessarily have been
defective if one were instead merely entertaining a hypothesis about S, or making a
judgment about Son the fly, taking some judgment about Son someone else's authority,
or engaging in make-believe. An applicable norm only applies to a given agent because a
reasoned practical judgment about S is being made.
VVhy care? But if applicable norms are just requirements of full understanding, why
care what they require? Korsgaard threatens such a skeptic with a kind of death: if you
go far enough against the internal norms of action, your very agency, and perhaps very
person, will come apart; there will be no "you" there to speak o£ The intellectualist
sees lower stakes, but can answer as follows. Like all judgment, ethical judgment
constitutively aims at truth. But aiming at truth requires concern for what generally
determines what is true. So if ethical truth is determined by the (idealized) output of
practical reasoning, ethical judgment must take account of it.
VVhy reason? Korsgaard's voluntarism supplies a reason why one has to make certain
moral or other judgments in the first place; that, too, is a condition of unified agency.
But if the act of judgment itself triggers the application of norms of reasoning, as
according to intellectualism, one can avoid those norms simply by steering clear of
judgment. Acceptance on the fly, make-believe, and careful empirical judgment might
help one get by without judgment of what is ethically true. Why then make reasoned
practical judgments at all?
Intellectualism can offer several replies. It can agree with Korsgaard that judgment
about what we ought or have reason to do is practically unavoidable. It can also agree
with Socrates that the unexamined life is not worth living, on substantive grounds.
Constructivism explains how there could be an objective fact of the matter of this kind.
One need only then make the case that good reasoning favors a life of reason over a life
of unreason, as urgent, needed, or worthwhile. And if all else fails, intellectualism can
be supplemented by voluntaristic ideas as the special normativity of a given domain of
ethics requires.
>• Intellectualist views also might be adequate. One explains f.illure of action in the face of a sufl-.cient
reason"sjudgment as a kind of incoherence (Scanlon 2004; Broome 2000).
Is constructivism distinctive tifter all? For all we have said so far, constructivism might still
be treated as a more conventional metaethical view. For instance, it becomes a form of
naturalistic reductionism if the norms of reasoned practical judgment can be fully
Is constructivism then distinctive after all?
I deny that the required enumeration could succeed, if only for reasons familiar from
the problem of rule following.
So constructivism does not collapse in the suggested
way. Yet much the same question of distinctness may arise in a different form. The
rule-following debate involves positions analogous to standard metaethical views. If no
distinctive account of rule following has been marked out, it is fair to ask, yet again,
whether or to what extent constructivism is a distinctive and interesting foundational
One reply is quietism about the rule-following problem. Constructivism is not then
simply a version of familiar metaethical views of rule following. Indeed, it would then
offer what Nadeem Hussain calls an "ambitious metanormative constructivism."
in Hussain's terms, a metanormative theory counts as "ambitious" as long as it so
transcends the metanormative issue in question. This is a natural extension of what we
called default quietism. We assume, as the default position, that a given meta-question
is misguided until we see that it is well motivated and admits of illuminating answers.
Constructivism grants that this burden of justification can be met for certain basic
questions of ethical truth, objectivity, and knowledge, and seeks to provide answers to
them. But it may also happily hold that rule following or other problems are not well
motivated (perhaps because illuminating answers are not forthcoming). With Wittgen-
stein, we may hold that the rule-following "problem" only seems well motivated
when a "conjuring trick" is pulled.
In any case, as Hussein will admit, constructivism offers an interesting philosophical
explanation. And that will be so even if it is not quietistic but merely agnostic about the
rule-following issue. Constructivism is of interest and significance because it provides a
protagorean conception of objective values, even if further questions in the realism/
anti-realism neighborhood are rejected, or simply never taken up.
Blackburn, Simon. 1993. Essays in Quasi-Realism (Oxford: Oxford University Press).
Broome,Jolm. 2000. Nonnative Requirements. Ed. Jonathan Dancy, 2000 (Oxford: Blackwell).
Burge, Tyler. 2005. Truth, Thought, Reason (Oxford: Oxford University Press).
Darwall, Stephen, Allan Gibbard, and Peter Railton. 1992. "Toward Fin de Siecle Ethics," in The
Philosophical Review 101, no. 1,January, pp. 115-89.
Dworkin, Ronald. 1996. "Objectivity and Truth: You'd Better Believe It," in Philosophy and
Public Affairs 25, pp. 87-193.
,. Jackson 1991!.
See McDowell \91!1, 191'14.
Huuain. this volume.
Enoch, David. 2011. "Can There Be a Global, Interesting, Coherent Constructivism About
Practical Reason?" In Philosophical Explorations 12, no. 3, pp. 319-39.
Feldman, Richard and Ted A. Warfield. 2010. Disagreement (New York: Oxford University
Gibbard, Allan. 1990. Wise Choices, Apt Feelings (Cambridge, MA: Harvard University Press).
Gibbard, Allan. 1999. "Morality as Consistency in Living," in Ethics 110 (1), pp. 1 4 ~ .
Gibbard, Allan. 2003. Thinking How to live (Cambridge, MA: Harvard University Press).
Harman, Gilbert. 1986. Change in View (Cambridge, MA: MIT Press).
Hussain, Nadeem. "A Problem for Ambitious Metanormative Constructivism," this volume.
Hussain, Nadeem and Nishi Shah. 2006. "Misunderstanding Metaethics: Korsgaard's Rejection
of Realism," in Oxford Studies in Meta-Ethics, ed. Russ Shafer-Landau, 2006 (Oxford: Clar-
endon Press), pp. 265-94.
Jackson, Frank. 1998. From Metaphysics to Ethics: A Difense of Conceptual Analysis (Oxford: Oxford
University Press).
James, Aaron. 2005. "Constructing Justice for Existing Practice: Rawls and the Status Quo," in
Philosophy & Public Affairs 33 no. 3, pp. 281-316.
James, Aaron. 2006. The Objectivity of Values: Invariance without Explanation," in Southern
Journal of Philosophy 44, no. 4, pp. 581-wS.
James, Aaron. 2007. "Constructivism About Practical Reasons," in Philosophy and Phenomenologi-
cal Research 74, no. 2, pp. 302-25.
Johnston, Mark. 1989. "Dispositional Theories ofValue," in Proceedings of the Aristotelian Society,
Supplementary Volume 63, pp .113-3 7.
Joyce, Richard. 2002. The Myth of Morality (Cambridge: Cambridge University Press).
Korsgaard, Christine. 1996. The Sources ofNormativity (Cambridge: Cambridge University Press).
Korsgaard, Christine. 1996a. Creating the Kingdom of Ends (Cambridge: Cambridge University
Korsgaard, Christine. 1996b. "Skepticism About Practical Reason," in Creating the [(jngdom of
Ends, 1996a (Cambridge: Cambridge University Press).
Korsgaard, Christine. 1997. "The Normativity oflnstrumental Reason," in Ethics and Practical
Reason, ed. Garrett Cullity and Berys Gaut, 1997 (New York: Oxford University Press).
Korsgaard, Christine. 2003. "Realism and Constructivism in Twentieth-Century Moral Philos-
ophy," in The journal of Philosophical Research, APA Centennial Supplement, pp. 99-102.
Korsgaard, Christine. 2009. Self-Constitution: Agency, Identity, and Integrity (Oxford: Oxford
University Press).
Lenman, James. "Expressivism and Constructivism," this volume.
McDowell, John. 1981. "Non-Cognitivism and Rule-Following," in Wittgenstein: to Follow a
Rule, ed. S. Holtzman and Christopher M. Leich, 1981 (London: Roudedge).
McDowell, John. 1984. "Wittgenstein on Following a Rule," in Synthese 58, pp. 325-63.
McDowell, John. 1985. "Values and Secondary Qualities," in Morality and Objectivity: A Tribute
to]. L. Mackie, ed. Honderich, 1985 (London: Roudedge).
Nagel, Thomas. 1997. The Last Word (Oxford: Oxford University Press).
Parfit, Derek. 2011. On What Matters (Oxford: Oxford University Press).
Rawls, John. 1996. Political liberalism (New York: Columbia University Press).
Rawls, John. 1999a. "The Independence of Moral Theory," in]ohn Rawls: Collected Papers, ed.
Samuel Freeman, 1999 (Cambridge, MA: Harvard University Press).
Rawls, John. 1999b. "Justice as Fairness," in john Rawls: Collected Papers, ed. by Samuel Freeman,
1999 (Cambridge, MA: Harvard University Press).
Rawls, John. 1999c. "Outline of a Decision Procedure for Ethics," in john Rawls: Collected Papers,
ed. Samuel Freeman, 1999 (Cambridge, MA: Harvard University Press).
Rawls, John. 1999d. "Kantian Constructivism in Moral Theory," in john Rawls: Collected Papers,
ed. Samuel Freeman, 1999 (Cambridge, MA: Harvard University Press).
Raz, Joseph. 2003. The Practice of Value, ed. R. Jay Wallace (Oxford: Clarendon Press).
Scanlon, T. M. 1998. What We Owe to Each Other {Cambridge, MA: Harvard University Press).
Scanlon, T. M. 2004. "Reasons: A Puzzling Duality?" in Reasons and Values: Themes from the
Work of joseph Raz, ed. Michael Smith, Philip Pettit, Samuel Scheffier, and R. Jay Wallace,
2004 (Oxford: Oxford University Press).
Scanlon, T. M. "The Appeal and Limits of Constructivism," this volume.
Schneewind,J. B. 1998. The Invention of Autonomy (New York: Cambridge University Press).
Shafer-Landau, Russ. 2003. Moral Realism: A Defense (New York: Oxford University Press).
Street, Sharon. "Coming to Terms with Contingency: Humean Constructivism about Practical
Reason," this volume.
Stroud, Barry. 1965. "Wittgenstein and Logical Necessity," in Philosophical Review 74, no. 4,
pp. 504-18.
Wiggins, David. 1987. "A Sensible Subjectivism," in Needs, Values, and Truth (Cambridge:
Blackwell Press).
Williams, Bernard. 1981. "Internal and External Reasons," in Moral Luck (Cambridge:
Cambridge University Press).
Williams, Bernard. 1995. "Internal Reasons and the Obscurity of Blame," in Making Sense of
Humanity (Cambridge: Cambridge University Press).
Wright, Crispin. 1992. Truth and Objectivity (Cambridge: Harvard University Press).
Yablo, Stephen. 1998. "Does Ontology Rest on a Mistake?" in Proceedings of the Aristotelian
Society, supplementary volume LXXII, pp. 2 2 ~ 1 .
Constructivism, Agency, and the
Problem of Alignment
Michael E. Bratman
According to Bernard Williams, a characteristic context for my judgment that you have
a normative reason to act is the context of "advice in the 'if I were you ... ' mode."
The idea, I take it, is that normative reasons for an agent are considerations that have
force from the "perspective" of the agent herself.
And this idea is built into important
versions of constructivism that are offered as a general account of normative practical
reasons. But what is the perspective of the agent? One lesson of work in the philosophy
of action is that there are hard questions here, including questions about alienation,
identification, depth, and commitment. The use by a constructivist of the idea of the
agent's perspective needs to be responsive to these complexities in the theory of
But it also needs to be responsive to the pressures that derive from the general
ambitions of such a constructivism. This leads to the question of whether, and if so
how, a constructivist theory can meet both kinds of demands-whether the pressures
from the general constructivism will align with the pressures from the theory of agency.
Call this the problem of alignment. The problem of alignment is a problem at the
intersection of philosophy of action and systematic reflection on normative practical
reasons;• and my hope is that it can be an occasion for increased understanding in both
areas. Toward that end, my plan here is to think in a preliminary way about this
problem by focusing on a specific essay of Sharon Street's: "Constructivism about
(Williams 1995, p. 36) This passage is emphasized by T. M. Scanlon in his "Appendix: Williams on
Internal and External Reasons" in Scanlon 1998, p. 372.
Talk of the agent's "perspective" is from Williams's next sentence. This is offered by Williams, in its
official version, as a necessary, but perhaps not a sufficient, condition for a normative reason. See Williams
1995, pp. 35-6. Williams also famously appeals to a connection between normative reasons and the potential
motivational explanation of action. I see the cited connection between normative reasons and what has force
from the agent's perspective or standpoint as a related but deeper consrraint on normative reasons.
' Harry Frankfurt connects these questions about alienation with issues about the nature of reasons for
action in Frankfurt 2006, pp. 8-13.
' Nadeem Hussain discusses related issues in Hussain 1999, pp. 151-90.
I do this both because this essay is clear in a way that helps advance the
discussion, and because this essay at least implicidy involves many of the ideas that are
my intended focus.
Street distinguishes between two kinds of constructivist views. The first she calls a
"restricted" constructivist view, versions of which she fmds in Rawls's political con-
structivism and Scanlon's contractualism (210). Such a view:
specifies some particular, restricted set ofjudgments about reasons, and says that the correctness of
a judgment about reasons falling within that set is constituted by the judgment's withstanding a
certain (specified) procedure of scrutiny from the standpoint of some (specified) set of further
judgments about reasons. (209-1 0)
Street here talks about reasons quite generally, but her focus is specifically on normative
practical reasons; and that will be my focus as well, though to save words I will
sometimes just talk about reasons. Such a constructivism about practical reasons in
the restricted set takes as given the correctness of the further set of judgments about
reasons from which the restricted set is to derive-to be, in this sense, constructed-by
way of "scrutiny." The purported correctness of these further, grounding judgments
about reasons is not seen as itself explained by way of such a construction.
The relevant procedure of scrutiny is a kind of function from some judgments about
reasons to other judgments about reasons. (Street also calls these judgments about
reasons "normative judgments," and I will follow her in this regard.) Street labels the
latter judgments the "target set" and the former judgments the "grounding set."
Street contrasts such a restricted constructivism with what we can call a non-restricted
constructivism. A non-restricted constructivist view sees the correctness of all judg-
ments about practical reasons as a matter of their construction from "a standpoint
constituted by some further set of normative judgments" (220). Because such a non-
restricted constructivism purports to provide, quite generally, "truth conditions of
judgments about practical reasons" (239), Street labels it "metaethical constructivism."
Street goes on to offer a specific formulation of such a non-restricted constructivism:
According to metoethica/ constructivism, the fact that X is a reason to Y for agent A is constituted by
the fact that the judgment that X is a reason to Y (for A) withstands scrutiny from the standpoint
of A's other judgments about reasons. (223)
This "metaethical constructivism" is a thesis about what constitutes facts about norma-
tive reasons: such normative facts are constituted by the fact that the corresponding
judgment stands in an appropriate relation to other relevant judgments of the agent's.
This thesis retains the idea, central to restricted constructivist theories, of grounding
' Street 2008. Parenthetical page references in the text are to this essay.
judgments about reasons that are in a target set in judgments about reasons that are in a
grounding set. But now the target set potentially includes any judgment about reasons
that is up for consideration. Given any such judgment, the grounding set "is the set of
all of the relevant agent's normative judgments, minus the normative judgment whose
correctness is in question" (226).
The idea of a constructivism that applies to all
judgments of reasons, and appeals in each case to scrutiny from other judgments
about reasons, involves in this way a Neurath's-boat structure: each judgment of
reasons,], is to be derived, by way of scrutiny, from a background, B, of judgments
about reasons that do not already include]. But given any particular judgment in B-
eall itj*-there is to be a similar construction in its support that draws on a background
that may include both] and many elements in B but does not include]*.
Street's view adds to this Neurath's-boat structure a special role for nonnative
judgments that are judgments of, in particular, the agent. On Street's view, that
there is a reason for agent A to act in a certain way is constituted by a construction
from a background that is characterized by two features:
(a) it is a background that is provided by the standpoint of the agent A; and
(b) this standpoint is constituted by, in particular, A's nonnative judgments.
I want to reflect on this appeal to (a) and (b).
Begin by stepping back a bit. Street's proposal involves a general idea, and a view
about how further to specify this idea. The general idea can be expressed roughly as
General idea: the correctness of a claim that X is a reason to Y for agent A depends on
facts about (1) a relevant, psychologically real standpoint, together with (2) the
outputs of "scrutiny" from that standpoint.
This general idea is motivated by Street's background conjecture that the existence of
practical reasons depends on the contingent existence of creatures whose psychic
economies involve thoughts about reasons, thoughts that help shape what the creatures
do (22o-3). And the hope is that by developing this background conjecture we can
carve a path between a realism that sees reasons as existing independendy of our
practical thinking, and skepticism about reasons.
As Street notes, this general idea can be developed in different ways depending on
whose standpoint is privileged by the account. Is what is basic the standpoint of the
agent? Or is what is basic the standpoint of the person making the judgment that X is a
reason for that agent? Street argues for the former. In her development of the general
idea she appeals not to the attitudes of the person making a judgment about reasons but,
There are puzzles about how to interpret the phrase "minus the normative judgment whose correctness
is in question." For example, if the judgment at issue is a judgment that X is a reason to Y for A, what should
we say about the judgment that if grass is white X is a reason to Y for A? or about the judgment that either X is
a reason to Yfor A or Cis a reason to D for A? I put t h ~ matters aside here.
rather, to relevant attitudes of the agent for whom the cited consideration is purported
to be a reason for action. That is, in developing the general idea, Street further specifies
(1) (in the general idea) along the lines of:
la. The relevant standpoint in (1) is the standpoint of the agent, A.
This is where Street's appeal to (a), noted earlier, appears in the theory.
Let me register a concern at this point. Even given the general idea, I would want also
to retain some form of the idea that when we judge that you have a reason, R, we
ourselves are in some way endorsing the significance of R. This is why, to use an
example from Allan Gibbard, we find it strained to suppose that given Caligula's
"horrendous" standpoint of judgments about reasons, he has normative, justifying
reasons to inflict suffering on many.
In giving the standpoint of the agent an exclusive
role in her non-restricted constructivism, Street's theory seems to be at least potentially
in conflict with this plausible reticence.
Street's explicit reasons in this essay for focusing exclusively on the agent's stand-
point do not seem sufficiently forceful to me. She says that "it accords much better
with our overall usage" (224). In reply, it does seem to me plausible that our ordinary
thinking about reasons for action supports Gibbard's denial that Caligula has a norma-
tive reason to torture even ifCaligula's standpoint wholeheartedly supports torture and
does not involve relevant and false non-normative belie( Street also says that "the
function of normative judgment is to get us to respond to our circumstances in ways
that are adaptive" (230), and she seems to think this argues for the exclusive focus on
the standpoint of the agent. But it does seem plausible to me that among the functions
of normative judgment are also the kinds of social coordination of thought, feeling, and
action that are highlighted by expressivists like Gibbard.
That said, I want to grant that there is something attractive about la insofar as it is
motivated by the Williams-friendly thought, noted earlier, that reasons for an agent are
considerations that have force from the agent's own perspective. To keep my discus-
sion manageable, then, I will focus here not on whether this exclusive focus on the
agent's standpoint is justified, but on how we are to understand the agent's standpoint
given that it plays, at the least, an essential role in normative reasons for action.
now we need to note that this Williams-friendly rationale for the move to la has an
important implication. When Williams considered what "the 'if I were you ... 'mode"
of thought appeals to, he settled for a broad idea of the agent's "motivational set." As he
said, this set can include "dispositions of evaluation, patterns of emotional reaction,
personal loyalties, and various projects ... embodying commitments of the agent."
Gibbard 1999, p. 145. For Gibbard's expressivist account of judgments about reasons see Gibbard 1990,
p. 163.
• I am, then, putting aside for another occasion the question of how we might coherendy bring together
the insights of Williams with those of an expressivist like Gibbard.
• Williams 1981, p. 105.
But there is philosophical pressure here for fmer-grained discriminations. The agent
may be in various ways alienated from certain elements in her motivational set; and
some of these elements may more deeply help constitute where that agent stands than
do others, and in that sense more deeply speak for the agent.
The thought that reasons
must have force from the perspective of the agent points to the need for a more
articulated understanding of what precisely constitutes that perspective. So if the cited
Williams-friendly thought is a basis for the move to la, we will want to see the
elements in the agent's psychology to which the non-restricted constructivism appeals
as the elements that constitute where-to return to Frankfurt's phrase-the agent
herself stands. This is a version of the problem of alignment.
I think Street would agree that there is a need for such alignment.
Consider her
insistence that "an agent A is, in an important sense, to be identified with her most strongly
and centrally held values" (235, n. 45). In talking of"values" here Street means to refer to
the agent's judgments of reasons. So this remark involves the idea in (b), an idea to be
challenged below. But here I just want to emphasize why, as it seems to me, this
identification matters to Street. It matters, I think, because her constructivism aims to
capture reasons that have force from, in particular, the perspective of the agent. So she
needs to be sure that what is at the bottom of her constructions of reasons is, indeed, the
perspective of the agent, and not merely some wiggles in the agent's psychic stew. And if
this is right then a defense of a non-restricted constructivist view along the lines proposed
by Street will also need an explanation of which attitudes constitute where the agent stands.
Now, in talking about an agent's standpoint we need to avoid a picture of a little
person inside-a homunculus-who steps back, reflects, and adjudicates. What we
want, rather, is a non-homuncular model of the agent's psychic economy such that we
can plausibly argue that certain elements in that economy constitute where the agent
stands. And Street supposes that this standpoint is constituted by, in particular, the
agent's judgments about reasons. This is to develop the general idea and la by adding the
yet further thought that:
1 b. The standpoint of the agent is constituted by that agent's normative judgments
about practical reasons.
What are these normative judgments, anyway? Street distinguishes them from
desires and from beliefS (23Q-1). Normative judgments "are by their nature motivat-
ing," in contrast, as Street sees it, with belie£ And, on Street's view, normative
judgments have certain constitutive features-features to which I will turn below-
that distinguish them from ordinary desires. What about other attitudes that also
See Harry Frankfurt's reflections on "where (if anywhere) the person himself stands" in Frankfurt
1988c, p. 166.
In addition to the passage to be noted in the main text, see also her appeal to issues about which
"standpoint is most deeply" the agent's (235), and her remark that the features of the agent's psychology that
are central to her constructivism shape "who we are" (245).
diverge in important ways from ordinary beliefs and desires-for example, intention or
certain emotions? Street's view, I take it, is that such attitudes help constitute the
agent's standpoint only insofar as they involve judgments about reasons.
Let's locate this view more explicidy within the debate about agency to which
I have already alluded. A classic moment in this debate occurred in the early 1970s in an
exchange between Harry Frankfurt and Gary Watson.
In the terms of our present
discussion, Frankfurt's proposal was that the agent's standpoint is constituted by certain
higher-order desires, whereas Watson's proposal was that it is instead constituted by the
agent's evaluative judgments. The details of these views shifted in important ways over
the years. Watson came to talk about valuing rather than value judgment-though
(with some qualifications) he saw valuing as tied to value judgment. Frankfurt came to
talk about what the agent cares about and loves.
But an important contrast remained.
This is the contrast between a view like Watson's that understands the agent's
standpoint in terms of attitudes that essentially involve judgments about values, and a
view like Frankfurt's that sees as fundamental certain conative commitments that are
not, at bottom, judgments about values (though they can affect one's reasons, and
thereby one's judgments about reasons). And on a natural reading, Street is opting for a
Watsonian view, though without the value realism that is also a part ofWatson's overall
view. This is a kind of Platonic psychology without the Platonic value realism.
Frankfurt's original concern was with acting of one's own free will. And Watson's
initial concern was specifically with free agency. So I think it is best to think of the
present issue about the agent's standpoint not as an issue about agency in general but,
rather, as an issue about a strong form of agency. Street does see the structures of
valuing that are central to her story as essential to agency itself (238). But we can
interpret this as a view of what is essential to a specific kind of agency, namely: one that
involves agents who have normative reasons to act.
And Street briefly indicates that she thinks that nonnative judgments have certain constitutive relations
with certain emotions (242, n. 57).
Frankfurt 1988b, Watson 1975. I discuss these essays further in Bratrnan 2007c.
Watson 1987, p. 150, Frankfurt 2006.
Street's Watsonian view about strong agency also shapes her approach to practical conflict. If we are
going to appeal to the standpoint described in la and 1 b to say what reasons A has for action, we are going to
need to say !IOIIlething about cases in which the agent's standpoint includes judgments that conflict, at least
with respect to a particular case. Perhaps A judges that there is reason for her to pursue philosophy, and also
judges that there is reason for her to pursue financial independence. Given certain non-nonnative facts, these
normative judgments support different judgments concerning a reason to go to graduate school in philoso-
phy. Street's resporue is that "the standpoint that determines what reasons (an agent] has is whichever
standpoint is most deeply hers" (234-5). Further, she avers that such depth "is a function of how strongly [the
agent] holds the nonnative judgments in question and how close to the center of her total web of nonnative
judgments they lie" (235).
Thilltrategy for treating conflict of judgments about reasons is an aspect of the idea that the attitudes that
ground reasons on the constructivist view are, as well, attitudes that constitute where the agent stands. And
Street here identifies depth within the psychology of agency with centrality within a web of normative
judgment. Since centrality within a web of judgment is a matter of relations of entailment and support
among those judgments (see p. 235), thi.• appeal to centrality is a kind of intellecrualism about such depth.
Let's now reflect more carefully on the nonnative judgments cited in 1 b. One worry
is that, even granting the coherence of a Neurath-type interdependence in justifica-
tion, appeal to such nonnative judgments at the bottom of the construction threatens a
vicious circularity in the non-restricted constructivist story about reasons. After all, we
seem to be appealing, at the bottom of the construction, to the very idea of a reason in
explaining what constitutes facts about reasons. One idea here might have been to try
to understand the idea of a reason, as it appears in the contents of the judgments in 1 b,
in terms of characteristic roles in practical thinking.
But this is not Street's strategy.
On her view the concept of a reason, as it appears in the contents of the judgments in
1b, is a concept we grasp by way of our experiential "knowledge of what it is like to
have a certain unreflective experience-in particular, the experience of various things
in the world as 'counting in favor of or 'calling for' or 'demanding' certain responses
on our part" (240). So we can say what the content is of the judgments cited in 1b prior
to our constructivist theory of what reasons there are. So there need be no criticizable
circularity. So we can add to the theory:
1c. Our understanding of the concept of a reason, as it is involved in the content of
the judgments in 1b, is given by our "nonnative experience" (240).
It is important here to distinguish two related but different issues. There is the worry
that the appeal to the very idea of a reason in specifying the contents of the judgments in
1b makes the constructivism circular. This is the worry to which 1c is a response. But
there remains a second, fundamental issue. We need to know what it is to judge that there
are certain reasons-in contrast, say, with merely having a thought or experience whose
content involves the concept of a reason. Not just any normative thought or "normative
experience" is a judgment about reasons.
We need to know what else-over and above
the appearance in the content of the idea of a reason-is involved in such a judgment.
Here Street's response is that this further idea of a judgment about reasons is
"supplied ... by our recognition of what is constitutively involved in the attitude of
judging something to be a reason" (241-2). In particular, Street proposes that what is
constitutive of normative judgment-and distinguishes it from ordinary desire even
though both can motivate-includes certain relations among such judgments. One of
Street's main examples involves relations between normative judgments about ends
and normative judgments about means:
[I]t is constitutive of taking oneself to have conclusive reason to Ythat one also, when attending
to the matter in full awareness, take oneself to have reason to take what one recognizes to be the
necessary means to Y. (228)
An idea in the spirit of Gibbard 1990.
Street is alive to this point. For example, in Street 2006, p. 110, Street explicitly contraSts "consciously
or unconsciously beld evaluative judgments, sucb as judgments about what is a reason for wbat" with an
"unreflective ... tendency to experience X as counting favor of or demanding Y."
" "Taking oneself to have" a reason is judging that one has that reason. Seep. 228, n. 37.
Note that Street is not appealing here to a connection between intending ends and
intending known necessary means. Street's appeal here is, rather, to connections
between judgments about reasons for ends and judgments about reasons for means.
And Street offers analogous constitutive claims concerning the relation between
judging X to be a reason to Y and judging X not to be a reason to Y, as well as the
relation between judging that only facts ofkind X are reasons to Y, recognizing that Z
is not of kind X, and judging that Z is not a reason to Y. Street offers these as "purely
formal statements about what is involved in the very attitude of taking something to be
a reason" (229). So we can add:
1 d. The features constitutive of normative judgment include certain formal rela-
tions among normative judgments (including the cited trio of relations).
Now, Street's view involves a kind of relativism: whether or not I have a reason to Y
depends on my own set of normative judgments; and even if I do have a reason to Y it
will not follow directly that you too have such a reason, for your relevant set of
normative judgments may diverge from mine. Might there be certain reasons any
reason-judging agent has? Such "universal" reasons would need to be "'legislated'
from within the standpoint of every creature who takes anything at all to be valuable"
(225). And Street is skeptical that there are such universal reasons since, on her view,
"depending on one's starting set of values one could in principle have a reason for
anything" (244). This is why she calls her view a "formalist"-in contrast with a
"substantive" -constructivism
(244 ).
Return now to the general idea; and consider in particular (2), the appeal to the
outputs of scrutiny from the agent's standpoint. What is scrutiny? Scrutiny is not free
association: scrutiny is a norm-guided mode of thinking. But what are these norms, and
why should the agent be guided by, in particular, those norms? These questions seem
to be asking for reasons to shape one's thinking in accordance with these norms. But
how should a constructivist think about such reasons? One idea is that these reasons for
these scrutiny-constituting modes of thinking are themselves at the bottom of the
constructivist story, and not themselves explainable along constructivist lines. But then
the constructivist proposal fails as a general, non-restricted account of practical reasons.
Street's specification of (2) (in the general idea) aims to block this objection by
drawing on 1 d. Here is what she says:
To decide whether a given judgment withstands scrutiny from the standpoint of A's other
normative judgments, we need not ourselves presuppose any substantive normative judgments;
we need only ask what further normative judgments are constitutively entailed by A's actual
" Street notes (n. 57) that there may well be other features that are constitutive of nonnative judgment.
This is a matter to which we will return.
.. Street, this volume, calls her view "Humean constructivtsm". In contrast, Aaron James builds it right
into the defmition of a constructivism about practical reasons that it supports such universal reasons. See James
2007, p. 303.
nonnative judgments when we take into account the non-nonnative facts as we know them.
A central example is the purported end-means constitutive entailment noted earlier:
roughly, it is constitutive of a judgment about conclusive reason that it entails judg-
ments about reasons for known necessary means. So we are to add:
2a. Scrutiny is a matter of determining "what further normative judgments are
constitutively entailed by [the agent's) actual normative judgments when we
take into account the non-normative facts as we know them."
And the claim is that 2a allows us to understand the relevant kind of scrutiny without
appealing to further, substantive normative judgments as essential to scrutiny in a way
that would undermine the completely general ambitions of Street's non-restricted
A problem here is that even such an apparently uncontroversial principle as the cited
end-means principle can turn out to be difficult to formulate and defend. Mter all,
different philosophers have offered different versions of such an end-means principle.
Joseph Raz, for example, highlights sufficient rather than necessary means. And John
Broome suggests that it is not clear that a reason for an end transmits to a reason for a
necessary means for that end in cases in which one knows one is not going to achieve
that end even if one pursues the cited necessary means.
Indeed, Street herself at times
seems unsure just what the relevant principle is: in formulating it she sometimes appeals
to a "conclusive" reason for the end, sometimes only to a reason for the end (229, 232).
So there is reason to be concerned that the mere appeal to what is "constitutive" of
normative judgment is not going to setde these issues, and that to setde them we will
need to engage in normative reflection; but that might threaten to undermine the idea
that these principles are formal in Street's sense.
For present purposes, however, I will put this concern to one side, though we will need
to continue to focus on related issues about the purported formality of scrutiny. What
I want to focus on now is the structure of the scrutiny to which the theory appeals.
We can think of scrutiny as an inference with psychological inputs and outputs. On
Street's picture the relevant inputs include normative judgments of the agent, taken
together with "the non-normative facts as we know them"; and the relevant outputs
are other normative judgments of the agent. And this picture of the judgment-infused
inputs and outputs of scrutiny seems deeply embedded in Street's non-restricted
Consider first the idea that the outputs of relevant scrutiny are, in particular, the
agent's normative judgments. This is an aspect of the parallel, highlighted by Street,
•• Raz 2005, p. 6 and Broome 2007, p. 176 (though Broome's remarks are explicitly aimed at the
somewhat different question of whether you ought to perform the necessary means). For an examination
of these issues see Kolodny unpublished. My own approach is sketched in Bratman 2009.
between restricted and non-restricted versions of constructivism--an aspect of what
Street sees as "the central distinguishing feature of all constructivist views in ethics"
(208). In each case, according to Street, the correctness of a normative judgment is a
matter of whether that very judgment withstands a relevant "procedure of scrutiny" -the
correctness of that judgment "is constituted by the fact that it withstands this scrutiny"
(208-9, second emphasis mine).
The next point is that if scrutiny is indeed to issue in normative judgment as an
output, without itself involving some substantive normative judgment internal to the
very process of scrutiny, then it seems that the scrutiny will need to begin with
normative judgment as an input. After all, suppose that the scrutiny were instead to
involve, as R. Jay Wallace might say, normative judgment out without normative
judgment in.
It seems that such scrutiny would itself need to invoke either some
substantive normative claim connecting non-judgment input with judgment output,
or a meta-normative view connecting non-judgment input with judgment output.
The former would compromise the generality of the constructivism: we would now be
appealing to a substantive normative claim in explaining what scrutiny itself is, and so it
would not be clear how to extend the constructivism to that substantive claim itsel£
The latter would involve appeal to a meta-normative thesis that would itself need a
defense, a defense that would then be prior to the constructivist story.
Let me briefly elaborate this last point. Consider a purported case in which there is
normative judgment out but not normative judgment in. Suppose that the input to
scrutiny were the agent's second-order volitions: desires that certain desires motivate.
And suppose the output of the scrutiny were a judgment about normative reasons. The
scrutiny would then involve a transition from second-order volition to a judgment
about reasons. What would support this transition? Well, we might try to appeal to a
substantive normative claim that when you have such a second-order volition you have
a normative reason. But then it will be unclear how the constructivism can apply to
that very normative claim. Could we say that it is constitutive of second-order volition
that if you have such a volition and, as Street would say, you attend "to the matter in
full awareness," you will have the cited normative judgment? Well, this seems prob-
lematic. It seems that one could have such a second-order volition and still wonder
whether one has a corresponding reason for action. Could we nevertheless insist that
the metaphysics of reasons, or the relevant concepts, ensure that if you have that
second-order volition then you have the corresponding normative judgment and/ or a
corresponding reason? But then we would be defending the constructivism by appeal
to a yet more basic meta-normative claim. In none of these cases does the constructiv-
ism clearly survive as the completely general and foundational theory Street seeks. And
it seems plausible that this will be a problem whenever the scrutiny has a nonnative
" Wallace 1990.
" For di.ocu..sion relevant to this Schroeder 2007, section 3.4.
,. Frankfurt 19KHb.
judgment as an output but no relevant nonnative judgment as an input. This is why it is
natural for Street to suppose, as she does, that the relevant scrutiny that issues in
nonnative judgments involves an input that includes nonnative judgments.
That, anyway, seems to me to be a structure of ideas that supports the assumption-
an assumption built into Street's account in this essay-that the relevant agential
scrutiny must begin with the agent's normative judgments and issue in "target"
normative judgments.
But we have also seen that in the background there is also
the agency-theoretic thought that the ground from which the scrutiny proceeds must
itselfhelp constitute the standpoint of the agent. The appeal to nonnative judgments in
1 b amounts, then, to part of a purported, unified solution to a pair of interrelated
problems: how should we think about the standpoint of the agent, and what can serve
as the basis for a non-restricted constructivist view of practical reasons? In this way, lb
involves an optimism that the pressures fiom the non-restricted constructivism will
indeed align with the pressures fiom the theory of agency. And we need some such
alignment if our non-restricted constructivism is going to tie reasons to what has force
from the agent's standpoint.
Now, as Nadeem Hussain has emphasized in conversation, if we were to make a
broadly Kantian assumption that an agent's standpoint is the standpoint of Practical
Reason, then we would have an alignment between the agent's standpoint and a
potential basis for the construction of reasons. (Though we would then need to worry
about how much can be derived simply from practical reason.) But Street rejects this
picture and sees the agent's standpoint as involving, inter alia, myriad and sometimes
idiosyncratic substantive elements. So we need to take seriously the question whether
what we need to appeal to in order to articulate a non-homuncular theory of agential
standpoint is what is needed by Street's non-restricted constructivism.
There are two kinds of considerations that point to a difficulty here. One is that we can
be alienated from certain normative judgments, just as we can be alienated from certain
desires. This is one way to understand, for example, Huck Finn's judgment that he
Drawing on Williams, Street says that her view is "Humean in that it understands each person's reasons
ultimately to be a function ofhis or her 'subjective motivational set' "(244). But she goes on to insist "that the
'elements' in that set are most profitably characterized first of all not as desires, but rather as ncmnative judgments
(and unreflective versions thereot)" (245). This is a difference between Street and Williams concerning the
inputs that are essential to relevant scrutiny or deliberation. There is also a difference about the outputs.
Williams says that "A has a reason to phi only if he could reach the conclusion to phi by a sound deliberative
route from the motivations he already has" (1995, p. 35). Williams allows that in some cases reaching a
"conclusion to phi" is deciding to phi, or merely being motivated to phi (1981, p. 109). So Williams does not see
nomtative judgment as essential to the outputs of the relevant deliberation or scrutiny.
This is why Williams, in insisting that all practical reasons are "internal," could agree with the general idea,
and with 1 a, but still reject 1 b. Street, in contrast, seems conunitted to saying that the output of the relevant
~ r u t i n y is a relevant normative judgment, and so she is under pressure to see the inputs as essentially involving
nom1ative judgments if she is to retain the merely formal character of scrutiny.
should tum in the runaway slave, Jim.2
If we were to say what constitutes Huck's
perspective-the perspective to which advice in "the 'ifl were you ... ' mode" should
appeal-it seems plausible that we would not include his judgment that he should tum
in Jim, but rather his attitudes in favor of protecting Jim. So not all normative
judgments help constitute the agent's relevant standpoint.
A second consideration is that much of what constitutes the agent's standpoint
involves attitudes other than normative judgment. In particular, a central Frankfurtian
idea is that what in large part constitutes an agent's standpoint are attitudes like caring
and love, and that these attitudes are not, at bottom, a matter of judging there to be
reasons. Given that the agent cares about or loves certain things he is, according to
Frankfurt, in a position to infer that he has certain reasons for action (though this
inference seems to depend on a substantive normative principle). But the ground of
these reasons is not itself a judgment of reasons but, rather, a distinctive kind of conative
In many cases what we care about and what we love have a ground-
level role in shaping where we stand, and this role goes beyond the role of any
normative judgments that are not themselves grounded in these attitudes. The role
of my love for my children in constituting my standpoint is not exhausted by judg-
ments about reasons that are not themselves grounded in my love. And this challenges
the idea in lb that the agent's standpoint is constituted by her normative judgments.
I will focus primarily on this second, Frankfurtian challenge. To assess this challenge
we need to reflect on how we might, consistent with the constructivist approach,
defend a specific story of the agent's standpoint. Of course, we cannot say, within such
a constructivist approach, that it is because normative judgments track independent
truths about reasons that they have the authority to constitute where the agent
stands: the constructivist does not countenance such independent truths. What then
can we say?
Well, one plausible approach here is to say that what constitutes the agent's practical
standpoint are those attitudes that play major roles in stably organizing the agent's
temporally extended life, including the motivation of action and the structure of
practical reasoning. There are threads in Frankfurt's work that point to something
like this strategy, and I myself have pursued a version of it in other work.
This strategy
could be developed in different ways. But the point to make now is that in its broad
outlines it seems to support the Frankfurtian challenge.
This is because it is plausible that much of the structure of a life is commonly induced
by commitments that go beyond judgments about reasons or value that do not
themselves depend on those very commitments. Return to Frankfurt's example of
Jove. My love for certain people (or ideals) shapes my life in fundamental ways. My
.. Arpaly and Schroeder 1999. For the general point about the possibility ofbeing alienated from one's
own value judgments see VeUernan 2000, p. 134.
n Frankfurt 2006, esp. pp. 25, 4D-2.
21 Frankfurt 1911!k, p. 175. This is a theme in many of the essays in Bratman 2007a.
love may be in part a response to, and even involve judgments about, reasons and
values; and once I do love these people (or ideals) I will, normally, have reasons I did
not have before. But the Frankfurtian challenge is that the role of my love in
constituting my practical standpoint, and grounding further reasons, goes beyond the
role of judgments about reasons that do not themselves depend on my love.
Sartre's case of the boy who must choose between the Free French and staying with
his mother points in the same direction: once the boy settles on one of these two
possible ways oflife, he has a commitment that is central to his practical standpoint but
that goes beyond his judgments of reasons that do not themselves depend on that
Less dramatic examples along such lines come from career decisions
between options no one of which the agent antecedendy judges to be uniquely best. In
all these cases there is a commitment that is not itself a judgment about reasons though
it seems normally to be a ground of further reasons.
If this is right then what constitutes the agent's standpoint is not limited to the
agent's judgments of reasons. (And, as we have noted, it may not include all such
judgments.) Central to that standpoint will be commitments that go beyond judgments
of reasons that do not themselves depend on those commitments.
It follows from this that the Williams-friendly thought that reasons must be con-
siderations that have force from the agent's standpoint does not yet lead to Street's idea
that reasons must have force from the standpoint of, in particular, the agent's other
normative judgments. So the assumption of a convergence of considerations about
strong agency with the purported role of the agent's normative judgments as the
fundamental inputs to constructivist scrutiny is challenged by these Frankfurtian
thoughts about the commitments that help constitute the agent's standpoint.
What are these commitments? Well, they do not simply motivate us: they shape our
practical lives in large part by shaping our practical reasoning. And this seems to involve
shaping what we at least treat as a reason in that reasoning. So perhaps we can see these
conunitments as essentially involving not judgments about reasons but, rather, com-
mitments to treat certain considerations as reasons-to give these considerations
weight-in one's relevant practical reflection. My love for my children constitutively
involves a conunitment to treat their interests as having weight in my deliberation. And
when the boy in Sartre's story settles on a life of aiding his mother his new commit-
ment is in large part constituted by a conunitment to treat certain considerations as
reasons-as having weight-in his relevant deliberation.
The idea here is not the idea that these commitments provide a ground for a
substantive inference to a further judgment about reasons. The present idea is, rather,
Sartre 1975, pp. 354-6.
that these commitments consist, in large part, in commitments to think and reason in
certain ways-in particular, to treat certain considerations as reasons, as having
But these commitments to treat as a reason need not themselves be judg-
ments that there are these reasons.
Why not? If you are committed to giving certain considerations weight-to treating
them as a reason-why does it not simply follow that you hold the corresponding
judgments about reasons? Here we need again to reflect on constitutive features of
What distinguishes judging that p from merely supposing that p, or taking it for
granted that p, or relying on the assumption that p? Here I think it is plausible to appeal
to an idea that goes beyond Street's appeal in ld to relations among the agent's
judgments. This is the idea that in judging that p one sees one's judgment as subject
to correction by relevandy situated thinkers (including, but not limited to, oneself at
other times)-as subject, in particular, to some sort of standard of inter-subjective
And the crucial point here is that you can have a commitment to treat a
consideration as a reason without seeing that commitment as subject to, or conforming
to, relevant demands of inter-subjective convergence.
Indeed, you need not even
think of your commitment as entering into a social dialogue that seeks such conver-
gence. So such a commitment to treat as a reason may not be a corresponding
judgment about reasons. The boy in Sartre's case, for example, can be committed to
giving overriding weight to aiding his mother without seeing this commitment as
subject to a standard of inter-subjective convergence. He need not think that someone
who in such a situation arrives at an alternative commitment in favor of the Free French
makes a mistake; and he need not aim at achieving social convergence on this decision.
The idea is that in loving or caring about something we are committed to giving
certain weights--to treating certain considerations as reasons-in deliberation. And
these commitments help structure our own practical thinking and thereby our practical
lives, and so help shape our agential standpoint. We can think of these commitments to
weights as a kind of valuing, as I have in other work.
Indeed, this may be a good way
to understand Street's own conception of valuing in her essay for this volume (though
her official view identifies such valuing with normative judgment).
But the important
point is that if we see judgment as subject to an associated constraint of inter-subjective
Frankfurt himself may be skeptical about this distinction when he writes: "Insofar as a person loves
something, he necessarily counts its interests as giving him reasons to serve those interests. The fact that his
beloved needs his help is in itself a reason for him to provide that help ... " (2006, p. 42).
" There will also be issues about intra-subjective, diachronic convergence; but for present purposes we
can put these issues to one side.
•• I discuss this idea in Bratman 2007b, esp. pp. 151-4. /'u I note there (n. 47) there is a parallel here with
Allan Gibbard's distinction between an ''existential conunitment" and acceptance of "a nonn as a require-
ment of rationality" Gibbard 1990, pp 166-70. In Bratman 2006 I discuss closely related issues about
Gibbard's theory of normative judgment as developed in Gibbard 2003 .
., See Bratman 2007d, pp. 295-!!.
•• Street. this volume. section 2.
accountability, we should say that these commitments to weights need not themselves
be normative judgments since they need not be seen by the agent as subject to demands
of inter-subjective convergence or even as contributions to a social dialogue that aims
at such convergence.
Of course, we could coin a term-"weak judgments," say--and say of such
commitments that they are weak judgments. But the issue is not the word
but the idea that the ordinary judgments about reasons which are the "target" of non-
restricted constructivism-the judgments whose truth conditions the construction aims
to articulate-are not weak in this sense, but are rather tied to relevant demands of
inter-subjectivity. And what we have seen is that there is pressure from the theory of
agency to resist the idea that what speaks for the agent consists solely in such ordinary
judgments. The agent's standpoint may well involve commitments to weights that are
not, strictly speaking, judgments about reasons.
Can we say more about the nature of these commitments? Well, we have available a
concept that seems well suited to help us here, the concept of intention. These
commitments to weights, we can say, are intentions to give these weights in relevant
deliberation, or-as we might say, in light of the normal generality of such commit-
ments-policies of giving such weights.
And such intention-like commitments to
treating as a reason need not themselves be judgments about reasons.
So the situation is this. According to lb, the standpoint of the agent that is needed for
non-restricted constructivism is constituted by that agent's normative judgments. But
reflection in the philosophy of action supports the idea that the agent's standpoint typically
involves intention-like commitments to treating as a reason, commitments that are not
themselves judgments about reasons. So there is a problem of alignment. What to say?
One response might be to be more liberal about the outputs of scrutiny. We agree that
an agent's standpoint typically involves commitments to weights that are not judg-
ments about reasons. But we also allow that the outputs of relevant scrutiny need not
themselves be judgments but can also be such commitments. We thereby support a
modified non-restricted constructivism along the lines of:
(Modified) The fact that X is a reason to Y for agent A is constituted by the fact that
either the judgment that X is a reason to Y (for A) or the commitment to treat X as a
reason to Y (for A) derives, by way of scrutiny, from the standpoint of A's other
normative judgments and commitments to treat as a reason.
The relevant scrutiny will not involve a direct transition from a non-judgment
commitment to treat as a reason to a judgment about reasons. It will instead involve
Bratman 2007c.
transitions either from nonnative judgment to nonnative judgment, or from non-
judgment nonnative commitment to non-judgment nonnative commitment. So the
scrutiny to which (Modified) appeals promises to respect both a judgment-out,
judgment-in principle, and an analogous commitment-out, commitment-in principle.
And, according to (Modified), there can be cases in which the fact that X is a reason to
Y for A does not consist in the fact that the judgment that X is such a reason is the
output of relevant scrutiny on the part of A. The fact that X is a reason to Y for A might
instead consist in the fact that a commitment to treat X as such a reason is the output of
relevant scrutiny.
It seems to me, however, that this would be a fairly fundamental revision in the
theory. We could no longer say, as Street does say, that "correctness of a judgment
about one's reasons must be understood as a matter of whether that judgment withstands
scrutiny from the standpoint of one's further judgments about reasons" (243, emphasis
added). But this idea-the idea that the correctness of a judgment about reasons is a
matter of whether that very judgment is the output of relevant scrutiny-seems central
to Street's theory and to the purported parallel between her metaethical constructivism
and the restricted constructivist theories from which she begins.
An alternative strategy would be to insist that a commitment to weights helps
constitute where the agent stands only if the agent herself judges that because of this
commitment she has a corresponding reason. And once that judgment about reasons is
an element in the agent's standpoint, the constructivist can appeal to it to explain how,
by way of formal scrutiny, the agent can arrive at relevant judgments of reasons. For
example: I love my children. So I have a policy of giving their interests great weight.
But if this love, and this policy about weights, is really to help constitute where I stand,
I must judge that they do indeed give me reason to promote the interests of my
children. But then this judgment about reasons is in a position to ground, by way of
formal scrutiny, other judgments about reasons. It is, once again, normative judgment
in and normative judgment out.
One might worry that the possibility of being alienated from one's judgments of
reasons will arise here, with respect to the judgment that one's commitment gives one
reasons. But the concern I want to highlight is not about the sufficiency of such a
background judgment about reasons for ensuring that the conunitment is an element
in where the agent stands, but its necessity. It seems that a person's standpoint can
include various personal commitments-commitments that shape her practical think-
ing-whether or not she herself goes on, strictly speaking, to judge that those
conunitments provide a reason, where this judgment of a reason is inter-subjectively
accountable. Perhaps she is too diffident--and/or too aware of disagreements about
such matters-to make such inter-subjectively accountable claims. Nevertheless, this is
still a part of her practical standpoint, where she stands.
The idea of the agent's standpoint is the idea of the agent's fundamental practical
conunitments. Such commitments may include but need not be limited to judgments
about reasons. A diffident agent with certain commitments to weights might not
herself be prepared to make corresponding inter-subjectively accountable judgments
about reasons. Yet these commitments may still be a central element in her practical
Further, once we acknowledge this point about the diffident agent's standpoint, it is
a short step to the substantive claim that such an agent can in fact have reasons for
action that are grounded in such personal commitments even if she herself is too
diffident to hold this judgment about reasons. But that is a substantive claim about
reasons that is difficult to bring within the ambit of Street's constructivism.
A final response might be to say that the relevant inputs to scrutiny can simply be
commitments to weights even though the outputs of the scrutiny are judgments about
reasons. But the problem here, as we saw in section 2, is that it seems that the scrutiny
will then need to bring to bear substantive normative judgments or meta-normative
principles of a sort that will prevent the scrutiny from being appropriately formal.
My tentative conclusion, then, is that once we try to do justice to the problem of
alignment, and take seriously the cited Frankfurtian challenge, a non-restricted con-
structivism, along the lines sketched by Street in this essay, faces a trilemma. It seems to
need to settle on one of the three modes of response just outlined; but no one of these
responses seems fully satisfactory.
Arpaly, Nomy and Schroeder, Timothy. 1999. "Praise, Blame and the Whole Self," in Philo-
sophical Studies 93, pp. 161-88.
Bratman, Michael E. 2006. "Thinking How to Live and the Restriction Problem," in Philosophy
and Phenomenological Research 72, pp. 708-14.
Bratman, Michael E. 2007a. Structures of Agency: Essays (New York: Oxford University Press).
Bratman, Michael E. 2007b. "A Desire of One's Own," in Bratman 2007a, pp. 137-61.
Bratman, Michael E. 2007c. "Three Theories of Self-Governance," in Bratman 2007a,
pp. 222-53.
Bratman, Michael E., 2007d. "Shared Valuing and Frameworks for Practical Reasoning," in
Bratman 2007a, pp. 283-310.
Bratman, Michael E. 2009. "Intention, Practical Rationality and Self-Governance," in Ethics
119, pp. 411-43.
Broome, John. 2007. "Is Rationality Normative?" in Disputatio 2, pp. 161-78.
Frankfurt, Harry. 1988a. The Importance of "What We Care About (Cambridge: Cambridge
University Press, 1988).
Frankfurt, Harry. 1988b. "Freedom of the Will and the Concept of a Person," in Frankfurt
1988a, pp. 11-25.
An earlier version of this essay was presented and usefully discussed at the 2009 University of Sheffield
Workshop on Constructivism. Special thanks to Nadeem Hussain and David Plunkett for helpful discussion,
and to James Lenman and Y onatan Shenuner for detailed and helpful written comments.
Frankfurt, Harry. 198&. "Identification and Wholeheartedness," in Frankfurt 1988a,
pp. 159-76.
Frankfurt, Harry. 2006. Taking Ourselves Seriously and Getting It IGght (Stanford: Stanford
University Press).
Gibbard, Allan. 1990. Wise Choices, Apt Feelings (Cambridge, MA: Harvard University Press).
Gibbard, Allan. 1999. "Morality as Consistency in Living," in Ethics 110, pp. 140--64.
Gibbard, Allan. 2003. Thinking How to uve (Cambridge, MA: Harvard University Press).
Hussain, Nadeem. 1999. Creating Values: Appropriating Nietzsche for a Fictionalist Theory of Value,
PhD Dissertation (University of Michigan).
James, Aaron. 2007. "Constructivism about Practical Reasons," in Philosophy and Phenomenologi-
cal Research 74, pp. 302--25.
Kolodny, Niko. Unpublished manuscript. "Instrumental Reasons."
Raz, Joseph. 2005. "The Myth of Instrumental Rationality," in Journal of Ethics and Social
Philosophy 1, pp. 1--28.
Sartre, Jean-Paul. 1975. "Existentialism is a Humanism," in W. Kaufmann (ed.), Existentialism
from Dostoevsky to Sartre rev. and expanded (New York: Meridian/Penguin), pp. 345--69.
Scanlon, T. M. 1998. Mat We Owe to Each Other (Cambridge, MA: Harvard University Press).
Schroeder, Mark. 2007. Slaves of the Passions (Oxford: Oxford University Press).
Street, Sharon. 2006. "A Darwinian Dilemma for Realist Theories of Value," in Philosophical
Studies 127, pp. 109--66.
Street, Sharon. 2008. "Constructivism about Reasons," Oxford Studies in Meta-Ethics 3,
pp. 207--45.
Street, Sharon. 2012. "Corning to Terms with Contingency: Humean Constructivism about
Practical Reason," this volume.
Velleman,J. David. 2000. "What Happens When Someone Acts?" in his The Possibility of Practical
Reason (New York: Oxford University Press).
Wallace, R.Jay. 1990. "How to Argue about Practical Reason," in Mind 99, pp. 355--84.
Watson, Gary. 1975. "Free Agency," in journal of Philosophy 72, pp. 205--20.
Watson, Gary. 1987. "Free Action and Free Will," in Mind 96, pp. 145--72.
Williams, Bernard. 1981. "Internal and External Reasons," in his Moral Luck (Cambridge:
Cambridge University Press), pp. 101--13.
Williams, Bernard. 1995. "Internal Reasons and the Obscurity ofBlame," in his Making Sense of
Humanity (Cambridge: Cambridge University Press), pp. 35--45.
A Puzzle for Constructivism
and How to Solve It
Dale Dorsey
1. Constructivism
The topic of this essay is the metaethics of practical reasons, with special focus on a view
I call "constructivism." Because this term is rather promiscuous, it seems best for me to
characterize the view I'm talking about right off the bat. As I understand it, construc-
tivism is characterized by three central tenets. The first of which runs as follows:
Truth-Apt: normative judgments are truth-apt.
For constructivism, normative judgments can be adequately described as true or false.
The second feature of constructivism is:
Relational: that which makes a particular normative judgment nj true is nj's bearing of
a favored relation to other normative judgments.
Relational offers a substantive truth condition for normative judgments: a normative
judgment nj is made true by the fact that nj bears a favored relation to other normative
judgments, or some suitably specified subset thereof.
Third, constructivism as I use the term accepts:
Unrestricted: Relational applies to all normative judgments.
Unrestricted holds that all normative judgments permit of the same truth conditions, viz.
the bearing of a favored relationship to other normative judgments.
Constructivism, so defined, permits of a number of different specifications. One can
identify the proper "relation" in many different ways. One can insist that normative
judgments are true if and only if they bear the proper relation to other normative
judgments after some suitably identified round of cognitive idealization, for instance,
I would like to thank James Lenman, Cory Wright, Jack Bricke, and an anonymous reader for helpful
and challenging feedback on an earlier draft of this chapter.
perhaps articulated counterfactually. Constructivism as a class, however, contrasts with
a number of alternative metaethical categories, most importantly realism and expressi-
vism. The chief argument between constructivists and their rivals will be over Relational.
Realists will claim that normative judgments are made true by judgment-independent
normative facts.
Whether normative judgments bear such a favored relationship to
other normative judgments is, for realism, irrelevant to their truth. Expressivists-if
they accept Trnth-Apt at all-will insist that the truth of normative judgments is strictly
minimal, and does not permit of substantive truth conditions of the sort endorsed by
The best articulated, but certainly not only, account of constructivism as
I understand it is offered by Sharon Street. Street's account runs as follows: "the fact
that X is a reason to Y for agent A is constituted by the fact that the judgment that X is a
reason to Y (for A) withstands scrutiny from the standpoint of A's other judgments
about reasons."
For Street, r is a reason for x to 4> if and only if the judgment "r is a
reason to x to rf>" "withstands scrutiny" from the perspective of x's other judgments
about reasons. Hence judgments about reasons will permit of relational truth condi-
tions: whether they are true will be determined by whether they withstand the scrutiny
of x's other normative judgments. For the purposes of the current chapter, I will
assume Street's account; those who prefer an account stated in a different way (e.g.
with an alternative characterization of the favored relationship) are free to make the
proper translations.
2. Normative judgments
Reasons, for constructivism, are a product of the normative judgments people make.
However, a number of questions arise when it comes to understanding the nature of
these judgments. Such questions include, but are not limited to, the nature of judg-
ments in comparison to other propositional attitudes like beliefS and desires, the nature
of judgments in comparison to other truth-bearers like sentences or propositions, the
distinguishing characteristics of non-normative judgments in comparison to normative
judgments, etc. Answering these questions is of the ftrst importance for any construc-
tivist view. But I lay discussion of them aside to focus on a further question: What,
according to the constructivist, is the semantic content of a normative judgment?
What, in other words, do normative judgments mean?
Constructivism cannot hold that normative judgments mean just any old thing.
They cannot, for instance, ascribe natural or non-natural judgment-independent
nonnative properties (in Roland Dworkin's terminology, "morons," or, given our
See, for instance, Shafer-Landau 2003, p. 16.
' See, importantly, Gibbard 2003, p. 42, n. 3.
• Street 200H, p. 223.
focus on nonnativity, "nonnons"
). They also cannot be taken to refer to individual
perceptual experiences, or conative or affective states.
But why not? The answer, of course, starts with a straightforward claim about trnth.
Generally speaking, we hold that a truth-bearer is true if and only if that truth-bearer's
meaning bears "the right relation" to that bit of the world, or state of affairs, that would
make it true (i.e. its truth condition).
("The right relation" can be given any number of
interpretations, such as correspondence, satisfaction, mirroring, picturing, etc., etc.
I won't commit to any one of these here.) For the sake of brevity, I will call any
theory of truth that accepts this claim a "semantic" theory oftruth.
Thus the semantic
analysis of any judgment, or any other truth-bearer for that matter, has to properly
"match" its truth condition, sufficient to render it the case that whenever the truth
condition holds, the judgment's meaning "bears the right relation" to it. On this
general thought, Alston writes that for a given bit of extra-linguistic reality (or
"fact") to make a given proposition (or other truth-bearer) true, "the proposition
and the fact that makes it true share the same propositional content. What the fact is a
fact that, is the same as what the proposition is a proposition that.''
Given a semantic
theory of truth, then, if nonnative judgments refer to nonnons, "r is a reason for x to cp"
would be made true not by its scrutiny-withstandingness, but rather by facts about
nonnons. If nonnative judgments refer to perceptual experiences, or affective or
conative states, "r is a reason for x to cp" would be made true not by its scrutiny-
withstandingness, but rather by facts about perceptual experiences or affective or
conative states. And so on.
So far, so platitudinous. However, a problem arises here for constructivism. Insofar as
constructivism accepts Relational, constructivists must offer a semantic analysis of
nonnative judgments that properly "matches" the bit of the world that makes them
true, viz. the fact that they withstand scrutiny. But it would seem that this can occur
only if the semantic content of a nonnative judgment is specified in terms of its
scrutiny-withstandingness. To put this in other words, any semantic analysis of"r is a
reason for x to cp" must be put in terms of that judgment itself, i.e. that that very judgment
withstands scrutiny. And here we find a problem. The semantic analysis of a nonnative
Dworkin 2011, Chapter 1.
• In referring to states of affairs, "bits" of the world, or,later on, "facts,"! do not intend to be committing
myself to a controversial metaphysical position with respect to the existence of such states of affairs, or facts, or
their efficacy in a theory of truth. Rather, I use these terms as efficacious shorthand in identifying the extra-
linguistic truth conditions of individual truth-bearers.
I do not intend a "semantic theory of truth" to be identical to Tarski's so-called "semantic conception of
truth," as spelled out by Tarski in Tarski 1999. Rather, I simply mean any theory of truth that holds that, to
be true, a given judgment's meaning must bear the right relation to its truth condition. Though Tarski's view
is. as I understand it, a semantic theory, it needn't be the only one.
• Alston 1996. p. 38. Alston defmes this as a "realist'"theory of truth, but his conception of a realist theory
is identical to what I have identified as a "semantic" theory. Alston claims that realiJt theories hold that: "A
statement (proposition, belief ... ) is true if and only if what the statement says to be the case actually is the
case"" (Alston, p. 5).
judgment must be elucidated in terms of that which is to be analyzed: that very
judgment itself. Any semantic analysis of normative judgments, on a constructivist
view, must be circular.
Take an example. Imagine that I judge the following:
1. That "Jumpin' Jack Aash" is on the radio is a reason for me to pump up the
Given a semantic theory of truth, we must offer a semantic analysis of (1) that
suitably matches its truth condition. Given constructivism, however, the truth
condition of (1) is the scrutiny-withstandingness of (1) itself. Hence, (1) must be
analyzed, at least in part, as follows: "'that "Jumpin' Jack Flash" is on the radio is a
reason for me to pump up the volume' withstands scrutiny from the standpoint of
my other normative judgments." But this analysis is clearly problematic. We were
asking what the semantic content (1) is. We were given an answer in terms of (1)
itself. To properly understand the propositional content of (1), one must again ask
what the propositional content of (1) is. But, quite obviously, a regress now looms.
The only additional information we seem able to convey is that (1) should be now
be analyzed as: "'"that 'Jumpin' Jack Aash' is on the radio is a reason for me to
pump up the volume" withstands scrutiny from the standpoint of my other norma-
tive judgments' withstands scrutiny from the standpoint of my other normative
judgments." And on and on. So far no informative semantics of this judgment
has been given, because the semantic analysis of (1) is circular: the semantic analysis
of (1) presupposes a prior semantic analysis of (1).
A circular semantic analysis of normative judgments is a problem for any construc-
tivism of the sort I mention. To see this, consider the following argument:
A. For every judgment}, the semantic analysis ofj must "match" the truth condi-
tions ofj (semantic theory of truth).
B. For some noi:mativejudgrnent of the form "ris a reason for x to¢" nj, njis true if
and only if nj bears a favored relation to a set of normative judgments SNJ (by
C. Hence (by (A) and (B)), the propositional content of nj must be that nj bears a
favored relation to a set of normative judgments SNJ.
D. Hence any semantic analysis of nj must be specified in terms of nj.
E. Hence any semantic analysis of nj is circular.
One might wonder why I have restricted my discussion to views that accept Unrrstrictrd. insofar as this
principle seems to play no role in the derivation of (E). It could very well be that constnKtivist views that
deny Unrrltrictrd alao face a problem here. But I restrict my discus.•ion insofar as my solution to this problem
1ucceeds only for views that accept it. I leave aside any potential solutions. or indeed any potential problems.
for restricted venions of constn1ctivivn.
3. So what's the problem?
One might reasonably wonder why commitment to a circular semantics of normative
judgments is a problem. There are two reasons that strike me as decisive.
First, circular semantic analyses are themselves unacceptable. As Ralph Wedgwood
writes, following Kripke: "it is precisely the task of an account of the reference of [a
name] to explain what it is for someone to be the person referred to by [the name).
Hence, for an account of the reference of the name 'Socrates' to be acceptable, it must
not use the notion of "being called 'Socrates'" or "being referred to by the name
'Socrates', on pain of simply presupposing what is to be accounted for."
Of course,
constructivism is not giving a circular semantics of a name. But it is committed to an
explicitly circular semantics of a class of judgments. And the problem is not much
different. In the case of normative judgments, what we seek is a semantic analysis
sufficient to determine their propositional content, what they refer to, that to which
the judgment is committed, etc. But a circular semantic analysis--one that implies the
sort of regress I explored in the previous section--cannot accomplish this task. Without
a prior understanding of the propositional content of "r is a reason for x to !fo," we
cannot understand the propositional content of"r is a reason for x to !fo." Call the first
problem the "problem of circularity."
A second problem has to do with the structure of constructivism itsel£ Constructiv-
ism about reasons requires us to be able to determine when a given normative
judgment withstands scrutiny from the standpoint of other normative judgments.
But it is difficult to see, without a substantive analysis of what these judgments mean,
whether and when any individual normative judgment will survive such scrutiny.
Hence without providing a substantive semantic analysis, or at least a semantic analysis
that says more than the one constructivism is stuck with, constructivism about practical
reasons, morality, or any other domain offers a set of truth conditions that is, at best,
entirely indeterminate.
The best way to see this problem is to consider an explicit example. Imagine that my
set of normative judgments contains only two, specified as follows:
2. That it will kill everyone is a decisive reason for me not to start a nuclear conflict
between the United States and Russia.
3. That it will kill everyone is a decisive reason for me to start a nuclear conflict
between the United States and Russia.
Does (2} withstand the scrutiny of (3)? One would expect that the answer should be
"no." As Street notes:
As soon as one takes anything whatsoever to be a reason, one thereby "legislates" standards
according to which, by one's own lights as a valuing agent, one is making a mistake, whether one
Wedgwood 2007, p. 19.
I thank Jimmy Lenman and an anonymous reader for inspiring further attention to this problem.
knows it or not, if one endorses certain other nonnative judgments ... For one nonnative
judgment to withstand scrutiny from the standpoint of other nonnative judgments, then, is for
that judgment not to be mistaken as determined by the standards of correcmess that are
constitutively set by those other nonnative judgments in combination with the non-nonnative
Stated in this way, it certainly seems as though (2) and (3) do not withstand the scrutiny
of each other. From (2), we are to take the proposition "starting a nuclear conflict
between the United States and Russia will kill everyone" to be a decisive reason not to
start such a conflict. (3) violates that standard of correctness by insisting that the very
same proposition is decisive reason to start such a conflict.
But there are two problems here for constructivism. First, insofar as constructivism is
unable to offer any informative semantic analysis of normative judgments, it is difficult
to see how one could support the claim that (2) and (3) do not withstand scrutiny of
each other. After all, to know whether they do, in fact, survive such scrutiny, one has to
understand their semantic or propositional content. Otherwise we have no idea
whether (3) "makes a mistake" by (2)'s lights. Hence without an informative account
of the propositional content of (2) and (3), whether I actually have a reason to start a
nuclear conflict seems indeterminate.
Second, if the semantic analyses of (2) and (3) must match their truth conditions,
(2) and (3) must be analyzed, respectively, like this:
4. "That it will kill everyone is a decisive reason for me not to start a nuclear conflict
between the United States and Russia" withstands scrutiny from the standpoint
of"That it will kill everyone is a decisive reason for me to start a nuclear conflict
between the United States and Russia."
5. "That it will kill everyone is a decisive reason for me to start a nuclear conflict
between the United States and Russia" withstands scrutiny from the standpoint
of "That it will kill everyone is a decisive reason for me not to start a nuclear
conflict between the United States and Russia."
But if (2) and (3) are properly analyzed by (4) and (5), to know whether (2) and (3)
withstand the scrutiny of each other, we must ask: Does (4) withstand the scrutiny of
(5)? Answer: absolutely! (4) merely says that a particular judgment, viz. "That it will kill
everyone is a reason for me not to start a nuclear conflict between the United States and
Russia," withstands the scrutiny of another judgment, viz. "That it will kill everyone is
a decisive reason for me to start a nuclear conflict between the United States and
Russia." (5) is merely the vice versa. But these judgments clearly withstand scrutiny
from the standpoint of each other: there is no "standard of correctness" set up by (4)
that could possibly be violated by (5). Indeed, one might be tempted to make an even
stronger claim: (4) and (5), far from failing to withstand each other's scrutiny, positively
•• Street 2008, p. 230.
support each other. So if (4) and (5) withstand scrutiny, and the proper semantic analysis
of(2) and (3) is (4) and (5), respectively, we cannot-or so it would appear--accept the
claim that (2) and (3) do not withstand the scrutiny of each other.
One might complain that (4) and (5) are false, and hence (2) and (3) are also false.
Surely, so it may be claimed, this is precisely the verdict that constructivism intends.
But one can plausibly say that (4) and (5) are false only if one offers a semantic analysis
of (2) and (3), sufficient to render (2) and (3) non-scrutiny-withstanding. But if
constructivism requires that (2) and (3) are analyzed as (4) and (5), then (4) and (5)
are false only if (4) and (5) do not withstand scrutiny. But they do, and hence (4) and (5)
are, or so it would seem, true.
The problem on display here traces direcdy to the semantic analysis of normative
judgments forced on constructivism by a semantic theory of truth. To see this, consider
the reading of (2) and (3) described by David Brink as follows: "As many have
observed, moral discourse is typically declarative or assertive in form. We say things
like 'The government's tax plan is unfair,' 'Waldo is just,' 'It would be wrong to work
for that cause,' and 'My obligation to Maurice is greater than my obligation to
Malcolm.' This language is putatively fact-stating (because it is declarative in form)
and certainly seems to ascribe moral properties to persons, actions, policies, and so
forth. "
Call this a "surface" semantic analysis of normative judgments. This surface
semantics seems committed to metaphysically significant moral (or normative) proper-
ties, such as reasons, obligations, and so forth. On a surface reading of (2) and (3), they
certainly do not withstand the scrutiny of each other: one cannot ascribe the properties
ascribed in (2) and (3) as they are ascribed by both (2) and (3). The property ofbeing a
decisive reason to rp simply rules out the possibility ofbeing a decisive reason not to tf>.
Hence, on this surface analysis, (2) and (3) do not withstand each other's scrutiny.
But there are two important differences between the semantic analysis forced on
constructivism by a semantic theory of truth and the surface semantics described by
Brink. First, though normative judgments ascribe properties, they ascribe only the
property of scrutiny-withstandingness. Second, these judgments ascribe no property to
"persons, actions, policies, and so forth,'' but rather state facts about the relation
between individual normative judgments. But if this is correct, constructivism cannot
vindicate the plausible thought that (2) and (3) do not withstand scrutiny because they
do not ascribe the sort of property that would entail a failure of (2) to withstand the
scrutiny of (3). They ascribe only scrutiny-withstandingness to themselves, and in so
doing (as we've seen with (4) and (5)), they do survive such scrutiny. At best,
constructivism seems, given its inability to provide a substantive semantic analysis,
unable to provide any determinate answer to the question of what reasons one has. At
worst, constructivism--given that normative judgments assert only that they withstand
Brink 1989, p. 25.
scrutiny--seems unable to declare any normative judgments false. For the sake of
charity, call the second problem the "problem of indeterminacy."
4. Constructivism: metaphysical, not alethic
A response to the above argument should be considered here. Recall Street's account
of metaethical constructivism: "the fact that X is a reason to Y for agent A is constituted
by the fact that the judgment that X is a reason to Y (for A) withstands scrutiny from the
standpoint of A's other judgments about reasons."
Here Street says very litde about
what makes a normative judgment trne, or what a normative judgment means. Rather,
Street offers what might be called a metaphysical account of reasons: what must be in place
for r to be a reason for x to cp.
But if this is right, constructivism can avoid the problem of circularity. On this view,
normative judgments refer, simply, to normative entities: reasons, full stop. In so doing,
it shares a truth condition with realism: normative judgments of the form "r is a reason
for x to f' are made true by the fact that r is a reason for x to if>. But this doesn't render
the resulting view any less constructivist: constructivism can be reinvented as a
distinctive metaphysical position. As Street urges, the fact that r is a reason for x to if> is
constituted by the scrutiny-withstandingness of the relevant normative judgments. If
we accept this version of constructivism, constructivism needn't be committed to a
circular semantic analysis of normative judgments.
There are two problems with this response. First, even if constructivism is construed
as a metaphysical thesis, to avoid this problem it must also reject Relational. But though
this is a position in logical space, many constructivist views do, in fact, accept Relational.
If so, though this needn't be a problem for every constructivist view, it remains a
problem for a wide range of them.
Indeed, Street herself suggests explicit commit-
ment to Relational at various points:
According to constructivism ... for a normative judgment ... to be correct is for it to stand up to
the specified sort of reflective scrutiny; the normative judgment's correctness is constituted by the
fact that it withstands this scrutiny. To speak more metaphorically ... the standards that deter-
mine the correctness or incorrectness of the normative judgment in question are thought to be
"given from within," or "legislated by," some further practical standpoint: to be correct is to
withstand scrutiny from that standpoint.
Street 2008, p. 223.
See Street 2008, pp. 241-2.
" Indeed, it seems to me that specifying constructivism as a theory of what makes nonnative judgments
true rather than as a metaphysical thesis of that which makes up reasons is independently motivated: the
fonner view can claim an additional advantage of ontological parsimony over nomtative realism.
Street 2!Xl8, p. 209. Of course, Street here uses the term "correctness" rather than "truth," but this point
is a red herring. Even if correctness is distinct from truth, it seems to me no less platitudinous to claim that the
correctness of a given judgment is a funt·tion of a proper relation between the meaning of that judgment and
extra-linKUistJc reality.
Second, even if we allow constructivism to reject Relational, and hence to avoid
semantic circularity, a constructivist metaphysics of reasons still has problems at the
semantic level: given that (2) and (3) ascribe a property that is metaphysically con-
stituted by the scrutiny-withstandingness of (2) and (3), the problem of indeterminacy
arises again. The problem is this: any standard of correctness established by the
ascription of a normative fact, as in (2), must be determined by that which constitutes
this normative fact. But the metaphysical substratum, as it were, of the normative fact
ascribed by (2) is the scrutiny-withstandingness of (2) itself. Hence any "standard of
correctness" set up by (2) must be set up by (4), i.e. the claim that (2) withstands
the scrutiny of (3). But, as we have seen, the claim that (2) withstands the scrutiny of
(3) sets up no standard of correctness whatsoever that conflicts with the property
ascribed in (3), which is itself constituted by the fact that that (3) withstands the scrutiny
of (2). If so, it is again difficult to see how (2) and (3) might fail to withstand the
scrutiny of each other: they do not ascribe properties that are mutually inconsistent or
even in tension.
The root of this problem is semantic. To vindicate the claim that (2) does not
withstand the scrutiny of (3), one must offer a semantic analysis of (2) and (3) that
renders (2) and (3) in some sort of tension. On the current proposal, (2) and (3) refer
to "reasons." But if reasons are metaphysically constituted by the simple scrutiny
withstandingness of (2) and (3), problems arise again. In referring to a property
constituted by (4), (2) can establish no standard of correctness that is incompatible
with (3). After all, to establish such a standard of correctness, the metaphysical
constituent of the property ascribed in (2) must be in tension with the metaphysical
constituent of the property ascribed by (3), i.e. (5). But (4) and (5) are most certainly
not in tension. If so, (2) fully withstands the scrutiny of (3) and vice versa. To solve
this, one must adopt a different semantic analysis of (2) and (3), and hold that they
refer to a more robust property than "reasons" (on the current analysis), e.g. some
judgment-independent property possessed by the fact of mass deaths, sufficient to
rule out (3) on the assumption of (2). But this semantic proposal, though it is
sufficient to deliver the plausible claim that (3) does not withstand the scrutiny of
(2), is obviously incompatible with constructivism. On such a proposal it would be
hard to see how constructivism's metaphysical analysis of reasons would avoid
obsolescence: normative judgments do not refer to properties that are made up by
the scrutiny-withstandingness of judgments. In addition, if normative judgments like
(2) and (3) refer to robust, judgment-independent properties, and if we accept the
semantic theory of truth, (2) and (3) will be true if and only if these judgment-
independent properties hold. Their scrutiny-withstandingness be would simply be-
side the point.
To sum up: rejecting Relational does not solve constructivism's problems. For the
remainder. then, I return to consideration of variants that accept Relational.
5. Good news and bad news
So long as the meaning of a judgment must bear "the right relation" to its truth
condition, a constructivist, who defines the truth conditions in terms of a normative
judgment's bearing a favored relation to other normative judgments, is forced to offer a
semantic analysis of normative judgments that, even independent of its circularity, is
problematic indeed. For constructivism to survive, we must break the link between the
meaning of normative judgments and their truth conditions. And, as far as I can tell, the
only way to do this is to reject a semantic theory of truth.
This is good news and bad news. The good news is that one can reject a semantic
theory of truth without absurdity, and can thus maintain a constructivist metaethic in
the wake of the problems I note here. I argue for this latter claim in the next two
sections. The bad news, however, is that a semantic truth predicate has been thought to
be simply platitudinous, a feature of the mere concept of truth. If so, many will claim that
despite the non-absurdity of the alternative I propose, any revision to our understand-
ing of truth is just too radical a move to be justified simply to save constructivism. For
my purposes, however, I am satisfied with a conditional conclusion: if we are to accept
constructivism, we must accept my alternative approach to truth in the normative
domain, or an alternative equally incompatible with a semantic theory. In addition,
I hope to have shown that for constructivism there is a way forward, whether or not
there are others, and whether or not the going is too rough.
6. Constructivism, coherence, and truth
Constructivists must reject a semantic theory of truth. What, then, should they accept?
To answer this question, it is helpful to say a bit more about the nature of constructiv-
ism as a general view of practical reasons. Though this categorization is imprecise,
constructivism is best thought of as a form of idealism about normativity. For the
constructivist, practical reasons are simply a product of the various relationships
between our own ideas or judgments about normative facts. Constructivism's general
idealism helps to guide the choice of truth predicates, however. Idealists have generally
been attracted to a coherence theory of truth. I suggest we explore this traditional pairing.
On this view, (1), for instance, is true if and only (1) is coherent with the other
normative judgments to which I am committed.
Before I investigate the advantages of a coherence truth predicate for the normative
domain, a few words must first be said about the coherence theory itself. First, such a
view would defme truth for normative judgments in terms of their coherence.
However, a problem arises. What is coherence? An adequate explanation of coherence
has been a classic sticking point for coherence theories in all domains. However, this
problem is less daunting given that we seek to offer a coherence theory in a way that
conforms to the general requirements of constructivism. If so, what it means for a set of
nonnative to be coherent will be defined by whatever favored relation is
referred to by Relational. For Street, for instance, coherence will be defined in tenns of
"scrutiny-withstandingness. "
I should address an immediate form of skepticism about a coherence theory of truth.
As Richard Kirkham writes: "If we believe that there is a world independent of our
thoughts, then no proposition that purports to describe that world can be considered
true if it is inconsistent with that world, no matter how well it coheres with other
propositions. And if it does express that world accurately, then it cannot be false, no
matter how much it fails to cohere."
Given, however, that we certainly believe that
constructivism as a general view is not true of, say, the domain of physics, or any other
non-normative domain, wouldn't it seem a wildly radical maneuver to adopt a
coherence theory on the basis of constructivism about normative judgments?
In response, the constructivist will say that Relational is only applicable to the
normative domain, and hence the coherence theory of truth itself is acceptable only
for the normative domain. This will commit constructivism to a position that has come
to be known as "alethic pluralism," the view that alternative domains of discourse can
permit of different truth-constitutive properties. The problem on the table arises only if
we assume that the property of coherence constitutes truth for judgments in all
domains of discourse. But we should not do this. A constructivist about normativity
is not committed to being a constructivist about all domains, including, say, the domain
of physics. In domains for which a form of ontological realism is appropriate, the
coherence theory of truth is inappropriate.
7. Questions
My proposal, then, is to reject a semantic theory of truth-not on the whole, but
simply for the normative domain-and replace it with a coherence theory of truth.
I hasten to add that this view is not novel. Similar views have been explored by
° Crispin Wright,
and Michael Lynch, who writes:
If, as we are assuming, moral judgment takes place within the space of reasons-we provide
reason and evidence for our moral beliefS--our moral judgments are subject to significant
rational norms. Consequently, a natural suggestion is that moral judgments are made true by a
property that is constructed out of those episternic norms ... Given coherentism's plausibility as
an account of the structure of warrant in morality it seems natural for anyone attracted to an
epistemic account of the property that makes moral judgments true to appeal to it.
For a more thorough description of how to make one's actual set of normative judgments into a
coherent set, see Dorsey 2006 and 2010.
Kirkham 1995, p. 111.
zu Quine 1986b, p. 63.
" Wright 1999, p. 227.

Lynch 2009, p. 164, p. 168.
Lynch offers one reason to accept a coherence theory in the nonnative domain: that
nonnative judgments are made true by the conditions by which they are warranted,
and they are warranted by their coherence. I am stricdy neutral on this motivation, as
I explore another: a method by which to salvage constructivism about practical reasons.
But whatever the motivation, there surely remains a number of unanswered questions
about this project. I address the most important ones here.
Question 1: It's all well and good to reject a semantic truth predicate. But don't
constructivists still have to accept a circular semantics?
No. Recall that constructivism was required to accept a circular semantics as a result of a
semantic truth predicate, which holds that a judgment is true if and only if its meaning
bears the "right relation" to extra-linguistic reality. But if we reject this claim, there is
no pressure to offer a semantic analysis for any normative judgment that "matches" that
normative judgment's truth conditions. If we know that a normative judgment is true,
and we know that this normative judgment is true if and only if it coheres with other
normative judgments, we are unlicensed to make any claims about its semantics. We
are licensed only to say that it coheres with other normative judgments, whatever their
semantic analysis may be.
Question 2: But then what sort of semantics should constructivists accept?
Given the problems stated in section 3, we must be given a semantics: what normative
judgments mean that doesn't simply reduce (2) and (3) to (4) and (5). And though
I hasten to add that constructivism cannot be totally neutral with regard to the semantic
choices it makes, it has wide latitude. Constructivists are licensed to select any semantic
programme they wish that can also deliver the proper verdicts concerning which
normative judgments withstand scrutiny by, or bear the favored relation to, other
normative judgments.
In particular, and most plausibly, constructivism is licensed to accept the "surface"
semantics as described by Brink, i.e. that normative judgments "ascribe (normative)
properties to persons, actions, policies, and so forth," and ascribe properties that are more
robust than simple scrutiny-withstandingness. Constructivists, for instance, could accept
the "normonic" semantics discussed in section 2: one could hold that "r is a reason for x to
c/)' means that a certain property--a "normon"-holds of r. The "normon" could be
understood naturally or non-naturally, perlups as a form of sui generis normative property.
The ascriptions of"normons" needn't be committed to any metaphysics that refers back to
normative judgments themselves, and hence these ascriptions can set up "standards of
correctness" just as any standard judgment-independent property ascription would.
One note: this semantic analysis seems to render a constructivist"s metaphysics of reasons, on<·e again,
obsolete: no nom1ative judgments actually refer to any entities made up of the scrutiny-withstandingness of
judgment•. But no matter. Constructivism remains a distinctive metaethical position in virtue of its a<·cep-
taru:e of Rtlat/cma/.
Question 3: Wouldn't this semantic analysis yield an error theory for normative
No. Judgments in a particular domain d pennit of an error theoretic analysis if and only
ifjudgments in domain dare truth-apt, and all judgments in domain dare false. Indeed,
as I have argued elsewhere,
far from succumbing to an error theory, accepting a
coherence theocy of truth for normative judgments is a novel method by which to
avoid an error theocy. Arguments for error theories generally start by offering a
semantics of judgments of the domain in question. In this case, normative judgments
refer to "normons." Next, error theories offer a metaphysical claim: "normons" do not
But the inference from the claim that no such metaphysically queer properties
exist to an error theocy-that is, that all normative judgments are false-goes through
only if we accept that nonnative judgments are judged true or false on the basis of their bearing the
"right relation" to such nonexistent metaphysically queer properties. But normative judgments,
on this view, are not judged as true or false in this way. Rather, they are judged true or
false on the basis of their coherence. Hence even if such normons don't exist, judg-
ments that ascribe them needn't be false if we judge the truth or falsity of the judgments
in question on the basis of their coherence.
Question 4: Isn't this a realist semantics? And if so, doesn't this mean that a coherence
theocy of truth-by your own admission-is inappropriate?
A "normonic" semantics of normative judgments is paradigmatically realist-a seman-
tics that refers to judgment-independent normative properties. If so, it would seem that
accepting a coherence theocy of truth for normative judgments--and their realist
propositional content-is explicitly accepting something I have so far deemed inap-
propriate: the combination of realism with a coherence theocy of truth.
This analysis, however, is mistaken. A coherence theocy of truth is inappropriate for a
domain d if and only if we believe that facts about domain dare facts about a "world
independent of our thoughts," to borrow Kirkham's phrase. To put this in another way,
realism and the coherence theocy are an inappropriate mix given realism's metaphysics,
not realism's semantics. Just because we accept a realist semantics of normative judgments
doesn't mean we have to conunit to evaluating normative judgments according to a
semantic theocy of truth. Rather, because constructivists do not believe that facts about
normativity are facts about a "world independent of our thoughts," they are licensed to
choose a coherence theocy, and evaluate normative judgments differendy. The seman-
tic analysis of normative judgments, then, says nothing about whether the truth of such
judgments is itself semantic. That is setded by further, metaphysical, questions about the
normative domain. Insofar as constructivism retains a realist semantics while rejecting a
normative "world independent of our thoughts," the constructivist is licensed to accept
a coherence theory of truth for that domain.
See Dorsey 2010.
Cf. Mackie 1977, pp. 36-42.
Question 5: Isn't the defmition of coherence circular?
Constructivism needn't be committed to a circular semantics. But a serious problem
looms. Consider the nature of coherence. Surely any plausible account of coherence is
going to be put in terms of the entailment relations between judgments, i.e. whether
judgments are inconsistent with others, whether they can be inferred from a set of
other normative judgments, etc. But entailment relations are understood in terms of
truth: for p and q to be inconsistent means that p and q cannot be true together. That
p entails q means that p cannot be true and q false, and so forth. Hence it would appear
that coherence is defmed in terms of truth. But because truth, at least for normative
judgments, is defmed in terms of coherence, any account of coherence will be viciously
This is a serious problem. After all, constructivism relies on a sensible understanding
of what it means for two judgments to "cohere" or "withstand scrutiny" or bear
whatever favored relation to each other. But if truth is defmed in terms of this favored
relation, and if this favored relation is defmed in terms of truth, we seem to be led to the
problem of indeterminacy by a back road: we seem unable to say with any certainty
whether two judgments will bear this favored relation, because we don't have any
informative account of what this favored relation is.
Accepting alethic pluralism, however, allows us to avoid a circular analysis of
coherence. Whether the members of any given set of judgments bear any particular
relation to each other is not a question that is properly evaluated in the normative
domain: the coherence of a belief set is a purely non-normative question. However,
because non-normative domains make use of a non-coherence truth predicate, the
coherence of any particular belief set is evaluated given a non-coherence account of truth. Let
me put this point in a slightly different way. Assume that non-normative truth-bearers
are susceptible to a semantic truth predicate. Because a judgment of the form "system
of belief S is coherent" is not a normative judgment, this judgment is evaluated by
means of a semantic truth predicate; this judgment is true just in case its meaning bears
the right relation to extra-linguistic reality. But if that is correct, the coherence of Sis
determined by the various entailment relations between its constitutive judgments, i.e.
whether they can be true together, whether they set up "standards of correctness" that
others survive, etc. But because the domain in question is non-normative, the applica-
ble truth predicate is semantic. Of course, if we accept constructivism, all normative
judgments, as evaluated by a semantic truth predicate, will come out false (assuming a
"surface" or "metaphysically queer" semantics). But this does not mean they cannot be
evaluated for their various entailment relations given a non-coherence truth predicate.
Hence the defmition of "coherence" is not viciously circular. Truth, for normative
judgments, is understood in terms of coherence, which is thereby understood in terms
Kirkham 1995, p. 107.
of the ability of particular judgments to be true together assuming a non-coherence, or
semantic, truth predicate.
By way of a conclusion to this section of the chapter, let me sum up my positive
proposal and its rationale. The problems of circularity and indeterminacy are foisted
upon constructivism so long as constructivism accepts a traditional, semantic, theory of trnth.
Hence to avoid them, constructivist views must reject a traditional, semantic truth
predicate in favor of an alternative. My proposal: accept a metaphysically robust
"surface" semantics of normative judgments, together with a coherence theory of
truth for normative judgments. This account captures the heart of a constructivist view.
After all, a coherence theory of truth for normative judgments guarantees both Truth-
Apt and Relational. For the coherence theory, normative judgments are true if and only
if they are coherent with an agent's other normative judgments, where "coherence" is
defined according to the favored relation as specified by a particular constructivist view.
Thus a coherence theory of truth can capture the essential element of constructivism:
that the truth of a normative judgment is set by its bearing of a favored relation
(coherence) to other normative judgments.
8. The problems of alethic pluralism
Recall that constructivism need not claim that judgments in all domains permit of a
coherence truth predicate. Rather, the constructivist must claim that non-normative
judgments permit of a different truth predicate than normative ones.
This result is significant in itsel£ As it turns out, to avoid the problem of semantic
circularity, the constructivist must embrace alethic pluralism, and so must address the
various problems that confront it. This might be thought an unattractive predicament.
Alethic pluralism is accompanied by familiar problems. Three are worth noting here.
First, it is unclear that pluralism about truth can plausibly accommodate inferences
involving judgments permitting of different truth predicates.
Second, pluralism
cannot seem to provide a coherent analysis of logical connectives including e.g.
conjunctions, disjunctions, etc. Third, pluralism would seem to leave open the possi-
bility that semantically identical judgments can be evaluated according to different
truth predicates. But if so, how are we to determine when a particular judgment is to be
evaluated by e.g. a coherence or semantic truth predicate?
The point of this section is to introduce these problems as problems for a construc-
tivist view that require solutions: constructivists must view these as their problems.
I attempt to sketch solutions to the problems, but I stress that these are mere sketches,
and are certainly not the last word on the subject. Indeed, my solutions require
substantial retooling of some important bits of our philosophical conceptual scheme.

See Dorsey 2006, pp. 502-4. .. Cf. Tappolet 1997.
As I note in section 9, whether such retooling is ultimately acceptable in light of the
plausibility of constructivism is as yet up for grabs.
8.1. Mixed inferences and mixed connectives
Mixed inferences are a problem for truth pluralism because truth pluralism appears
unable to accommodate the sense in which valid inference is "truth preserving."
Tappolet writes:
(T]here is a simple and equally powerful objection to the claim that there is a plurality of truth
predicates. Consider the following inference:
(1) Wet cats are funny.
(2) This cat is wet.
Ergo, this cat is funny.
The validity of an inference requires that the truth of the premisses necessitates the truth of the
conclusion. But how can this inference be valid if we are to suppose with Crispin Wright that
two different kinds of truth predicates are involved in these premisses? For the conclusion to
hold, some unique truth predicate must apply to all three sentences. But what truth predicate
is that?'
This problem applies to a constructivist's truth pluralism. Take the following inference:
A. That a particular action causes pain is a reason for anyone not to do it.
B. Torturing babies causes pain.
C. Hence, anyone has a reason not to torture babies.
This inference is certainly valid. But notice that, given the proposal I suggest here, its
premises do not permit of the same truth predicate. But if an inference starts from
normative and non-normative premises, and results in a normative conclusion, what is
it preserving? Certainly not semantic truth, because the conclusion is evaluated accord-
ing to a coherence truth predicate. Certainly not coherence truth, either. One of the
premises is semantic.
Take now the problem of mixed connectives. Consider:
6. That cookies are delicious is a reason for me to eat them, and snow is white.
(6) is certainly true. But under which predicate? To say that it is susceptible to the
semantic truth predicate would seem to render the left-hand conjunct false (at least on a
constructivist view). To say that it is susceptible to a coherence truth predicate would
seem to render us idealists about the whiteness of snow. This presents a problem.
Which truth predicate evaluates mixed connectives? It seems the answer is: none at all.
One possible solution to this problem is to accept a version of alethic pluralism
Michael Lynch calls "truth functionalism." According to Lynch, to be true is to bear a
property that plays the "truth-role," where the "truth-role" can be distinct in different
Tappolet 1997, pp. 209-10.
domains of discourse.
Here truth is a JUnctional property that is preserved in valid
inference. It just so happens that what plays the "truth-role" may be distinct for
different premises of valid mixed inferences. (Lynch's solution to the problem of
mixed connectives requires holding that a different property plays the truth-role for
connectives: the property of its truth-value being "grounded" in its atomic proposi-
I have no particular problem with Lynch's solution. But I want to try out a
somewhat different suggestion. My proposal begins by introducing a new truth-like
predicate. Call this predicate "truth*. "
A truth-bearer is true* if and only if that truth-
bearer is true as evaluated by the truth predicate appropriate for its domain. A non-normative
truth-bearer is true* if and only if extra-linguistic reality is as the truth-bearer says it is.
A normative judgment is true* if and only if that judgment coheres properly with the
rest of the speaker's normative judgments.
Now take the problem of mixed connectives. The introduction of truth* can solve
this problem with a mere tweak of the semantics of connectives. Instead of claiming
that a conjunction means that both conjuncts are true, we say that a conjunction means
that both conjuncts are true*. Hence (6) is true if and only if each conjunct is true*, i.e.
"That cookies are delicious is a reason for me to eat them" coheres with my other
normative judgments, and whether "snow is white" bears the right relation to extra-
linguistic reality. Given this analysis of connectives, any connective can properly be said
to be susceptible to a standard, semantic truth predicate. Whether any particular
judgment, or other truth-bearer, is true* is a non-normative question, settled simply
by facts about the coherence of my normative judgments. Hence it would appear that a
slight alteration in the semantic analysis of connectives allows us to plausibly assign a
single truth predicate, i.e. a semantic truth predicate, to all connectives (including mixed
connectives). Though, on a semantic theory of truth, "That cookies are delicious is a
reason for me to eat them" might very well be false (depending on its semantic
analysis), it is, nonetheless, true*, i.e. true as assessed by the predicate appropriate to
its domain.
If we accept this account of mixed connectives, the proper account of mixed
inferences is trivial. We can perfectly preserve the claim that valid inferences are
truth preserving if we understand that which inferences preserve to be truth* rather
than truth. However, given the definition of truth*, this guarantees that, for valid
mixed inferences, whenever the premises are true, the conclusion is true as well.
Though I believe either solution succeeds, I am neutral regarding the relative merits of
'" Lynch 2009, Chapter 4.
Lynch 2009, pp. 90-1.
" See Dorsey 2006, pp. 504-{); for what it's worth, I take my solution to the problem of mixed inferences
expressed there to be unnecessarily comp!t:x, though the thmst of my solution stands .
.., It is important to note here that this way of understanding mixed inferences puts constraints on a proper
understanding of coherence. In particular. it must be the case that coherent nomtative judgments must also be
consistent with true non-nomtative claims. But I take this to be an uncontroversial requirement.
my proposal versus Lynch's truth functionalism. Suffice it to say, however, there is at
least a solution.
8.2. Normative and non-normative normative judgments
The third problem runs as follows. Consider:
7. That this suit is well made is reason for me to purchase it.
If we accept a "normonic" semantics for judgments like (7), (7) would appear to mean:
8. Normonic property "reason for me to purchase this suit" holds of the fact that
this suit is well made.
But (8), quite obviously, can be assessed non-normatively. And if we wish to deny that
normons exist, (8) is false. Hence (7) is false. But, if we assume that (7) withstands the
scrutiny of my other judgments about reasons, (7) is true. But if this is correct, (7) is
both true and false. How could this be?
Of course, (7) is both true and false. But this isn't as strange as it sounds. Such a claim
is absurd only if the truth predicate used to evaluate a proposition as true and false is
identical, i.e. if the domain of discourse is the same. But, quite obviously, (7) is
evaluated as true in the normative domain, false in a non-normative domain. In this
way, (7)'s truth and falsity might be compared to those of another sentence:
9. Sherlock Holmes lives at 221b Baker Street.
In discussing.fiction, we are inclined to say that (9) is true. But when in a non-fictional
domain of discourse, we are tempted to say that (9) is false. But this is not mysterious.
Neither is it in the case of (7). Of course, one might wonder what distinguishes the
normative use of (7) from a non-normative use of (7). Though I don't have much of a
theory to offer here, it seems to me that the same thing said about (9) should be said
about (7), viz. that it depends not on the semantics of a given judgment, but rather its
pragmatic elements, as well as the context of utterance.
Though I don't have the
space to develop this suggestion any further, it seems to me that something like this is
roughly correct.
9. Conclusion: Isn't this all just wildly ad hoc?
Before I conclude, a nagging objection should be dealt with. I have offered a solution
to the problem of semantic circularity for constructivism that rearranges our under-
standing of the nature of truth in the normative domain. Not only this, to accommo-
date the truth pluralism this requires, I suggested a number of further semantic
>• See, fOr instance, Azzouni, 2010, pp. 251-2, for a helpful analysis of the fictional case.
maneuvers, including an alternative semantic analysis of connectives. Isn't all this just ad
hoc? Is there any independent reason to believe that any of this stuff is actually true?
This objection should be broken down into two different questions. First: is there
any reason, independent of the prior acceptance of constructivism, to accept a coher-
ence theory of truth in the normative domain, and its theoretic accoutrements?
Second: if the answer to the first question is no, is this an objectionable feature of
the solution I propose?
Take the first question first. I think there is, indeed, independent reason to believe
that a coherence theory of truth in the normative domain, along with its alethic
pluralism, succeeds. As I note above, I have argued elsewhere that such a move
plausibly eliminates the spectre of an error theory.
Furthermore, I have argued that
a coherence theory for normative judgments can be motivated by certain Quinean
reflections about the relationship between normativity and observation.
there are a number of independent motivations for a more general alethic pluralism,
beyond constructivism.l
Assume for the purposes of argument, however, that these
prior motivations fail. Let's say that the only genuine motivation one might have for
accepting the view I propose here is a prior commitment to constructivism. This raises
the second question: is this problematic for the solutions I propose?
Well, yes and no. It may very well be that there is pressure not to alter, e.g. the
semantic truth predicate, the semantics of connectives, etc. But it is important not to
overstate this pressure. The conceptual, linguistic, or logical apparatus proposed for any
domain is only as acceptable as its ability to allow us to say things we wish to say about
the domain in question. In other words, if we find constructivism about the nature of
practical reasons plausible, this should be all the evidence we require to reject a
semantic theory of truth. If so, the acceptability of my positive proposals will depend
on the strength of the metaethical argument for constructivism. But this is as it should be.
Allow me to put this point in slightly different terms. The problem for constructiv-
ism is serious: under a semantic theory of truth, constructivism must accept a semantic
analysis of normative judgments that is seriously problematic for the constructivist
project as a whole. Hence, to salvage constructivism, we should accept an alternative
theory of truth for normative judgments. Any such alternative (including my own
favored proposal) raises a number of complications which, though I believe they can be
adequately dealt with, require some reorganization of our philosophical conceptual
scheme. At least for the sake of argument, I concede that none of these maneuvers is
motivated independently of constructivism. Hence at this point we have a choice: we
can either accept these technical suggestions as the price of accepting constructivism, or
we can reject these technical suggestions and with them reject constructivism. Which
alternative we should choose depends fully on the strength of the argumentforconstruc-
tivism. If this argument is good enough, there should be no hesitation to reject a
" See Dorsev 201 0. " See Dorsey 2006.
Lynch 2009, Chapter 2.
semantic truth predicate in the nonnative domain. Such a move is not ad hoc, but is
simply a necessary corollary of a metaethical view we have reason to accept. Never-
theless, as I do not mean to assess the strength of the argument for constructivism here,
I leave its all-things-considered fate an open question.
Alston, William P. 1996. A Realist Theory ofTrnth (Ithaca, NY: Cornell University Press).
Azzouni, Jody. 2010. Talking about Nothing (Oxford: Oxford University Press).
Blackburn, Simon and Simmons, Keith. 1999. Trnth (Oxford: Oxford University Press).
Brink, David 0. 1989. Moral Realism and the Foundatiom of Ethics (Cambridge: Cambridge
University Press).
Dorsey, Dale. 2006. "A Coherence Theory of Truth in Ethics," in Philosophical Studies 117,
pp. 493-523.
Dorsey, Dale. 2010. "Truth and Error in Morality," in Wright and Pedersen 2010, pp. 235-48.
Dworkin, Ronald. 2011.Justicefor Hedgehogs (Cambridge, MA: Harvard University Press).
Gibbard, Alan. 2003. Thinking How tc Live (Cambridge, MA: Harvard University Press).
Kirkham, Richard. 1995. Theories of Trnth (Cambridge, MA: MIT Press).
Lynch, Michael. 2009. Trnth as One and Many (Oxford: Oxford University Press).
Mackie, J. L. 1977. Ethics: Inventing Right and Wrong (New York, NY: Penguin).
Quine, W. V. 1986a. Theories and Things (Cambridge, MA: Harvard University Press).
Quine, W. V. 1986b. "On the Nature of Moral Values," in Quine 1986a.
Shafer-Landau, Russ. 2003. Moral Realism: A Defence (Oxford: Oxford University Press).
Shafer-Landau, Russ. 2008. Oxford Studies in Metaethics, volume 3 (Oxford: Oxford University
Street, Sharon. 2008. "Constructivism about Reasons," in Shafer-Landau 2008, pp. 207-45.
Tappolet, Chrisrine. 1997. "Mixed Inferences: A Problem for Pluralism about Truth Predicates,"
in Analysis 57, pp. 209-10.
Tarski, Alfred. 1999. "The Semantic Conception ofTruth," in Blackburn and Simmons 1999,
pp. 115-43.
Wedgwood, Ralph. 2007. The Nature ofNormativity (Oxford: Oxford University Press).
Wright, Cory and Pedersen, Nicolaj. 2010. New Waves in Trnth (New York, NY: Palgrave
Wright, Crispin. 1999. "Truth: A Traditional Debate Reviewed," in Blackburn and Simmons
1999, pp. 203-32.
Constructivism and the
Argument from Autonomy
Robert Stern
My aim in this paper is to consider a particular line of criticism that has been used by
constructivists to argue against moral realism, which is to claim that if moral realism
were true, then this would threaten or undermine our autonomy as agents. I will call
this the argument from autonomy. The basic criticism can be found in the work of many
constructivists, where the core idea is helpfully summarized by Stephen Darwall as
follows: "For a Kantian such as Korsgaard, the idea of an independent order of
normative fact is inconsistent with the autonomy of the will. Practical reasoning is not a
matter of orientating oneself properly in relation to some external source of value.
Rather it is a self-government or autonomy-the agent determining herself in accord
with principles she can prescribe for herself as one rational person among others."
Now, I take it that if this line of criticism could be successfully substantiated, this
would be of great philosophical significance in relation to the debate between con-
structivists and realists. For, while the constructivist can raise and has raised other kinds
of criticism against the realist, most especially perhaps the familiar objection of being
"anti-naturalist", such criticisms have brought equally familiar responses, where either
the charge is accepted with equanimity,
or the whole naturalism/anti-naturalism
distinction is questioned by the realist,
so that in this way the objection is sidestepped.
On this and other issues, therefore, the constructivist/realist controversy may perhaps
be said to have petered out, as each side seems to just talk past the other. The interest of
the argument from autonomy, therefore, might be felt to lie in its potential to offer a
criticism of realism that will be more difficult for the realist to shrug ofT, as autonomy
looks on the face of it to be something that the realist will want to preserve as much
as will the constructivist, making it harder for her to respond by biting the bullet,
or denying that the notion of autonomy makes any clear sense. Of course, in this
Darwall1997, p. 310. • Cf. Lamtore 2008. • Cf. McDowell1994.
• This is not to deny that the realist can also respond more directly, by deknding the claim that realism is
compatible with naturalism.
debate, too, such evasive tactics can be adopted by the realist;
but the price of adopting
them is perhaps higher and more worrying to the neutral than it is when it comes to the
naturalism/non-naturalism issue, thereby making the argument from autonomy of
great potential value to the constructivist.
However, despite this potential, when one comes to consider the argument from
autonomy in any detail, and just what it amounts to, it is in fact surprisingly hard to find
it spelled out in any significant way in the literature; and likewise, it is correspondingly
hard to find any detailed response to it by realists. I think the reason for this is that,
curiously, each side takes it as so obvious that they are right, that they feel no real need
to say much more in defence or elaboration of the argument. Thus, constructivists just
take it as obvious that realism is a threat to autonomy and that this is a serious count
against it, while realists take it to be just as obvious that this threat is non-existent, and
that the argument can be ignored. So, on the one side, constructivists such as Rawls,
Schneewind, Korsgaard, and Reath have often deployed the argument from auto-
nomy,6 but not done much to explain it or defend it against criticisms, while on the
other side realists have not put themselves to much trouble to dismiss it.
Thus, debate
on this important issue has not in fact really got very far.
My aim in this paper, then, is to try to flesh out in more detail than is usual what the
basis for the argument from autonomy might actually be, and to consider then what
strategies the realist might use in response. I will therefore consider three different
broad types of "argument from autonomy" that might underlie the constructivist
position on this issue, and what line the realists could best take to deal with arguments
of these types. I will call the three types in question: Kantian arguments from autonomy;
Rortyean arguments from autonomy; and arguments from obligatoriness.
A final preliminary remark: While obviously relevant, I do not have the space here
to consider in any detail how the constructivist/realist distinction is best elaborated. But
just to give some general and basic characterisation of the distinction as I am under-
standing it for the purposes of this paper, I think the following account by Sharon
Street is helpful:
Thus, some have responded to the worry about autonomy raised in this and other contexts, by arguing
that what matters instead is a weaker notion that can more easily be accommodated by their view, such as self-
governance, orthonomy, or theonomy. See, for example, Hare 2001, pp. 114-15.
• Rawls's position will be discussed further below. Cf. also Schneewind 1986, p. 66: "The defining feature
of an autonomous agent, in Kant's view, is its ability to guide its own action by the choices of a will that is
such that whatever it wills is good simply because it is willed by it"; Kongaard 1996, p. 5: "If the real and the
good are no longer one, value must find its way into the world somehow. Fomt must be imposed on the
world of matter. This is the work of art, the work of obligation, and it brings us back to Kant ... The ethics of
autonomy is the only one consistent with the metaphysics of the modem world, and the ethics of autonomy
is an ethics of obligation"; Reath 2006b, p. 164, n. 17: "A constructivist account of Kant's moral theory offers
one way to speD out what it is for the 'rational wiU' to have autonomy."
So, for example, Ruu Shafer-Landau dismisses it in one page, calling it the weakest of the constructivist's
arguments against realism: see Shafer-Landau 2003, p. 44.
(T]he key point at issue between realists and antirealists is the answer to the central question of
Plato's Euthryphro (in roughly secular paraphrase), namely whether things are valuable ultimately
because we value them (antirealism), or whether we value things ultimately because they possess
a value independently of us (realism). In the final analysis, in other words, is normativity best
understood as conferred or recognized?
Metaethical constructivism fulls squarely on the antirealist side of this divide ... [M]etaethical
constructivism asserts a counterfactual dependence of value on the attitudes of valuing creatures;
it understands reason-giving status as conferred upon things by us. According to metaethical
constructivism, there are no facts about what is valuable apart from facts about a certain point of
view on the world and what is entailed from within that point of view.•
Working with this general characterization of the two sides in this debate, I will now
consider a first type of argument from autonomy that the constructivist might use
against the realist, which has a broadly Kantian inspiration.
1. The Kantian argument from autonomy
There is widespread consensus amongst constructivists that Kant should be credited as
holding a constructivist position in ethics at least partly on the strength ofhis commitment
to autonomy, and thus that Kant was committed to something like the constructivist's
autonomy argument. This is pretty clearly Rawls's view, and I think it is his view which
has influenced so many others. Thus, Rawls comments in Political Liberalism:
Another and deeper meaning of autonomy (than "doctrinal autonomy"] says that the order of
moral and political values must be made, or itself constituted, by the principles and conceptions
of practical reason. Let us refer to this as constitutive autonomy. In contrast with rational
intuitionism, constitutive autonomy says that the so-called independent order of values does
not constitute itselfbut is constituted by the activity, actual or ideal, of practical (human) reason
itsel£ I believe this, or something like it, is Kant's view. His constructivism is deeper and goes to
the very existence and constitution of the order of values. This is part of his transcendental
idealism. The intuitionist's independently given order of values is part of the transcendental
realism Kant takes his transcendental idealism to oppose.•
and his position is echoed in passages from others such as the following:
Kant's conception of autonomy precludes such a (realist] conception of morality. Human beings
cannot be dependent upon anything distinct from their will for the moral law that binds
• Street 2010, pp. 37(}-1. For similar accounts of the distinction, cf. also Cullity and Gaut 1997b, p. 4, and
Hills 2008, pp. 182-3.
• Rawls 1993, pp. 99-100. Cf. Neiman 1994, p. 33: "Having declared that reason is in the world, Leibniz
is stuck with the fact that reason is in the world-to be read off of, rather than put into, the objects of
experience. Naturally, those objects are not the everyday ones to which empiricists appeal but the supersen-
sible truths of an intelligible world. For Kant, however, the detemrination of reason by eternal truths is as
fundamentally heteronomous as its detennination by any other object"; and Lafont 2004, p. 28: "[Kantians]
agree with the anti-realist that our normative judgments do not purport to describe a pre-given moral order,
heteronomously imposed on us independently of our practical reason."
them ... If the distinctive feature of Kant's moral theory is autonomy, and if autonomy requires
the dependence of moral principles upon the human will, and if this dependence on the human
will is idealist, then the distinctive feature ofKant's moral theory is its idealism. Kantians ought to
embrace moral idealism as the distinctive feature of Kant's moral theory.
We might therefore expect to find in Kant some material by which the argument from
autonomy is to be elaborated.
Now as it happens, however, the interpretative position taken here by the con-
structivists has in fact been strongly challenged by realists, who have argued instead that
Kant is best understood as being on their side in metaethics, at least to a significant
If this is indeed so, then of course constructivists are wrong to look to Kant for
their inspiration for an argument from autonomy. While myself sympathizing with the
realist reading,
however, I do not intend here to enter into this complex interpreta-
tive debate, and will take it for granted that the constructivist can appropriately look to
Kant as some sort of ally in this area.
A first argument we might consider, then, under the broad heading of Kantian
constructivism, is one that has been put forward by Robert johnson, which is modelled
on the argument for voluntarism that has been put forward by theists:
First, let me give you an intuitive sense of why autonomy commits Kantian ethics to denying that
value is a source of reasons. Consider a parallel example: the Divine Command theory's
resolution of a Euthyphro-style dilemma. That dilemma begins with the assertion that God
loves (or responds in some appropriate way to) all and only good things. This raises the question,
Why? Is it because their value provides a reason for God to give them their due, His love? But if
God loves a good thing only because its goodness gives him a reason to love it, then its goodness
explains the appropriateness of God's love for it, and this is incompatible with God's omnipo-
tence. The value that provides a reason for God to love it would be a constraint on God's love in
the sense that God must respond to reasons provided by the value of things or else fail to have the
requisite response. The alternative is to say that those things are good because of God's love for
them. God's love explains their value. But then goodness looks like an arbitrarily distributed
shadow cast by God's attention. For if God loved some entirely different set of things, then those
other things would have been good. The Divine Command theorist opts for the second hom,
and then is saddled with the problem of explaining why goodness isn't arbitrarily distributed
after all.
Kantians must resolve a similar Euthyphro-style dilemma in the same way as the Divine
Command theory. What possess value on the Kantian view are all and only the objects of rational
agency. Now if value is the source of the reasons for the pursuits of rational agents, then the
authority governing rational agency is external to that agency itself, in the value of things that are
its objects. But on the Kantian view, rational agency must be autonomous, in the sense that the
requirements binding it are wholly self-generated and self-imposed. The autonomy of reason,
the central guiding idea behind Kantian moral theory, is thus the very foundation of the case
•• Rauscher 2002, p. 496.
See, for example, Ameriks 2003b; Hare 2001, pp. H7-ll'J; Hills 2008; Kain 2004, 2006; Langton2U07.
" See Stem 2012, esp. ch. 1.
against the claim that there is some value that provides reason to conform to moral obligation.
Autonomy requires that value not be a source of reasons.
The central claim seems to be, then, that just as value realism would clearly undermine
God's onmipotence in the theistic case, so it would clearly undermine our autonomy
in the case ofhuman beings.
However, the realist can, I think, argue in response to this challenge that there is a
significant difference between omnipotence and autonomy, however we exactly spell out
the latter, and that this is apparently being overlooked by Johnson. Thus, while it may
indeed be plausible to hold that God's omnipotence would be undermined if values
obtained independently of his will (though of course even this could be denied), given
that autonomy is a much weaker notion that onmipotence, nothing so far shows that
the parallel argument has any such plausibility and that to model the argument on the
theistic case is highly misleading for that reason. The realist may therefore grant that
Johnson is right that realism would undermine onmipotence in the theistic case, but
still feel she has been given no grounds on which to think it undermines autonomy in
ours. For, suppose the goodness of something did indeed provide God with a reason to
love it, such that he would be rationally criticisable were he not to do so; there would
then be some sense in which he "must" respond to that thing with love, based on this
reason. Perhaps Johnson is right that then God would then have to fulfll certain
requirements in a way that would limit his "powers" in some way and thus threaten
his onmipotence; but (the realist will argue) how can it be that having to follow a
reason in this manner in his actions would make him less autonomous, any more than
having to follow what there is reason to think does so?
The constructivist could respond, however, by suggesting that if we look a bit more
closely at Kant's position, a better argument from autonomy can be constructed on
Kantian materials-and in particular, on Kant's discussion of the heteronomous nature
of what Kant calls material moral principles which (in contrast to his own formal
formula of universal law and related formulae) employ some notion of value and the
good as their basis-for example, the good ofhappiness or perfection. These principles,
it could be argued, involve a realist notion of value, so we might look to Kant to show
how it is that this realism would lead to heteronomy were such principles to be
What, then, makes these material moral principles heteronomous for Kant? The
answer seems to be that in acting from them, the will is determined by our inclinations
and desires, not solely by our reason, so that it is because we want or desire our
happiness or perfection that we are led to follow the principles of a hedonistic or
perfectionist ethics. So we can take Kant as arguing that because certain ethical
positions ground their position in some conception of what is good for us, and
motivate our attachment to that good by appeal to our desires, that they result in an
" Johnson 2007, pp. 140--1.
outlook that is heteronomous. The constructivist's claim might be, therefore, that any
form of realism will have to end up giving a priority to desire over reason in this way,
and thus end up in a heteronomous position. An argument of this sort has been
proposed by Christine Lafont, where she writes that:
The standard reason that Kantian constructivists adduce against any kind of moral realism is
always the concern that any concession to realism unavoidably involves introducing heterono-
mous considerations about what human beings happen to want or desire which are incompatible
with the crucial role that the notion of autonomy plays in Kantian moral theories.••
Thus, following someone like Lafont, Kant might be interpreted as saying that if we
adopt value realism, then one will be forced to claim that this value gives one reason to
act and motivates one's actions only because of the satisfaction to be gained thereby;
but then reason can only play a subordinate role, in working in the service of desire to
help it fmd the best means of attaining this end; so reason loses its autonomy; so the will
as a whole is no longer autonomous.
Now, Kant certainly does seem to have thought that if reason is just used by a person
to "service" or in subordination to their desires, such a person is not fully autono-
mous;15 and this would seem to be one premise that the Kantian constructivist needs
here. Another premise she needs is that the value realist will must end up treating
reason in this way, as a general claim about all values the realist might propose. Thus, in
response to the Kantian constructivist's argument here, the realist can object to either
of these claims.
To resist the first premise, the realist might adopt a more Humean picture of reason
as "the slave of the passions", and argue that there is nothing in this picture that
undermines our autonomy as agents, provided other more familiar Humean constraints
relating to our desires are met (for example, that our second-order desires are in line
Lafont 2004, p. 35.
C[ Kant 1996b, p. 90 [4:441]:
only when it does "not merely administer an interest not belonging to it" can practical reason "show its own
commanding authority as supreme lawgiving".
and Kant 1996a, p. 189 [5:61]:
"[The human being] is nevertheless not 50 completely an animal as to be indifferent to all that reason says on
its own and to use reason merely as a tool for the satisfaction of his needs as a sensible being" (cf. Kant 1996b,
pp. 5G-2
and Kant 1997, p. 266 [27:499]:
"[I]f, for example, the principle of universal happiness were to be the basis for detennination of the moral law,
it would be a question of how far our needs were satisfied in their entire totality by following these laws; but
here laws of nature are involved, and the moral laws would have to be subject to them, 50 that reason would
have to obey the laws of nature and sensibility, and that in a necessary fashion (for in the physical order this is
50 anyway). But this would obviously put an end to the autonomy of reason, and thus be heteronomy."
and fmally Kant 1997, p. 263 [29: 625-6]:
''Reason attends either to the interest of the inclinations, or to its own interest. In the first case it is
subservient, but in the other, legislative."
[Note: the references in square brackets give the detail• from the Akademie edition of Kant's works, by
volume and page number, where this is reproduced in the margin$ of most translations. I
with our first-order ones). Of course, it may be true that (so to speak) the faculty of
reason itse!fis not autonomous here, because it is working "in subordination" to desire;
but the Humean response might be that this is irrelevant to whether the person as a
whole is autonomous, where it is this that really matters here. Now, obviously, this raises
a host ofhighly complex issues in the philosophy of mind and action, which I do not
propose to go into here: but it does suggest that there is a possible line of response open
to the realist, at least, even if it is one that the Kantian might fmd objectionable.
When it comes to the second premise, however, some Kantians themselves have
seemed to question it: namely, so called "teleological" Kantians who think Kant
himself was a realist who had a conception of the good underlying his moral posi-
tion-namely the good of freedom, rational agency or the good will which forms the
basis of the principles of morality.
Kantians of this sort therefore hold that there are
some things of objective value, where these things make Kant's principles of morality
valid, but where the will is not thereby rendered heteronomous, as it is not related to
these values through desire but through reason instead. Thus, on this "teleological"
reading of Kant, it is the value of rational nature that gives you reason to respect it in
yourself and others; and as a rational agent, and hence respecter of reasons, this gives
one the motivation to act accordingly, without appeal to the heteronomous motiva-
tions of desire and inclination.
Now, again, I do not want to enter into the complex interpretative debate between
"teleological" interpreters of Kant and their opponents.
But in general, these teleo-
logical Kantians are simply exploring an option that seems to be open to the realist,
whether or not it is one Kant himself took: namely, of denying that the realist can only
relate value to the will through desire, in a way that would inevitably push realism into
this kind of heteronomy. The constructivist might say, of course, that without some
appeal to desire then the realist response here will leave moral action mysterious in some
way; but again there are familiar realist responses to this sort of worry, which hold that
only a misplaced Humeanism concerning the sources of moral motivation could make
it seem that any sort of gap here really arises. If this is right, therefore, there would again
seem to be resources that the realist can use to resist this form ofKantian argument from
autonomy, once the relation between reason, desire and action are seen in the way the
realist is likely to favour.
We are therefore yet to fmd a convincing argument from autonomy that can be built
on Kantian resources; but it may be that the constructivist can look elsewhere, so that
I now tum to some ideas found in the work ofRichard Rorty.
Cf. Hills 2008; Langton 2007. Cf. also Wood 1999, pp. 46--7, pp. 111-55, and Guyer 2000. Wood
makes the general view clear when he writes: "The content of the [moral] law is not a creation of my will, or
the outcome of any constructive procedure on my part. The law of autonomy is objectively valid for rational
volition because it is based on an objective end-the dignity of rational nature as an end in itself" (Wood
2008, p. 108).
For some responses, see Dean 2006 and Sensen 2009.
2. The Rortyean argument from autonomy
Rorty has argued that realism in general (not just moral realism) is authoritarian, as it puts
something over and above us to which our beliefs or actions are required to conform,
thereby leaving us in the position ofbeing subordinated to something else outside and
beyond us. Rorty argues that just as we should learn to give up thinking that we have a
duty to follow God's will as a sign of our greater intellectual maturity and self-
confidence, so too we should learn to give up thinking that we should conform our
thoughts or actions to "how the world is" or Reality (with a capital R), which is
somehow prior to us and governs how we should act or think. Rather, Rorty claims,
what we should really care about is not this independent order of things, but our fellow
human beings, and how we get along with them and secure agreement with them-in
his slogan, we should look for solidarity, not objectivity. Thus, Rorty writes:
There is a useful analogy to be drawn between the pragmatists' criticism of the idea that truth is a
matter of correspondence to the intrinsic nature of reality and the Enlightenment's criticism of
the idea that morality is a matter of correspondence to the will of a Divine Being. The
pragmatist's anti-representational account of belief is, among other things, a protest against
the idea that human beings must humble themselves before something non-human, whether the
Will of God or the Intrinsic Nature of Reality.'"
Given his general claim about the authoritarianism of realism in general, then, we can
see quite easily how he might extend this claim to moral realism in particular, and thus
argue that the latter is a threat to our autonomy.
It is not clear, however, that Rorty would take constructivism to be much better
than realism in this regard, at least when it comes to Kantian constructivism. In general,
Rorty tends to see Kant as a "transitional" figure, whose own notion of pure practical
reason is still tainted with the same motivations as realism, to set something "above"
us-where in Kant's case this is the rational or noumenal self:
We are often told by contemporary moral philosophers that Kant made a breathtaking discovery,
and gave us a vitally important new idea, that of moral autonomy. But I suspect that when Kant is
given credit for this discovery, we are using the term ambiguously. Everybody thinks autonomy
in the sense of freedom from outside imposition is a fine thing. Nobody likes either human or
divine tyrants. But the specifically Kantian sense of autonomy- having one's moral decisions
made by reason rather than by anything capable of being influenced by experience - is quite a
different matter ...
As Dewey saw these matters, the Kantian split between the empirical and the non-empirical
was a relic of the Platonic distinction between the material and the immaterial, and thus of the
theologico-metaphysical distinction between the human and the divine. Dewey thought this
Rorty 2006, p. 257. See also Rorty 199lb, pp. 21-34 and Rorty 2000, pp. 217-tll. Cf. Nietzsche 1974,
IC(;tion 344, p. 283: "[E]ven we seekers after knowledge today, we godless anti-metaphysicians still take our
fire, too, &om the Rame lit by a &ith that is thousands of yean old, that Christian faith which was also the faith
of Plato, that God is the truth, that truth is divine."
"brood and nest of dualisms," as he called it, should be swept aside, taking Plato and Kant with
For Rorty then, it seems, Kantian constructivism is at root as authoritarian in its
outlook as moral realism.
However, it might be said, some Kantian constructivists (such as Onora O'Neill)
could be held to be closer to Rorty than others (such as Christine Korsgaard, who does
seem inclined to hang on to the hierarchical view of the self that Rorty wants to
Moreover, some constructivists (such as Sharon Street
) are closer to Hume
than to Kant, and so may be more palatable to Rorty in this regard as a result (though
whether this move towards Hume makes it hard for the constructivist to avoid
collapsing their position into some sort of expressivism or subjectivism is currently a
subject of much discussion).
But what about Rorty's argument as such, whether or not constructivism itself
would fall foul of it? Is moral realism "authoritarian" and undermining of our autono-
my because it treats values in a realist manner? What one thinks about this depends, as
we have seen, on how plausible one finds Rorty's argument against realism in general,
and his claim that any position that makes it necessary to conform our minds or actions
to a subject-independent world is authoritarian as a result. John McDowell has argued
against Rorty that this claim does not hold, and that realism only becomes an
authoritarian position if it becomes a form of transcendent realism, that treats what is
real as an unknowable beyond to which we have no cognitive access, thereby cutting
us off from how things are:
The world as it figures in mainstream epistemology is a counterpart, not to just any idea of the
divine as non-human and authoritarian, but to the conception of deus absconditus, God as
withdrawn into a mysterious inaccessibility. A telling Deweyan protest against epistemology, as
practised in the Cartesian and British-empiricist style, can be cast as a protest against the idea of
philosophy as priestcraft, supposedly needed to mediate between this mundus absconditus and
ordinary human beings who aspire to knowledge of it.
The idea that inquiry is answerable to the world does not by itself commit us to believing that
there is a need for philosophy as priestcraft. We can accept that inquiry is answerable to the things
themselves and still suppose, correctly, that the resources of ordinary investigative activity can
suffice to put us in touch with the subject matter of investigation, without need of special
philosophical mediation. That is: we can follow Dewey in rejecting philosophy as priestcraft,
without needing to abandon the very vocabulary of objectivity. What we need to dislodge is the
idea of the world as withdrawn into inaccessibility, and that is quite another matter.
Rotty 2007a, pp. 188-9.
Cf. Rorty 2007a, p. 195, who quotes Korsgaard as writing that "the relation of the thinking self to the
acting self is one of legitimate authority" (Korsgaard 1996, p. 165).
" C( Street 2001!.
" Cf. Street 2010, and the discussion of her position in Ridge, this volume.
" McDowell 2000, pp. 11 o-tt.
So, McDowell argues, there is nothing authoritarian in realism as such, just a transcen-
dent realism which treats reality as a distant, unreachable realm, to which we are
expected to conform without knowing quite how or whether we have done so. If we
abandon realism in this form, however, McDowell argues that the realism we are left
with is thoroughly unobjectionable, and indeed a normal part of common sense. As he
puts it: "Acknowledging a non-human external authority over our thinking, so far
from being a betrayal of our humanity, is merely a condition of growing up."
sympathies are with McDowell in this debate, and indeed it is notable than when
Rorty himself comes to spell out what he finds objectionable about realism, it is the
inscrutable nature of reality as conceived of by the realist that he seems to be picking up
on, rather than its subject-independence per se.
However, even if this sort of response to Rorty is right, it might still be felt that a
problem can be raised here for the realists along Rortyean lines. For, it could be argued,
it is a feature of realism that, because it treats what is real here as mind-independent, the
possibility of our being in radical error always arises, so that moral realism must always
leave open the sceptical worry that our beliefS about moral matters are in error, given
the independence of the moral domain-and doesn't this make the moral epistemically
transcendent in just the way that raises the authoritarianism worry for Rorty? Con-
versely, the constructivist might argue that it is an advantage of their position that no
such gap can arise.
Now, of course, the realist can respond in turn that this is precisely one thing that
makes constructivism (and other forms of anti-realism) in ethics implausible, because
any sensible view must in fact allow for the possibility of error of this kind. I do not
wish to pursue this line here, however, as it is unlikely that either side will feel that its
opponent can avoid begging the question in their favour--the intuitions one has on
this matter tend to reflect too closely the intuitions that make one a realist or anti-realist
in the first place, in my experience. I think, however, the realist can take a less
ambitious line, and argue that while it is indeed true that according to realism, we
have no a priori guarantee that there is no gap between our moral convictions and
moral reality, this in itself is too weak a basis to generate any genuine moral scepticism
and thus any real doubt that the moral realm is in fact transcendent, in a way that would
lead to Rorty's authoritarian worries. For, the sceptic here is raising a mere logical
possibility that follows from realism, and not giving any evidence that we have in
fact got things wrong, for which some arguments to establish a source of error in
our moral convictions would have to be provided-where here it is not realism
simpliciter that would be generate the sceptical doubt, but those arguments themselves.
McDowell 2000, p. 120. Cf. also Larmore 1996, pp. 87-8.
» Cf. Rorty 2006, p. 257 and p. 258 .
.. C£ Rauscher 2002, p. 496: "Kant's notion of autonomy distinguishes Kant's cognitivism from other
cognitivist moral theories ... [On these latter theories) liJt might be the case that humans are wrong about
every moral belief they hold, since the truth of mo12l propositions is independent of human knowledge.
Kant's conception of autonomy precludes such a conception of morality."
Thus, just as in the external world case, realists have responded to sceptical worries
based on the mere fact of mind-independence by dismissing such doubts as unreal and
groundless in themselves (even though the possibility of error cannot be ruled out by
the realist as a logical possibility),
so the same strategy (I am suggesting) can be applied
here, in a way that shows how one can endorse realism, without spiralling into the kind
of sceptical doubt that would generate the concern about epistemic opacity that might
trigger and substantiate Rorty's worry about authoritarianism. To that extent, there-
fore, it would seem that the Rortyean argument from autonomy against realism falls
3. The argument from obligatoriness
We can now then turn, finally, to the third type of argument from autonomy that
I wish to consider, which I take to be more promising than any of those we have
discussed so far, but where again I think the realist has resources with which to
challenge the argument.
The intuitive idea behind this sort of argument, I take it, comes from a challenge
posed by the apparently special obligatory nature of morality and the moral "must":
morality is taken by many to be a matter of obligation, and to that extent a matter oflaws
or principles that bind or command us, and thus have imperatival force. It may therefore
seem to follow that morality as a system oflaws requires a legislator, so that morality can
only come about as a result of some act oflegislation which this legislator undertakes-
as Korsgaard has put this view, "Obligation must come from law, and law from the will
of a legislating sovereign; morality only comes into the world when laws are made."
But then, if we accept this line, it may seem that we are left with only two options:
either morality is legislated by us, or by some other sovereign authority, such as God.
Now, if we opt for the second alternative, the link with heteronomy may look very
because now it will seem that in following the dictates of morality as these
are conceived to exist independently of us, we are following the directions of another
will, as the institutor of that moral law. Faced with that alternative, it may then appear
that the Kantian constructivist is right, and that the only way to preserve our autonomy
is to endorse the first option, and to make us the legislators of the moral law, in a way
This approach can be found, for example, in the fallibilist pragmatism inspired by C. S. Peirce, and his
distinction between "real" and "artificial" doubt: see Peirce 1992b.
Korsgaard 1996, p. 23. Cf. also G. E. M. Anscombe's discussion of the "special moral sense" of"ought"
and its connection with the "law conception of ethics" in Anscombe 1991, p. 30.
Even taking this option, however, I do not deny that the divine command theorist may find a way to
preserve some conception of our autonomy; but for the sake of the discussion here, I will not consider the
issues any further, taking it that the worry has at least a prima facie force, and might well persuade people to
think they need to opt for the second, constructivist, option-where I am mainly concerned below with
challenging the exhaustiveness of this dichotomy and so avoiding the dilemma, rather than ways dealing with
the fir.;t (theistic) hom of it. For some further discussion of how the divine command theorist might response
to the problemli, see Hare 2001.
that takes us from considerations of autonomy to constructivism. Thus, as Robert
Pippin has put the general thought here: "Laws, to be laws, require legislators, and
once a divine legislator is excluded, 'we' are the only candidates left."
The realist
would thus seem to face the problem of accounting for the obligatoriness of morality,
in a way that makes this obligatoriness compatible with our autonomy, if they reject
constructivism and the Kantian idea of "self-legislation" and so deny that this obliga-
toriness stems from our own legislative act.
I now want to tum to a consideration of ways in which I think the realist can
respond to the argument from autonomy that takes this form. Some sorts of responses
are "offensive", in trying to show that the constructivist option here is itself incoherent
or deeply problematic in some way, so that there is really no constructivist route out of
the problem either, and so no grounds in this respect to prefer constructivism to
realism. Other sorts of response are "defensive", and try to show that really the issue
of obligatoriness does not arise as a problem for realism in the way that the argument
from autonomy tries to suggest.
Going on the offensive, the realist can try to claim that the constructivist position is
itself unable to show that it makes sense to treat self-legislation as a source of obligatori-
ness, and thus as a coherent alternative to realism. The first problem often raised
by realists here is the problem of emptiness: if moral reasons and norms do not exist
prior to the legislative acts of the will, how is such a will to be guided or constrained in
any way in its legislative activities, and this how can those activities be anything other
than empty and arbitrary?
However, constructivists have perhaps found a reasonable
answer to such concerns in the idea that some constraints are constitutive of the
will or practical reasoning itself, so that there is some content built into the
activity of legislating by such a will from the outset.
What those constitutive
constraints exacdy are, and whether they are still sufficiently contentful to do much
guiding, will still remain an issue for the constructivist to resolve; but in principle, at
any rate, this might be thought to provide an adequate response to the emptiness
At least two further problems for the constructivist remain, however, relating to the
autlwrity of norms that are self-legislated in this way. The first of these remaining
problems concerns whether self-legislating can ever really amount to a genuine kind of
legislating or binding-for if I bind myself, can't I unbind myself at will, leaving it unclear
how I was ever really bound in the first place? And the second, related, problem is raised
by Elizabeth Anscombe in the claim that "the concept of obligation requires superior
'" Pippin 2005b, p. 219.
" C£ Lannore 2008, p. 44: "[W]ben we do impose principles on ounelves, we presumably do so for
rea10111: we suppose that it is fitting for us to adopt them, or that adopting them will advance certain of our
interests. Self-legislation, when it does occur, is an activity that takes place in the light of rea.'Klns that we must
antecedendy recognize, and whose authority we therefore do not institute but rather find ounelves called
upon to acknowledge."
'" Cf. Korsgaard 2006, pp. 235-6.
power in the legislator".
But if it is right that some such hierarchical relation is needed
here, how can that be the case if I am legislating over myself, where it would seem that
both sides are on a par, so making the act oflegislation impossible?
In response to the f!rst of these worries, the constructivist might, I think, appeal again
to the idea of the norms in question being constitutive of reason or practical agency in
some way, where as a result ofbeing constitutive, it is not clear that reason or practical
agency could ever "withdraw" or alter them; but nonetheless, the constructivist could
argue, it still makes sense to think of them as self-legislated, as they get their normative
force through being constitutive of our agency.
In response to the second of these worries, Kantian constructivists have often appealed
to the hierarchical structure of the self that Kant himself seems to employ, which puts the
rational aspect of the person in a position of"superior power" over other aspects, which
enables it to have this legislating role. Thus, in this vein Korsgaard writes: "We might say
that the acting self concedes to the thinking self its right to government. And the thinking
self, in tum, tries to govern as well as it can. So the reflexive structure of human
consciousness establishes a relation here, a relation which we have to ourselves. And it
is a relation not of mere power but rather of authority. And that is the authority that is the
source of obligation."
A difficulty here, however, that might well be raised by
Anscombe, is that Korsgaard still talks about one part of the self conceding authority to
the other, which Anscombe might think gready dilutes the kind of power required,
arguably again making talk of legislation inappropriate. Nonetheless, Korsgaard could
reply that we are familiar with democratic models of sovereignty, where legislative
authority arguably is transferred from the people to a democratically appointed legislative
bodies in this way, and perhaps this is the model that the Kantian could use here.
We have seen, then, that notwithstanding the criticisms of the constructivist's "self-
legislation" model put forward by Anscombe and others, it can perhaps be made
coherent, so that the realist cannot rely on ruling out constructivism as a strategy for
defending her own view, and must instead seek to defend it from the argument from
obligatoriness in its own right. I will here discuss two possible lines of defence, both of
which take the general approach of denying that the realist must be committed to there
being an external lawgiver as essential to their realist conception of morality; so the
claim is that it is possible to remain a realist in ethics while denying that this realism
commits the realist to holding that the obligatoriness of morality has its basis in any
external law-giver, and thus in any source that might seem to threaten our autonomy.
The f!rst such approach is in many ways the simplest, and just questions the
argument's assumption that obligation requires a legislator at all, where the claim is
that just as values are independent features of the world, so too are the duties and
obligations to which they give rise. This would seem to be the position adopted by
Samuel Clarke:
" Anscom be 191! 1 b, p. 27. " Korsgaard 1996, p. 104.
rrJhese eternal and necessary differences of things make it fit and reasonable for creatures so to
act; they cause it to be their duty, or lay an obligation on them, so to do; even separate from the
consideration of these rules being the positive will or command of God; and also antecedent to
any ... particular private and personal advantage or disadvantage, reward or punishment, either
present or future.
Clarke seems to see no difficulty in being a realist about duties and obligations in a way
that does without the need for any legislating will, and thus without a lawgiver who
might threaten our autonomy; for, he held, just as it makes sense to speak of laws in
arithmetic and geometry without postulating a legislator for them, so the same can be
said of morallaws.
By dispensing with the role of an external lawgiver in this way,
therefore, the argument from obligatoriness based on the threat of such a lawgiver is
simply inapplicable to a position of his sort.
Now, the constructivist will no doubt feel, that just as there is something peculiar
and "queer" in the realist's conception of moral values, so there is something even
more peculiar and "queer" about treating duties and obligations in this realist manner,
where any explanation or account of this remarkable feature of the world would appear
to be lacking on the realist's picture. This, however, is to move from considerations of
autonomy, to the more standard naturalistic and explanatory concerns which (as we
have noted at the outset) have for some time formed part of the realist/anti-realist
debate in ethics, where the outcome of this controversy remains to be seen. What is
more relevant to our purposes here, is to ask whether, giving the realist the benefit of
the doubt on these questions, there are any special considerations of autonomy that can
be used against a position like Clarke's, once the role for an external lawgiver has been
dropped from the realist's picture and the need for any such lawgiver has been rejected? It
would seem that the argument from obligatoriness itself would not bite here, as that did
rely on an appeal to such an external lawgiver; and given the realist's apparendy cogent
responses to the other arguments from autonomy we have considered, it would therefore
appear that a realist's position such as Clarke's is immune to the constructivist's concern
with autonomy--at least until an argument somehow different from the one's we have
discussed comes along.
Nonetheless, we may assume that many will find Clarke's treatment of the obliga-
toriness of morality unenlightening at the very least, in seeming to have nothing to say
about how it comes about and what it consists in. Can the realist therefore do any
better on this front, without on the one hand postulating some sort of external lawgiver
•• Clarke 2003, p. 295. Cf. also Price 1948, pp. lOH: " ... virtue, as suth, has a real obligatory power
antecedently to all positive laws, and independently of all will; for obligation, we see, is involved in the very
nature of it." And cf. also Roger Crisp: "Oddly, Anscombe appean never to consider the view that claims
that we have such obligations might be self-standing. requiring no justification from elsewhere, though she
does c:onsider, as alternatives to divine legislation, the norms of society, self-legislation, the laws of nature.
Hobbesian contractualism, and the virtues. Perhaps, like the early Greeks. she also felt that a nomos had to be
rwmiztllli ('dispensed')" (Crisp 2004, p. 86, n. 33).
,. Clarke 2003. pp. 29H.
and thus raising the threat of heteronomy, or on the other hand treating morality as
stenuning from some sort of self-legislation and thus abandoning realism?
Now, ironically enough given his influence on constructivism, it could be argued
that one option for the realist to explore here is offered by Kant's treatment of the
contrast he draws between the holy or divine will, and one such as our own. For, in
drawing this contrast, Kant takes himself to be explaining the way in which for us, what
is morally right takes on an obligatory force, as we are faced by inclinations and desires
that go against the correct action, and which makes it appear to us that this right action
is demanded of us in an irnperatival manner ("You must not tell lies," and so on). Thus,
on the one hand, Kant offers an account of obligation that does not rest on any appeal
to an external lawgiver, while treating it as a result of our limited and human moral
perspective, where from the perspective of a divine or holy will, morality has no such
character; to that extent, therefore, his position might be seen as anti-realist. On the
other hand, however, unlike constructivism this is arguably not an anti-realism that
goes "all the way down" but incorporates an important element of realism-not about
how it is that morality appears obligatory to us, but about what it is that so appears,
where this is held in common between us and the holy will, as the thing that it is right
for both wills to do (tell the truth, example), but where this is a something that we
feel obliged to do but which the divine will does not, as it lacks the non-moral
inclinations that create the sense of obligation that we experience in our moral lives.
Kant can therefore be held to offer what might be called a "hybrid" view that
contains both anti-realist and realist elements, the former at the level of obligation and
the latter at the level of the right, where this enables the realist to avoid the threat of the
argument from obligation on the one hand, while avoiding anti-realism "all the way
down" on the other, while also offering some sort of account of the nature of
obligation in a way that eluded Clarke. The realist level concerns the content of
morality, what is right and wrong, and the value of freedom on which this rests.
Kant is insistent in many passages that not even God can determine by an act of will or
choice what this content is to be, as this is as fixed and necessary as the fact that a
triangle has three comers, which obtains independendy of any relation to an agent.
C£ Kant 1997, p. 76 [27:282-3):
"The lawgiver is not always simultaneously an originator of the law; he is only that if the laws are contingent.
But if the laws are practically necessary, and he merely declares that they conform to his will, then he is a
lawgiver. So nobody, not even the deity, is an originator of moral laws, since they have not arisen from
choice, but are practically necessary; if they were not so, it might even be the case that lying was a virtue. But
moral laws can still be subject to a lawgiver; there may be a being who is omnipotent and has power to
execute these laws, and to declare that this moral law is at the same time a law ofHis will and obliges everyone
to act accordingly. Such a being is then a lawgiver, though not an originator; just as God is no originator of
the fact that a triangle has three comers."
and Kant 1996c, p. 381 [6:228):
"A (morally practical) law is a proposition that contains a categorial imperative (a command). One who
commands (imp<Tans) through a law is the lawgiver He is the author of the obligation in
acconbnce with the law. but not always the author of the law. In the latter case the law would be a positive
He also speaks in tnany passages as if he conceives of the value of the free rational
agent in realist tenns.
However, as we have seen, Kant does not treat the obligatcriness
of what is right and wrong as independent in this way, for we give the content of
morality its obligatory form, in so far as this depends on our limitedness as finite
creatures, so that this obligatoriness is just the way in which what is right and wrong
presents itself tc us, from our human (all too human) perspective; from the perspective
of a divine will, and so from the "absolute standpoint", there is no duty and obligation,
but only what is right and wrong, because it has none of the non-moral inclinations
which Kant thinks render us subject to duties and obligations, and thus the moral
Kant is thus able to side with the realist about the right; but he is able to side
with the anti-realist about the obligatory, and thus avoid much of the "queerness"
associated with the idea that the world in itself makes demands on us, and also avoid the
threat to our autonomy that any account of such a demand that postulates an external
lawgiver would seem to imply.
Interpreted in this way, therefore, Kant might provide
a model for the sort of position that can claim to incorporate a significant element of
realism on the one hand, while on the other hand showing how (if this Kantian
account of obligatoriness can be accepted) the argument from autonomy based on
obligation can be avoided.
(contingent) and chosen law. A law that binds us a priori and unconditionally by our own reason can also be
expressed as proceeding from the will of a supreme lawgiver, that is, one who has only rights and no duties
(hence from the divine will); but this signifies only the idea of a moral being whose will is a law for everyone,
without his being thought of as the author of the law."
and cf. also Kant 1997, p. 302 (27:544]:
"Were we to conceive of the legislator as auctor legis, this would have reference only to statutory laws. But if
we ascribe an auctor to laws that are known, through reason, from the nature of the case, he can only be an
author of the obligation that is contained in the law. Thus God, too, by the declared divine will, is auctor legis,
and precisely because natural laws were already in existence, and are ordained by Him."
For further discussion of this material and related passages, see Hare 2001, pp. 94-7; Ameriks 2003b; Kain
2004; Irwin 2004; Kain 2006; Wood 2008, pp. 106-16.
Cf. Kant 1996b, pp. 78-9 (4:428):
"But suppose there were something the existence of which in itself has an absolute worth, something which as
an end in itself could be a ground of determinate laws; then in it, and in it alone, would lie the ground of a
possible categorical imperative, that is, of a practical law.
Now I say that the human being and in general every rational beings exists as an end in itself, not mmly as a
means to be used by this wiD or that will at its discretion."
Cf. Kant 2000, p. 273 (5:403-4]: " ... it is dear that it depends only on the subjective constitution of our
practical faculty that the moral laws must be represented as commands (and the actions which are in accord
with them as duties), and that reason expressed this necessity not through a be (happening) but through a
should-be ... "
I think Terence Irwin also sees Kant as adopting this sort of hybrid view: "[Kane] recognizes inainllic
rightness without any acts of commanding or obligating ... In Kant's view, commands and act ofbinding are
(only) relevant to fmite rational agents, who are also subject to other incentives and 10 have to be instructed
and urged to follow the moral law" (Irwin 2004, p. 149).
4. Conclusion
In this paper, I have set out to explore in some detail the threat to realism posed by
considerations of autonomy. I have argued that the best way to understand the
argument from autonomy is to relate it to the issue of obligatoriness; but that there is
a variety of strategies to be explored concerning obligation before it is clear that the
right response to this issue is a constructivist one, or that the realist is hereby compelled
to surrender their position.
Ameriks, Karl. 2003a. Interpreting Kant's "Critiques" (Oxford: Oxford University Press).
Ameriks, Karl. 2003b. "On Two Non-Realist Interpretations of Kant's Ethics," in Ameriks
2003a, pp. 263-82.
Anscombe, G. E. M. 1981a. The Collected Philosophical Papers of G. E. M. Anscombe: Volume Three:
Ethics, Religion and Politics (Oxford: Blackwell).
Anscombe, G. E. M. 1981b. "Modem Moral Philosophy," in Anscombe 1981a, pp. 26-42.
Brandom, Robert B. (ed.). 2000. Rorty and his Critics (Oxford: Blackwell).
Clarke, Samuel. 2003. A Discourse Concerning the Unchangeable Obligations of Natural Religion, and
the Truth and Certainty of Christian Revelation, partially reprinted in Schneewind (ed.), 2003,
pp. 295-311.
Crisp, Roger. 2004. "Does Modem Moral Philosophy Rest on a Mistake?" in O'Hear (ed.),
2004, pp. 75-94.
Cullity, Gareth and Berys Gaut (eds). 1997a. Ethics and Practical Reason (Oxford: Oxford
University Press).
Cullity, Gareth and Berys Gaut (eds). 1997b. "Introduction," in Cullity and Gaut (eds), 1997a,
pp. 1-28.
Darwall, Stephen. 1997a. "Reasons, Motives, and the Demands of Morality: An Introduction,"
in Darwall, Gibbard and Railton (eds), 1997b, pp. 305-12.
Darwall, Stephen, Allan Gibbard and Peter Railton (eds), 1997b. Moral Discourse and Practice:
Some Philosophical Approaches (Oxford: Oxford University Press).
Dean, Richard. 2006. The Value of Humanity in Kant's Moral Theory (Oxford: Oxford University
Guyer, Paul. 2000. Kant on Freedom, lAw, and Happiness (Cambridge: Cambridge University
Hare, John E. 2001. God's CaD: Moral Realism, God's Commands, and Human Autonomy (Grand
Rapids: Eerdmans).
Heller, Thomas C., Morton Sosna and David E. Wellbery (eds). 1986. Reconstructing Individual-
ism: Autonomy, Individuality and the Self in Western Thought (Stanford: Stanford University
Hills, Alison. 2008. "Kantian Value Realism," in Ratio 21, pp. 182-200.
Hyman, John and Helen Stewart (eds). 2004. Agency and Action (Cambridge: Cambridge
University Press).
Irwin, Terence. 2004. "Kantian Autonomy," in Hyman and Stewart (eds), 2004, pp. 137-64.
Johnson, Robert N. 2007. "Value and Autonomy in Kantian Ethics," in Shafer-Landau (ed.),
2007, pp. 133-48.
Kain, Patrick. 2004. "Self-Legislation in Kant's Moral Philosophy," in Archiv .for Geschichte der
Philosophie 86, pp. 257-306.
Kain, Patrick. 2006. "Realism and Anti-Realism in Kant's Second Critique," in Philosophy
Compass 1, pp. 449-65.
Kant, Immanuel. 1996a. Critique ofPractical Reason, translated by Mary J. Gregor, in Kant, 1996c,
pp. 133-272.
Kant, Immanuel. 1996b. Groundwork of the Metaphysics of Morals, translated by Mary]. Gregor, in
Kant, 1996c, pp. 37-108.
Kant, Immanuel. 1996c. The Metaphysics of Morals, translated by Mary]. Gregor, in Kant, 1996d,
pp. 363-604.
Kant, Immanuel. 1 996d. Prllcticlll Philosophy, translated and edited by Mary]. Gregor (Cambridge:
Cambridge University Press).
Kant, Immanuel. 1997. Lectures on Ethics, edited by Peter Heath and J. B. Schneewind, translated
by Peter Heath (Cambridge: Cambridge Univenity Press).
Kant, Immanuel. 2000. Critique of the Power of judgment, translated by Paul Guyer and Eric
Matthews (Cambridge: Cambridge University Press).
Korsgaard, Christine M. 1996. The Sources of Normativity (Cambridge: Cambridge Univenity
Lafont, Christine. 2004. "Moral Objectivity and Reasonable Agreement: Can Realism be
Reconciled with Kantian Constructivism?" in RAtio juris 17, pp. 27-51.
Langton, Rae. 2007. "Objective and Unconditional Value," in Philosophical Review 116,
Larmore, Charles. 1996. The Morals of Modernity (Cambridge: Cambridge University Press).
Larmore, Charles. 2008. The Autonomy of Morality (Cambridge: Cambridge University Press).
McDowell, John. 1994. Mind and World (Cambridge, MA: Harvard University Press).
McDowell, John. 2000. "Towards Rehabilitating Objectivity," in Brandom (ed.), 2000,
pp. 109-22.
Neiman, Susan. 1994. The Unity of Reason (Oxford: Oxford Univenity Press).
Nietzsche, Friedrich. 197 4. The Gay Science, translated by Walter Kaufinann (New York: Vintage
O'Hear, Anthony (ed.). 2004. Modem Moral Philosophy (Cambridge: Cambridge University Press).
Peirce, Charles Sanders. 1992a. The Essential Peirce, ed. Nathan Houser and Christian Kloesel,
two volumes (Bloomington and Indianapolis: Indiana University Press).
Peirce, Charles Sanders. 1 992b. "Some Consequences of Four Incapacities," in Peirce 1992a,
volume 1, pp. 28-55.
Pippin, Robert B. 2005a. The Persistence of Subjectivity: On the Kantian Aftermath (Cambridge:
Cambridge Univenity Press).
Pippin, Robert B. 2005b. "Postscript: On John McDowell's Response to 'Leaving Nature
Behind'," in Pippin 2005a, pp. 206-20.
Price, Richard. 1948. A Review of the Principal Questions in Morals, edited by D. D. Raphael
(OxfOrd: Oxford University Press).
Rauscher, Frederick. 2002. "Kant's Moral Anti-Realism," in Journal of the History of Philosophy
40, pp. 477-99.
Rawls, John. 1993. Political Liberalism (New York: Columbia University Press).
Reath, Andrews. 2006a. Agency and Autonomy in Kant's Moral Theory: Selected Essays (Oxford:
Oxford University Press).
Reath, Andrews. 2006b. "Autonomy of the Will as the Foundation of Morality," in Reath,
2006a, pp. 121-72.
Ridge, Michael. 2012. "Kantian Constructivism: Something Old, Something New," this volume.
Rorty, Richard. 1991a. Objectivity, Relativism, and Truth: Philosophical Papers Volume 1 (Cambridge:
Cambridge University Press).
Rorty, Richard. 1991b. "Solidarity or Objectivity?" in Rorty, 1991a, pp. 21-34.
Rorty, Richard. 2000. "Response to Michael Williams," Brandom (ed.), 2000, pp. 213-19.
Rorty, Richard. 2006. "Pragmatism as Anti-Authoritarianism," in Shook and Margolis (eds),
pp. 257-67.
Rorty, Richard. 2007a. "Kant vs. Dewey: The Current Situation of Moral Philosophy," in
Rorty 2007b, pp. 184-202.
Rorty, Richard. 2007b. Philosophy as Cultural Politics: Philosophical Papers Volume 4 (Cambridge:
Cambridge University Press).
Schneewind, J. B. 1986. "The Use of Autonomy in Ethical Theory," in Heller, Sosna and
Wellbery (eds), 1986, pp. 64-75.
Schneewind, J. B. (ed.) 2003. Moral Philosophy from Montaigne to Kant (Cambridge: Cambridge
University Press).
Sensen, Oliver. 2009. "Dignity and the Formula of Humanity," in Timmermann (ed.), 2009,
pp. 102-18.
Shafer-Landau, Russ 2003. Moral Realism: A Defence (Oxford: Oxford University Press).
Shafer-Landau, Russ (ed.). 2007. Oxford Studies in Metaethics, volume 2 (Oxford: Oxford
University Press).
Shafer-Landau, Russ (ed.). 2008. Oxford Studies in Metaethics, volume 3 (Oxford: Oxford
University Press).
Shook, John R. and Joseph Margolis (eds). 2006. A Companion to Pragmatism (Oxford: Blackwell,
Stem, Robert. 2012. Understanding Moral Obligation: Kant, Hegel, J(jerkegaard (Cambridge:
Cambridge University Press).
Street, Sharon. 2008. "Constructivism about Reasons," in Shafter-Landau (ed.), 2008,
pp. 207-46.
Street, Sharon. 2010. "What is Constructivism in Ethics and Metaethics?" in Philosophy Compass
5, pp. 363-84.
Timmerrnann,Jens (ed.). 2009. Kant's "Groundwork of the Metaphysics of Morals": A Critical Guide
(Cambridge: Cambridge University Press).
Wood, Allen W. 1999. Kant's Ethical Thought (Cambridge: Cambridge University Press).
Wood, Allen W. 2008. Kantian Ethics (Cambridge: Cambridge University Press).
Kantian Constructivism: Something
Old, Something New
Michael rudge
In this paper I argue that in its metaethical guise, constructivism is best understood not
as a new and entirely freestanding approach, but instead as providing an interesting sort
of species of each of the familiar genera of metaethics. I initially argue for this view in the
context of Christine Korsgaard's familiar Kantian version of constructivism (section 1).
I then test my hypothesis by exploring a novel and ambitious version of metaethical
constructivism developed by Sharon Street (section 2). I argue that Street's more
ambitious conception of constructivism can be understood in a number of different
ways, but that none of these readings can vindicate her claim to be offering an approach
to metaethics which is at once thoroughgoing, novel, and plausible.
1. Nothing new under the sun?
Constructivists typically take their inspiration from Kant, but the philosopher most
closely associated with the label is John Rawls. In Political Liberalism, Rawls distin-
guishes between "political constructivism" and "Kantian constructivism," making it
clear that his project is a species of the former. Both political and Kantian constructiv-
ism assert that some set of moral principles is appropriately represented as the outcome
of a suitable procedure of construction. I shall take this admittedly schematic formula-
tion as defining moral constructivism in the broadest sense. This formulation poses
more questions than it answers, for we now need to know what it is for something to
be a procedure of construction, what it is for such a procedure to be suitable, and what
it is for a set of principles to be appropriately represented as the outcome of such a
Many thanb to Michael Bratman, Dale Dorsey, Nadeem Hu!Sain, Aaron James, James Lenman,
T. M. Scanlon, Sharon Street, Andrew Williams, and the participants at the conference on Constructivism
in Practical Philosophy at the University of Sheffield in August 2009 for helpful feedback and comments on
an earlier draft of this material.
Sharon Street proposes a rather different definition of" constructivism." It is worth
pausing over her proposed definition because she has, in my view, done the best job of
articulating a version of constructivism which can with some plausibility be construed
as a dramatically new and thoroughgoing approach to metaethics. Here is how Street
defmes "constructivism in ethics":
Constructivist views in ethics understand the correctness or incorrectness of some (specified) set of
normative judgments as a question of whether those judgments withstand scrutiny from the
standpoint of some (specified) set of further normative judgments. (Street 2008, p. 208)
Street then distinguishes what she calls "restricted constructivism" from "metaethical
Restricted Constructivism in ethics specifies some particular, restricted set of judgments about
reasons, and says that the correctness of a judgment about reasons falling within that set is
constituted by the judgment's withstanding a certain (specified) procedure of scrutiny from the
standpoint of some (specified) set of further judgments about reasons. (Street 2008, p. 21 0)
According to metaethical constructivism, the fact that X is a reason to Y for agent A is constituted by
the fact that the judgment that X is a reason to Y (for A) withstands scrutiny from the standpoint
of A's other judgments about reasons. (Street 2008, p. 223)
One immediate point is that Street's metaethical constructivism is a species of con-
structivism in my broad sense. For on Street's account, we can construe moral
principles as being appropriately represented as the outcome of a suitable procedure,
where the suitable procedure is simply withstanding scrutiny from the standpoint of the
relevant agent's other judgments about reasons. The appropriateness of representing
moral principles in this way will simply consist in the fact that such a representation
displays the facts which constitute the truth of those principles.
While Street's metaethical constructivism is a species of constructivism in my broad
sense, she defmes the genus differently. For Street builds the idea that the relevant sort
of procedure is from the standpoint of some set of normative judgments into the very
definition of constructivism. This seems unduly narrow. Perhaps some forms of
constructivism will characterize the relevant procedures in non-normative terms.
Indeed, on some ways of reading Kant, the procedures in question are arguably not
best understood in terms of withstanding scrutiny from the standpoint of some set of
normative judgments. Rather, the categorical imperative is itself constructed from the
standpoint of freedom; that is, from our conception of ourselves as autonomous. This
conception of ourselves as autonomous is, on one way of reading Kant, more basic than
and prior to any set of normative judgments, and from the very idea of freedom, and
what it presupposes, we construct our most fundamental normative principle. Perhaps
this is not the right way to read Kant, but it is in any event an interesting and
recognizably constructivist view. It would therefore be a shame to define "construc-
tivism" so that a view of this sort is excluded.
Rawls himself distinguishes political constructivism from Kantian constructivism,
where the fonner is meant to remain neutral on crucial metaethical questions, such as
the nature of moral truth. Kantian constructivism is more ambitious. According to
Kantian constructivism, moral reality is itself constituted by the procedures of practical
reason which define the procedure of construction. This, in tum, seems to entail that
the "suitability" of a procedure is on the Kantian account a deeper notion. The Kantian
constructivist claims that the relevant procedure is suitable in the strong sense that the
relevant moral truths are constituted by these procedures, whether actual or hypothetical.
This reflects the Kantian idea of morality as autonomy, where autonomy requires the
idea of a moral law legislated by the agent's own reason rather than anything external to
that agent.
Kantian constructivism now begins to look like a view in metaethics. For Kantian
constructivism is a view about the nature of moral truth and reality, and these are surely
metaethical topics. How, though, might the Kantian constructivist be understood as
offering a novel approach to metaethics, ideally one which somehow transcends the
traditional debates in which metaethics has been mired at least since G. E. Moore's
Prindpia Ethica? One strategy here is usefully discussed by Stephen Darwall, Allan
Gibbard, and Peter Railton. They suggest that the constructivist's idea might be to
articulate some concern which explains why we care about morality in the ftrst place,
and then to construct a procedure which somehow embodies that concern. The
constructivist might then suggest that if the principles generated by such procedures
are ones which answer adequately to that concern, then· we might "no longer care
what our moral questions originally meant" (Darwall, Gibbard, and Railton 1992,
p. 142). However, as they go on to point out, this attempt to transcend traditional
metaethics faces a powerful dilemma. On the one hand, the constructivist can articulate
the concern in moral terms, in which case the meaning of those terms needs to be
articulated. On the other hand, the constructivist might articulate the concern in non-
moral terms, but then many people will reasonably doubt whether this was ever their
own concern, and even if it was whether it constitutes the moral concept being
elucidated (Darwall, Gibbard, and Railton 1992, p. 143). This strikes me as a powerful
dilemma for any version of constructivism which is meant to transcend the traditional
metaethical debates.
Korsgaard's discussion in her "Realism and Constructivism in Twentieth-Century
Moral Philosophy" provides a good illustration of the force of this dilemma. Korsgaard
tries to articulate a form of constructivism which rejects what she takes to be the moral
realist idea that the function of normative concepts is to describe reality. Of course,
traditional forms of so-called "anti-realism" in metaethics, most notably expressivist
views, can be understood as rejecting this assumption as well. However, Korsgaard
claims that her view is also no form of expressivism either.
Instead, she argues that we
• Though, confusingly, by the end of her paper she goe5 on to claim that both realism and expressivism are
each true in their own way, but in a way that makes each of them "boring." She daims that while
should understand the function of moral concepts in tenns of our need to solve certain
practical problems. Moral concepts do refer, at the end of the day, on this account. In
particular, moral concepts refer to whatever provide the best (or "only" -though this
clause is redundant in a way; if there is only one solution it is thereby trivially the best)
solution to the problem they function to help us solve; the idea is that while they refer,
it is not their function to refer or describe reality.
With this rough sketch in hand, consider again the dilemma articulated by Darwall,
Gibbard and Railton. Is Korsgaard's constructivism impaled on one or the other of the
horns of this dilemma, or does her view provide, as she claims, "a genuine alternative"•
to cognitivism and non-cognitivism? Actually, in one sense it is relatively clear that her
view does not offer a genuine alternative to cognitivism, since cognitivism as such (as it
is usually understood) does not claim that the function of moral concepts is to describe
reality. Typically, cognitivists claim only that moral concepts do in fact describe reality;
they are silent on the question of whether that is the Junction of those concepts, and
indeed generally silent on such functional questions full stop. Moreover, even if
cognitivists did claim that one function of moral concepts is to describe moral reality,
it is hard to see why this is inconsistent with them also having the function of naming
solutions to practical problems. Perhaps moral concepts have both functions, for all
Korsgaard has said.
After all, Korsgaard does hold that moral concepts do in fact refer
to moral properties, which are on her view the solutions to practical problems. In any
event, cognitivism as such does not require such a functional claim, so Korsgaard's view
is a species of cognitivism.
expressivism will seem correct from the descriptive or explanatory perspective of scientific or social-scientific
inquiry, those who use normative language will indeed seem just to be expressing their values. She then
remarks that "The trouble with expressivism is that it describes moral language from the outside, as if we were
not ourselves the creatures who &ce practical problems, but only someone making anthropological claims
about them. Behind that stance is the idea that so long as we are reasoning we must remain at this
anthropological level, and behind that view is the same error that animates moral realism-the view that
the business of cognition is describing the world" (Korsgaard 1993, p. 122, n. 49). This is one of the few
passages in which Korsgaard discusses expressivism, but it is hard to know what to make of it. On the one
hand, the point that the expressivist takes a detached and anthropological approach to the meaning of moral
discourse is hardly an objection to expressivism so much as an explication of the view. Moreover, if one wants
an objective account of the meaning of moral discourse then it is hard to see what is inherendy misguided
about taking a perspective which is detached in this way. Compare: would one need acrually to be in love or
to be a racist to study the narure of romantic love or the meanings of racial slurs? The very last sentence seems
to be the only objection, but against contemporary "quasi-realist" forms of expressivism, it is a non sequitur.
For the quasi-realist does not deny that we can and do reason about moral questions. The expressivist's point
is not that we do not reason about morality, but rather that reasoning about morality is distinctively practical,
and quite different from theoretical reasoning. Such a view should actually be quite congenial to Korsgaard's
own view, who in this paper emphasizes the differences berween theoretical and practical reasoning. For an
argument that Korsgaard's approach is most plausible if understood as a form of expressivism, see Gibbard
' See Korsgaard 1993, p. 115.
' Korsgaard 1993, p. 106.
' A similar point is made by Nadeem Hussain and Nishi Shah. See Hussain and Shah 2006, p. 289.
Korsgaard's view might, however, still provide an importantly new species of the
cognitivist genus, and one which might allow us to transcend the traditional metaethi-
cal debates. It is here that the dilemma posed by Darwall, Gibbard, and Railton is
germane, though. In some passages, Korsgaard characterizes constructivism in a way
that leaves it impaled on the first horn of this dilenuna. For example, her canonical
statement of constructivism characterizes moral concepts as referring to whatever is the
only or "best" solution to "problems." The concept of what would be the "best"
solution to a problem. and indeed the concept of what it is for something to be a
"problem" in the first place in the relevant sense, are normative concepts. For intui-
tively, for something to be a practical problem for someone just is for it to be something
which that person ought or must (in some practical sense of "must") regard as a
problem-as something she ought to solve. What, though, does it mean to say that
something is a problem, or that one of a number of possible solutions to that problem is
best? Since these are normative questions, it looks like all the standard metaethical
disputes will emerge again.
In other places, Korsgaard gestures in the direction of a robustly Kantian form of
constructivism which might be seen as aspiring to avoid the first horn of the dilemma
by offering a statement of the relevant problem in such a way that no rational agent as
such could fail to regard as a problem. The problem is that the free will as such needs
some law it can regard as non-arbitrary and binding (for freedom is not simply
randomness or arbitrariness), yet which is also not imposed on the free will by anything
external. This is a problem because there seems to be nothing left from which the free
will could derive such a law. Kant then argues that the only solution to this problem is
given by the Formula of Universal Law. The idea is that the free will has no basis for
choosing any particular content for its law; all it has to be is a law. The Formula of
Universal Law is meant to solve this problem because it too as such requires no specific
content. For as Korsgaard reads it, the Formula of Universal Law requires only that that
the choice of the free will must have the form of a law. Korsgaard seems to think that
this Kantian form of constructivism can avoid reliance on any further and contestable
normative notions. This particular Kantian version of constructivism comes out in her
statement of how the most ambitious forms of constructivism can in her terms go "all
the way down":
Can even our most basic reasons themselves be constructed? Kant's view, as I understand
him ... is that they can ... If this sort of Kantian argument doesn't work, then constructivism
cannot go "all the way down." I of course think that it can. (Korsgaard 2003, p. 118)
Does this ambitious Kantian form of constructivism transcend the dilemma posed by
Darwall, Gibbard, and Railton? Two points are germane here. First, it is worth noting
just how much this gives hostages to fortune. For now it turns out that constructivism
can provide the sort of genuine alternative to traditional metaethics that Korsgaard has
in mind only if something very much like Kant's derivation of the categorical impera-
tive from pure practical reason as such is sound. Needless to say, whether any such
derivation is possible is deeply controversial at best.
Second, even such a deeply
Kantian form of constructivism will still need to claim that we do indeed face the
"problem" of needing a law with the features Kant says the free will needs, and perhaps
that the "best" solution to this problem is the categorical imperative (I say "perhaps"
because the Kantian might argue instead that it is the only solution).' The crucial point
is that talk of what constitutes a problem, and of what would be an adequate or ideal
solution to a problem, is itself normative talk. So once again we will get controversy
about normative questions-now about what problems we face, and what the best
solution to those problems are. It is hard to see how the traditional metaethical
questions will not still arise. For what does it mean to say that the need for the sort of
law that Kant had in mind is a "problem" for the free will as such, or that the
categorical imperative is the "best" solution to that problem?
If even such a robusdy Kantian form of constructivism cannot transcend traditional
metaethics, then the hope that any form of constructivism could do so seems slim. This
does not, however, mean that constructivism makes no important contribution to
metaethics, or is plausible only as a first-order view as Rawls might be understood as
offering. For constructivism does introduce a distinction which might, so to speak,
carve "metaethical nature" at its joints. That is, perhaps the distinction between truths
that are constituted by suitable procedures and those that are not is a fruitful one for
metaethics to incorporate. On this way of thinking, constructivism provides a sort of
supplement to traditional metaethics, rather than a way of transcending it. Perhaps my
negative conclusion is premature, though. Perhaps constructivism can, after all, provide
something more like a genuine alternative to the traditional metaethical questions and
categories. For there is another and very interesting form of constructivism which
I have not yet considered. Here I have in mind the ingenious version of constructivism
developed by Sharon Street, to which I now tum.
2. Metaethical constructivism and primitive
normative judgments
As Street herself notes, her approach is in one way deeply Kantian, and in another way
deeply unKantian. It is deeply Kantian in that it attempts to derive moral truth from
autonomy, where autonomy is understood in terms oflaws one legislates to onesel£
This is because, on Street's view, moral truth is constituted by one's own judgments
about reasons for action, and such judgments are on Street's view self-legislated laws
which express an agent's autonomy. On the other hand, the view is deeply unKantian,
and instead rather Humean, in that it accepts the Humean idea that practical reason is
• For my own grounds for scepticism about Kongaard's attempt to pull the Kantian rabbit out of the hat of
practical reason, see Ridge 2005.
See Hussain and Shah 2006. p. 291.
purely instrumental.' Whereas for Kant, it is a necessary truth that rational agents as
such value humanity, on Street's view, "depending on one's starting set of values, one
could in principle have a reason for anything" (Street 2008, p. 244).
The motivation for Street's approach is the idea that more restricted forms of
constructivism fail to offer a thoroughgoing metaethical view because, qua restricted
constructivism, they take the normative judgments which are the "materials of con-
struction" as primitives. Street thinks, nonetheless, that these views "contain the seeds
of" a thoroughgoing metaethical view. For reflection on why such views fail to offer a
thoroughgoing metaethical view naturally invites the thought, "what happens when
we take the account of correctness proposed by these views for their limited target set
of normative judgments, and try to apply it across the board?" (Street 2008, p. 220).
Street begins with a thought experiment in which the first two valuing creatures
ever were hom, and in a relatively sophisticated form. The ftrst valued its own survival,
and nothing else, while the second one valued its own destruction and nothing else.
These creatures behaved accordingly, the first enthusiastically pursuing whatever
would, by its lights, conduce to its survival, and likewise with the second, mutatis
mutandis with regard to its destruction. Predictably, the creature who values its own
survival lives and leaves progeny; the creature who values its own destruction gets what
it wants. As Street points out, the realist will want to claim that the creature who valued
its own destruction is thereby making some kind of mistake. She then asks us to
recognize why we are tempted to think this. Her suggestion is that we find it much
easier to relate to a creature who values its own survival because we are the descendants
of creatures who valued their survival. That Darwinian explanation, though, does not
suggest that the other creature was making any sort of mistake. As she puts it, it got just
what it valued, after all, its own destruction.
On the strength of this line of argument, Street invites the reader to share the
constructivist intuition that "these two creatures' normative judgments ... were nei-
ther true nor false ... No independent standards existed (nor do any exist now) to give
any sense to truth and falsity when it comes to these two creatures' values; their
normative judgments merely popped into existence in a universe which until that
moment had been utterly devoid of standards" (Street 2008, p. 222). At this point,
though, Street asks us to consider a variant on this thought experiment. We now
imagine a third creature who values two things for their own sake-its own survival
and the survival of its offspring. Now suppose that the creature continued to value the
survival of its offSpring, but stopped valuing its own survival. Street suggests that in this
case the creature would be making a mistake. The idea is that the existence of the
creature's second value provides a non-arbitrary standard by which its first can be
assessed. The standard is non-arbitrary in that it reflects the agent's own normative
perspective. Street then takes this basic approach and generalizes it, so that the truth of a
Street makes this observation about her own view at Street 2001!, p. 244.
given normative judgment is a function of whether that judgment can withstand the
scrutiny of the agent's other normative judgments tout court.
Does this represent a novel and thoroughgoing approach to metaethics? An obvious
worry is that the approach on offer is incomplete. For the theory invokes normative
judgments in order to give an account of their truth conditions in terms of the agent's
other normative judgments. So far, though, we have no account of what it is to make a
normative judgment. What, in other words, are we really being asked to imagine in the
original thought experiment? What sort of state of mind was the "valuing" of those
creatures we were asked to imagine? A natural construal, though as we shall see one
that Street explicitly rejects, would be that their valuing simply amounted to their
normative beliefs. In that case, though, two closely related worries would immediately
spring to mind. First, the theory really is radically incomplete. For now we need some
account of the content of those original judgments. Familiar metaethical worries about
how a mere belief could motivate an action would spring to mind, for example.
Second, we would need some account of how the content of those beliefs could be
squared with the proposed account of their truth conditions. What content could those
belie& have, such that the belief is true if and only if it is endorsed by one of the agent's
other normative judgments? It is hard to see what an intelligible answer to this question
might look like. For unless the content just is that the judgment is endorsed by the
agent's other normative judgments, it is hard to see how there could be any sort of
guarantee that the belief is true if and only if endorsed by the agent's other normative
judgments. If, however, the content of the judgment is that it is endorsed by the agent's
other normative judgments then the account of content is viciously circular.
Let us call normative judgments as they figure as the materials of construction,
"primitive normative judgments", since our reflective concept of a reason for action is
"built out of" this more primitive notion. Street quite rightly rejects the idea that
primitive normative judgments are beliefS. I have emphasized the problems with
characterizing those judgments as beliefs simply to underscore that this is not really a
matter of theoretical discretion. Given the sort of theory that Street is developing, it is
hard to see how primitive normative judgments could be belie&; that way leads to
vicious circularity. What, though, are primitive normative judgments?
Here we must be careful, for Street distinguishes two notions of normative judgment.
First, there are what I am calling "primitive normative judgments," which function as
the "materials of construction" on Street's theory. These are the judgments such that
their withstanding the scrutiny of one another fixes the agent's normative reasons.
Second, there are normative judgments qua "results of construction" which are about
which of the agent's normative judgments in the first sense withstand the scrutiny of the
agent's other normative judgments. As Street puts the point, "the word reason as it
appears in the definiens is not understood in the same sense as being defmed by the
constructivist proposal as a whole; this would make the proposal uninformative and
viciously circular" (Street 2008, p. 241). Normative judgments qua "results of construc-
tion," then, presumably are to be construed as belie&. For the natural reading ofStreet's
view is that such judgments are belie& about which primitive nonnative judgments can
withstand the scrutiny of the agent's other primitive nonnative judgments.
Clearly, a suitable account of primitive nonnative judgments is essential to the
tenability .of this approach. for we need some idea of what primitive normative
judgments could possibly be such that they are not simply belie& about an agent's
reasons for action; otherwise the worries raised above are fatal. Street is well aware that
her theory needs to say something informative about primitive nonnative judgments,
and indeed she tells the reader quite a bit about normative judgments in the more basic
sense. At the same time, primitive nonnative judgments are on her view about reasons
in a sense of"reason" which is irreducible, so there is a limit to just how much Street
can consistendy say about such judgments. At least, there is a limit to how much she can
say about them qua their content. This point bean emphasis, as it brings out the ways in
which Street's view is and is not reductionist.
Street agrees with T. M. Scanlon that in at least one sense "reason" cannot be
explicated in any way which does not come back to the idea of a consideration that
counts in favour of something. However, Street does think we can say quite a lot more
here than Scanlon allows. fmt, she tells us quite a lot about what is constitutively
involved in making primitive judgments about reasons, and also tells us quite a lot
about what judgments are not. Second, unlike Scanlon, she goes on to develop a theory
of nonnative judgments in another sense--qua the "results of construction." She
ugues that we can reduce reasons as they figure in normative judgments in the second
sense to facts about judgments about reasons in the ftrst sense. Moreover, this reduction
does serious metaethical work by putting the notion of a reason for action into an
intelligible and naturalistic framework.
A great deal depends on precisely how we understand primitive nonnative judg-
ments here. I shall ugue that no matter how we understand such judgments, there will
be trouble making sense of Street's approach. fmt, though, I must review the various
things that Street does tell us about primitive nonnative judgments. Indeed, she does say
quite a bit about them, particularly given that these are judgments the normative
content of which is irreducible. Stringing together a number of comments and
commitments which emerge at different points in her discussion, Street tells us the
following about primitive normative judgments (page references all to Street 2008):
(1) A primitive normative judgment that something is worth pursuing for its own
sake, when made by a lone agent who values nothing else for its own sake is
"neither true nor false." (222)
(2) A primitive normative judgment that something is worth pursuing for its own
sake, when made by a lone agent who values nothing else for its own sake, can
explain mistakes in instrumental reasoning. (222)
(3) The standard for correctness for a primitive normative judgment that something
is worth pursuing for its own sake is whether that judgment can withstand
scrutiny from the perspective of the agent's other normative judgments. (223)
( 4) It is constitutive of a primitive normative judgment that if an agent A judges (in the
primitive sense) that there is reason for her to ¢ then for any 'l' such that A judges
(simultaneously, in full awareness) that 'P-ing is a necessary means to her ¢-ing,
A also judges (in the primitive sense) that there is reason for her to '1'. (227-8)
(5) It is constitutive of a primitive normative judgment that if an agent A judges (in
the primitive sense) that F is reason for her to ¢ then A does not (simultaneously,
in full awareness) judge that F is not a reason to ¢. (229)
(6) It is constitutive of a primitive normative judgment that if an agent A judges (in
the primitive sense) that only facts of kind X are reasons to ¢, and who
recognizes that F is not a fact of kind X, then A cannot (simultaneously, in
full awareness) judge that F is a reason to¢. (229)
(7) (Primitive) normative judgments are different from beliefS in that they are by
their nature motivating, such that if one judges that one has reason to </J, then
one is thereby necessarily at least somewhat motivated to </J (this explains why
such attitudes evolved in the first place). (230)
(8) (Primitive) normative judgments are different from desires in virtue of the
kinds of constitutive involvements laid out in (4)-(6) above. (231)
(9) There might be a broad sense of "desire" in which (primitive) normative
judgments are desires, but this is a very broad sense of "desire," unlike the
ordinary sense of"desire" in which someone who needs to have his leg sawed
off with no anaesthetic in order to survive might be truly said "to have no
desire" to have his leg sawed off. (231)
(10) Primitive normative judgments can be more or less strongly held. (234)
(11) Primitive normative judgments can in some sense be closer or further to the
centre of the agent's total web of (primitive) normative judgments. (235)
(12) The idea of one thing's being a reason for another (as this idea figures in
primitive normative judgments) cannot successfully be reduced to thoroughly
non-normative terms. (239)
(13) Our understanding of the idea of one thing's being a reason for another (as this
idea figures in primitive normative judgments) is given by our knowledge of
what it is like to have a certain unreflective experience--in particular, the
experience of various things in the world as "counting in favor of" or "calling
for" or "demanding" certain responses on our part. (239-40)
(14) The experience of something "counting in favour of" or "calling for," etc. is
given by such experiences as "how we feel when ... a tractor trailer swerves
toward us on the highway or we see a stranger threatening our child." These
examples illustrate the experience of evasive action or a protective response as
utterly "demanded" or "called for" by the circumstances. (240)
(15) The experiences outlined in (14) are ones we surely share with other animals. (240)
(16) There is no detenninate answer to the question of whether our normative
experience (as described in (14) and (15)) attribute the property of"counting in
favour of' to objects as they are in themselves, utterly independent of us and
our attitudes. To think there is a determinate answer to this question is to
overly intellectualize and distort our understanding of such experiences, which
we share with other animals, just as we share colour experience with some of
them. (241, n.)
(17) If normative experience did attribute the property of" counting in favor of" to
objects as they are in themselves, utterly independent of us and our attitudes,
then we should embrace an error theory about the content of that experience.
(241, n.)
(18) There are constitutive connections of some kind between (primitive) norma-
tive judgments and emotions like fear, hope, regret, sadness, joy, anxiety,
frustration, and so on. (242, n.)
(19) If primitive normative judgments were just desires then it would be obscure
exacdy how standards of correctness in the normative domain are generated.
What, then, can we conclude about primitive normative judgments given this exten-
sive and somewhat morley set of characteristics? We know that the content of such
judgments is irreducibly normative, so it would be misguided to try to characterize that
content in non-normative terms. Putting the question of the content of a primitive
normative judgment to one side; though, what sort of state of mind is a primitive
normative judgment?
One important and central strand of traditional metaethics distinguishes states of
mind in terms of their "direction of fit." The metaphor comes from Elizabeth
Anscombe. She illustrates the basic idea with the memorable and often cited example
of a man doing his shopping with a shopping list and a detective following the man,
writing down what he buys. Here is Anscombe:
If the list and the things the man buys do not agree, and this and this alone constitutes a mistake,
then the mistake is in the man's performance, and not in the list ... whereas if the detective's
record and what the man buys do not agree then the mistake is in the record. (Anscombe 1957,
p. 63)
The shopping list has a world-to-word direction of fit, in that the world is meant to fit
the list. By contrast, the detective's record has a word-to-world direction of fit, in that
the record is meant to fit the world. Very roughly, the idea is that belief is in this regard
like the detective's list, while desire is like the shopping list. The metaphor of a
direction of fit can be spelled out non-metaphorically in a number of ways, but a
discussion of these different elucidations of the idea here would take us too far afield.
Here I shall simply assume that some way of elucidating the distinction is both well
fotmed and philosophically illuminating. It is worth noting that this assumption is
• For some discussion of the options, see Hurnbentone I 'J'J2.
widespread in contemporary metaethics. If Street's account is distinctive only if the
very idea of a direction of fit is ill conceived, then that would itself be an interesting and
somewhat surprising result. Simply helping ourselves to this distinction, though, how
should we understand Street's primitive normative judgments?
So far as I can see, there are five options worth considering here, but none of them
sits well with Street's project, construed in the way she intends it to be construed. First,
primitive normative judgments might simply have no direction of fit, but instead be
pure feelings, understood only in terms of their phenomenology, and perhaps some
further nominal essence specified by the constitutive claims Street makes about primi-
tive normative judgments. Second, primitive normative judgments might simply be
belief-like states which represent the world as being a certain way. Third, primitive
normative judgments might be desire-like states-pro-attitudes in some broad sense.
Fourth, primitive normative judgments might at one and the same time have both
directions of fit. This could be because primitive normative judgments are a unified
state of mind which has both directions of fit-what is sometimes called a "besire. "
Alternatively, this could be because primitive normative judgments are actually hybrid
states, being constituted by some combination of belief-like and desire-like states.
I shall discuss each of these in tum, but shall save the option of primitive normative
judgments construed simply as desire-like states until last, as I take that approach to be
the most promising one for Street.
Suppose primitive normative judgments have a belief-like direction of fit. We know
from Street's discussion that they are not actually beliefS, and moreover the objection
raised above suggests that they had better not be so construed. However, not only
beliefs have a belief-like direction of fit; shopping lists, for example, can have that
direction of fit. Here the question would be what other states of mind might have the
same direction of fit, and one candidate comes readily to mind: perception. Perception
represents the world as being a certain way, and so has a belief-like direction of fit. It is
an interesting question in the philosophy of mind why perceptions are not beliefS, but
there are a number of ways of distinguishing them.
So suppose primitive normative judgments just are perceptions. The obvious ques-
tion is then, "What is their content?" Given Street's endorsement of Scanlon's view
that the notion of a reason is irreducible as it figures in primitive normative judgments,
presumably the content is some sort of irreducibly normative one. This reading seems
to face four objections. First, perceptions are typically understood as passive-as states
we just "find ourselves with." This does not fit well with the Kantian motivation for
Street's project, which construes primitive normative judgments as laws we autono-
mously give ourselves. Second, and more troubling, perceptions as such are not
motivating. Someone can perceive something as being thus and so without necessarily
being motivated one way or the other. However, Street is quite explicit that primitive
A term coined by Altham; see Altham 1986.
nonnative judgments differ from beliefs in that they are motivating. Third, representa-
tional content is enough for truth-aptness. Once we have a representational content,
there is nothing ill-formed about asking whether that content is veridical, after all. This
is problematic, for recall that an essential motivation for Street's constructivist machin-
ery in the first place was the intuition that the primitive normative judgments of the
two creatures who made no other normative judgments were "not properly called
correct or incorrect" and that they are therefore "neither true nor false." This does not
fit well with a perceptual gloss on those judgments. This is an important point, for it
goes to the motivation of Street's theory. For if primitive normative judgments are
already truth-apt all on their own, and indeed either true or false depending on the
match between their content and the world, then it is hard to see why we should adopt
the elaborate constructivist machinery Street outlines to define "reason" in some more
sophisticated sense in terms of these primitive judgments.
Fourth, it now looks like the theory will now tum out to be a species of error theory.
For the perceptions which are primitive normative judgments will on Street's account
have an irreducibly normative content. Qua representational states, then, they will
purport to refer to some irreducibly normative property-presumably of "counting
in favor of". Given Street's naturalism, though, and rejection of such sui generis
normativity (which anyway does not fit well with constructivism) there are no such
properties. Notice that this point emerges not from the idea that any such normative
property would have to be construed by agents as mind-independent, which Street
does briefly discuss. Rather, it emerges simply from the fact that any such property
would, it seems, have to be irreducible (actually, there is room to dispute this; I return
to this below).
In fact, Street at one point considers the possibility that our primitive normative
judgments might purport to refer to irreducible and even mind-independent norma-
tive properties. In that context she allows that, although it is not her view, there might
tum out to be some compelling argument for this view (Street 2008, p. 241, n. 55). In
that case, she says she would adopt an error theory about primitive normative judg-
ments, but does not seem to think this creates any trouble for her view. If that were
correct then the fact that the perceptual reading also commits her to an error-theoretic
view of the primitive normative judgments would also be unproblematic. The idea
would be that judgments about reasons in the constructed sense need not be in error even
if they are constructed on primitive normative judgments which are systematically in
error. This, however, seems too quick. For the idea behind Street's constructivism is
that the agent's other primitive normative judgments provide a sort of intelligible
criterion of correctness for a given primitive normative judgment. This idea no longer
seems intelligible if we suppose that each of these primitive normative judgments
anyway has a content, and that moreover they are all (trivial cases like tautologies to
one side) false. The systematic falseness of primitive normative judgments does not
seem to sit well with a constructivist account which purports to provide a criterion by
which many of them are deemed correct, and then constructs a more sophisticated
concept of a reason on the basis of this construction. For once it is clear that all of these
primitive normative judgments are systematically false anyway, we seem to have
"garbage in, garbage out", to borrow a phrase from computer programming. Why
should the fact that one false primitive normative judgment in some sense "endorses"
another false primitive normative judgment inspire any confidence of the correctness
of either, given their outright falsity? This sort of constructivism would seem to be
building castles in the sky.
On the whole, then, the perceptual reading seems deeply problematic. Suppose,
then, that we went for a sort of hybrid view, according to which primitive normative
judgments had both a belief-like and a desire-like direction of fit. Actually, there are
two ways of understanding the hybrid approach, as mentioned above. The first would
be to understand primitive normative judgments as unified states of mind in their own
right, but which somehow have a dual direction of fit. Here there are two main models
in the literature which could be deployed to make sense of this idea. First, there is the
idea explored in the work of John McDowell of normative judgments as "besires,"
though McDowell himself does not use this word. The idea, very roughly, is to model
normative judgments on judgments of secondary qualities like color, and to insist that
normative judgments bear an internal and conceptual relation to affect and motivation
in the same way that colour judgments bear such a relation to having had suitable visual
experiences. Someone not subject to the relevant sort of affect or motivational
tendencies simply could not, on this view, count as making a judgment with the
relevant representational content. So there are certain states of affairs which necessarily
can only be represented insofar as one is suitably motivated. Hence it is not hard to see
why one might suppose such a representation should be construed as having both a
belief-like and a desire-like direction of fit.
Alternatively, one might allow that there is no necessary connection between the
representational content of a primitive normative judgment and a suitable motiva-
tion. Perhaps it is possible to make a judgment with precisely that content but
without any attendant motivation or even any attempt to understand why someone
might fmd it motivating. One theoretically fruitful way of spelling out this idea
would be to cash out the idea of "direction of fit" in terms of biological function.
One could then argue that our primitive normative judgments have evolved with
two functions, a representational one and a motivating one. They would in this sense
be similar to primitive representations found throughout the animal kingdom, such
as the bee dance which plausibly has evolved with the dual functions of representing
a location and functioning like an imperative Qoosely speaking, of course), "telling"
the other bees to go to that location. Ruth Millikan has discussed such states of
mind, understood in just these functional terms, at some length. She calls them
"Pushmi-Pullyu" states of mind.
Kim Sterelny discusses the same phenomena, and
See Millikan 2005. ch. 9 and Millikan 2004, ch. 13.
calls them "coupled" representations because of the way in which they are coupled
with motivation. On Sterelny's account, one of the evolutionary "neat tricks" that
humans have picked up is our ability to decouple representational content from
motivation--this is what is distinctive about belief as opposed to the more primitive
coupled representations we find in bees and other parts of the animal kingdom.
Will either of these models provide a helpful account of Street's primitive normative
judgments? Let us first consider a besire-like version of Street's view, on which
primitive normative judgments are representational judgments with a content one
can only grasp if one is motivated in the right way. This approach improves on the
perception-based view with respect to the first two objections discussed above. For
besires need not be passive in the way perceptions arguably are, and could therefore
perhaps in some sense express our autonomy. Since they are by defmition motivating
they avoid the second objection discussed above. However, this approach faces a
battery of powerful objections, all the same. First, the objection that such a state
would already have a truth condition all on its own, in virtue of its representational
content, still applies. Once again, the motivation for the constructivist machinery (to
provide an otherwise absent standard of correctness) vanishes. Second, the objection
that this seems to entail an error theory also seems to carry over, and for the same
Moreover, the McDowellian rifF on Street's view raises new objections of its own.
A third objection is that it is very controversial whether there are or even could be
besires in the needed sense. That is, it is unclear how there could be a possible state of
the world that one could only grasp and represent to oneself if one were suitably
motivated. Certainly, on some standard ways of understanding belief and desire, it will
always be conceptually possible to divorce the content from the motivation.U Nor,
indeed, do McDowell's own examples inspire confidence. McDowell's leading exam-
ple is the judgment that someone is shy and sensitive. He suggests that someone lacking
in the relevant moral virtues does not really know what it means to be shy and sensitive.
However, he is quick to add that his notion of "not knowing what it means" is
compatible with, "all ordinary tests" indicating that the person does indeed know what
it means to be shy and sensitive (McDowell1978, p. 22). As many have noted, it is hard
to know what to make of this suggestion. It does not seem to get much support from
common sense, even by McDowell's own lights, since all "ordinary tests" would
suggest that an unkind person could perfecdy well understand what it means to be
shy and sensitive. Indeed, common sense seems to license the attribution of the
concepts of shyness and sensitivity to such people. The schoolyard bully is precisely
undentood by common sense as someone who understands what it means to be shy
and sensitive, and exploits this to better humiliate his victims.
See Sterelny 2003. " See Smith 1994. ch. 4.
What, though, about the rather different approach suggested by Millikan's idea of a
"Pushmi-Pullyu"? This approach would avoid the objection raised against McDowel-
lian besires, for there is no suggestion that it is impossible to represent the content of such
a state without the associated desire. The idea is the more modest one that this is not
how our primitive normative judgments function. Moreover, this approach would fit
well with Street's idea that other animals can and do make primitive normative
judgments. However, the first two objections raised above would apply to this view
as well. That is, the objections that this construal makes constructivism both unneces-
sary and either incoherent or at best a somewhat schizophrenic variation on a familiar
view (reductive naturalism) surface again. Again, those objections required only that
we understand primitive normative judgments as having representational content.
Finally, the objection that these approaches do not preserve the Kantian idea of
autonomy also seems to resurface, albeit for different reasons. As I noted above, besires
need not be understood as passive in the way that perceptions are. Perhaps the same
point can be made about Pushmi-Pullyu states, though this is less obvious given the
standard examples given of such states of mind-the bee's dance hardly seems like an
expression of autonomy, e.g. both seem not to express the agent's autonomy in that
one cannot freely choose one's beliefS; belief is famously not under one's voluntary
control. Given that one deploys the concept used in a primitive normative judgment,
on these accounts, one's application of that concept will be determined by the world.
In particular, one's ends will be set by whatever content is associated, whether
contingently or necessarily, with one's motivation, and in Kantian terms this seems
One complication I have been overlooking so far is that Street also characterizes
primitive normative judgments in terms of various constitutive involvements. For
example, it is constitutive of the primitive normative belief that there is reason for
one to survive that if one at the same time and in full awareness believes that one will
survive only if one does not eat this plant, then one thereby also believes that there is
reason not to eat the plant. If one lacks the latter belief then on Street's account one
does not really have the former belie£ It might seem hard to see why any given
perception, besire, or Pushmi-Pullyu state of mind would, of necessity, respect this
constraint. I think the best strategy for Street here is to insist that these constraints
provide a nominal essence for primitive normative judgment. That is, she could claim
that the concept "primitive normative judgment" simply does not apply to any state of
mind which does not satisfy these further constraints. Insofar as the essence is purely
nominal and not metaphysical, she would then not need to claim that there could not
be something otherwise just like the judgment that there is reason to survive, save that it
did not respect these various constraints. This would make her account intelligible, but
it does raise the question of why we should accept these constraints. Without some
intelligible connection to the content of primitive normative judgments, these
constraints will inevitably seem somewhat arbitrary. Even if there were an intelligible
and even conceptual connection to the content of the judgments, one might wonder
why someone could not make such a judgment without drawing an obvious inference.
So much, then, for readings of Street's view on which primitive normative judg-
ments have both belief-like and desire-like directions of fit in virtue of being identical
to some single and unified state of mind with a dual direction of fit. Suppose, instead,
that Street glossed primitive normative judgments as hybrid states-states of mind that
are really combinations ofbelief-like state(s) of mind and desire-like state(s) of mind? In
that case, we would then need to know how the content of the relevant belief is fixed.
The obvious move would be to gloss the content of the belief in irreducibly normative
terms, for that would fit with Street's endorsement of Scanlon's line on the conceptual
irreducibility of the normative. Perhaps Street could give up this line and claim that the
belief component of such a hybrid has a descriptive content, and is only normative in
virtue of the accompanying desire. Either way, though, the account once again faces
two by now familiar objections. For such a view threatens to make constructivism
otiose (we again have a standard of correctness already) and potentially incoherent or at
best a schizophrenic version of a familiar view. Moreover, this approach does not sit
well with Street's claim that primitive normative judgments are not beliefS.
So much for views according to which primitive normative judgments have the
direction of fit of belief and for hybrid views which posit dual direction of fit. What of
views which deny that primitive normative judgments have a direction of fit at all?
Such a view might insist that primitive normative judgments are essentially phenome-
nological--pure feelings, or rather dispositions to have such feelings when presented
with the idea of that which we thereby value (or disvalue). The move to dispositions
would accommodate the platitude that one continues to make a normative judgment
when not actively thinking about it. This approach would have the virtue of fitting
well with Street's characterization of primitive normative judgments in terms of a
certain sort of "unreflective experience," although arguably this fits even better with
the perceptual reading. In any event, this way of glossing primitive normative judg-
ments also faces a series of powerful objections. First, it is not all that phenomenologi-
cally plausible that there is any "pure" feeling of normativity. I agree with Street that
we have certain characteristic responses to danger and threats to our loved ones, but
these responses do not seem to me well characterized as a pure feeling of normativity.
To the question, "How does it feel to see your child in danger?," "reasonesque" is
hardly a compelling answer.
Second, dispositions to have pure feelings in response to the idea of possible object of
evaluation are not in any sense well understood as judgments about those objects.
Purely causal relations are not intentional. The thought of something I regard as
worthless might for some psychoanalytic reason cause me to have feelings of arousal,
pleasure, and the like. It would be strained at best to say that I thereby value it. Third,
pure feelings seem to be beyond our control. To this extent, this characterization does
not fit well with the Kantian idea of primitive normative judgments as expressions of
our autonomy. Fourth, pure feelings need not motivate us; even the prospect of
pleasure need not be motivating. Yet Street's view seems to require a necessary
connection between primitive normative judgments and motivation. Fifth, it is hard
to see how to as much as make sense of the idea that dispositions to have such feelings
might be seen as being structured into a "web of judgments" in any sense. That
metaphor is apt only when we have something that looks much more like a genuine
judgment which might bear conceptual and epistemic relations to other judgments.
Sixth, this characterization also does not fit well with Street's remarks about instru-
mental error. Someone who is disposed to have certain pleasant feelings on contem-
plating some outcome, and believes that some action is a necessary means to that
outcome, but is not disposed to have similar feelings on contemplating that action, is
not thereby making a mistake.
The seventh objection to this reading is perhaps the most telling, though. For if
primitive normative judgments are really just suitably structured dispositions to have
distinctive feelings then Street's constructivism collapses into a familiar form of cogni-
tivism. If such feelings can be understood in naturalistic terms, then the view collapses
into a naturalistic cognitivism which defmes "reason" in terms of such feelings. If such
feelings cannot be understood in naturalistic terms, then the view likewise collapses
into cognitivism, but here a non-naturalistic version which defmes "reason" in terms of
feelings which cannot be understood as part of the natural world. Either of these
consequences spells trouble for the idea that constructivism is a novel and thorough-
going alternative to traditional approaches to metaethics.
This leaves us with only one remaining interpretation of primitive normative
judgments, as being desire-like states with a world-to-mind direction of fit. In my
view, this is by far the most plausible approach for Street to take, and probably the
approach she would favour. It has a number of virtues. First, it fits well with Street's
remark that primitive normative judgments, while not desires in a narrow ordinary
language notion of"desire," are perhaps best construed as "pro-attitudes" in a broader
sense. On one reading this just is the claim that primitive normative judgments have the
direction of fit of desire. Second, it can make sense of the idea that the concept of a
reason as it figures in a primitive normative judgment is irreducible. For expressivists
have long endorsed the idea that normative concepts are irreducibly precisely on the
grounds that to make such a judgment is to adopt a pro-attitude. Hence no cognitive
representation of a possible object of evaluation as having such-and-such descriptive or
naturalistic properties will entail that one ought to promote it, care about it, or
whatever. For one might simply be left cold by such descriptions. Third, this approach
makes it easy to see how primitive normative judgments are essentially motivating, as
Street emphasizes and needs given her evolutionary account. Fourth, this approach
seems to avoid the objections levelled against the other approaches canvassed here.
One might object that desire-like states as such do not, in general, respect the
constitutive constraints Street laid down--see (4)-(6) above. However, as I pointed
out above, these constraints can be understood in a deflationary way-as picking out a
sort of nominal essence.
The only aspect of Street's discussion which might seem not to fit well with this
construal is her phenomenological construal of primitive nonnative judgments in terms
of certain characteristic experiences. Fortunately, we can already capture the idea that
"reason" is irreducible as it occurs in primitive nonnative judgments. This can be
accommodated in the expressivist way glossed above. So there is no theory-driven
rationale for positing a purely nonnative experience in any sense that would involve a
genuinely nonnative representational content. Moreover, Street's examples seem to
me to be open to a more deflationary understanding. When I contemplate a tractor
trailer veering toward me on the highway, I do imagine myself having a distinctive
experience--stark terror! Similarly, when I contemplate someone threatening my
child, I again imagine myself having a distinctive experience--anger. These emotions
can in some intuitive sense be glossed as ones which present their objects as "calling
for" some course of action, but not in any sense which requires us to posit any
genuinely representational nonnative content. Furthermore, this would also fit very
well with Street's suggestion that these feelings are ones we share with other animals.
What should we make of Street's view, so understood? First, the view no longer
looks like a genuinely new and thoroughgoing approach to metaethics. For in effect
the view is a form of sophisticated subjectivism according to which a person's reasons
for action are a function of what he or she would want if a privileged subset ofhis or her
desires (the ones which are primitive nonnative judgments) were in a recognizable
sense more fully coherent. Indeed, the view seems like a variation of Bernard Wil-
liams's account of reasons as what he calls "internal reasons. "
To be sure, there are
differences. For example, Williams is more ecumenical in terms of which pro-attitudes
are relevant; not only those pro-attitudes which are also primitive nonnative judg-
ments in Street's sense can ground reasons for action for Williams. Williams's account
of sound practical reasoning is also more vague and probably more ecumenical than
Street's. Again, this is an interesting view, and Street's version of it is certainly
interesting. Construed in this way, though, it is a riff on a familiar tune rather than a
whole new song.
Second, it is not clear what is gained by calling the relevant desire-like states plus
emotional/imperatival dispositions "judgments." An expressivist would argue that we
should call certain pro-attitudes "normative judgments" on the grounds that they are
precisely the states of mind expressed by our normative claims. Street's view cannot
give this rationale for her proposed terminology, though. For on her view, ordinary
•• If this does not e m like enough to capture the phenomenology, I would add that in many such cases,
we would experience ourselves as unreftectively giving ourselves imperatives. Seeing an oncoming tractor
trailer I can imagine myself thinking to myself, "Get the hell out of the way!" Similarly I can imagine myself
telling ntyldf; "Stop him" or "Hurt him" when presented with someone threatening my child. Or rather,
I can imagine myself having such thoughts if there was time and I kept my head enough to think and not just
react "in•tinctively". These imperatives might not be thoughts we share with non-human animals, but then
our normative experiences might be richer than theirs.
" See Williams 1981b.
language normative claims will instead be construed as expressing beliefS about such
desires (etc.) rather than those desires (etc.) themselves. It is hard to see what is gained
by going on calling them normative judgments rather than a particular sort of desire
(and associated emotional and imperatival tendencies). It would seem less confusing
and more straightforward simply to characterize these states of mind as pro-attitudes,
emotions, intentions, etc. This is in a way a minor and terminological point. It is,
however, worth underscoring precisely because the idea that we did really have
judgments in the materials of construction was, it turns out, essential to the idea that
Street's constructivism represented a new and thoroughgoing approach to metaethics.
Once we gloss the materials of construction as pro-attitudes of a certain sort, this idea
no longer looks plausible or well motivated. Which is just to underscore that the view
is ultimately a sophisticated species of meta ethical subjectivism, rather than an entirely
novel metaethical theory.
In this chapter I have defended a schematic characterization of constructivism which
includes both its first-order and meta-ethical species. I have then argued that, so
defined, constructivism cannot offer an approach to metaethics which is at once
fundamental, truly thoroughgoing, and prima facie plausible. To be clear, I do not
mean to imply that constructivism is therefore only a first-order view. For construc-
tivism can provide a sort of supplement to traditional metaethics, rather than a way of
transcending it. For, after all, there will be constructivist and non-constructivist ver-
sions of naturalism, non-naturalism, and expressivism. Indeed, one might even say that
there will be naturalist, non-naturalist, and expressivist species of constructivism and
anti-constructivism. Which one takes to be genus and which one takes to be species
here is to some extent a matter of philosophical taste. What is clear, though, is that
constructivism cannot hope to offer a prima facie plausible approach to metaethics
which is not dependent, at some level, on one of the familiar metaethical doctrines.
Constructivism will not, at the end of the day, allow us to transcend or avoid those old
and difficult debates.
Altham,J. E.J. 1986. "The Legacy ofEmotivism," in McDonald and Wright 1986 .
.Anscombe, G. E. M. 1957. Intention (Oxford: Basil Blackwell).
Darwall, S., Gibbard, .A., and Railton 1992. "Toward Fin de siecle Ethics," The Philosophical Review
v. 101, pp. 115-89.
Gibbard, .A. 1999. "Morality as Consistency in Living," Ethics 110, pp. 140-64.
Humberstone. L. 1992. "Direction of fit," Mind 101, pp. 59-83.
Hussain, N. and Shah, N. 2006. "Misunderstanding Metaethics: Korsgaard's Rejection of
Realism," in Shafer--Landau, Ross 2006, Oxford Studies in Metaethics, volume 1 (Oxford:
Oxford University Press), pp. 265-94.
Korsgaard, C. 2003. "Realism and Constructivism in Twentieth-Century Moral Philosophy,"
journal of Philosophical Research AP A Centennial Supplement, pp. 99-122.
McDonald, G. and Wright, C. (eds). 1986. Fact, Science and Morality: Essays on A]. Ayer's
lAnguage Truth and Logic (Oxford: Basil Blackwell).
McDowell, ]. 1978. "Are Moral Requirements Hypothetical Imperatives?" Proceedings of the
Aristotelian Society 52, pp. 13-29.
Millikan, R. 2004. Varieties of Meaning: The 2002 Jean Nicod Lectures (Cambridge, MA: MIT
Millikan, R. 2005.1Anguage: A Biological Model (Oxford: Oxford University Press).
Rawls, J. 1993. Political Liberalism (New York: Columbia University Press).
Ridge, M. 2005. "Why Must We Treat Humanity With Respect? Evaluating the Regress
Argument," European journal of Analytic Philosophy 1, pp. 57-74.
Smith, M. 1994. The Moral Problem (Oxford: Basil Blackwell).
Sterelny, K. 2003. Thought in a Hostile World (Malden, MA: Blackwell).
Street, S. 2008. "Constructivism About Reasons," in Shafer Landau, Russ. 2008, Oxford Studies
in Metaethics, volume 3 (Oxford: Oxford University Press), pp. 207-45.
Williams, B. 1981a. Moral Luck (Cambridge: Cambridge University Press).
Williams, B. 1981b. "Internal and External Reasons," in Williams 1981a, pp. 101-13.
Constructing Coherence
Yonatan Shemmer
1. Introduction
According to constructivists the process of norm construction is governed by the
principle of practical consistency. The principle of consistency is a thin principle of
rationality that prohibits agents from both adopting and rejecting the same goal at the
same time. However, the principle of consistency is too thin to account for the kind
of structural restrictions that agents impose on the dynamic process of goal manage-
ment. To account for these restrictions constructivists must add another principle of
rationality-a broad principle of coherence. In this chapter I explain why the principle
of consistency is insufficient, offer an account of the broad principle of coherence that
together with the principle of consistency governs our goal adoption process, explain
how constructivists can account for the normativity of this broad principle by showing
that it also can be the object of construction, and consider some worries regarding this
A friend asks you if you want to join a hiking club. Club members go out for hikes
on the weekends. You like hiking and you think it would be nice to meet some new
people. You moved to your town four years ago and you still haven't explored the
natural surroundings; joining the club would give you the opportunity to do that. But
you are really busy. You have two young kids and you can only really spend time with
them during the weekend. If you go on a hike you would have to split the weekend
with your wife such that each one of you takes care of the kids during one weekend
day, but then you would not see much ofher. And on top of it you were planning to
spend more time reading and it's not clear how that fits in with the idea of joining the
club. You ask yourself what to do and on reflection you realize that this is not merely a
question of whether to add another membership card to your wallet. Rather it is a
question about what kind of person you want to be. A decision to join is a decision to
give a certain importance in your life to spending time in nature, even at the expense
of time with your family. It is a decision to treat your outdoor t i m ~ v e n while
away from the family-as a goal, or, in other words, as a reason in your future
Should you do that? The decision to join the hiking club and to spend more time
outdoors away from your family is in conflict with your existing reasons. So you tell
your friend that you'd really like to, but you don't have the time. Of course, you could
join the club and then skip many of the hikes. When we say that the decision to join
the club would be in conflict with your current reasons we do not mean that it would
be strictly inconsistent with those reasons. If you skipped many of the hikes you could
both join the club and act on your existing reasons. Your decision not to join the club is
not grounded in any strict inconsistency with your current reasons. Rather you
recognize that joining the club would have stretched you too thin. If you made this
decision without dropping any of your existing goals you would do everything you
wanted to do, but worse. Your life would become more incoherent. Beyond a certain
point you would find such incoherence unacceptable.
Even though such incoherence does not amount to strict inconsistency we see it as
inational. When we say that a potential new goal is in conflict with our existing goals
what we often mean is that these goals are a package without integrity. We consider
decisions that lead us to such incoherence in breach of a fundamental principle of
rationality; a principle of rationality whose purpose is to guide our adoption of new
ends: the principle of coherence.
Principles of coherence are nonnative principles. They are therefore in need of
justification. One concern of this chapter is to explain how constructivists may account
for the construction of principles of coherence and thus provide them with a justifica-
tion. A second concern is with the priority given in the process of construction, to
reasons and principles that one is already committed to, over reasons and principles that
one deliberates about constructing or adopting. On the one hand such a priority seems
necessary for the diachronic stability of the self, on the other hand it legitimizes a
despotism of our present self over our future self-a despotism we are hard pressed to
Before I get to my main tasks I will say a bit more about principles of coherence.
I will also explain which constructivists should be interested in the construction of
principles of coherence.
2. Rationality, coherence, and constructivism
Principles of coherence are part of a larger family of principles of rationality. Principles
of rationality are formal principles that guide or constrain transitions between mental
states. Principles of coherence are formal principles that guide the process by which we
adopt new goals and drop old ones.
I will refer to these principles as Principles of Coherence in the Construction of
One's Reasons but sometimes simply as principles of coherence (PC) or as reasons of
coherence. The principle of consistency is itself a strict principle of coherence, but
I will mostly be interested in a broad principle of coherence (PCB)-one that was
implicitly appealed to in the story above. What is common to all principles of
coherence is that they constrain our set of goals in such a way that makes it possible or
easier to satisfy these goals.
Not all philosophers think that principles of rationality are normative.' Some
philosophers think that they are psychological rules of thumb without normative
force (Kolodny 2005). Of those that think that they are normative, some are not
constructivists. However, if one is a constructivist, one will assign a special role in one's
nonnative system to these principles. According to constructivists, a putative norm is a
sound norm if it "passes a test of construction". Most constructivists think of this test as
provided (at least partly) by principles of coherence. Thus on these views principles of
coherence guide the construction of reasons.
The possibility of justifying principles of coherence will be of interest to anyone who
thinks that they are normative but it will be of special interest to constructivists. Since
constructivist views rely on principles of coherence in their attempt to construct the
nonnative domain they depend on the success of the justification of these principles.
Some constructivists depend in particular on the success of our attempt to justify
principles of coherence in a constructivist way. To see which constructivists those
are, why they owe us an account of the normativity ofPC, and what kind of account
they are committed to, we need to introduce some distinctions between different types
of constructivists.
3. Types of constructivism
1. A local constructivist thinks that only part of the normative domain is constructed.
A global constructivist thinks that all our norms are constructed.
2. According to certain constructivists, when we construct norms we guide the
process of construction by our non-constructed norms.
I will call someone who
holds that view a dependent constructivist. A dependent constructivist is necessar-
ily a local constructivist.
An independent constructivist thinks that when we construct norms, the process
of construction is not guided by any non-constructed norms. An independent
constructivist may be a global constructivist, but she may also be a local con-
structivist if she thinks that the norms we do construct are independent of the
norms that exist without the process of construction. A global constructivist is
necessarily an independent constructivist.
John Broome thinks they are but is unsure how to defend that view (Broome 2008).
I speak here of the construction of norms as if it were an active process. This is particularly suitable to my
Humean form of constructivism according to which active endorsement plays a crucial role in the constitu-
tion of reasons. Not all constructivists think of the construction of norms as an active process. For some the
only sense in which a norm is constructed is that its status as a norm is contingent on it passing a test given by
some norm of construction. According to these constructivists, speaking of the activiry of construction is
merely a metaphor.
3. According to certain constructivists, as Heidegger would have put it, nonns are
always already there. When we construct nonns we are always guided in the
process of construction by the very set of norms we are constructing. If you think
that, you are a Neurathian constructivist.
A foundational constructivist, on the other hand, thinks that we construct
norms solely from material that is outside the domain of construction.
independent foundational constructivist thinks we construct norms from non-
normative material. A dependent foundational constructivist thinks we construct
norms from non-constructed norms.
Both the foundational constructivist and the Neurathian constructivist may be
local or global constructivists.
It might help us see what these distinctions amount to if we consider some construc-
tivist theories and see where they are located in this three-dimensional organizational
Hobbes is the classic example of a local, foundational, dependent constructivist. On
his view norms of rationality are not constructed. This makes him a local constructivist.
Political and moral norms are constructed by appeal to norms of rationality. This makes
him a dependent constructivist. When we construct political norms, Hobbes claims,
we can imagine a world without such norms and see how political norms would
emerge and be justified in light of our norms of rationality. This makes Hobbes a
foundational constructivist.
James Lenman is a global, independent, and Neurathian constructivist. He thinks
that all norms are constructed, and naturally this construction does not appeal to any
non-constructed norms. This makes him a global, independent constructivist. Fur-
thermore, on his view we start the process of construction in the midst of the
nonnative domain; our norms are justified by appeal to norms that are already
there. In "Expressivism and Constructivism" Lenman says: "Normative inquiry, on
this account, works much as Neurathian boat repair does. So thinks Street and so do
many, including myself" (this volume, p. 217). This makes him a Neurathian
Korsgaard is a global, independent foundational constructivist. She thinks all norms
are constructed and naturally this construction does not appeal to any non-constructed
norms. In "Realism and Constructivism in Twentieth-Century Moral Philosophy" she
says: "if this sort ofKantian argument doesn't work, then constructivism cannot go 'all
the way down'.) of course think that it can" (Korsgaard 2008a, p. 322). This makes her
a global, independent constructivist. But on her view we need not appeal to our
existing norms to justify the normative domain. This makes her a foundational and not
a Neurathian constructivist.
, One may further distinguish between Neurathianim1 and foundationalism about particular norms and
about the constructed domain as a whole.
All constructivists I know of think that the construction of new norms must be
governed by some principle of coherence: our constructed norms should cohere with
each other and with the norms that ground their construction. This may not be
obvious at first glance, but note that all. constructivists think that a putative norm is
(or becomes) a sound norm if it is approved of or endorsed by, or stands to scrutiny
from, the agent's
existing norms. Saying this is tantamount to imposing some norm of
coherence on the process of construction of reasons. So principles of coherence play a
crucial role in all constructivist accounts. This is not to say that all constructivists
understand or must understand these norms of coherence in the same way. In particular
a constructivist may understand the kind of coherence that must exist between our
norms to be a strict consistency requirement or he might understand it to be a broader
notion of coherence of the kind that played a role in the story with which I opened this
Need all constructivists think that norms of coherence are constructed? All con-
structivists can think that norms of coherence are constructed. But not all of them must
think that. Dependent constructivists may think that norms of coherence govern the
process of construction but are themselves not constructed norms. Independent con-
structivists, on the other hand, must think that norms of coherence are constructed, and
so they must explain how these norms would satisfy the constraints of the constructivist
But despite what seems like a commitment of all independent constructivists to
provide us with an account of the construction of norms of coherence that would
justify these norms, such an account is often not provided. Moreover, most indepen-
dent constructivists seem to assume that norms of coherence exist prior (logically
speaking) to the process of construction. They assume that the demands of coherence,
though they are clearly normative, are somehow presupposed by the idea of construc-
tion. These norms are taken to play the role of an underlying rule that governs the
dynamics of construction. Different constructivists refer to principles of coherence by
different names but the role they play in their theories is essentially the same. They are a
precondition for any construction. Thus Korsgaard says: "So reason's own principle just
is the principle of acting in a way that constitutes you into a single wtified agent"
(Korsgaard 2008b, p. 119). In other words, according to Korsgaard, only such a unified
agent most truly has reasons for actions. And Lenman says that "To call a consideration
C a moral reason in favour ofsomeone's </J-ing is to favour that consideration guiding
our deliberation and action in furtherance of concerns, desires, and aims to which it
speaks in virtue of the fact that we might all, insofar as we are reasonable, agree in
endorsing it and them as able to withstand scrutiny in the light of our other reasons"
(this volume, p. 222). Thus both, who intend to develop an independent constructivist
view, assume that normative principles of coherence are presupposed by the procedure
• Or "a group of agents' existing nonns".
of construction. This assumption makes it especially hard for them to justify these
norms by appeal to the process of construction, since the fact that the norms of
coherence are presupposed by the process of construction gives rise to a worry of
Independent constructivists seem to adopt one of three strategies when it comes to
an account of the norms that govern the procedure of construction:
1. Some ignore the need to justify these norms.
2. Some take these norms to be constitutive of our notion of a reason or of the
procedure of construction. The construction of other norms is dependent on the
prior existence of these constitutive norms.
3. Finally, some constructivists try to offer a story that explains how the norms that
govern the procedure of construction are themselves justified by the procedure
of construction.
Strategies 2 and 3 are both legitimate approaches to the justification of fundamental
norms of coherence-though you may think that 2 involves dropping the aspiration of
developing an independent constructivist view.
In the rest of the chapter I will focus
on the third approach. This is primarily because the particular norm of coherence
I want to discuss seems to me not to be constitutive of our idea of a reason, but also
because the third approach fits more naturally with the idea of independent construc-
tivism. As I said earlier, there is a strict version of the principle of coherence-this is the
principle of consistency6--and there is a broader version of this principle, one that calls
for an integrity or unity that goes beyond strict consistency. I assume that we are bound
by a strict principle of consistency, but I will say nothing here about the possibility of
justifying this principle. My aim is to see whether we can give a constructivist
justification of the broader principle of coherence-PCB.
PCB requires that we limit the number of ultimate goals we adopt and that, other things being
equal, we prefer goals that share the resources (temporal, mental,.financial . .. ) required for their
satisfaction over goals that require distinct resources.
PCB is an interesting principle. First, it is a formal principle that imposes structural
constraints on the construction of reasons. It determines which goals will be normative
for us without a bias as to the content of these goals. Second, PCB is not a single
principle but a schema: the constraints it imposes vary from agent to agent. Such a
subjective formal principle cannot be part of our understanding of what reasons are and
thus is not amenable to a constitutivist approach.
• Not everyone thinks that. Wallace (this volume) c l a i m . ~ that a justification by appeal to a constitutive
argument is itself a fonn of oollltrllction.
I do not suggett that there is only one possible undemanding of what the principle of consistenq'
requires. It may prohibit one from having a reason to do always do A and having a reason always not to do A;
it may prohibit treating a certain desire or fact as reason giving while at the same time treating that fact as a
reason not to do that same thing and other understanding are possible as well.
4. A central question and two worries
The main question I will aim to answer is how we can construct, and thus justify,
instances of PCB. To answer this question we will need a constructivist account of
reasons. I will describe mine shortly, but flrst let us make explicit the potential problems
with such a constructivist justiflcation of PCB.
The prospect of such construction raises two worries:
1. How can you construct a norm that is supposed to govern the process of
2. What gives norms of coherence a normative priority over newly constructed
I will provide an account of reasons that both explains how to construct norms (or
reasons) of coherence and will serve as a basis for answering the worries raised above.
I will not defend my account of reasons; I will simply assume it.
5. Reasons
On my view reasons are complex mental states that an agent has, or would have had were
she to reflect in a certain way, whose role is to guide the agent in deliberation and action
(Shemmer 2007). Such mental states have a few features: they must be motivating or
capable of being motivating, they must be guiding in form and capable of guiding the
agent in deliberation, they must be elements in the set of mental states that constitutes the
identity of the agent, and they must be stable on reflection. These are the desiderata. Below
I offer one story about how these desiderata can be satisfled. Since there may be other ways
to satisfy them, my account is only intended to provide us with sufficient conditions for
being a reason. Here are the details. Start with pro-attitudes. Those are motivating, guiding
in form, and in some sense capable of guiding the agent in deliberation. The complex
mental state that constitutes the agent's reasons includes pro-attitudes. However, some
pro-attitudes are not part of the agent's identity. Some pro-attitudes are seen by the agent
as forces external to herself that need to be ignored or suppressed in deliberation. Some are
mere urges, attempts to highjack deliberation, rather than to guide it, and others are ones
the agent does not care about. To distinguish between pro-attitudes that are part of the
agent's identity and those that are not, we need a theory of identification or endorsement.
I will assume a view ofidentiflcation that is close to Bratman's view ofidentiflcation. On
that view an agent identifies with a pro-attitude if (in the basic case) she decides to adopt a
self-governing policy to treat the pro-attitude as a reason in deliberation, is satisfied with
such decision, and in fact treats it as reason giving in deliberation, or is fully prepared to treat
it as such if the occasion were to arise (Bratman 1996).
Satisfaction on this account requires that the self-governing policy in question not be in
conflict with the agent's existing self-governing policies (Bratman 2007). This notion of
conflict is related to the notion of coherence we are trying to investigate. The exact nature
of the notion of conflict is thw crucial to my account and I will return to it shortly.
However, a pro-attitude that is part of the agent's identity and that is treated by the agent
as a reason in deliberation is not yet a reason.
An agent may have such complex mental state because she is unaware of some
crucial non-normative facts. She may, for example, decide to treat her desire to eat red
fruits as reason giving because she thinks it will promote her health, but in fact eating
red fruits will kill her. Such a mental state could not be reason giving. We need to add a
further condition to our account. It must be the case that the agent would have
identified with the pro-attitude in question even if she were informed about the
non-normative facts. We now have a pro-attitude that is part of the agent's identity,
is treated as a reason, and which the agent would have endorsed even if she were
informed about the relevant non-normative facts.
Such an attitude is almost a reason. A set of reasons would not be a set of reasons if as
a result of reflecting on its structure, content, and causal origins the agent would lose
confidence in its normative force. A reason must be a member of a set of reasons that is
jwtificatorily stable under reflection.
My account of reasons draws on Bratman's account ofidentification. A central element
ofBratman's account ofidentification is the thought that an agent identifies with a pro-
attitude (henceforth, desire) only if her self-governing policy to treat that desire as goal
setting for action
is not in conflict with the agent's other policies about what to treat as
goal setting for action. The decision to adopt goal A is in conflict with the decision to
adopt goal B if adopting the one will undermine the role that the adoption of these goals is
meant to play in the constitution of the agent's identity (Bratman 2007, p. 35). In other
words, the adoption of one goal is in conflict with the adoption of another if the adoption
of the one makes it impossible for the adoption of the second to play a role in determining
what kind of a person one is. On tim understanding most goals are not conflicting--even
if they pull in different directions. If two goals conflict on this strict understanding of
conBict, I will say that they "strictly conflict".
Here then is the resulting view about sufficient conditions for being a reason.
A complex mental state that satisfies the following conditions is a reason:
1. It includes a pro-attitude.
2. It includes a self-governing policy to treat that pro-attitude as reason giving
which is not in strict conflict with the agent's other self-governing policies, that
the agent has decided to adopt, or would have decided to adopt were she to
sufficiendy reflect about her situation.
3. The agent would adopt the self-governing policy in question even if she were
informed about the relevant non-normative facts.
4. The complex described in 1, 2, and 3 is stable under reflection.
' Bratman often IIICS the expression "reason in deliberation"' instead of"goal setting for action'".
I The ronnulation here is inexact. 1!"1 a place holder for 3 COndition that deals With the problem of partial
or misleading
6. Reasons and coherence
On this view of reasons agents are not bound by an a priori normative requirement to
have, and therefore to construct, a coherent set of reasons or goals (of the kind required
by PCB). An agent may have reasons to achieve a multiplicity of goals each preventing
her from satisfying her other goals in the way she would if they were her only goals.
She may have reasons to have a family of eight kids, to be the head of a successful firm,
to be a marathon runner, to spend time in nature, to be spontaneous, and to be
organized, all at the same time. When deciding whether to perform a particular action
she will certainly not be able to work towards all of these goals, but since she will treat
all of these goals as reasons in deliberation she is likely to try and satisfy each one of
them at different times and not to satisfy any of them to the degree that someone with
fewer goals could do.
It is often assumed that there is something wrong with such an incoherent set of
reasons. Both constructivists and non-constructivists think that a single principle
requires all agents to try and be coherent, well integrated, unified-all to the same
degree. I believe that this is not the case: many of us, maybe most of us, have reasons to
be coherent, integrated, and unified, but not all of us do and certainly not all of us do to
the same degree. Some people thrive in less organized, less coherent lives. They enjoy
doing a bit of this and a bit of that. They don't want to be great at any particular thing,
and they don't mind if each one of their activities eats away at the full realization of
their other goals. It is also often assumed that the principle of consistency is sufficient to
ensure the requisite coherence. As I will explain in section 8, to the extent that
coherence is important, consistency cannot ensure it.
Proponents of the view that our reasons should form a tighdy coherent set may
worry that if our reasons were not coherent, we would end up undermining our own
projects. Taking care of our kids, we would not prepare for the marathon; and
attempting to be spontaneous, we would fail to run our business effectively. We
would, in short, walk backwards and forwards at the same time, and would not be
able to bring any action to completion. The worry is exaggerated. As far as I know,
people cannot walk in different directions at the same time and doing a bit of this and a
bit of that is not the same as being paralysed. Maybe an incoherent life is not the best
way to become a Nobel laureate, but not everyone wants to win the Nobel Prize.
7. Constructing reasons of coherence
In the previous section I have concluded that the broad principle of coherence (PCB) is
not a necessary constraint on the construction of reasons. Nevertheless I think that most
of us, though to varying degrees, are guided by reasons of coherence. And when we are
guided by reasons of coherence these reasons do constrain the construction of all of our
other reasons. How can we come to have reasons of coherence and how can they come
to govem the construction of reasons?
The answer is quite simple. We come to have reasons of coherence just as we come
to have all of our other reasons. We have them in virtue of the fact that they pass the
test of construction.
Imagine an agent with a multiplicity of reasons (a reason to eat, a reason to work, a
reason to travel, a reason to have sex, a reason to be healthy, a reason to have physical
pleasure) but without reasons of coherence (without PCB). Under what conditions
would such an agent come to have reasons of coherence?
First, the agent must have a pro-attitude in favour of coherence. Such pro-attitude
may have a variety of sources. I assume that for most of us its main source is innate. We
are born with a certain discomfort when we face incoherence beyond a certain degree.
But innate dislike of incoherence is not its only source. Society and in particular our
parents instil such a pro-attitude in us through the usual mechanisms of socialization.
Coherence has advantages for the people we live with and our parents may think that it
also has advantages for us. The perception of these advantages leads them to inculcate in
us a preference for coherence. Finally, we omselves might see the advantages of
coherence. From the point of view of our existing desires and reasons the fact that
we are not guided by a reason of coherence in our further acquisition of reasons means
that over time the satisfaction of each one of our existing desires will become more and
more diluted. We will acquire reasons to do more things that will pull us in different
directions. Thus even though the acquisition of one more goal that does not cohere
with the rest of our goals will not yield a strict conflict with any of our existing goals,
the trend of acquiring incoherent goals will eventually bring about a level of incoher-
ence that goes beyond what we are willing to live with. This may lead us to decide
(possibly as a preventive measure) to inculcate a pro-attitude in favour of coherence.
Second, the agent needs to decide to adopt a policy to treat his pro-attitude in favour
of coherence as reason giving in deliberation, or it must be the case that he would
decide to adopt such a policy ifhe were to sufficiendy reflect on the matter. Note that
the deliberation in question is not (or at least not primarily) deliberation about how to
act, but rather deliberation about what pro-attitudes to endorse and therefore about
what reasons to adopt. If the agent adopts the policy of treating her pro-attitude in
favour of coherence as reason in future deliberation, then when she deliberates in the
future whether to adopt a new reason or not she will be bound (to the extent that no
other reasons will push her to be incoherent) not to adopt reasons that will make it the
case that her set of reasons exceeds a certain level of incoherence. The agent's current
desires can put pressure on the agent to adopt such a policy.
Third and fourth, the decision must be such that it wouldn't be changed under
conditions of increased information and that it would be stable under reflection. If all
four conditions are satisfied the complex mental state characterized by these conditions
is a reason of coherence.
• People do not fonn up their set o f ~ a s o n by deliberating all at once from a reason-les.' state. Usually part
of the package of reasons they fmd thenL'Ielves with alreadv includes a reason of coherence. But this does not
8. Is a special norm of coherence necessary
in order to achieve coherence?
You might think that the pressure from our existing reasons will prevent the future
acquisition of diverging goals anyway and therefore that we have no need for a special
reason of (and therefore a special pro-attitude for) coherence. I do not think this is the
case. To see why, we need to clarify the notion of a "pressure from our existing
reasons". Our existing reasons include our specific reasons or goals and the strict
principle of consistency. To show that our existing reasons put pressure on us to be
coherent we need to show that incoherent reasons/goals would be prohibited by the
principle of consistency because they would yield strict conflict.
Goals may strictly conflict with other goals directly or indirectly. Directly: when
achieving one goal entails not achieving the other. Indirectly: when the resources and
time needed to achieve one goal are consumed by the attempt to achieve the other.
Very few goals conflict directly. The goal ofbeing on top ofEverest at noon on August
3"d 2010 conflicts with the goal ofbeing in Nazareth at the same time. But the goal of
reaching the top of Everest and that of visiting Nazareth do not in themselves conflict.
Nor do the goals of being a good parent, travelling, reading books, having an
interesting job ... conflict directly with each other.
Could incoherence between our goals be banned because it leads to indirect strict
conflict? Most goals depend for their satisfaction on a limited pool of resources.
Therefore most goals are potentially in tension with each other. On the other hand
the mere fact that two goals share the same pool of resources does not mean that they
are in indirect strict conflict. Agents may have enough time, energy, money, and
computational capacities to achieve both. But then again our limited resources set a
limit to the number of goals we may satisfy in a given time. Isn't that constraint
sufficient to restrict the extent to which we may allow our set of goals to be incoher-
ent? And can't this constraint be determined solely by appeal to our specific goals, the
size of our resources, and the principle of consistency?
The answer is no. Our goals are rarely well defmed. In fact, for the most part their
boundaries are wide open. In particular, the amount of resources needed to satisfy them
is under-determined. My goal ofbeing an attentive parent does not specify the amount
of money or time I have to dedicate to my kids. This open-endedness is not a mere
epistemic artefact. It is not as if there is an amount of money or time that I ought to
dedicate to my kids and of which I am ignorant. Rather the goal of being attentive to
my kids is bounded only by an acceptable satisfaction of my other goals. But this just
means that my goals by themselves do not determine when they are acceptably
satisfied. The notion of acceptable satisfaction is a global notion that is determined
by the degree of coherence one hopes to have in one's life. Some people prefer unity;
mean that our reasons of coherence are necessary. Even though to a large extent the exercise of construction
is hypothetical, it derives its legitimacy from the nonnative possibility of incoherence.
they want few goals with a great degree of overlap and mutual support and a high
degree of satisfaction. Others prefer variety; they want to pursue many goals even
though they pull in different directions and even though most of them will not be
gready satisfied. The extent to which one cares about unity is not determined by one's
individual goals together with the principle of consistency. Rather it is a further
constraint imposed by a separate norm.
What it takes to satisfy our goals does not become more specific once, due to
consideration of coherence, we narrow our range of goals. Even if we have fewer goals
their boundaries are not well defined. But having fewer goals and having goals which
support each other make it easier for us to satisfy them to a greater extent.
9. Two worries
I mentioned two worries that one might have about the possibility of the construction
of reasons of coherence. The first worry was that it would be impossible to construct a
reason that governs the process of construction. Mustn't one already have such a reason
prior to its construction?
The central answer to that worry was already given above: though for most of us the
broad principle of coherence does govern the process of acquisition of new reasons it
does not govern it necessarily. PCB is contingent. It is possible to imagine agents that
are not governed by PCB and so it is possible to explain how it can be constructed from
the perspective of someone who does not possess it. Thus the construction of PCB
does not presuppose PCB. PCB governs only the further construction of reasons.
This leaves us with a residual question. It is easy to see how one can find oneself with
a justified set of coherent reasons that includes a reason of coherence. That is, it is easy
to see how an agent who reflects on her existing reasons will conclude that a set of
coherent reasons that includes a reason of coherence satisfies the constraints imposed by
the procedure of construction. But can a person who does not have a reason of
coherence come to have such a reason?
In particular, can an agent come to adopt a normative principle of coherence if his
other reasons are incoherent? For simplicity's sake let us assume that the agent already
has a pro-attitude in favour of coherence. The question is whether this pro-attitude can
be 'converted' into a reason of coherence given the incoherence of his existing reasons
without yielding strict inconsistency.
Consider an agent with an incoherent set of reasons who deliberates about whether
he should adopt a principle of coherence. Such agent may have good reasons for
deciding to adopt a principle of coherence (or a stricter principle of coherence). Mter
all, each one of his current goals is being negatively affected by the current incoherence
among his reasons and he might fmd that this effect on the satisfaction of his goals is
•• I am mainly concerned here with logical priority, but as we shall see, one may also decide to be more or
less coherent and then the question will be one of temporal priority as well.
unacceptable to him. The problem is that he cannot adopt a principle of coherence
without dropping some of his existing reasons,
and his current set of reasons does not
give him an oveniding reason to drop some of his reasons rather than others. It is
methodically useful here to think of each one of his existing reasons as a homunculus
and think of the situation from their point of view. Each one of them would have had an
interest in the agent adopting a principle of coherence if the burden ofbeing streamlined
fell on the other reasons; however, none of them would like to sacrifice its own normative
authority in order to allow the adoption of a principle of coherence. Could such an agent
rationally adopt a reason
to be more coherent than he currendy is?
Stricdy speaking, the answer is no. But agents can adopt an intention to adopt such a
reason and may implement a variety of strategies for getting there. In particular, they
may inculcate in themselves a greater aversion to such incoherence, or they may wait
till certain grounds for their existing reasons have become weaker and plan to drop
them in a manner that increases coherence when the time is ripe. So an agent may
"work towards coherence" even without adopting a reason to be coherent.
10. A second worry: are reasons of coherence despotic?
The process of construction of reasons as I understand it requires that the agent decides
both to treat a pro-attitude as a reason and that such decision will not stricdy conflict
with previous decisions of the agent about what to treat as reason giving. The second
requirement imposes a normative constraint on our decisions. Even if you decide to
treat a certain pro-attitude as a reason, it is not a reason if it conflicts with your existing
reasons. This raises the worry that there might be a tension between your actual
decision and the decision you should make or would make were you to reflect
properly about your situation. What shall we say about an agent who decides to treat
a certain pro-attitude as reason giving even though the resulting reason is in conflict
with her existing reasons? The problem as I presented it arises out of the particular
structure ofBratman's view of endorsement. But in fact this is a general problem. Any
theorist who thinks that the process of construction of reasons depends on the one
hand on the agent's actual mental states and on the other hand on some normative
constraints that the agent would have abided by if she were fully rational must face this
Two standard ways of dealing with this problem are either to give priority
By "dropping a reason" I mean stop treating it as a goal one takes into account in deliberation. That in
certain remote circumstances it may again be treated as a goal is neither here nor there.
Our reasons change over time. According to my account of reasons, an agent may on occasion decide to
endorse a certain pro-attirude. If he does and the other conditions for there being a reason are satisfied he will
be gaining a new reason, or in other words a new reason will be constructed. When I use the expression "to
adopt a reason" I will mean it as shorthand for "endorsing a pro-attitude in a way that makes it the case that
the agent gains a new reason".
" See Jay Wallace's discussion in this volwne of a similar tension in Korsgaard's thought (Wallace, this
to the nonnative constraints, or to give priority to the agent's actual deliberation or
actual mental states.
This general problem is also a particular problem for someone who thinks that agents
may have reasons of coherence whose role is to govern the acquisition and manage-
ment of one's reasons. How should we think of an agent who ignores her reasons of
coherence in her decision about the adoption of new reasons? Consider a person who
in her actual deliberation endorses a desire that is in significant tension with her other
endorsed desires, while having a reason to keep her set of endorsed desires coherent.
Imagine that she has an existing reason to finish her PhD and that she has just endorsed
a desire to join a performing band whose practice sessions and performances will take a
significant part of her time. There is no strict conflict between the two. It is possible
with an effort to do both, but most likely as a result of joining the band her PhD will
not be of the quality it would have been if she were not to join the band and her
performances will not be as good as they would be if she weren't in the middle of
writing a PhD. Of course, she may decide that it is more important to her to do both
somewhat well than it is for her to do one of these two activities excellently. But
imagine further that she has a reason to keep her set of reasons coherent-and recall
that this reason to be coherent is there precisely to prevent her from adopting reasons
that are in great tension with her current reasons. Such agent if she fully and clearly
reflected would not have endorsed the desire to join the band and therefore would not
have adopted a reason to join the band. But as we are imagining, after reflection
(though maybe not ideal reflection), she did decide to adopt the desire to join the band
as a reason. How should we think of such agent? Note that the problem for this agent is
not that her putative new reason to play in the band is in strict conflict with her reason
to fmish her PhD, rather it is the incoherence between her reason for finishing her PhD
and her reason to play in the band that is in strict conflict with her requirement of
coherence. Either she satisfies the goal of having coherent reasons or she keeps the two
reasons (for PhD and band), but she cannot do both. Should we say that in virtue of this
conflict the putative reason to play in the band is not a reason for her after all, or should
we say that her adoption of a reason to play in the band is a de facto change in her
reason of coherence?
We might be tempted to think that the proposed dilemma is no dilemma at all. If the
agent has a reason to be coherent and she decided against this reason, then the agent is
weak willed and her decision is not one we can truly see as hers. But cases of weakness
of will are often cases in which one acts in ways that one wouldn't endorse on
reflection. The case at hand is a special case in which what one endorses on reflection
conflicts with what one has previously endorsed on reflection. The question is why we
should prefer the conclusions of a person's previous reflection over those ofher current
The worry, in other words, is that my account of the construction of reasons of
coherence under-determines which reasons a person has since it introduces an irre-
solvable tension between the idea that a person's reasons are determined by what she
decides and the idea that her reasons are, at the end of the day, detennined by their
relations with her previous reasons.
I take it that we have two conflicting intuitions concerning this question. We think
that an agent's reasons should not be inconsistent. A set ofincoherent
reasons that also
includes a reason for coherence is an inconsistent set. The putative reasons in this set
cannot all be true reasons. If we were to allow our present decision to determine what
reasons we have, we would end up either with a set of conflicting reasons or with a
constant change in our set of reasons--since the adoption of new reasons would
displace old reasons that the new ones conflict with. In other words, we will end up
with no diachronic stability. Since we think that strict conflict between our reasons is
unacceptable and that diachronic stability is desirable we conclude that the agent's
previous reasons have priority over her actual decisions in the process of determining
new reasons.
But there is also an intuition that pulls us in the other direction. We think that the
actual decision says something about who the p e ~ o n really is as opposed to who she
was or maybe even who she should have been. At most, we think, the agent's previous
reasons tell us something about what reasons she should have had, but, given our
thought that reasons are grounded in the agent's identity, her actual deliberation
determines what reasons she does have. And this intuition suggests that we should
give priority to the agent's actual decision.
I think both of these alternatives are too radical. The answer is somewhere in
between. To see exactly where the answer falls between these two alternatives I will
discuss two analogies.
The thought that we should give priority to the agent's actual decisions about what
to treat as reason giving is supported by an analogy with real construction, construction
of houses and roads and toys. On the view I have been developing, reasons are
justificatory because they are grounded in the agent's identity.
If we then think
that the construction of the agent's identity and therefore the construction of reasons is
like the construction of houses we might be led to the conclusion that our actual
decision should have priority in the analysis of reasons. The agent's identity, we think,
is determined by her mental constitution, by the way her mental states are, not the way
they should have been. When an agent decides to adopt a new reason she introduces a
change in her mental constitution. This change is real change. It is similar to the
changes we make in the world when we build a new house. When you build a house
the house is a new thing, above and beyond the prior reasons for building it. Similarly
you might think when an agent decides to treat certain pro-attitudes as reasons, the
agent is making changes in her constitution, changes that go above and beyond the
In the sense defined by PCB.
" See Michael Brat man's chapter in this volume for a discussion of the problem of alignment between ow
account of the mental states that participate in the constitution of the agent's identity and the mental states
that serve as inputs to nomt construction (Bratman, this volume).
normative constraints on the process of construction. The agent's decision creates a
new fact, which is independent of the reasons for its creation. When you build a house
the house exists even if you did not have a good reason to construct it.
The analogy with real construction suggests that we should give priority (in the
construction of reasons) to our actual decisions and not to the way they should have
been made. We have seen, however, that there are strong reasons to resist this analogy:
if we allowed our current decisions to determine our reasons, we would either lose our
diachronic stability or end up with a conflicting set of reasons.
I think that a different analogy might help us better understand cases of putative
construction of reasons against one's principle of coherence: an analogy with legisla-
tion. The idea is that a decision to adopt a new reason against one's principle of
coherence is analogous to the decision of a political body to adopt a law that is in
conflict with previous laws whose role is to govern the legislative process.
Consider first the case of standard legislation. A parliament makes a law that requires
all future laws not to impede citizens' freedoms of speech--call this the freedom of
speech rule. Such rule should prevent the parliament from passing future laws that
undermine freedom of speech. The parliament may, of course, abolish the freedom of
speech rule, but until it is abolished any attempt to pass a law which conflicts with it
will be considered invalid. The problem is that if the freedom of speech rule has the
status of a regular law it doesn't have much force. After all, if there is in the parliament a
majority for a law that would conflict with the freedom of speech rule and a willingness
to pass such a law, there may very well also be a majority for abolishing the freedom of
speech rule. Indeed, efficient legislators will package the new law and the abolition of
the freedom of speech rule into one act oflegislation.
The analogy with legislation developed so far does not give us the resources to
explain how the principle of coherence gains the special status that enables it to
override decisions to endorse that would lead to incoherence. As far as I can see,
what is at the root of the problem is the view that new acts oflegislation constitute the
will of the legislator in a way that is independent of the history of legislation. The
decision to adopt a new reason, on this version of the legislative analogy, constitutes the
will of the agent in a way that cannot be limited by previous acts of endorsement.
We do have, however, an alternative legislative model. This is the constitutional
model. On this model new legislation constitutes the will of the legislator only if it is
consistent with the existing constitution. A constitution can be understood as a set of
laws that are given a special status, whose role is to govern and limit future legislation.
In a constitutional system a majority of the votes does not in itself constitute the will of
the legislator; regular voting does not in itself construct binding laws. The analogy with
the constitutional model suggests that the principle of coherence function as a special
constitutional law. The idea is that such laws have a special status both practically and
normatively. Even in a non-constitutional system a new law must be consistent with
previous laws, but without the constitutional apparatus we may think of any legislation
that would conflict with old laws as a decision to abolish them, not only de facto, but
de jure as well. The constitutional apparatus blocks this understanding of the legal
process; and if the analogy works then the same is true of the process of construction of
new reasons. On the constitutional model both constitutional laws and certain previous
reasons have a special status. You cannot make a regular law which conflicts with a
constitutional law without abolishing the constitutional law first. Moreover, on the
constitutional model you cannot abolish a constitutional law merely by way of
legislating a conflicting regular law. To abolish a constitutional law you need a special
majority, greater than the one needed to pass regular laws.
If this is the right analogy we can retain our initial thought that a decision to adopt a
new reason that is inconsistent with old reasons--or at least with those old reasons that
have constitutional status--does not constitute the construction of new reasons.
Should we then accept the constitutional analogy? You might think that the best
reason for adopting this analogy is the fact that if we give priority in the construction of
reasons to the agent's actual decision we will end up with an inconsistent set of reasons
and that such set is unsatisfiable. But this argument is too quick. Insisting that our
reasons should be co-satisfiable only supports a wide scope principle of consistency.
A wide scope requirement of consistency is a requirement to: either not adopt a new
reason if it is inconsistent with an old one, or adopt it and drop the old one. Insisting
that our reasons should be consistent may justify rejecting the analogy with the
construction of toys. Considerations of consistency tell us that we have to insist on
some normative standards in the construction of reasons. However, these considera-
tions cannot explain why old reasons and in particular reasons of coherence have
priority over the agent's actual decisions; thus they do not fully support the constitu-
tional analogy.
There is, however, something else to be said for this analogy. Ask yourself: why do
we have reasons? We need reasons to regulate our desires. We know that our desires
sometimes take us where we do not want to go (among other things, they prevent us
from achieving goals that require long-term planning and goals that require social
coordination) and we have mechanisms for preventing this from happening. Reasons
are constituted by these mechanisms. If we want our reasons to be able to resist
temptation we must grant them immunity from easy abolition. Otherwise every
tempting desire will take the guise of a new reason and will reject our old ones just
so it can be satisfied. We have reasons in part to stabilize and prevent quick changes in
our identity: changes that are motivated by passing urges, whose consequences have
not been properly considered. It is important that temptation often disguises itself as
dear-headed rational reflection. One of the functions of our reasons is to give ourselves
enough time to get back to our senses. Our notion of reasons has therefore a lot in
common with the constitutional model. What we want is not mere consistency, but
consistency and (at least some) diachronic stability.
I said earlier that the dilemma we are facing applies to all of our existing reasons. It is
important to see, however, that it is of special relevance to reasons of coherence. While it
would undennine our identity if ordinary reasons would not be stable over time, it would
not be a disaster if some of them enjoyed only limited stability. Reasons of coherence, by
their very nature, affect our entire system of reasons; if these could be rejected merely
because they do not fit with a new prospective reason they would not be able to play the
organizing role they are meant to play. The constitutional model is particularly relevant to
our understanding of the normative status of reasons of coherence.
But the traditional constitutional model is problematic, and if our notion of reason is
analogous to this model it too is bound to be problematic. Rousseau identified the
central problem correctly. The idea of an autonomous people governed by a constitu-
tion is incoherent. In The Social Contract he says: "if the established order is bad, why
should the laws which prevent its being good be regarded as fundamental? Besides, a
people is in any case entirely at liberty to alter its laws, even the best laws" (Rousseau
1968, p. 99).
You might think. that Rousseau overstated the requirement of liberty. To be
autonomous, a people needs to be self-governed. But why think that the identity of
a people is determined solely by the votes of its current members and not also by its
The idea that a constitution partly expresses the identity of a people might seem
acceptable as long as it can be changed by a high percentage of the present population.
Under this condition a constitution seems like a device for preventing hasty changes
that nevertheless lets the present population have the fmal say.
But the idea loses its appeal when you take a step back and think not of a specific
constitution with a fixed rule for introducing constitutional changes but of the consti-
tutional mechanism that is at the background of any constitution. This mechanism is a
generic tool for a people to impose its views on future generations. After all, what
prevents a legislator from deciding that a constitutional law cannot be changed without
the consent of 90 or 99 per cent of future votes? Rousseau was right in principle: a
people that is constrained by a constitutional mechanism is not autonomous.
The same is true of our reason-formation mechanism. If we can construct reasons
that are immune to change we are in the grip of a mechanism that threatens to limit our
freedom. On the other hand, as we have seen above, the whole point ofhaving reasons
is that they can limit and guide our actions and to do that they must be also capable of
limiting and guiding the process by which we form new reasons.
Go back to the political case. One solution to our dilemma is to restrict the
percentage of the population that is required in order to change a constitutional law.
For example, restrict it to 66 per cent of the population. And naturally you might think
the same is true about reasons. Reasons, we might say, should never be constructed in
such a way that it is unreasonably hard to change them.
But who has the authority to decide exactly how hard it should be to change our
reasons? If we claim that there is an a priori fact about how hard it should be, we are
giving up on the idea of independent constructivism. If on the other hand, we leave
it to the agent to decide then we are again under the threat of losing our future
Of all the models we have surveyed, the constitutional model seems to me to be the
best analogy for the process of construction of reasons. It does, however, leave us with a
dilemma. How can we justify the claim that certain constructions are too limiting?
An agent may develop her reasons in such a way that she will not leave herself
nonnative leeway for change. She may decide to adopt a narrow scope, strict standard
of coherence that will prevent her from adopting any new reason that does not cohere
with her current reasons. Imagine that our agent has three reasons: a reason to pray, a
reason to do what is necessary for survival, and an extreme coherence principle that
prevents her from adopting any reason that will take time from her praying if it is not
for the sake of survival. Such a person is an extreme ascetic. Finally, following the
constitutional analogy, imagine that reasons of coherence always have priority over the
agent's decisions about what to treat as reason giving.
Some such ascetics lose their zeal after a while, they stop caring about praying or they
stop caring about coherence. They are not ascetics any more because they lose the
motivation for their asceticism. But some ascetics are tom. They hold fast to their
reasons for asceticism but also feel a strong pull for having company or seeing a movie
or eating chocolate mousse or having sex. If we understand their reasons of coherence
along the lines suggested by the constitutional model, these agents will have to see their
desires for company or sex as outlaws, urges that must be repressed. They will have to
see them as outlaws because their principle of coherence will prevent them from
adopting any reasons they do not already have and will also prevent them from
abolishing this very principle. Indeed, this is how many ascetics see them.
The constitutional model tells us that the ascetic we are imagining has no reason to
have sex or indulge in a good meal regardless ofhow strong her desires to do so are.
So far I think the constitutional model has got it right. The constitutional model also
tells us that on reflection such agent would not adopt reasons that conflict with her
ascetic tendencies. And fmally, the constitutional model tells us that if our agent in her
actual deliberation purports or attempts to endorse the desire for sex or food, such
endorsement would not be real endorsement and therefore would not ground the
adoption of new reasons. Here I think that the constitutional model goes too far.
Sometimes a purported endorsement that conflicts with what the agent would have
endorsed if she fully subscribed to her previously endorsed reasons is indeed no
endorsement at all. But sometimes an agent finds herself incapable of endorsing in
accordance with her previously constructed reasons. She has no reason to reject her
reasons of coherence-indeed, it is not clear what could count as such a reason given
the constitutional model-but she is also not willing to ignore her new-found interests.
Note that most of us do not find ourselves in this position for reasons that are orthogonal to our current
concerns. Most of us have reasons to seek physical pleasure and avoid physical displeasure. Thus our desires
for food or sex always have some--even ifindirect-roots in our system of reasons. But this is not the case of
the ascetic we are imagining now. She has no reasons, let us stipulate, to have physical pleasure or to avoid
physical displeasure.
The same situation arises in the political arena: sometimes a segment of the popula-
tion-larger than a regular majority but smaller than the majority needed to change
the constitution-finds itself incapable of following an overly constricting constitution.
In these cases we have a dilenuna. On the one hand, giving overriding authority
to the constitutional laws or to reasons we have adopted in the past seems to give
despotic powers to these laws and reasons. On the other hand, we cannot, for reasons
we rehearsed above, ignore the constitutional laws or the agent's past reasons
and ascribe overriding authority to a people's current majority or an agent's actual
I think the only way to understand these cases is as cases in which the legal or
normative system collapses. Cases in which the tension between one's previously
adopted reasons and one's ability to accept these reasons in her actual deliberation is
too strong are cases of dissolution of one's normative system.
In these cases neither her
old reasons nor her new decision have nonnative authority. She has no reason to
pursue new goals that are incoherent with her old ones but neither does she have a
reason to abide by her reasons of coherence. Most often after such collapse a new
normative system will emerge, a new stable set of reasons that do not conflict with each
other and that the agent is capable of endorsing in her actual deliberation.
According to Frankfurt, St Augustine thought that the transition from volitional
division to psychic unity (that is, coherence) requires a miracle and therefore he prayed
for conversion (Frankfurt 1992, p. 16). If the view I have presented here is correct
then when our reasons of coherence lose their nonnative priority, we can pray for
coherence but we cannot have a reason to pray. In these circumstances a normative
system and with it the agent's reasons of coherence must be constructed anew.
Brannan, Michael. 1996. "Identification, Decision, and Treating as a Reason," in Philosophical
Topics 24(2), pp. 1-18.
Brannan, Michael. 2007. "Reflection, Planning and Temporally Extended Agency," in Strudure
of Agency (Oxford: Oxford University Press), pp. 35-61.
Broome, Jolm. 2008. "Is rationality normative?" in Disputatio, 11, pp. 153-71.
Frankfurt, Harry, 1992. "The Faintest Passion," in Proceedings and Addresses of the American
Philosophical Association 66(3), pp. 5-16.
Kolodny, Niko. 2005. "Why Be Rational?" in Mind 114:455, pp. 509-63.
Korsgaard, Christine M. 2008a. "Realism and Constructivism in Twentieth-Century Moral
Philosophy," in The Constitution of Agency (Oxford: Oxford University Press), pp. 302-26.
Korsgaard, Christine M. 2008b. "Self Constitution in the Ethics of Plato and Kant," in The
Constitution of Agency (Oxford: Oxford University Press), pp. 100-28.
'' Or pans of one's normative system.
11 S ~ c i a l thanks to Michael Bracman, James Lenman and Valerie Tiberius for detailed and helpful
Lenman, James. 2012. "Expressivism and Constructivism," this volume.
Rousseau, Jean-Jacques. 1968. The Social Contract (London: Penguin).
Shemmer, Yonatan. 2007. "Desires as Reasons," in Philosophy and Phenomenological Research 75
(2), pp. 326-48.
Wallace, R.Jay. 2012. "Constructivism about Normativity: Some Pitfills," this volume.
A Problem for Ambitious
Metanormative Constructivism
Nadeem]. Z. Hussain
1. Introduction
We can distinguish between what I will call ambitious metanormative constructivism and a
variety of other constructivist projects in ethics and metaethics.
To put the point
roughly for now, ambitious metanormative constructivism is the project of either
developing a type of new theory, worthy of the label "constructivism," that is distinct
from the existing types of metaethical, or metanormative, theories already on the
table--various realisms, non-cognitivisms, error theories and so on-or showing that
the questions that lead to these existing types of theories, or the discussions that involve
them, are somehow fundamentally confused. Natural ways of pursuing the project of
ambitious metanormative constructivism lead to certain obvious, and related, worries
about whether the ambitions are really being achieved-that is, whether we really are
being given a distinctive theory-and whether such accounts are viciously circular.
I will argue that responding to these initial worries pushes ambitious metanormative
constructivism towards adopting a kind of position that I will call "constructivism all
the way down." Such a position does see off most, though not all, of the above initial
worries. However, it faces a distinctive objection that is a descendent of Bertrand
Russell's "Bishop Stubbs" objection against coherentist theories of truth. As it turns
out, it is hard to tell whether the objection is fatal for ambitious metanormative
constructivism; however, it shows, nonetheless, that a distinctive philosophical task
needs to be carried out by such a constructivist before we will have any reason to think
that such a theory can overcome the objection.
The theories of concern to me are actually theories that aim to provide a unified metanonnative and
metaevaluative account. However, I will be using two shorthands in this chapter. I will use "ambitious
metanoi1J12tive constructivism" to refer to such unified theories, and, instead of using some neologism such as
metan011J12tive and metaevaluative studies, or what have you, I will just refer to the relevant sub-discipline as
For reasons of space, I will assume a general familiarity with the spirit of construc-
tivist projects that take themselves to be, in some way or the other, challenging
mainstream metaethical theories or theorizing. In what follows, I will be relying on
this sense of the spirit of ambitious metanormative constructivism in order to rule out
various ways of developing the metaphor of constructivism, or various ways of
responding to objections to constructivism, as not being in this spirit. I will not have
much to say here to someone who thinks that I am simply wrong about what motivates
ambitious forms of metanormative contructivism. Similarly, constructivists who do not
share those motivations will also have no reason to fmd much force in the objection
I try to articulate.
Let us start with a very abstract schema for constructivist accounts for some domain D
of judgments:
(1) For any claim Pofsome domain D, CONSTRUCTS (outcome Opis produced
when a certain class of subjects S follow procedure R in conditions C, P)
Iffor some proposition the CONSTRUCTS relationship is actually instantiated, then
we can conclude that P. What precisely the CONSTRUCTS relation is will of course
be part of what needs to be specified by some particular instance of the schema. For
ease of exposition, I will occasionally talk of the "constructor" and the "constructee."
The "constructor" is the first relatum in the above schema-it is what constructs P, the
I have put the schema in this abstract way in part because the English word
"constructs," or the expression "is constructed by" forces one, like most terms for
relations, to use a noun phrase to state the relata. And this in tum inevitably leads to
expressions of the form such-and-such constructs the fact that P. One might worry that,
despite whatever protestations to the contrary, talk of the fact that P is already
smuggling in realist assumptions. Instead the schema officially tries to leave open
how we are to take the constructed item, and the constructor for that matter. I have
also put the subscript "P" on "0" to mark that the outcome in question has to be
appropriately discriminating between P's being the case and P's not being the case.
For the purposes of my argument below, we can also allow the construction procedure
C£ Mark Johnston's requirement that the response be "x-directed" (Johnston 1993, p. 103). I also do
not mean to rule out accouna in which a single domain, in some intuitive sense, is divided into relevant sub-
domains with different subjeca, procedures, and circumstances being relevant. For example, the fact that in
general a batsman is not out when the ball touches the wicket as long as the bail is not completely removed
from the top of the stumps is constructed by the relevant vote of the MCC; however, whether a particular
batsman is dismissed, say, on a particular occasion can be a matter of whether the umpire has given him out on
appeal. (Of course, that is not a necessary condition, since the batsman can be dismissed if he is out under any
of the Jaws and chooses to leave his wicket.)
to be indexed to the particular proposition or its subject matter. Thus, perhaps, the class
of subjects, the procedures they follow, and their circumstances could all be topic
specific. Nothing in the argument below turns on this not being the case; I have left out
the relevant subscripts for simplicity's sake.
Finally, for the kinds of accounts that concern us, the CONSTRUCTS relation will
usually be a trans-world relation since the claim will not be that the outcome OP is
produced in the actual world, but rather that it would be produced in the relevant
possible world where a certain class of subjects followed the required procedure.
Without this, depending upon the details, there will be hardly any facts of the matter
for the domain or most claims will tum out to be false.
3. Specifying ambitious metanormative constructivism
One initial constraint on the relationship CONSTRUCTS is that the constructor should
determine whether Prather than simply reflecting an independendy determined fact that
P. In other words, there needs to be some sense in which P is the case because the
procedure has yielded the relevant result. We do not want it to be the case that the
procedure yields this result because it is independendy the case that P. To put the intuitive
point a third way, we need to have the relevant kind ofEuthyphro contrast maintained.
Let me immediately rule out one relation, or perhaps family of relations, which
would not give us ambitious metanormative constructivism. Plausibly for almost any
normative or evaluative predicate F, if x is F, then x is Fin virtue of, given some other
predicate G, x's being G. G is F-making. These are forms of what Gideon Rosen calls
the "grounding idiom" (Rosen 2010). The serial killer is evil because he intends to kill
the next hitchhiker he comes across. Take even the predicate " ... is a reason." Some
consideration, that the grass is brown, say, can have the property ofbeing a reason, say
being a reason to water the grass. However, intuitively, the consideration will have the
nonnative property of being a reason in virtue of some other property it has, say, the
property of being what I desire not to be the case.
Why do I say almost any normative
or evaluative predicate? Well, we might think that there are some cases in which it is a
bare or brute fact that something is good or bad. Perhaps, for example, pleasure's
goodness. Is pleasure good in virtue of some other property of pleasure? Perhaps not. In
any case, some such grounding relationship is one that every metaethical theory grants
though different theories will have different accounts of it.
Compare the distinction Wright draws between "the Pinceon biccmditiortal for P: P is true if and only if
were P applaised under conditions, U, P would be believed," where U are generally ideal conditions for
inquiry, on the one band, and a "corresponding PNmamitln biccmditiortal for P: Pis true if and only if were
P appniled under topic-!peci6cally rufficiently good conditions, P would be believed" (Wright (1 CJ87) 2003,
p. 293).
• a. Wright tm, PP· to&-JCJ.
• Cf. Darwall 1998, p. 6.
Though not all may be able to do an equally good job (Dancy 2004).
I will also set aside some versions of constructivism that aim to provide metanorma-
tively neutral constructivist accounts of relatively limited domains. Examples might be
the constructivisms of Rawls and Scanlon. Whatever is the right way to interpret them,
the accounts I will put aside are accounts in which (i) the constructor is expressed in
normative or evaluative terms, (ii) no metanormative account is given of the con-
structor, and (iii) the CONSTRUCTS relationship is not metanormatively insulating. In
other words, for all the theory says, the constructee inherits whatever metanormative
account we give of the constructor. Such accounts do not attempt to carve out a new
type of metanormative theory and so are not, for my purposes, metanormatively
ambitious. (Which is, of course, not to say that they cannot be very ambitious, and
very interesting, in other ways.)
Instead the constructivisms I am concerned with are those that are metanormatively
ambitious in the precise sense that they aim to provide a new type of metanormative
theory. In principle such constructivisms could be metanormatively ambitious about
some particular domain; they could claim to be providing a metanormative theory that
instantiates a new type of metanormative theory even though it is a theory for only
some limited domain of normative or evaluative claims. However, I will focus here on
theories that attempt to provide a constructivist account of the entire domain of
practical normative judgments?
4. The relation and the relata
What then are plausible accounts of the CONSTRUCTS relation for ambitious
metanormative constructivism? An intuitive thought here is that the CONSTRUCTS
relation is either identity or some close relation of constitution.
We could try to
capture this relation with identity combined with a subjunctive conditional:
(2) P = Op would be produced were S to follow R in C.
Or perhaps a biconditional and some accompanying commentary:
(3) P iff OP would be produced were S to follow R in C.
Both of these strategies generate further puzzles. Instead, I will use the subjunctive
conditional on the right-hand side of these schemas but embed it in a negative
One rationale for this focus is that most reasons for providing an ambitious metanonnative constructiv-
ism for some particular domain of nonnative or evaluative claims are also reasons for providing such an
account for the entire domain of practical normative claims. For reasons of space, I will not try to defend this
claim here.
• An important additional possibility is the idea that the CONSTRUCTS relation actually provides
somehow, or in effect, a distinctive account oft be truth predicate for the domain in question. The discussion
below will not directly address this possibility though much of the discussion will be relevant to assessing any
such account.
• Recall the point I made earlier that the CONSTRUCTS relation would need to be trans-world.
condition, a condition tailor-made for the specific purposes of ambitious metanorma-
tive constructivism:
(4) There is nothing more to its being the case that P than that OP would be
produced were S to follow R in C.
Even though I will use (4) in the discussion below, I will talk of the left-hand side
{LHS) and the right-hand side (RHS) of such claims, where the left-hand side goes into
the first slot in the following and the right-hand side into the second:
(5) There is nothing more to its being the case that ... than that ....
I should make clear that just because P appears as a subscript to 0 does not immediately
mean that we have some kind of problematic circularity here. The theory needs a
specification of the relevant outcomes, but it would require further argument to claim
that such a specification does have to involve problematically appealing to P. I am also
not asswning that the outcome would have to be a propositional attitude like a belief or
a judgment. For example, we could allow for the outcome to be a non-cognitive
attitude, an act, or even just behavior. What is required is that the theory be able to
specify a sufficient number of these, so to to give us outcomes for all
the claims of the domain. If we do take the outcome to be a judgment or a belief with
the content P--the claim on the LHS-then there are some special puzzles that arise
but I will again mosdy ignore those puzzles here.
Now if the RHS or constructor is specified in non-normative and non-evaluative
terms (and assuming that no non-natural terms are used instead) we presumably just
have a naturalistic reduction.
Such reductions are quite ambitious, are metanormative
accounts, and probably do deserve the label "constructivism" precisely because they
might appropriately satisfy our schema.
Most contemporary naturalist realisms in
metaethics, such as those of Peter Railton or Michael Smith, might well be such
forms of constructivism. Presumably most of us are such constructivists about the laws
of cricket, for example-putting aside important concerns, for now, about whether the
act of voting by a member of the MCC can be stated in non-normative and non-
•• Assuming also that no special problems are raised for the naturalism by the subjunctive conditional. Part
of what would need to be assessed in further detail is whether the reduction is really a reduction to the
features of the actual world that are in tum the reduction basis for the subjunctive conditional. Cf. Railton's
approach in his reductive account of individual good: "what makes some or other end or activity be part of an
individual's good is not the fact that he would, were he ideally informed (and so on), desire that his actual self
pursue it, but rather the existence of the reduction basis for that counterfactual, namely, the particular
constellation of law-governed features of the actual individual and his circumstances in virtue of which these
claims about idealized hypothetical desires hold. Thus, the truth-condition of the claim that such-and-such is
good for a given individual is direcdy given by the existence of this constellation of features, without detour
through idealized desires" (Railton 2003, pp. 62-3).
Note that the strategy of fmding a reduction basis for the subjunctive in actuality is obviously not restricted
to the naturalist.
Appropriate satisfaction would require ensuring that the Euthyphro contrast is present. Cf. Wright
1992, pp. 108-39.
evaluative terms. However, any such constructivism would not be ambitious in the
sense I am after since it would not be attempting to be an instance of a new type of
metaethical theory (or be in the business of showing how mainstream metaethics is
confused). Thus in order to have the kind of ambition I am looking for, we can put
aside constructivist accounts in which the RHS of (4) can be stated in non-normative
and non-evaluative terms.
Even if the RHS needs to be stated in normative or evaluative terms, this still leaves
three basic options open. Assuming for the moment-controversially, no doubt-that
there is a fundamental distinction between the normative or evaluative concepts of
practical reason and those of theoretical reason, we can distinguish between RHSs that
can be stated only in the nonnative or evaluative terms of practical reason, those of
theoretical reason, or both. If the RHS could be stated only in terms of the normative
or evaluative concepts of theoretical reason, then we would, in effect, have a reduction
of practical nonnativity to theoretical normativity. What kind of metanormative
account, if any, that would yield would then depend, not surprisingly, on what kind
of metanonnative account we gave of theoretical normativity. I, however, am not
going to investigate that branch further explicidy because I suspect that most of us do
not think that the practical can be reduced to the theoretical in this way. Such accounts
are also not in the spirit of much that goes under the label of constructivism. In any
case, much of what I will discuss in what follows can be applied to such accounts even if
I do not take the space to do it explicidy here.
The accounts of interest here, then, are those in which the RHS above can only be
stated using, in part, normative or evaluative terms of practical reason. Now, such an
account seems not to be metanonnatively ambitious since we still seem to need a
metanonnative account for the RHS. After all if the RHS is given, say, a non-
cognitivist reading, then, it seems, we will give the LHS a non-cognitivist reading.
Here, however, we need to note that our constructivist is committed to a certain
claim anyway that might well provide a solution to just this worry. Recall that (4) is
meant to hold of any nonnative or evaluative claim (of practical reason, that is, but
I will leave off that qualification from now on). This now includes the RHS of (4).
Thus the constructivist is committed to:
(6) There is nothing more to its being the case that Op would be produced were S
to follow R in C than that OQ would be produced were S to follow R in C
where Q is just the claim that OP would be produced were S to follow R. in C. (We see
again how important it is that the theory be able to generate all the different outcomes
it needs and how this might not be a trivial requirement.) The new RHS in (6) will be
nonnative too, of course, and thus some version ofthe schema (4) will hold for it too
•• I am also putting aside mixed accounts, though, again, much of what I say below would be quite
relevant to assessing them.
with another normative RHS. And so on. In this way we have a form of"constructiv-
ism all the way down. "
AJ I said, it looks as though our constructivist is required to say this, but it also looks
as though we might fmally have the needed kind of ambition. When challenged with
whether the RHS of any such identity is to be given a realist, non-cognitivist, or error-
theoretic interpretation, the constructivist replies that the interpretation to be given is,
of course, a constructivist one. There is no place-or no need--110 the constructivist
claims to tell any of the other stories.
Note that the goal here is not to provide an epistemic story on the behalf of the
constructivist. The opponent whose challenges we are here concerned with is not
necessarily one who questions the justifications of the relevant normative belie&.
Consider an analogy. I may completely grant that indeed there is a table in front of
us and that you and I both have sufficient justification for our beliefs that there is such a
table. However, I might think there are further questions about the constitution of
tables or &cts about tables. I might think that the following is a possibility we should
(7) There is nothing more to its being the case that there is a table in front of us than
that certain collections of atoms are arranged thusly.
You and I could disagree, or so I claim, about (7), but this is not a disagreement about
whether there is a table in front of us and (7) need not be intended as an answer to any
such potential disagreement.
Rather, disagreement about (7) is the kind of disagree-
ment we want to capture here: disagreement, say, about whether human practices of
treating the relevant objects in certain ways are part of what make it the case that there
is a table in front of us. The schema in (5) is selected precisely so that our ambitious
metanormative constructivist can engage with the analogues issues in metaethics.
The ambitious metanonnative constructivism I am attempting to articulate here is thus
also not a form of quietism. The quietist constructivist, as I am, perhaps stipulatively, using
the term, would deny that anything more than (4) needs to be part of the theory. Recall,
however, that the pressure to include (6) comes from the apparent need to give a
metanonnative account of the RHS. AJ I pointed out earlier, the natural thought to
have is that the constructee inherits whatever metanonnative account we give of
the constructor. And without a metanonnative account of the RHS on behalf of
constructivism, we seem to have left the field open to all the standard metaethical
competitors. Since the relation represented by the schema in (5) is, as far as we have
seen, not, to use my earlier locution, metanonnatively insulating, whoever wins this
" Surely, one might claim, lots of non-nonnative facts are going 10 be part of what detennines some
normative matter in most Ca3es-«her than perhaps some subset of pure normative principles. We should
not, however, worry that a schema like (6) must inevitably make the mistake ofleaving these non-nonnative
facts out. These can easily be written into the procedure R and the circumstances C.
•• My thanks to a conversation with Elijah Millgram for prompting me to add this clarification.
competition will then get to tell the metaethical story for the nonnative and evaluative
claims of practical reason. Our constructivist would tum out, then, not to be metanonna-
tively ambitious. Now there may well be some sound argument on the behalf of quietist
constructivism that denies the need to go beyond (4); however, for the purposes of this
chapter, I will not be developing or assessing that kind of view.
The form of constructivism "all the way down" that we have developed in this
section, then, fmally has the kind of metanormative ambitions I have been looking for.
The question, of course, is whether there is a philosophically plausible account here.
5. The problem
It may seem obvious that there must be some VICious circularity or regress here;
however, I think that any such charge takes some work to make it stick. To start
with, note that (4) and (6) are not in the business of providing an informative analysis of
normative concepts in the way in which that project is traditionally understood. There
is no claim, for example, that someone who did not understand the concepts being
used to state the LHS could nonetheless acquire such an understanding by seeing the
presentation of the RHS.
Obviously, there is an episternic question that any such constructivism must face: it
must explain how we come to know (4), (6), and so on-the claims that constitute the
constructivism "all the way down." Here the constructivist could appeal to conceptual
truths but not ones that are meant to provide an analysis in the traditional sense. But
note that the constructivist does not have to if he or she has some other account ofhow
we could come to know these truths. For now let us just add the task of giving the
needed epistemology or appealing to the epistemology of the conceptual to our to-do
list. There are problems that will emerge here, but they are not the ones I want to
focus on.
Instead, let us take the claim that:
(8) Bishop Stubbs was irrational.
(Practically, that is.) Now any particular way offilling out the schema in (4) will give us
a way of saying what the RHS is. The general schema would be:
" Perhaps the easiest response to epistemic worries is to claim that these are conceptual truths, but then
any such account will face open-question style arguments. If these are not conceptual truths, then some other
epistemic story will need to be told. Since, in that case, these truths will themselves, for all that has been said so
far, have to receive a constructivist account, the epistemic story will have to fit with this constructivist
account. How all that would be made to work out is not something I will address in this chapter.
For the purposes of this chapter, I will also ignore a range of potential problems lurking under the
traditional heading of the "Conditional Fallacy." For useful discussion of such matters in the context of
rejections of realism, see Wright (1987) 2003 and Wright 1992, pp. 117-20. However, Wright's own
response will not work, I suggest, for the purposes of ambitious metanormative constructivism precisely
because it yields a framework within which it is only the case that, as he puts it, ''truth within a particular
discourse is partially detennined by best opinion" (Wright 1992, p. 120).
(9) Oc
l would be produced were S to follow R in C.
And the relevant RHS for this in tum would be:
(10) 0(9) would be produced were S to follow R in C.
And so on.
If asked why we do, or do not, think (8) in some everyday context, the natural
response is of course to point to what Bishop Stubbs thought, did, or said.
A metaethicist will, however, also want to know what kind of fact, if any, (8) is
supposed to be, or what precisely we are doing when we apparently assert something
like (8). In other words, the metaethicist will want to know what is the correct
metanormative account for (8). As I mentioned above, certain kinds of quietism are
certainly, to repeat the locution I used earlier, in the spirit of constructivism; however,
our constructivist claims to have a positive, ambitious response to the metanormative
question. One could simply assert an identity of (8) with propositions or facts about
what Bishop Stubbs thought, did, or said. And, as I noted earlier, this would just mean
that the metanormative question would have to be answered by looking closely at
what we make of claims about what people think, do, or say-the metanormative task
would just have been put off. Our earlier range of possibilities looms again. If we can
give a non-normative account of what Bishop Stubbs thought, did, or said, then it
looks like we have a reduction of the normative to the non-normative. If, on the other
hand, it is a normative matter what Bishop Stubbs thought, did, or said, then, for all we
have seen, the metanormative account of (8) will be determined by the metanormative
account of what it is for someone to think, do, or speak. No distinctive metanormative
account yet.
Our ambitious metanormative constructivist, however, is committed to construc-
tivism all the way down. Thus, either right off the bat, or as an account of a supposedly
normative account of thinking, saying, and doing, such a constructivist will point to
something like (9). And when pressed on (9), again the constructivist will point to (10).
And so on.
And again the constructivist has to, in order to avoid one of the standard
metaethical accounts.
So fur, no surprise. However, the problem is that a similar sequence can be
constructed for the claim that:
Recall that the issue here is not epistemic.
Notice that this is why our constructivist cannot simply claim that the fact that the relevant procedure
yields the required result is constituted just by that very same procedure yielding that very same result: for
example, where q8) and, the relevant procedures, etc. would just be identical to 0('1)• and so on. This would
be equivalent to not having a response to the metanormative question about (9). Our ambitious metanor-
mative constructivist has to be able to say something new-and recognizably constructivist--at each stage in
order to defend the claim that he or she has a distinctive metanormative view. My thanks to Michael Bratman
for raising the issue.
Of course, it should also be pointed out that in any case we would still have the symmetry that is the source
of the problem I lay out in what follows.
(11) Bishop Stubbs was not irrational.
(12) q
) would be produced were S to follow R in C.
(13) q
) would be produced were S to follow R in C.
And so on.
Now the intuitive thought, of course, is that both (8) and (11) cannot be the case.
A natural reaction would be that what explains this is that only one of(9) or (12) is the
The natural reaction to have here is that surely as a matter ofjact either O(s) or
> is produced when the relevant procedures are followed.
But, crucially, all that there is for it to be the case that either outcome is produced on
the constructivists account is (10) or (13) respectively. And for these in turn all there is
for them to be the case is that the next element in the sequence is the case. And so on.
5.1 Russell's Bishop Stubbs objection to coherentism about truth
We can now begin to see why there might be an objection here that is a descendant of
Bertrand Russell's "Bishop Stubbs" objection to coherentist theories of truth. Russell
... the objection to the coherence theory lies in this, that it presupposes a more usual meaning of
truth and falsehood in constructing its coherent whole, and that this more usual meaning, though
indispensable to the theory, cannot be explained by means of the theory. The proposition
'Bishop Stubbs was hanged for murder' is, we are told, not coherent with the whole of truth or
with experience. But that means, when we examine it, that something is known which is
inconsistent with this proposition. Thus what is inconsistent with the proposition must be
something true: it may be perfectly possible to construct a coherent whole of false propositions
in which 'Bishop Stubbs was hanged for murder' would find a place. In a word, the partial truths
of which the whole truth is composed must be such propositions as would commonly be called
true, not such as would commonly be called false; there is no explanation, on the coherence
theory, of the distinction commonly expressed by the words true and false, and no evidence that a
system of false propositions might not, as in a good novel, be just as coherent as the system which
is the whole of truth. (Russell 1906-7: 33-4)
Ralph Walker puts the natural response on the behalf of the coherentist as follows:
Russell thought this an objection to the coherence theory, but it is not, for the coherence theory
is concerned with coherence not amongst arbitrary propositions, but amongst beliefs. (Walker
1997, p. 310)
There is the separate question of whether the theory embodied by the two sequences is the correct
theory for such claims and how we would know that. Here I am just assuming that we have the correct
theory and, as I emphasized earlier, I am, for the most part, putting aside epistemic issues.
What about construction procedures that need to actually be carried out? It is hard to imagine an even
remotely plausible account along these lines that would not lead to massive indeterminacy. More crucially,
however, for reasons already discussed, ambitious metanonnative constructivism along these lines would still
have to require that there is nothing more to its being the case that the procedure has yielded the relevant
outcome than that that, or some other, procedure has yielded the further relevant outcome. And so on.
As Crispin Wright emphasizes in his assessments of Walker's discussions of coherent-
ism, this is a particular instance of a general strategy that responds to Russell's worry by
"earmarking certain propositions as in some way privileged" (Wright 1999, p. 220).
So, for example, given our beliefS it just will not turn out that both "Bishop Stubbs was
hanged for murder" and "Bishop Stubbs died of natural causes" are coherent with
them. The privileging response thus disposes of the original Bishop Stubbs objection.
Or so at least Walker seems to think.
As Wright points out in his critical study of
Walker's book The Coherence Theory of Truth, this will depend in part on how
coherence is eventually characterized (Wright 1995, p. 282). I will put this aside in
order to focus on the objection Walker goes on to make against the privileging
response. This is what Wright calls Walker's Master Objection to coherentist theories.
As Wright's interpretive attempts make clear, there may well be several different
objections lurking under this heading.
What lies at the heart of these concerns, though, is that if the coherentisrn is "pure"
enough--to use Walker's phrase--a descendant of the Bishop Stubbs objection applies to
the statements that make some propositions privileged. In the cases that Walker takes up,
these are claims that a certain belief in one of the privileged propositions is actually held.
For a pure coherentist, this too is a claim whose truth "must consist in coherence ... with
the other beliefS that are also held" (Walker 1997, p. 316). However, the same applies to
claims about these other beliefs. For Walker this means straightforwardly:
that the Bishop Stubbs objection recurs after all. We can easily denominate an arbitrary set of
internally coherent propositions including "Bishop Stubbs was hanged for murder", such that for
each proposition Pn in that set, "It is believed that Pn" coheres with the original set .... What the
theory requires is that it should be a fact that certain things are believed, a fact that obtains in its
own right and not in virtue of some further coherence. A pure coherence theory of truth, which
holds that truth always consists in coherence, cannot accommodate this. (Walker 1997,
pp. 31Cr17)
As Kirkham points out, the argument can be generalized so that it holds against any
"pure" coherence theory with any account of what the designated set is. Thus we get
the sequence:
p coheres with the designated set
'p coheres with the designated set' coheres with the designated set
etc. (Kirkham 1992, p. 115)
•• For the point that it is not clear why Walker thinks that the "privileging response demands invocation of
belief," see Wright 1995, p. 282.
•• James Young calls the objection the "specijialtion probkm" and calls the privileging system of beliefs the
"specified system" (Young 2001, p. 93).
•• & does Young 2001, p. 93.
5.2 Bishop Stubbs and ambitious metanormative constructivism
Now, I should make clear that I do not think that the constructivist has to be
committed to a coherence account of the property of being true, let alone a "pure"
coherence theory. Indeed, it would probably be best for the constructivist to avoid any
such controversial view. Thus our ambitious metanormative constructivism could be
construed as the view that, for example, the proposition that Bishop Stubbs was always
irrational is true because things are as it says they are, namely, that in fact Bishop Stubbs
was always irrational. Constructivism would then be an account of the fact that Bishop
Stubbs was always irrational. That fact that Bishop Stubbs is irrational just is the fact that
the relevant outcome would be produced by the relevant procedure.
In any case, what I do want to argue is that though our ambitious metanormative
constructivist need not be a coherentist about truth, there is still a version, or a
descendant, of the Bishop Stubbs objection that does apply to such a constructivism.
Return to our two sequences. All there is to its being the case that:
(8) Bishop Stubbs was irrational
is that it is the case that:
(9) O(s) would be produced were S to follow R in C.
And all there is to its being the case that (9) is that:
(10) 0(9) would be produced were S to follow R in C.
And so on.
Similarly all there is to its being the case that
(11) Bishop Stubbs was not irrational
is that:
(12) O(lt) would be produced were S to follow R in C.
And all there is to its being the case that (12), is that:
(13) 0<
) would be produced were S to follow R in C.
And so on.
The temptation is to say that only one of these sequences can be the case.
But all
there is for each member in the sequence to be the case is that the next member is the
Put in this context, I do not think the term "fact" smuggles in ontological commitments of any kind
that should acrually be anathema to the spirit of constructivism.
Again this does not immediately make what Bishop Srubbs thought, did, or said irrelevant since they
could play some role in making it the case that the relevant outcome is produced by, for example, being
considerations some set of ideal agents take into account in their deliberations. What our ambitious
constructivist needs to avoid is some reduction just in these terms.
" Again, we are just assuming that we have the right constructivist theory of all the claims of the fonn (4).
case. In this the two sequences are exactly the same. Thus by hypothesis there is no
further feature in which one sequence's being the case, but the other's not being
the case, could possibly consist. There is nothing this further fact of the matter could
consist in.
Indeed, the ambitious metanormative constructivist may well, despite accepting a
correspondence theory of truth, be in a situation worse than the coherentist about
truth. James Young tries to argue against Crispin Wright, that once we allow the
coherentist to make a distinction between giving an account of the truth of proposi-
tions and the obtaining of facts or states of affairs, then the coherentist does have a
response to Walker's objection. The coherentist can claim that there are facts that
determine the privileged system of propositions that are the propositions coherence
with which is the property ofbeing true. In order to maintain the purity of his or her
coherence, such a coherentist would have to deny that, to put it in Kirkham's terms,
the proposition that p coheres with the designated set is true in virtue of corresponding
to the fact that p coheres with the designated set ry oung 2001, p. 96). I suspect that this
an odder thing to say than Young lets on, but I will put that aside for now.
For our
purposes it is important to see that such a view will not help the ambitious metanor-
mative constructivist since the relevant version of the Bishop Stubbs objection can be
construed as applying directly to what is the case.
6. Clarifications
Further clarification of the kind of objection being aimed at the ambitious construc-
tivist comes from attempting to use Frege's regress-which was aimed at any attempt
to defme truth--as the inspiration for an argument that attempts to show that there
must be something wrong with my version of the Bishop Stubbs argument against
constructivism. Note the sequence:
It is true that p
It is true that it is true that p
Surely such a sequence can be constructed for any p including both p and not-p.
However, it cannot be both the case that p and the case that not-p.
Whatever the right thing to say about Frege's purported regress is in the context of
giving an account of truth, I do not think this particular objection to my argument
against ambitious constructivism does raise any genuine worry that we may have
shown too much. Even if we take the fact that p to be the same fact as the fact that
it is true that p and so on, we should grant that it is true that p does not tell us what is to
•• Walker rejects this option by claiming that the "coherence theory is a theory of facts as well as of truth"
be the case that p. That is why this sequence does not give us any insight into what kind
of fact p is, say, whether it is mind independent and so on. It presupposes some
independent account of the fact that pP The rest of the sequence is simply parasitic
on whatever it is for p to be the case. For our ambitious metanormative constructivist,
however, the sequence is meant to be an explanatory articulation of what it is for p to
be the case. There is no more fundamental account of what it is for each step in the
sequence to be the case other than the next step in the sequence. And this is why a
modified version of the Bishop Stubbs objection still applies.
In conclusion, I want to emphasize a couple of other points about the objection I am
raising. First, the objection applies because our ambitious metanormative constructivist
is, so to speak, "pure" enough. That is to say that though the constructivist may not aim
to give a constructivist account of all claims-and thus is not the straightforward
analogue of the "pure" coherentist about truth--our constructivist must give a con-
structivist account of all the claims in the sequence. This is because they all belong to
the domain of practical normative and evaluative claims. Recall that were the con-
structivist to deny this then we would have a reduction to either the non-normative
and non-evaluative or a reduction to theoretical normative and evaluative truths.
I have not, of course, given any argument against such reductions but the loss of a
particular kind of ambitiousness in taking those ways out is obvious.
Second, the objection is not that we have some unacceptable form of relativism.
Charges of relativism do apply to specific versions of constructivism, including possible
versions of ambitious metanormative constructivism. However, that is not the charge
being placed here. The worry is not that Bishop Stubbs will be rational for some and
not for others, but rather he will, so to speak, be both or neither for all of us.
Third, the objection is not that some purported analysis is circular or not explanatory
in the way traditional analyses were meant to be. I am granting that what is being
presented by the ambitious metanormative constructivist is not such an analysis. The
problem is that at the most fundamental level there is nothing to make it the case that
one normative claim is correct rather than another.
The above version of the Bishop Stubbs worry for ambitious metanormative
constructivism has been pitched at a level abstract enough to allow for hope that
some response could well be found. A lot more work would have to be done to show
that we are even close to an impossibility proof. Nonetheless, I suggest that the
constructivist should be quite concerned about how one might respond to this worry.
Dancy, Jonathan. 2004. "Discussion on the Importance of Making Things Right," in Ratio 17,
pp. 229-37.
Darwall, Stephen. 1998. Philosophical Ethics (Boulder, CA: Westview Press).
Or, rather, it is not in the business of giving any further account of what it is for p to be the case.
Johnston, Mark. 1993. "Objectivity Refigured: Pragmatism Without Verificationism," in
J. Haldane and C. Wright (eds), Reality, Representation, and Projection (Oxford: Oxford
University Press).
Kirkham, Richard. 1992. Theories of Truth: A Critical Introduction (Cambridge, MA: MIT Press).
Railton, Peter. 2003. "Facts and Values," in Facts, Values, and Norms: Essays toward a Morality of
Consequence (Cambridge: Cambridge University Press).
Rosen, Gideon. 2010. "Metaphysical Dependence: Grounding and Reduction," in Modality:
Metaphysics, Logic, and Epistemology, ed. Bob Hale and Aviv Hoffinann (Oxford: Oxford
University Press), pp. 109-36.
Russell, Bertrand. 1906-7. "On the Nature of Truth," in Proceedings of the Aristotelian Sodety 7,
pp. 28-49.
Walker, Ralph C. S. 1997. "Theories ofTruth," in B. Hale and C. Wright (eds), A Companion to
the Philosophy of L:mguage (Oxford and Cambridge: Blackwell).
Wright, Crispin. 1992. Truth and Objectivity (Cambridge, MA and London: Harvard University
Wright, Crispin. 1995. "Critical Study ofWalker's The Coherence Theory of Truth," in Synthese
103, pp. 279-302.
Wright, Crispin. 1999. "Truth: A Traditional Debate Reviewed," in S. Blackburn and
K. Simmons (eds), Truth (Oxford and New York: Oxford University Press).
Wright, Crispin. 2003. "Truth as Sort of Epistemic: Putnam's Peregrinations," in Saving the
Differences: Essays on Themes from Truth and Objectivity (Cambridge, MA: Harvard University
Young,James 0. 2001. "A Defence of the Coherence Theory ofTruth," in]oumal of Philosophi-
cal Research 26, pp. 89-101.
Constructivism and Wise Judgment
Valerie Tiberius
In this paper I introduce a version of constructivism that relies on a theory of practical
wisdom. Wise judgment constructivism is a type of constructivism because it takes correct
judgments about what we have "ail-in" reason to do to be the result of a process we
can follow, where our interest in the results of this process stems from our practical
concerns. To fully defend the theory would require a comprehensive account of
wisdom, which is not available. Instead, I describe a constructivist methodology for
defending an account of wisdom and outline its main features. This gives us enough to
see what wise judgment constructivism would look like, why it might be an attractive
theory, and how it is different from other versions of constructivism.
1. Introduction
Should you brush your dog's teeth? My dogs absolutely hate teeth-brushing, but not
doing so means they have to undergo general anesthesia every couple of years to have
them cleaned at the vet, which they also hate. How hard should you try to avoid eating
eggs from caged chickens? Is it enough to buy free-range eggs for home or should you
also never eat another commercially baked cookie? Every member of a university
department should do their share of service, but what exactly is your share? Must you
accept nominations to important university committees in addition to your de-
partmental service? Helping out family and friends is important too, but what if you
are asked for financial help from a family member who has a history of gambling or
drinking away his money? Or what if you are asked for your "honest opinion" about a
friend's disastrous new dating partner? Or asked to choose sides in a conflict between
two friends, one of whom has deeply hurt the other but in such a way that you cannot
entirely blame her? What to do?
This question-"What to do?" or "What should I do?''-is a starting point in ethical
theory. Kant thought it was the primary practical question. And for good reason:
surveying the examples above, it does seem obvious that we confront this question
on a regular basis. When we do--when we ask ourselves what we should do with
family members who need help, or dogs who need teeth-brushing-we very quickly
confront the need to know what is at stake or what matters. What matters to dogs: not
being tortured by their own people or not having to spend a day at the vet's? What
matters to your friend: your honest opinion or your support? Is not contributing to the
market for eggs worth refusing a piece ofbirthday cake at a friend's party? Is meeting a
family obligation worth the risk of enabling an unreliable relative? These are the kinds
of questions we ask ourselves when we think about what we should do: questions
about what matters to us, what matters to those affected by our actions, and what
matters more or less.
What I've just described are practical problems. There are many things that matter,
more or less, and we are therefore subject to many competing demands. These various
demands are often of different kinds (moral, prudential, and so on). When trying to
decide what kind of professional service to do there are reasons of fairness, friendship,
contractual obligation, and one's own mental health to be considered. When trying to
decide what to eat, there can be considerations of animal welfare, environmental
impact, physical health, and courtesy to the people who have prepared the food. We
are often in situations in which we need to know what we should do, all things
considered. What ought we to do in the face of competing (and sometimes apparently
incommensurable) demands? Or, in other words, what do we have the best all-in
reason to do? Before we can answer this question in specific cases, we need to know
what all-in reasons are, how to understand them in an illuminating way.
In this chapter I explore an answer according to which good all-in reasons are the
product of wise judgment. Lest this seem like an exercise in explaining the obscure in
terms of the incomprehensible, let me explain that "wise" here does not refer to a
mysterious faculty of perception.
Rather, wise judgment is just judgment in accor-
dance with the norms of excellent judgment, and though this does tum out to be
complicated, I don't think it is obscure. As I will explain (in section 4), the theory we
end up with is a kind of constructivism, which I will call wise judgment constructivism
or WJC. An important feature ofWJC is the acknowledgement that the norms of good
judgment are multiple and cannot be reduced entirely to non-normative terms. I spend
some time (in section 3) discussing what a good account of wisdom is for the purposes
of playing a role in constructivism. I can't defend that theory here, but I do try to
explain why doing so isn't at all hopeless. But first, in the next section, I'll say more to
motivate this tum to wisdom in the ftrst place.
2. Why wisdom?
Even in the brief descriptions of our opening examples one thing we can notice is the
messiness or complexity of our questions about what to do. Is there anyone else in the
family who could help out the troubled relative? Are you in a family that takes turns
Some interpretations of Aristode (for instance, McDowell1979) make wisdom seem mysterious in this
helping him and is it your tum? How much help does he really need? Is there some
reason to hope that he might be setting out on a different path this time? Is it time for
an intervention? Would an intervention work? What would you have to sacrifice to
help? Who else would this sacrifice affect? "What should I do?" when a dissipated
relative asks for help leads to many other complicated questions. Even if we start with
certain basic assumptions about what matters, we aren't out of the woods. If we believe
that respecting people's rational capacities is what matters, we invite questions about
how to respect someone whose current choices are not rational. If we are convinced
that happiness is what matters, we invite questions about what happiness really is and
how to weigh one person's happiness against another. These are not simple questions,
but there's an even bigger issue lurking, which is that most of us are likely to think that
many things matter in this case: happiness, respect, family obligation, and self-interest
all seem to be important to making a good decision here.
We might think that moral philosophy, which has much to say about what matters,
should help us figure out what to do in the kinds of examples we've been discussing.
But ethical theories synthesize and systematize. Because of this, they promise to
simplify choices; this is their advantage. If we approach problem like the ones we
have considered so far with a single theory in hand we risk filtering out some of the
relevant mess. To focus on happiness is to miss the issues of fairness and obligation. To
focus on respect misses the important contribution that friendship makes to happiness.
It isn't that Utilitarians cannot account for the value of respect, or that Kantians cannot
make room for the value of happiness. They can and they have, but insofar as they do,
they lose the advantage of simplifying matters. Moreover, insofar as some of the
relevant considerations in our cases are not moral, moral theories will not take them
into account. My point here is to motivate an approach to thinking about ali-in reasons
that makes sense of the complexity of things by design, not to argue that traditional
ethical theories are untenable.
In addition to complexity, there is another feature of our practical problems that we
should notice. It is not just that we need to know what to do; we need to know what
matters and therefore what we should do. This means that the answers to our practical
problems have to be ones we can recognize as good answers, something authoritative
that really answers our questions rather than just offering an option to be considered.
We need something that puts an end to consideration so that we can act with
some confidence that we're doing the right thing, or at least that we're doing
the best we can.
One thing we sometimes do when we're in search of a good answer to a complex
problem is to look for advice. We look for an advisor who is good at apprehending the
nuanced details of a situation and understanding what's at stake. In other words, we
could say, we look for someone with some wisdom. Wisdom might be a strange word
• Cf. Korsgaard: "There [in Kant's argument for the categorical imperative) the problem is the one set by
the fact offree agency. It's nothing less than the problem of what is to be done" (2003, p. 116).
to modem ears. We don't use the word much and we don't go around talking about
who is wise or unwise. But if wisdom is just excellent judgment about what matters,
then we do look for wisdom and it is important. Indeed it is just what's needed to
answer our various "What should I do?" questions in a way that we can have
confidence in the answers.
The trouble with this solution is that there don't seem to be many wise people.
I suspect that most of us can identify a few people in our lives whose counsel we would
seek in difficult circumstances, but I also suspect that we recognize the limitations of
these advisors. We often ask advice from more than one person in order to correct for
the biases and weaknesses of a single advisor. Most of us would probably not be willing
to say that even our most trusted advisors have anything approaching perfect wisdom.
Now we might think that we don't really need wisdom or wise people, but instead
we need to know what wise people know (or what a wise person would know, if one
existed). Perhaps if we knew what values are associated with wisdom, we could then
apply this knowledge to our practical questions and fmd good answers to them without
thinking any further about wisdom. For instance, if we knew that the wise always
valued other people's happiness more than their own, this would give us some
guidance about questions about what to do when what we want most conflicts with
what our loved ones want most. We could even see this as one way of describing what
moral theories are doing: specifying particular values or principles is like articulating the
knowledge of the wise.
The problem with this strategy is that it's very unclear how we would figure out
what values wisdom entails, particularly at a shatp enough level of detail that this
information would actually be useful. We might be able to say that wise people value
other people's happiness, but we don't have grounds for concluding that in every set of
circumstances, a wise person will put someone else's interests above her own. This is a
strong claim about the value of happiness (and, I suspect, a false one) that requires
argument to establish; it isn't something we can know about wisdom without thinking
more about what wisdom is.
Not that there aren't any arguments we can use to gain insight into the values that
are associated with wisdom. We could argue that certain values are part of the concept
of wisdom, or that because wisdom is a virtue that is conducive to flourishing, it must
be associated with values that are tied to flourishing. Or we could argue on empirical
grounds that people who have certain wisdom-related skills also have a certain set of
values. But none of these kinds of arguments will lead us to insight about "wise values"
without making some rather substantial assumptions about what wisdom is. We would
end up making assumptions about the concept, or about flourishing, or about the
wisdom-related skills. If we want to answer our ethical questions by appeal to wisdom,
we can't bypass thinking about what it is to make judgments about what matters in an
excellent way.
In this section I hope to have demonstrated the rationale for attending to wisdom.
What we need now is to think about what wisdom is.
3. Wisdom and good judgment
One might think that wisdom is such a difficult notion that it is hopeless to think that
we could use it to illuminate other normative notions. I think this is pessimistic.
Certainly, it would be overly optimistic to think one could defend a comprehensive
account of wisdom in one section of a single paper, but we can take some first steps that
I hope will demonstrate the promise of the project. What I can do in this section is to
articulate an approach to defining wisdom and sketch the likely results of applying it
that will meet the needs of wise judgment constructivism.
The methodology I propose is a version of the method of wide reflective equilibri-
um (Daniels 1979) that constructs an account of wisdom by integrating both normative
argument and empirical data.
Psychological accounts of what people actually think
wisdom is--so-called "implicit theories" of wisdom, or "folk theories" as I shall call
them-play the role that moral intuitions or considered judgments play in standard
applications of wide reflective equilibrium (WRE). Four key components of wisdom
emerge from the research on folk theories of wisdom: deep understanding (about
complex human problems, in particular), reflective capacities, problem-solving capa-
cities, and the motivation to make good decisions and to help others do so too.
Beginning with the folk theory ensures that our account of wisdom will be impor-
tantly related to an ideal that people actually have. But it would certainly be a mistake to
think we can stop at the folk theory. The folk theory may not be at an equilibrium point,
after all. To reach equilibrium, WRE requires that we be prepared to jettison or revise
components of the folk theory that are misguided or in conflict with each other. Hence
the folk theory needs to be examined and, possibly, refmed and supplemented.
The folk theory can be supplemented by philosophical arguments that test and
elucidate it against various theoretical constraints, which take the place of moral
principles in standard applications of WRE. Given our purposes, there are three
important theoretical constraints on the process of finding an equilibrium about
wisdom. First, if our account of wisdom is going to be responsive to the rationale for
our interest in it, it must be a theory that makes wise judgment something that we
(those of us who seek answers to the question "What should I do?") recognize as
authoritative. Wisdom, in other words, must be normative for us. Second, an account
of wisdom should be detailed enough to make action-guiding recommendations.
A good account of wisdom will have enough substance to ground prescriptions that
provide guidance in particular situations.
Third, an account of wisdom should be
' This methodology was developed with jason Swartwood in our paper "Wisdom Revisited" (2011).
• For an excellent overview of the implicit theory research in psychology, see Bluck and Glueck 2005 ..
This distillation of the folk theories is already somewhat theoretically laden. In order to make sense of their
data, psychologists must use some criteria in order to group responses together and make sensible general-
' One might take this to be implied by the requirement of normativity, but I think it worth separating
action guiding from reason giving to emphasize the point that the theory needs to avoid vagueness so that it
wiU generate specific prescriptions (which is a distinct matter from whether its prescriptions have authority).
empirically sound; it should not make implausible claims about human nature or false
assumptions about our capacities. For example, it should not prescribe processes of
reasoning that, as a matter of psychological fact, undermine a person's ability to make
good choices.
So the best account of wisdom will be normative, action guiding, and empirically
sound. The first constraint is the most important for the purpose of developing wise
judgment constructivism, because the conception of wise judgment must answer to
our practical concerns about reasons if it is to provide a helpful understanding of what
our ali-in reasons are. As Lenman (this volume, p. 215), quoting the Confucian Chung-
Yung, says, "If the Way were remote from humanity, it would not be the Way."
Starting with the facts about the folk concept of wisdom helps to ensure that the way is
not remote from humanity. The idea here is that starting with the folk theory of
wisdom ensures a sound deliberative route from the ideal of wisdom back to something
people actually care about and this helps to ensure that the judgments of the wise
(understood according to our theory) are ones we find authoritative or reason giving.
We will fmd them authoritative because they represent what our own judgments
would be if they were improved along the lines that we regard as improvements.
We arrive at a more refmed account of wisdom by thinking about how the
components of the folk theory can be realized together as a coherent ideal composed
of wise objectives (goals) and the norms of good judgment the following of which will
accomplish these objectives. In particular, the conception of wisdom that results from
this methodology includes three basic goals that a wise person has: to decide well, to
help others decide well, and to develop and maintain the understanding, abilities, and
motivations associated with wisdom. We then consider how to meet these goals and
how to understand the four components of the folk theory in the context of the vast
amount of information that is relevant to making good choices and the serious
limitations to which our understanding is subject. Given these factors, the kind of
deep understanding possessed by a wise person cannot be near-omniscience, rather it
must have to do with strategies for dealing with how little we can ever know.
Similarly, given the ways that explicit reflection can distort decision making, the
kind of reflection that is part of wisdom must acknowledge the limits of our higher-
order cognitive processes. Some research on decision making suggests that reflecting on
the reasons for a choice right before making it tends to lead to decreased satisfaction
with that choice later (e.g. Wilson, Hodges, and Lafleur 1984). This kind of research
can contribute to the case for a strategy of reflecting on values when one is not
confronted with a choice that has to be made and going with your gut in the heat of
the moment. Notice that the argument for this strategy would be informed by
background empirical theories about good decision making. Given the importance
of emotions to values, another useful strategy would be one that helps you tune into
your feelings-for example, by paying attention to physiological clues. In short, the
wise person engages in reflective thinking at the right time so that the act of reflection
does not end up undermining her convictions or distorting her preferences.
With these thoughts in mind, we can say that wisdom includes several broad,
interrelated norms the following of which is necessary for deciding well: overcoming
biases; humility about what one knows, and open-mindedness to other people and the
lessons of experience; empathy with others; and appropriate reflectiveness about our
Wise judgment, then, is characterized by a complex set of norms many of
which must be themselves characterized in normative terms. A wise person follows
these norms in order to identify what matters and to reach and justify a decision about
what to do. Often, following these norms requires cultivating and maintaining long-
term policies and short-term strategies for coping with our limitations. For example,
we tend not to overcome cognitive biases upon learning that we have them; rather,
what works is following rules of thumb, which require the adoption of a long-term
strategy or policy.
The wise person identifies the relevant considerations by making
sure that her judgments conform to all the norms of good judgment in the best possible
way. She puts these reasons together to form a justification for her judgment about
what one has best all-in reason to do in the circumstances. For example, if the wise
person were to decide that she ought to brush her dog's teeth, she would have a
justification for this judgment in terms of the reasons that favor this action (for example,
preventing pain and not shirking an obligation in the face of minor inconvenience).
Now, one might agree that a wise person needs to be able to make these various
discriminations among values, but disagree that this requires any justification or
explanatory ability. Wise people, one might argue, feel or intuit their way to good
decisions, they do not need to be able to justify their judgments by appeal to supporting
reasons. One might ground this criticism in the philosophical tradition that emphasizes
the importance of sentiment and emotion to moraljudgment, or one might ground it
in recent psychological research that highlights some of the shortcomings of reflection
and reasoning (see Tiberi us 2009 for a critical discussion of this literature).
We can admit that emotions and intuition are very important to wise judgment
without abandoning reflection and explanatory ability altogether. Emotions can be an
important source of reasons for valuing what we value, and emotional responsiveness
to others is absolutely essential to understanding what matters to them. Further, we are
sometimes better off deciding intuitively than by engaging in a process of counting pros
and cons. But these admissions do not obviate the need for a justification about what
matters in general. Certainly we shouldn't reflect on what matters all the time, but
justification is needed in the background if a person is to be able to aim at cultivating
wisdom, to make decisions when her heart is conflicted, and to advise others who need
reasons to trust her advice.
• This last requires deliberative skill, which is a component of wisdom emphasized by Rosalind Hurst-
house 2006. It also requires achieving and maintaining proper perspective, a part of wisdom which I focus on
in The R{fkctive Ufr (2008). The importance of empathy to wisdom is emphasized by Stohr 2006.
For a review of the psychological literature on overcoming biases, see Lilienfeld et a!. 2009.
We now have a sketch of an account of wisdom. An example might help to make
the preceding rather abstract discussion more concrete. Imagine Professor Smith trying
to decide whether to take a new job as the director of a research center at her
university. Insofar as Smith is wise, she has to this point been open to many sources
of information and humble about what she knows. This means that she has learned
some things about her own skills from paying attention to others' reactions toward her.
In particular, she has learned that she is more organized than most and actually thrives
in positions of great responsibility. Smith also has the ability to empathize with her
&mily (who would get less of her attention if she took this job}, her philosophy
colleagues (who would get even less attention), and the university administrators
who are asking her to step up. Further, she sees that her family's concerns about her
time are legitimate and represent values that are also important to her, but her
colleagues' concerns are rather selfish and fuil to acknowledge the greater value of
the work she would do at the center. Smith understands the importance of work-life
balance and has strong, stable commitments to the values of family relationships,
challenging herself, and contributing to causes she believes in. Finally, her emotions
are in tune with these values--she doesn't long for escape or daydream about becom-
ing a professional windsurfer. She takes these basic values to structure her life and she
sees no reason to make revisions. Smith has some wisdom. Whatever she decides, she
will justify her decision about what she has most reason to do by appeal to the particular
reasons that stem from her stable and reflective assessment of what matters. Anticipating
the role that wisdom plays in a constructivist theory, we can say that whatever Smith
decides, insofar as she is wise her judgment about what she ought to do will be
warranted and she will indeed have ali-in reason to do what she judges she ought.
With this sketch of wisdom in hand, the next step in reaching wide reflective
equilibrium would be to evaluate the preliminary account of wisdom (and, eventually,
the more comprehensive conception) against background theories and theoretical
constraints. The constraints I mentioned at the beginning of this section were that a
good account of wisdom should be normative, empirically sound, and action guiding.
The fact that the basis for the refined theory is the theory of wisdom makes it plausible
that the refmed theory is reason giving for anyone who endorses the standards of wise
judgment in the ordinary conception. The refmed theory is empirically sound by
design, because it takes into account information about what we are like in order to
specify the relevant norms, policies, and strategies. The practical action-guidingness of
the account will be difficult to evaluate before we have a more comprehensive version,
but there seems to be no reason to think that wise judgment as understood here would
not result in specific conclusions in practical contexts. Turning to the matter of
background theories, an account of wisdom ought to be acceptable from the stand-
point of compelling theories of the good life. Though I can't make the argument here,
it seems reasonable to suppose that an account of wisdom that highlights deep under-
standing, reflective capacities, problem-solving skills, and appropriate motivations will
fare well when seen from the standpoint of either objective or idealized-subjective
theories of the good life.
There is much more work to be done. My purpose in this paper is not to defend an
account of wisdom; rather, it is to sketch the methodology for developing such an
account in order to explain its role in constructivism. Certainly, the account of wisdom
would have to be developed far beyond its current form for wise judgment construc-
tivism to be thoroughly evaluated. But even at this early stage of development there are
a few points in favor of the approach. First, notice that many of the components of
wisdom extend to the case of theoretical judgment, which is a bonus. For example, it is
a familiar thought that good belief formation requires open-mindedness, freedom from
bias, and epistemic humility.
This adds to the case for thinking that wisdom might
shed light on the question of what counts as a good ail-in reason, because it means that
the norms of wisdom have a different (perhaps broader) base than the norms for our
target notion. Second, an account of wisdom that is grounded in the folk theory will
provide norms for judgments about good reasons that people can see as authoritative,
and therefore as answering to their actual practical concerns. In the context of
constructivism about good reasons, this is a plus because according to constructivist
theories, speaking to our practical concerns (rather than "representing reality" or
"capturing the truth") is at the heart of normative theory.
4. Wise judgment constructivism
The basic idea behind wise judgment constructivism is that we understand good ail-in
reasons by appeal to the deliberations of a wise person, that is, by appeal to judgment
that is in accordance with all the norms of good judgment. Wise judgment construc-
tivism is a normative theory that elucidates one normative notion (good ail-in reasons)
by appeal to others (wisdom and the norms of wise judgment). According to wise
judgment constructivism, good ail-in reasons are the reasons that are arrived at through
a process of wise judgment.
Wise judgment constructivism builds on the insight that made ideal observer
theories attractive. Here's Firth: "In analyzing ethical statements, for example, we
must try to determine the characteristics of an ideal observer by examining the
procedures which we actually regard, implicitly or explicitly, as the rational ones for
deciding ethical questions" (1952, p. 332). The advantage of taking seriously what we
actually regard as the features of good judgment is that we will end up with an account
that answers our practical problem by furnishing reasons that we can act on with
confidence. But because ideal observer theories have typically wanted to reduce ethical
notions to natural ones, they took this insight to an extreme that was unsustainable.'
• These are indeed virtues emphasized by virtue epistemologistl. See Zagzebski 1996, RobertS and Wood
• This restates an argwnent I made in Tiberius 1997. See also Rosati 1995.
Moreover, this extreme does not accurately reflect what we actually regard as rational
procedures for deciding: if the research on folk theories of wisdom is any evidence, the
omniscient, omnipercipient, disinterested, dispassionate, consistent, and otherwise
normal creature does not, in fact, represent our views about good judgment.
Instead, we take wise judges to care about other people and about giving them good
advice, we think that wise people have knowledge about certain important topics but
not about everything, and we take wisdom to include skills of decision making and
problem solving that one develops with experience. A theory that begins with what we
actually take excellent judgment to be will be better suited to meet our practical
concerns and to provide satisfying answers to our questions about what to do.
4.1. Locating wise judgment constructivism on the map
It will be helpful to distinguish wise judgment constructivism from some other views
with which it might be confused.
First, consider the classic ideal observer theory just alluded to above, according to
which correct ethical statements are identified with the ethically salient reactions of an
omniscient, omnipercipient, disinterested, dispassionate, consistent, and otherwise
normal judge (Firth 1952). If the right account of wisdom were one that took the
above features as the defining features of wisdom, WJC would be an ideal observer
theory. However, this is not the case. The best account of wisdom does not take
wisdom to be comprised of characteristics that can be defined non-normatively,
because norms must be employed in attributions of wisdom. For example, wisdom
requires that people reflect in the right way, at the appropriate time. Humility and open-
mindedness are also normative insofar as each must be appropriately engaged, for
example, at the mean between the extremes of arrogance and self-abnegation, or
pig-headedness and irresolution. Further, since wisdom is a virtue, the components
of wisdom include skills, policies, and strategies that are cultivated and employed by an
agent (rather than just characteristics possessed by an agent as in the ideal observer
theory). The best account ofwisdom makes it normative "all the way down". WJC
does not reduce claims about all-in reasons to claims about natural facts; rather, it
explains one kind of normative claim (about all-in reasons) in terms of another (about
practical wisdom).
The recognition that wisdom is a virtue invites the thought that wise judgment
constructivism is an Aristotelian theory. Indeed, my analysis of wisdom as a virtue is
Aristotelian in some respects. A virtue is a set of dispositions, skills, and habits that
comprise an ideal that it makes sense to take as a goal of character development.
Further, wisdom is importandy related to human flourishing: one criterion of success
for an account of wisdom, on my view, is how well wisdom conceived in its terms
contributes to an overall good life for a person. Nevertheless, the theory I am proposing
also differs from a traditional Aristotelian theory in important respects. I do not defme
wisdom by appeal to an independendy normative conception of human nature, nor do
I rely on any claims about a human telos. Moreover, I do not assume that there is such a
thing as perfect wisdom to which correct judgments about reasons are tied. Rather, on
my view, the warrant we have in our judgments varies with the degree to which they
are the result of wise deliberation, which itself comes in degrees.
The fact that WJC understands good reasons in terms of a process of deliberation
characterized in normative terms is some evidence that it is indeed a form of construc-
tivism. The idea that sound normative judgments are to be understood in terms of the
results of a process of reasoning (rather than the other way around) is a core element of
constructivist theories. Another key feature of constructivist theories is that they take
our interest in the results of these processes of reasoning to be practical rather than
theoretical. This is true of wise judgment constructivism: it is our practical problem of
needing to know what to do in the face of competing considerations that gives us an
interest in the results of wise judgment.
Finally, it is worth saying something about the difference between WJC and other
versions of constructivism. The best-developed versions of first-order, normative
constructivism are theories of our moral duties to other people. As the main proponent
of the view has acknowledged, what we owe to each other does not exhaust the
normative domain and hence contractualist theories might not answer all of our
questions about what we ought to do (Scanlon 1998). We have reasons to do many
things that don't have to do with our moral duties to others: personal reasons,
perfectionist reasons, reasons of friendship, aesthetic reasons, and so on. How these
reasons bear on what we ought to do in a particular situation is often an important
question. There may also be some moral reasons that do not fit nearly into the
contractualist framework (I'm inclined to think the reasons we have to be vegetarian
are reasons of this sort). Further, unless one thinks that all moral reasons are necessarily
overriding, moral reasons will at least sometimes have to be compared to these other
kinds of reasons in order to determine what one ought to do. The version of
constructivism I am proposing is intended to answer the first-person question about
what we ought to do, where the relevant considerations appear pluralistic and messy.
It is not a theory about our duties to each other nor does it make the justification of
judgments about reasons essentially social. Instead, it is a normative theory about good
Not all Aristotelian views make these assumptions, of coune. Insofar as Aristotelian views do not need to
make such assumptions, WJC might count as a kind of Aristotelianism. For a constructivist theory that is
explicidy Aristotelian, see LeBar 2008.
WJC is therefore a constructivist theory according to Lenman's (this volume, p. 216) careful defmition
of constructivism: "Constructivist views understand correct normative views of the relevant kind (political,
ethical, normative) as those which are the upshot of some procedure or criterion, where (a) that procedure or
criterion is one followable or applicable by human beings where (b) that procedure or criterion is itxlf
characterized in normative terms invoking ideals of e.g. rationality or reasonableness and (c) applying the
procedure or criterion is taken as determining or constitutive of that correcmess rather than as tracking a
correcmess conceived as prior and independent to it and (d) where the rationale for our taking an interest in
whatever the procedure or criterion in question delivers is conceived of as speaking to distinctively practical as
opposed to theoretical concerns."
So, my project here is in the same vein as Thomas Hill's (1989) attempt to extend constructivism to a
larger normative domain. Hill takes a Kantian path, relying on the ideal of the Kingdom of Ends.
reasons that takes the wise judgment of an individual deliberator as central to the
understanding of correct judgments about good ali-in reasons.
The comparison to other versions of constructivism invites a question about how
deeply constructivist WJC really is. In Yonatan Shemmer's terms (this volume, p. 161),
is it local (according to which only part of the normative domain is constructed) or
global (according to which all our norms are constructed)?
WJC is indeed a local
version of constructivism because it explains one normative domain in terms of
another: good reasons are explained by appeal to norms of good judgment. But it is
local in an interesting way for two reasons. First, the norms that are presupposed are
virtue-epistemological norms, rather than norms from another part of the moral
domain. Second, the theory of wisdom itself does not make realist metaethical
assumptions that are in tension with the spirit of constructivism. I have not character-
ized wisdom as a capacity to perceive the reasons that there are; rather, I have proposed
that we use wide reflective equilibrium to identify the norms of wise judgment,
beginning with the commitments to ideals that we already have. WJC is Neurathian,
then, (again, in Shemmer's terms) because we start with the norms we already have (in
our folk theory) and work from there, improving, modifying and revising as needed.
WJC lays out the pattern that our various norms need to be in to sustain stable claims to
normativity; in other words (to continue with the metaphor), WJC provides instruc-
tions for Neurathian boat repair that doesn't sink itsel£
An important feature of constructivism is, as Lenman (this volume, p. 216) puts it,
that "applying the procedure or criterion is taken as determining or constitutive of that
correctness rather than as tracking a correctness conceived as prior and independent to
it." That wise judgment constructivism takes this to be so has been implied, since I have
not explicitly assumed any independent truth that wise judgment is tracking. However,
one might object that the notion of wisdom does implicitly assume an independent
normative truth to which wise judges are responsive. This would be true if the only
way to understand wisdom were by appeal to a special sensitivity to reasons that is
analogous to visual perception. If the view is that the wise person "sees" all the relevant
considerations in the circumstances, then there is something to be seen or grasped that
must be identified prior to the judgment of the wise. Part of the point of providing a
sketch of an account of wisdom that does not take Aristotle as its starting point was to
show that the perceptual model is not the only way to think about wisdom.
Wise judgment constructivism suggests a particular understanding of the construc-
tivist thesis we have just been considering (i.e. the thesis that the procedure is
constitutive of correctness rather than tracking correctness). When a wise person
aims to figure out what to do, her answer will refer to reasons. For example, the
wise person who decides to lend money to her dissipated nephew may justifY this
decision by appeal to reasons of family obligation or preventing harm to the nephew. If
Street 2009 makes a similar distinction between restricted and unrestricted fomu of constructivism.
giving money to the nephew is the right thing to do in this case, then it is the right
thing to do because it avoids harm and meets an obligation. At this point we might ask what
makes these reasons the right ones to follow in this case. The answer WJC gives to this
question is that these are the right reasons in virtue of their being identified by the
process of wise judgment. Wise judgment constructivism has two stages: (1) good
reasons for action are identified through a process of wise judgment and (2) norms of
wise judgment are identified through a process of wide reflective equilibrium. WJC
can say, then, that what makes it the right thing to do to lend money to the nephew is
that it prevents harm and meets a family obligation. We do not have to say that the
reason to give money to the nephew is "constructed" as if the wise person makes it the
case by fiat. The wise person does not make it up; rather, she must offer reasons for her
judgment. It is because her judgment follows the norms of good judgment that her
judgments have authority, but this does not mean that the reasons for helping the
nephew must themselves make reference to the wise person's process ofjudgment.
4.2. Some questions about J.t]C
Good all-in reasons are reasons that are arrived at through a process of wise judgment.
This formulation raises a question: Whose wise judgment is relevant? Does the theory
say that what I have good reason to do is what I would judge that I have reason to do if
I were wise? Or does it say that what I have good reason to do is what would be judged
that I have reason to do by the perfecdy wise person? Is there even a difference
between the two views? The frrst thing we need to make clear is that either answer
can allow for significant variation in what reasons different people have. Even if you
think that the status of some consideration as a good all-in reason depends on its being
arrived at by a perfecdy wise agent, you could still think that perfecdy wise agents judge
that different people have different reasons depending on their values and goals. So, to
take our example from before, a perfectly wise person may judge that Professor Smith
has most reason to accept the administrative position, while one of her colleagues has
most reason to refuse.
The second point to make clear is that given the role wisdom is supposed to play in a
constructivist theory, the wise judgment in question has to be one that the agent whose
reasons are at issue can recognize as wise. In other words, the important thing is that it
must be a process that follows standards that are recognized by the agent as standards of
excellent judgment. If this is not the case, the agent won't have the right kind of
interest in the results of the process.
With these two qualifications in mind, we can see that the question of whose wise
judgment is relevant is relatively inconsequential. The important variables are whether
all the standards of excellent judgment are being observed and how well or to what
degree. An agent looking to the theory for guidance could apply these standards to her
I take the argument in this paragraph to be a version of the dialectical strategy Wallace attributes to
Korsgaan:l (this volume, p. 27).
own judgment or to the judgment of another person (perhaps someone she is looking
to for advice). Further, in articulating an account of wisdom we aim to define a theory
on behalf of many who share important commitments. We aim to answer shared
questions on the assumption that the answers will speak to everyone for whom the
question has significance. This also makes the question of who the wise judge is less
Of course, WJC cannot rule out the possibility of disagreements in practice among
people who are not fully wise. How does WJC treat these disagreements? When there
are differences in practice, WJC can explain these differences by appeal to the failure to
follow the right norms. It might seem that the account of wisdom I have sketched will
not have the resources to explain disagreement in this way, because it does not include
a list of wise values. One might think that if the account of wisdom does not specify the
values of the wise, then some disagreements will just result from different inputs (in the
form of different values) as opposed to any failure to adhere to the norms of excellent
judgment. But the fact that the account of wisdom does not specify what values a wise
person must have does not mean that there are no constraints on what values are
included in wisdom. This is so because the procedural norms of wisdom also constrain
judgments about basic values in such a way that when there are differences in judg-
ments about reasons that result from differences in basic values, these differences can be
explained by reference to norms of, say, open-mindedness, humility, or responsiveness
to the facts. The account of wisdom does not specify substantive values, but people
who make judgments about what counts as wise do make assumptions about values.
When we perceive differences in judgment that seem to be due to differences in basic
values, we can assume either that these basic values were not arrived at wisely, or that
these different values are equally wise and hence that there is a set of reasons that are
equally good.
Granted, WJC cannot rule out in-principle disagreements that are due to different
initial commitments that will not change when all the norms of good judgment are
followed. So, there might be disagreements between wise people that are not merely
practical (that is, differences that cannot be removed by ensuring that all the norms of
excellent judgment were followed). The question about this possibility is whether it
undermines the authority of ail-in reasons as characterized by WJC. It seems to me that
it does not undermine the authority of a claim about what someone has ail-in reason to
do from the point of view of that person, because the person would have no reason to
reject the wise judgment of the person with her own initial commitments who follows
all the norms of good judgment. If she did have such a reason, it would be because there
was some norm that reconunends rejecting or revising something about her position.
So, WJC might imply a certain amount of relativism, but the range of cases that must be
interpreted as allowing for relativism will be limited by the norms of good judgment,
and what relativism is left need not undermine the authority of reasons on the view.
To this point, my aim has been to accommodate the idea that we need to be able to
explain disagreements in practice in a way that doesn't undermine the authority of the
judgments we take to be the product of wisdom. But one might think that the
disagreement of others is something to be heeded, not explained away. The observa-
tion draws attention to the fact that the version of constructivism I have described
differs from more familiar versions insofar as it does not make justification to others an
essential ingredient of the process. Is this a problem? One might think so because it
means that correct norms according to WJC are less constrained than correct norms
according to a contractualist version of constructivism and so WJC will be less likely
than social versions of constructivism to capture the apparent objectivity of claims
about good reasons. Correct norms, according to WJC, float more freely (one might
think) and therefore it amounts to a less compelling version of constructivism. More-
over, the apparent indeterminacy about reasons in this version of constructivism seems
to undermine the confidence that wise judgment is supposed to secure. There are
several ways to respond to this challenge.
The first is to point out that social, contractualist versions of constructivism have their
own problems, given the claims about reasonableness on which they must rely. The
constraint on correct moral judgment that is imposed by social versions of constructiv-
ism is not, after all, that your judgments must be justifiable to other actual people.
Rather, the constraint requires that your judgments be justifiable to reasonable others
who are similar to you insofar as they seek mutually acceptable grounds for their claims.
This has the result of weakening the constraint on appropriate judgment in such a way
that some of our questions about what we ought to do are not answered. For example,
consider the question of whether one ought to stop eating certain animals. In my
experience, there is disagreement among reasonable (in the above sense) people about
this question. Further, what I think is wrong with the judgment of others who do not
share my views (e.g. that there are good reasons not to eat pigs) does not have to do with
their reasonableness as usually understood by social versions of constructivism. Here is a
case where wise judgment constructivism allows us to think about the appropriateness
of some moral judgments in ways that seem more suited to the subject matter. I can say
that my reasons not to eat pigs are good reasons because people who disagree with me
are lacking wisdom in their judgments-they exhibit bias, fail to appreciate relevant
information, or fail to distinguish what matters more from what matters less.
Second, we can point out that the two forms of constructivism aren't mutually
exclusive. Indeed, WJC may need to be supplemented in order to account for certain
kinds of reasons. I am inclined to think that when it comes to judgments about what
we owe to each other, philosophers such as Scanlon and Lenman are correct. One way
for wise judgment constructivism to handle this point would be to argue that when it
comes to such judgments wisdom requires offering reasons that could not be reason-
ably rejected by others committed to coming to agreement with you. In other words,
the thought would be that the account of wisdom sketched here would have to be
supplemented to count as a theory of moral wisdom. If wise judgment constructivism
and contractualism can be put together in this way, then worries about WJC as an
alternative to contractualism are misplaced.
The third response to the challenge is to emphasize the significance of the constraints
that do come from a theory of wise judgment. Along these lines, a point that deserves
emphasis is that even though wisdom is not defmed in terms ofhypothetical agreement
with others, the wise penon certainly does not ignore the views of others. A wise
person must be able to see things from other people's penpectives in order to see what
matteiS to them and give good advice. She must be open-minded about new sources of
evidence and bumble about what she doesn't know. She must also be committed to
overcoming biases in decision making that are extremely difficult to detect in oneself
All of these facts about wisdom suggest that a wise person will be solicitous of and
attentive to the opinions of others.
Finally, the fact that the reasons anived at through the process of wise judgment are
good reasons upon which we are justified in acting does not rule out the possibility of
error, and the possibility of error is an important reason not to ignore disagreement
with others.
We can have enough confidence in our reasons to see our practical
problem as solved, even while we acknowledge the possibility that we've made a
mistake. Notice that room for this possibility is created by the fact that the norms of
wise judgment are themselves described normatively: even once we have anived at our
best judgment about what we have reason to do, it is open to us to think that we could
have reflected more appropriately, that our empathy was not as well tuned as it could
have been, that we misjudged the relevance of one fact and put too little weight on
another, and so on.
5. Conclusion
According to wise judgment constructivism, what you have the best all-in reason to do
is what you would judge younelf to have reason to do if you were judging wisely. You
get better at discerning good reasons by developing and applying the capacities
necessary for wisdom or by looking to those who have. We ensure that these wise
judgments answer to our practical concern-the concern to sort through the mess and
anive at a judgment we can act on with confidence-by taking people's actual norms
for good judgment, as revealed by the folk theory, as the starting point for theorizing
about wisdom.
All-in reasons, on this view, are best undentood in terms of other normative
notions, in particular, a process of wise judgment that itself can only be described by
employing normative terms and defended by employing wide reflective equilibrium.
Wise judgment constructivism doesn't answer all of our questions about reasons, of
course. One might want to know what judgments about reasons are semantically. One
might want to know how norms are represented psychologically. Or, one might want
to know what the explanation is for why reduction isn't feasible. I take it that WJC is
•• On the importance ofleaving room for the possibility of error, see Wallace, this volume.
compatible with at least a few different answers. An advantage of WJC is that it
promises to illuminate what all-in reasons are in a way that responds to our practical
concerns without our having to resolve the semantic and psychological debates.
Of course, at this stage of the development of wise judgment constructivism, even
many of the questions that the theory is supposed to answer are still unanswered.
Insofar as the view is promising, though, I hope to have made a case for shifting
philosophical attention to developing an empirically informed, but normatively signif-
icant, account of good decision making, a shift away from metaphysics toward
epistemology. If suitably developed, wise judgment constructivism would have the
same attractive features of any constructivist theory-metaphysical modesty without
simple reduction and the ability to explain the motivating force of normative claims-
with the added benefit ofbroader scope in terms of the kinds of reasons it can be used
to illuminate.
Bluck, S. and Gliick,J. 2005. "From the Inside Out," in Sternberg and jordan (eds), A Handbook
of WISdom: Psychological Perspeaives (New York: Cambridge University Press), pp. 84-109.
Daniels, N. 1979. "Wide Reflective Equilibrium and Theory Acceptance in Ethics," in]oumal of
Philosophy 76(5), pp. 256-82.
Firth, R. 1952. "Ethical Absolutism and the Ideal Observer," in Philosophy and Phenomenological
Research 12(3), pp. 317-45.
Hill. T. 1989. "Kantian Constructivism in Ethics," in Ethics 99(4), pp. 752-70.
Hursthouse, R. 2006. "Practical Wisdom: A Mundane Account," in Proceedings of the Aristotelian
Sodety 106:3 (May 2006), pp. 283--307.
Korsgaard, C. 2003. "Realism and Constructivism in Twentieth-Century Moral Philosophy," in
Philosophy in America at the Tum of the Century, APA Centennial Supplement, journal of
Philosophical Research, pp. 99-122.
LeBar, M. 2008. "Aristotelian Constructivism," in Soda! Philosophy and Policy 25, pp. 182-213.
Lenman,J. "Expressivism and Constructivism," this volume.
Lilienfeld, S., Ammirati, R., and Landfield, K. 2009. "Giving Debiasing Away: Can Psychologi-
cal Research on Correcting Cognitive Errors Promote Human Welfare?" in Perspectives on
Psychological Science 4(4), pp. 390-8.
McDowell,). 1979. "Virtue and reason," in Monist 62, no. 3, pp. 331-50.
Roberts, R. and Wood,]. 2007. Intelleaual Virtues: An Essay in Regulative Epistemology (Oxford:
Clarendon Press).
Rosati, C. 1995. "Naturalism, normativity, and the open question argument," in Nous 29(1),
pp. 46-70.
Scanlon, T. M. 1998. W'hat We Owe to Each Other (Cambridge, MA: Harvard University Press).
Shemmer, Y. "Constructing Coherence," this volume.
Stohr, K. 2006. "Practical Wisdom and Moral Imagination in Sense and Sensibility," in Philosophy
and Literature 30, pp. 378-94.
Street, S. 2009. "What is Constructivism in Ethics and Metaethics?" in Philosophy Compass 5,
pp. 363--84.
Tiberius, V. 1997. "Full Infonnation and Ideal Deliberation," in Journal of Value Inquiry Volume
31, No. 3, pp. 329-38.
Tiberius, V. 2009. "The Reflective Life: Wisdom and Happiness for Real People," in Lisa
Bortolotti (ed.), Philosophy and Happiness (London: Palgrave MacMillan), pp. 215-32.
Tiberius, V. and Swartwood, ]. 2011. "Wisdom Revisited: A Case Study in Normative
Theorizing," in Philosophical Explorations 14(3), pp. 277-95.
Wallace, R. Jay. "Constructivism about Normativity: Some Pitf.ills," this volume.
Wilson, T. D., Hodges, S. D., and Lafleur, S. J. 1984. "Effects of Analyzing Reasons on
Attitude-Behavior Consistency," in Journal cif Personality and Soda! Psychology 4 7 (1 ), pp. 5-16.
Zagzebski, L. 1996. Virtues cif the Mind: An Inquiry into the Nature cif Virtue and the Ethical
Foundations cif Knowledge (Cambridge: Cambridge University Press).
Expressivism and Constructivism
James Lenman
According to the tables of contents of two important recent American anthologies of
papers there is an important position in contemporary metaethics that is called construc-
tivism: both Stephen Darwall, Allan Gibbard, and Peter Railton's collection and Russ
Shafer-Landau and Terence Cuneo's devote entire sections to it. We Brits are less
susceptible: check out Andrew Fisher and Simon Kirchin's recent anthology or Miller's
Introduction to Contemporary Metaethics and you won't even fmd it mentioned in the
So are we Brits missing something? What is this constructivism?
The story--or the presendy interesting bit-begins with Rawls's classic paper
"Kantian Constructivism in Moral Theory", first published in 1980. There Rawls
writes as follows:
To justify a Kantian conception within a democratic society is not merely to reason correcdy from
given premises, or even from publicly shared and mutually recognized premises. The real task is to
discover and formulate the deeper bases of agreement which one hopes are embedded in common
sense, or even to originate and fashion starting points for common understanding by expressing in a
new form the convictions found in the historical tradition by connecting them with a wide range of
people's considered convictions: those which stand up to critical reflection ...
I should emphasize that what I have called the "real task" of justifying a conception ofjustice is
not primarily an epistemological problem. The search for reasonable grounds for reaching
agreement rooted in our conception of ourselves and in our relation to society replaces the
search for moral truth interpreted as fixed by a prior and independent order of objects and
' I am grateful for help with this to all the participants in the Sheffield workshops and conference through
the coune of2009 as well as to an audience at Cardiff' in March 2010. I am especially indebted to Yonatan
Shenuner and Valerie Tiberius.
See Darwall, Gibbard, and Railton 1997, Shafer-Landau and Cuneo 2007, Fisher and Kirchin 2006,
Miller 200]
relations, whether natural or divine, an order apart and distinct from how we conceive of
ourselves. The task is to articulate a public conception of justice that all can live with who
regard their penon and their relation to society in a certain way. And though doing this may
involve settling theoretical difficulties, the practical social task is primary. What justifies a
conception of justice is not its being true to an order antecedent to and given to us, but its
congruence with our deeper understanding of ourselves and our aspirations, and our realization
that, given our history and the traditions embedded in our public life, it is the most reasonable
doctrine for us. We can find no better basic charter for our social world. Kantian constructivism
holds that moral objectivity is to be understood in terms of a suitably constructed social point of
view that all can accept. Apart from the procedure of constructing the principles of justice, there
are no moral facts. Whether certain facts are to be recognized as reasons of right and justice, or
how much they are to count, can be ascertained only from within the constructive procedure,
that is, from the undertakings of rational agents of construction when suitably represented as free
and equal moral persons.'
I fmd in myself a considerable measure of agreement with what Rawls says here, so
perhaps I might be some kind of constructivist. But it would still be nice to be clearer
about what that is. This will take us some distance from Rawls but we will return to
him in the end.
Consider sets of norms. In order to be metaphysically as unassuming as we can let's
suppose that sets of norms are just sets of sentences in the imperative mood, so they are
things of whose truth or falsity there can be no question. Though sets of norms
are never true they are sometimes interesting. We could say that a set of norms is
]-interesting if it captures the requirements of justice in such a way that if and only if
you act in accordance with it you will act justly. And we could say that a set of norms is
M-interesting if it captures the requirements of morality in this way. And we could say
that it is R-interesting if it captures the requirements of rationality in this way.
And now there may be interesting theories to me proposed of the form:
A set of norms is !-interesting iff it is N.
where "!-interesting" is a general place holder where we might stick "]-interesting" or
"M-interesting" or "R-interesting" or any of the other ways ofbeing interesting that
we might, on a given occasion, be interested in.
There are some quite familiar such theories. Thus many of you will have met:
Nonnat: A set of norms isM-interesting iff the irreducible nonnatural moral facts say
it is.
Which is quite interesting. And many of you have also met:
' Rawls 19'J\I, pp. 306-7.
Util: A set of norms isM-interesting iff it leads us to maximally promote the welfare
of all impartially considered.
(Util is of course perfectly consistent with Nonnat. For someone might combine the
metaethical claim that what moral rightness consists in is possessing the irreducible
nonnatural property of moral rightness with the normative ethical claim that it is those
actions that impartially promote welfare, and only those actions, that have this property.
Equally someone who endorsed Util might mean thereby to express a form
of reductive naturalism which simply identifies the property of rightness with that of
impartially promoting we.lfure. The latter view of course would conflict with Nonnat.)
Here are a few more:
ConstA: A set of norms is J -interesting iff it is congruent with our deeper understanding
of ourselves and our aspirations, and our realization that, given our history and the
traditions embedded in our public life, it is the most reasonable doctrine for us.
ConstA': A set of norms is )-interesting iff it would be agreed to by rational agents
in the original position.
(Rawls, as I read him, believes both ConstA and ConstA', but takes A to be
fundamental in that A' derives its warrant from the application of A.)
ConstB: A set of norms isM-interesting iff is it not possible for anyone to reasonably
reject it as a basis for informed, unforced, general agreement.
ConstC: A set of norms isM-interesting iff it is possible for a rational agent to will
it as a universal law.
(Not every reading of ConstC is necessarily constructivist but Christine Kors-
gaard's is, as is Rawls's own.
What might make these, at least under certain constructions, forms of constructivism?
I think at least four things need emphasizing at the outset, though they are not all evident
from the formulaic summaries just aired. The ftrst is that it seems to be characteristic of
constructivist versions of our general formula that we fmd normative concepts, rational,
reasonable, etc., on the right-hand side of the biconditional.
So constructivism always
contrasts with reductionist views such as the more metaethically ambitious reading ofUtil.
The second feature that distinguishes constructivist views in normative philosophy is
that N, the thing on the right-hand side, is intended to specify some procedure we can
in principle follow or some criterion we can in principle apply. It thus disallows any
possibility that moral rightness might consist in something mysterious and epistemically
inaccessible and remote to us. As Confucius is said to have rather sensibly proposed, If
the Way were remote from humanity, it would not be the Way.
The third feature is that the order of determination
is, for typical constructivists,
intended to be read from right to left. N, the right-hand side, is not intended to pick
out some feature ofinteresting things of the appropriate sort that tracks their appropriate
• Kongaard 1996, 2009; Rawls 2000. ' Pace Ridge, this volume, section 1.
Chung-YunJ! (Doctrine of the Mean), 13. ' See Wright 1992, appendix to ch. 3.
interestingness where the latter is conceived as constintted prior to and independendy of
the procedure or criterion picked out by N.
A fourth distinctive feature of consntctivism is evident from the sentence with which
Rawls follows his statement of ConstA in the long quote I started with: "And though
doing this may involve setding theoretical difficulties, the practical social task is primary."
Constructivism takes the task to which moral and/ or other normative inquiry is addressed
to be fundamentally practical as opposed to theoretical. In fixing what principles of reason
or morality to accept, we are trying to address some distinctively practical problem that we
face. We are not, in anything at all like the way scientists are, trying to fmd stuff out.
The term "constructivism" is a slippery one with considerable variation in how
various moral philosophers characterize and understand it. But in the light of the
foregoing, I think the following is roughly accurate in capturing what is shared by
the main recent writers to represent themselves as some kind of constructivist and
successfully delineates a more or less distinctive family of positions.
Constructivist views understand correct nonnative views of the relevant kind (political, ethical,
nonnative) as those which are the upshot of some procedure or criterion, where (a) that procedure or
criterion is one followable or applicable by human beings, where (b) that procedure or criterion is
itself characterized in normative terms invoking ideals of e.g. rationality or reasonableness, and (c)
applying the procedure or criterion is taken as determining or constitutive of that correctness rather
than as tracking a correctness conceived as prior to and independent of it, and (d) where the
rationale for our taking an interest in whatever the procedure or criterion in question delivers is
conceived of as speaking to distinctively practical as opposed to theoretical concerns.
A particularly interesting recent version of constructivism is articulated in Sharon
Street's recent paper "Constructivism About Reasons". Street defends what she calls
metaethical constructivism, which she defmes thus:
According to metaethical constructivism, the fact that X is a reason to Y for agent A is constituted
by the fact that the judgment that X is a reason to Y (for A) withstands scrutiny from the
standpoint of A's other judgments about reasons.•
This looks open to an obvious worry. It certainly satisfies the constraint I placed on
counting a view as constructivist that N, the procedure or criterion invoked in order to
elucidate the normative concept that is our target for elucidation, should itself be a
• Enoch 2009, p. 331 insists on the further specification that constructivist views must appeal to an
ineliminable procedure. He means by this that "[t]he normative status of the consequence here arguably
depends on actually going through the procedure, not just on it being the result of some hypothetical
procedure". But it seems rather questionable to characterize a position in such a way that none of its best-
known defenden turns out to hold it, and Enoch's characterization mrely has just t h i ~ consequence.
• Street 2008, p. 223.
nonnative concept. But surely, it might be thought, the nonnative property on the
left-hand side had better not be the very same nonnative concept as we are seeking an
elucidation o£ Otherwise surely our theory is circular and explains nothing.
The worry can be tamed, as Street makes plain. According to metaethical construc-
tivism, a certain class of judgements, we might say, are each, in isolation, answerable to
a process of scrutiny in the light of the other judgements in the class. That is not
necessarily a vicious circularity but seems rather defensible as a benign holism of the sort
standardly taken on in preferring a coherentist to a more foundationalist account of
nonnative epistemology. Normative inquiry, on this account, works much as Neur-
athian boat repair does. So thinks Street and so do many, including mysel£
Really the idea is just the £uniliar idea of reflective equilibrium but with an
enhanced alleged significance. As Street writes: "In metaethical constructivism ... the
fact that a normative judgment withstands scrutiny in reflective equilibrium is under-
stood to be not only of epistemological significance but also of constitutive signifi-
cance; in other words: this fact is understood to be not only an indication that the
nonnative judgment is correct, but what it is for that judgment to be correct. "
But now that we have beefed up the ambitions of reflective equilibrium, a new
worry about circularity arises. If what we are trying to do is give an account of what it is
to be a reason, does the invocation of judgements about reasons in our account itself
simply show that we have failed to shed any noncircular light on this? For we already
need to know what a reason is to know what judgements about reasons are. Just as we
must already know what fish are in order to know what judgements about fish are.
Where fish are concerned, Street would, I am confident, agree with this. Where
reasons are concerned she does not. Her theory, she tells us, is distinctive in the way it
"reduces facts about reasons to facts about what we judge or take to be reasons, with
the latter understood in a way that is prior to and independent of the former". This, she
says, retains the attractions of naturalism-judging something to be a reason is after all
just a mental state of a certain distinctive kind-while avoiding an unappealing kind of
naturalistic reductionism.
At which point the reader is liable to think, Rullo, this sounds a bit familiar. For there
are other people out there who seek to tame the mysteries of metaethics by first
characterizing certain mental states of normative commitment or endorsement, the
states, we might say, that normative utterances express, and then explaining what is
conveyed by normative judgements in the light of this characterization. And these
people too claim to have pulled off the attractive double act of avoiding an implausible
naturalistic reductionism while avoiding metaphysical and epistemological mystifica-
tion by saying nothing not consistent with a broadly naturalistic metaphysics. And
these people are of course people like Simon Blackburn, Allan Gibbard, and mysel£
These people are expressivists. Street is not. Rather she takes the distinctive character of
IO Ibid., pp. 238-9.
For both the quoted passage and the point about reductionism, see ibid., p. 242.
Gibbard 19YO. 2003; Blackburn 1998; Lenman 2007, 2010.
normative judgement, of the state of mind of having something strike one as counting
in favour of something, to be primitive, not open to further explanation.
That won't satisfy expressivists who will want to press the following question:
Can't you at least tell us whether this primitive state of mind is a belief, a desire, or
something else? Does the judgement that something is a reason involve some distinc-
tively cognitive orientation towards that thing? In which case we surely want a
metaethical theory to tell us what we are believing about something when we believe
it to be a reason. And until it does we surely haven't really made any metaethical
progress at all. But if the state in question is fundamentally noncognitive in character
then the dialectical pressure to give in and go expressivist seems irresistible.
Street sees this problem and seeks to address it. Normative judgements, she says,
clearly differ from both beliefS and desires. They clearly differ from beliefs in that they
have a necessary connection to motivation that belieli; as such lack. So far so familiar.
And they differ from desires because, she says, while the normative judgement that
I have a reason to X constitutively involves judging that I have a reason to Y where Y is
some necessary means to Xing, desires as such have no such constitutive involvement.
There are two problems with the latter claim. First it is not so obviously correct. It might
be argued, and has recently been energetically argued by Stephen Finlay, that desire
does indeed constitutively involve desire to take the necessary means to its object.
Secondly and less controversially, nobody ever said that normative judgements and
desires were the same thing. The expressivist claim is that normative judgements are a
distinctive subspecies of desire, broadly understood. And the very thing that makes that
distinctive subspecies distinctive may well be some constitutive feature its members, as
such, must have that is not shared by the wider species as a whole.
Street's view, I suggest, is attractive but unstable. It is highly natural to recast it in
expressivist terms. Doing so might yield something like the following:
REASON: To think a consideration C a reason in favour ofsomeone's 4>-ing is to
favour that consideration guiding deliberation and action in furtherance of concerns,
desires, and aims to which it speaks in virtue of it and they being able to withstand
scrutiny in the light of other reasons.
This is still an attractive view but it is no longer a view in competition with expressivism.
Still further concerns for a view such Street's are aired by some other contributors to
this volume. Thus Scanlon urges that the idea of reflective equilibrium could never
u Street 2008, esp. pp. 239-42.
" Cf. Ridge, this volume, who develops a similar point.
" Finlay 2()()8.
itself be the basis of a constructivist account of reasons generally. The process of seeking
reflective equilibrium about a subject matter cannot itself supply the basis for a con-
structivist account of that subject matter because the process of following the procedure
specified in a constructivist account cannot itself require "repeated judgments about the
domain in question". That a judgement to the effect that something is a reason for me is
among my judgements in reflective equilibrium means that the thing in question is
indeed a reason for me only if the judgements I make along the way to that equilibrium
are themselves sound. But that fact cannot, without circularity, furnish the basis for a
constructivist account of what such soundness consists in. So while Scanlon thinks the
method of reflective equilibrium is our best method for ascertaining what reasons we
have, he thinks it cannot furnish us with a constructivist account of what a reason is.
This is a powerful objection. But something not unlike Street's view might perhaps
avoid its force. Consider:
RED: x is red iff x's appearing red survives the scrutiny of some appropriate
normative discipline.
This discipline might include appraising the stability of x's red appearance by checking
it against x's earlier and later appearances, appraising its commonality by comparing
notes with others about how x appears, checking that x's appearance is not distorted by
peculiarities or deficiencies in the lighting conditions or in the observer's visual
faculties, etc. Here we might think of things like x's appearing red as raw input states,
not so much judgements as experiences distinguished by a particular phenomenal
character. It is only through the application of disciplined scrutiny that they somehow
solidify into judgements. There might be a version of Street's constructivism that
sought to escape Scanlon's objection by similarly taking normative judgements, qua
inputs into reflective scrutiny, what Michael Ridge in his contribution to this volume
calls primitive normative judgements, as raw feelings of this sort.
But, as Ridge's own discussion of this possibility makes clear, this is not a promising
view. He makes a number of objections of which I here note two. First, I would
concur with his observation that "it is not all that phenomenologically plausible that
there is any 'pure' feeling of normativity". Second, as Ridge again rightly observes: "it
is hard to see how to as much as make sense of the idea that dispositions to have such
feelings might be seen as being structured into a 'web ofjudgments' in any sense. That
metaphor is apt only when we have something that looks much more like a genuine
judgement which might bear conceptual and epistemic relations to other judgments."
The normative discipline is to take the form of pressure towards coherence; there will
be no way for it to get to work at all on pure raw contentless experiences.
Far more promising to conceive our normative judgements as desires, broadly
understood. If we did this only for such judgements conceived as inputs, only for
Scanlon, this volume, p. 231.
All quotations from Ridge, this volwne, pp. 154 and 155.
primitive nonnative judgements, we would end up, as Ridge notes, with a familiar
fonn of subjectivism. But if we so regard all nonnative judgements, primitive or
otherwise, we will get something like the fonn of expressivism articulated in
REASON. Here the solidification into judgement imposed by normative discipline
can be understood in terms of ways in which the applicability of norms of stability, of
commonality, of coherence and consistency to what are, in the ftrst analysis, passions in
our souls, end up, if the quasi-realist project in metaethics can be made to succeed,
intelligibly construed also, in the fmal analysis, as truth-apt judgements capable of at
least some forms of objectivity.
A still further concern about circularity is raised by Ridge when he is discussing the
possibility of understanding primitive nonnative judgements as sharing a direction of ftt
with belie£ In which case, like myself above, he is keen to be told what content these are
supposed to have. The content of a given such judgement, he observes, better not be just
that it withstands scrutiny from the standpoint of other nonnative judgements or the
account will be circular. Street herself has briefly aired the possibility that we might
understand primitive normative judgements as referring to sui generis normative proper-
ties. This, she recognizes, would conunit her to an error theory about primitive nonnative
and Ridge is surely right to judge that there would be little prospect of
containing our scepticism to this level. If the raw materials are rotten, so, inevitably, will be
whatever we can make from them.
Once again, an expressivist tum, taking the raw
materials as desires, tames the problem nicely.
Among constructivism's most energetic critics are Nadeem Hussain and Nishi Shah,
notably in their 2006 paper, "Misunderstanding Metaethics". Here Hussain and Shah
focus on the version of constructivism defended by Christine Korsgaard. Their princi-
pal objection to Korsgaard is that she seems to suppose her constructivism is a
metaethical view offering a genuine alternative to such familiar metaethical positions
as nonreductive realism or expressivism. Thus consider Korsgaard's transcendental
argument in Sources of Normativity that seeks to show us that we are rationally required
to value our humanity.
Here Hussain and Shah protest, justly I think, that this doesn't
really add up to a distinctive position in metaethics until we have been told what the
devil this valuing is supposed to be. Is valuing my humanity a matter of believing
See Gibbard 1990, part III.
Street 2008, p. 241, n. 55.
ao Ridge, this volume, pp. 150--1. Dorsey, this volume, develops a similar proposal while seeking to avoid
error theory by divorcing semantic content from truth in normative contexts. But his view still finishes by
taking normative truth to be a matter of coherence among judgements that speak of nothing real. Perhaps it
might make sense to characterize normative truth in this way but it is puzzling why, so characterized, we
should take much interest in it.
21 Kongaard 1996.
something about it? If so, exactly what are we supposed to be believing when we value
our humanity? Or it is some other kind of psychological orientation towards humanity,
something perhaps like love? Whatever the answer, as long as it is missing, they deny
that Korsgaard is in the business of meta ethics at all.
My own account as sketched here is not vulnerable to this charge. By marrying a
form of constructivism to a form of expressivism I give an account that speaks to the
semantic and metaphysical concerns of metaethics: on my expressivist account, nor-
mative judgements are taken to be a distinctive subspecies of desires. But that is not to
say the constructivism bit is otiose and itself does no metaethical work. For the account
of reason given by REASON helps us, I hope, to see part at least of what is distinctive
about the distinctive subspecies. What does it tell us?
I think what it tells us is a bit complicated. REASON looks, to echo Street, like a
highly formal, far from substantive, account of normativity. Which is sort of right. But
REASON is a bit substantive and here's why. REASON takes normative judgements
as a species of desire that is constitutively responsible to some standard of coherence in
the light of other normative judgements. And normative judgements like that plausibly
must be governed by some sort of very weak norm of what we might call the unification
of agency. I stress that the sort of thing I mean by this is very modest, not a philosophical
hat from which I propose suddenly to pull the categorical imperative. (I would never
do that to my fellow Humeans.) It's simply a matter of how Neurathian boat repair
works. We fix this bit of the boat buoyed up by the bits we fixed yesterday and last
week. And this is possible, at least at the normative end of the analogy, only if we have
some trust in and accord some authority to those other bits. Without some degree of
normative community between myself now and at earlier times, without a degree of
diachronic normative stability between my deliberating reflective self now and at
earlier times, I doubt that normative thinking, as opposed to a much simpler kind of
practical thought-Jimmy want fish, jimmy grab fish--would get off the ground at all.
Normative thought is deeply concerned with such unification. The form of con-
structivism I'm defending takes seriously the idea that we should see normative
questions as addressed to distinctively practical as opposed to theoretical problems.
The fundamental practical problem to which normative thought is addressed is that of
conflict. I have many many desires and often they pull me in conflicting directions. The
way I, like all human beings, solve this problem is by reflection, by reflecting criticaUy
on my desires and seeking to distil from the chaos of brute first-order desires a far less
chaotic body of desires I reflectively endorse and so stand ready wholeheartedly to
promote to the status of intentions or plans. The idea is that this body of reflectively
sanctioned desires can, if I do my normative thinking right, be made to cohere, as the
chaotic mass of raw wants from which it emerges does not cohere. (That is why Gary
Hussain and Shah 2006; see esp. p. 274 .
., Cf. Gibbard 1990, ch. 9. See also Lenman 2011.
Watson's famous question-What's so special about second-order desires?-is not in
fact so hard to answer.
In order then to be in the game of normative thought at all, I need to be in the
business of seeking to unify my agency. And as well as unification playing some such de
jure role in my thought, my playing the game at all also requires at least a minimal level
of de facto unification, in particular, as I just urged, some minimal degree of diachronic
stability of normative response and judgement. But here I stress again all this is very
minimal. The unity of agency required to count as a maker of normative judgements at
all is fantastically modest. So modest that, to all practical intents and purposes, it is
highly appropriate to think of REASON, following Street as a formal rather than
substantive constructivist account of reasons. So we are a long way from Rawls-with
whom we began-and a long way from Scanlon. A long way from those philosophers'
rich and impressive accounts of the workings of moral or political justification.
But not such a long way. Take a look at this:
MORALITY: To call a consideration C a moral reason in favour of someone's </>-ing is
to favour that consideration guiding our deliberation and action in furtherance of
concerns, desires, and aims to which it speaks in virtue of the fact that we might all,
insofar as we are reasonable, agree in endorsing it and them as able to withstand scrutiny
in the light of our other reasons.
Here my reasons have become our reasons. I like REASON as a theory of reasons and
I rather like MORALITY as a theory of moral reasons. MORALITY embodies,
among other things, a particular view of moral epistemology that regards moral inquiry
as the pursuit of the kinds of normative commonalities needed to make life together as
a well functioning moral community possible for us, a view that puts what we can
justify to others at central stage.
It is sometimes asked to what is moral theorizing responsible in the way scientific
theorizing is to empirical observation. The question is hard and the standard answer,
intuition, is of course notorious in its feebleness unless we can find a lot more to say.
I like this rather different answer. My moral theorizing is responsible to, well, to you,
provided only you are willing to engage with me in reasonable ways. And likewise to
everyone else with whom I share a social world and seek to live in some kind of
normative community, including my own later selves. Intuitions are important here all
right but they're not best understood as beliefs about some independent order of moral
facts. Rather they are a sort of desire. I have a strong intuition that torture is wrong.
See Watson 1975; and cf. Bratman 2004.
» This paragraph recapitulates thoughts aired in Lenman 2007, section Ill. Cf. Lenman 2009 .
.. Who is included in "we" here is of coune a larp;e question but I do not addn:1• it here.
I really do, but the best way of understanding what that means is just that I am deeply
unwilling to accept as a set of moral norms governing the society where I live any such
set that permits torture.
If you want to live in moral conununitywith me, you have to
deal with that fact. It sits there on the table of our co-deliberations and there it stays till
you are able to persuade me to remove it. This pursuit of normative agreement takes
many forms, from the relatively rough-and-tumble, messy business we call politics to
the, at least in aspiration, more careful and rigorous business that we call moral
philosophy. It's a difficult business and one that can go badly wrong. When it goes
badly wrong we shouldn't see that as a case of failing to find something out we needed
to discover but as a distinctively political catastrophe where our urgent aim of arriving
at a shared set of moral understandings we are all willing to live with has failed.
Contrast this picture of moral epistemology with a more robusdy realist alternative,
naturalist or otherwise, according to which the relevant enterprise is simply one of
fmding stuff out. On that rival Platonist view, moral epistemology is, much as you
might expect, a distinctive part of metaethics. But on my account, that isn't so. On my
account, moral epistemology is, to a great extent, not part of metaethics at all. Thus
when I say my moral theorizing is responsible to you and to others, I am not engaged in
metaethics but simply moralizing, though I believe I am moralizing well. On this
account, moral epistemology turns out to be, in Mackie's tenns,
an almost wholly
first-order enterprise, one about which, at the level of second-order metaethical
theorizing, there is really precious litde to say.
That is very much in the spirit of the expressivist project. Back in the 1980s,
Blackburn, famously, took on one of the great challenges for an expressivist metaethics,
the challenge of explaining how, on an expressivist account, morality could be mind-
independent, and tamed it by effectively taking it away from metaethics altogether.
The mind-independence of moral value, he urged, is a matter of regular first-order
normative ethical theory. A virtuous normative sensibility is one that takes the answers
to normative questions as responsible to facts that are, at least usually and for the most
part, not facts about that sensibility itsel£ If I thought the wrongness of torture
depended on me and could be undercut simply by my changing my mind about it,
that would be a disastrously bad way for me to think about torture.
Another challenge for expressivism is to tell a credible story about moral epistemol-
ogy. I think the kind of first-order moral constructivism articulated by Rawls and
others suggests a beautifully credible way to answer this question. But that answer is
part of normative ethics and is moralized through and through. So the appearance of
normative concepts on the right-hand side of MORALITY is no objection to it. Nor is
moral constructivism the only available answer. Instead of joining with you in the
Cf. Lerunan 2007.
Cf. Lerunan 2010, sec. 2.
" Mackie 1977, p. 16.
Blackburn 191!4, pp. 217-20.
search for moral understandings we can all accept, I might propose that others simply
defer to my superior wisdom, offering, if need be, to secure your deference by coercive
means. There are extraordinarily good moral reasons to reject that understanding in
favour of its constructivist alterative but expressivism as a metaethical theory says
nothing against it. The reasons in question are substantive moral reasons, not part of
metaethics. At a strictly positive level, as opposed to the business of criticizing rival
accounts, the right story for an expressivist metaethics, as such, to tell about moral
epistemology is: no story at all.
Constructivism, if MORALITY is a form of constructivism. is a story about how
moral justification is possible without reference to what Rawls calls: "the search for
moral truth interpreted as fixed by a prior and independent order of objects and
relations, whether natural or divine, an order apart and distinct from how we conceive
of ourselves". As such it coheres beautifully with the anti-realist moral metaphysics we
expressivists like. It offers us a credible and attractive story to tell about how moral
epistemology might work. This story,like Blackburn's story of mind-independence, is
itself no part of metaethics even though, again like that other story, it addresses a
challenge that arises within metaethics. So expressivists need constructivism. Expressi-
vism, in tum, offers answers to questions about how the actions and utterances in
which we make and express judgements about reason and values make good philo-
sophical sense, questions with which, as Hussain and Shah have urged and as we saw
ourselves when discussing Street, constructivism, as such, offers little help. So con-
structivists need expressivism.
Simon Blackburn, Simon. 1984. Spreading the Word: Groundings in the Philosophy of LAnguage
(Oxford: Clarendon Press).
Blackburn, Simon. 1998. &ling Passions: A Theory of Practical Reasoning (Oxford: Clarendon
Bratman, Michael. 2004. "Planning Agency, Autonomous Agency," in J. S. Taylor ( ed.), Personal
Autonomy: New Essays on Personal Autonomy and Its Role in Contemporary Moral Philosophy
(Cambridge: Cambridge University Press), pp. 33-57.
Chung-Yung (Doctrine of the Mean) (many editions and translations).
Darwall, S., Gibbard, A., and Railton, P. (eds). 1997. Moral Discourse and Practice (New York:
Oxford University Press).
Dorsey, Dale. 2012. "A Puzzle for Constructivism and How to Solve It," this volume.
Enoch, David. 2002. "Can There Be A Global, Interesting, Coherent Constructivism About
Practical Reason?" in Philosophical Explorations 12, pp. 319-39.
Finlay, Stephen. 2008. "Motivation to the Means," in David K. Chan (ed.), Moral Psychology
Today: Essays on Values, Rational choice and the Will (Dordrecht; Springer), pp. 173-91.
" for useful other discussions about the relation between constuctivism and expressivism, cf Gibbard
1999, Magri. 2002. Lenman 2010, Ridge, this volume, Wallace, this volume.
Fisher, Andrew and Kirchin, Simon (eels) 2006. Arguing About Metaethics (London: Routledge).
Gibbard, Allan. 1990. Wise Choices, Apt Feelings: A Theory of Normative judgment (Oxford:
Clarendon Press).
Gibbard, Allan. 1999. "Morality as Consistency in Living: Korsgaard's Kantian Lectures," in
Ethics 110, pp. 140-64.
Gibbard, Allan. 2003. Thinking How to Live (Cambridge, MA: Cambridge University Press).
Hussain, NadeemJ. Z. and Shah, Nishi. 2006. "Misunderstanding Metaethics," in Oxford Studies
in Metaethics 1, pp. 265-94.
Korsgaard, Christine M. 1996. The Sources of Normativity (Cambridge: Cambridge University
Korsgaard, Christine M. 2009. Self-Comtitution: Agency, Identity and Integrity (Oxford: Clarendon
Lenrnan, James. 2007. "What is Moral Inquiry?" in Proceedings of the Aristotelian Sodety, Supple-
mentary Volume 81, pp. 63-81.
Lenman, James. 2009. "The Politics of the Self: Stability, N ormativity and the Lives We Can
Live With Living," in Lisa Bortolotti (ed.), Philosophy and Happiness (London: Palgrave),
pp. 183-199.
Lenman, James. 2010. "Humean Constructivism in Moral Theory," in Oxford Studies in Meta-
ethics 5, pp. 175-93.
Lenrnan,James. 2011. "Pleasure, Desire and Practical Reason," in Ethical Theory and Moral Practice
14, pp. 143-9.
Mackie, J. L. 1977. Ethics: Inventing Right and Wrong (Harmondsworth: Penguin).
Magri, Tito. 2002. "Freres Ennernis: The Common Root ofExpressivism and Constructivism,"
in Topoi 21, pp. 153-64.
Miller, Alexander. 2003. Introduction to Contemporary Metaethics (Oxford: Polity Press).
Rawls, John. 1999. Collected Papers (Cambridge, MA: Harvard University Press).
Rawls, John. 2000. Lectures on the History of Moral Philosophy (Cambridge, MA: Harvard Univer-
sity Press).
Ridge, Michael, 2012. "Kantian Constructivism: Something Old, Something New," this
Scanlon, T. M. 2012. "The Appeal and Limits of Constructivism," this volume.
Shafer-Landau, Russell and Cuneo, Terence (eds) 2007. Foundations of Ethics (Malden, MA:
Street, Sharon. 2008. "Constructivism About Reasons," in Oxford Studies in Metaethics 3,
pp. 207-45.
Watson, Gary. 1975. "Free Agency," in]oumal of Philosophy 72, pp. 205-20.
Wright, Crispin. 1992. Truth and Objectivity (Cambridge, MA: Harvard University Press).
The Appeal and Limits of
T. M. Scanlon
In 1982, I wrote the following.
There is such a subject as moral philosophy for much the same reason that there is such a subject
as the philosophy of mathematics. In moral judgments, as in mathematical ones, we have a set of
putatively objective beliefS in which we are inclined to invest a certain degree of confidence and
importance. Yet on reflection it is not at all obvious what, if anything, these judgments can be
about, in virtue of which some can be said to be correct or defensible and others not. This
question of subject matter, or the grounds of truth, is the first philosophical question about both
morality and mathematics. Second, in both morality and mathematics it seems to be possible to
discover the truth simply by thinking or reasoning about it. Experience and observation may be
helpful, but observation in the normal sense is not the standard means of discovery in either
subject. So, given any positive answer to the first question-any specification of the subject
matter or ground of truth in mathematics or morality-we need some compatible epistemology
explaining how it is possible to discover the f.!cts about this subject matter through something
like the methods we seem to use?
I went on to say that in the case of morality there is also a third question: "Given any
candidate for the role of subject matter of morality, we must explain why anyone
should care about it." I will call this the question of practical significance: why we should
regard moral demands as ones we have strong reason to accept as guides to conduct.
In the case of normative truths, worries underlying the first two questions I have
listed were famously stated by John Mackie. Stating a version of the question of subject
matter, Mackie wrote that, "if there were objective values, then they would be entities
or qualities or relations of a very strange sort, utterly different from anything else in the
And, addressing the epistemological question, he went on to say that if we
I am grateful to Peter Koellner and Charles Panons for very helpful conunents on an earlier vetsion of
this paper.
a Scanlon 1982, p. 104.
s Mackie 1977, p. 38. In fairness to Mackie, I should emphasize that, like most people discussing these
isues at the time he was writing, he was concerned with morality, not with practical rea.'iOns more generally.
were aware of these facts about such a subject matter "it would have to be by some
special faculty of moral perception or intuition, utterly different from our ordinary
ways of knowing everything else." He continued
When we ask the awkward question, how we can be aware of this authoritative prescriptivity, of
the truth of these distinctively ethical premises or of the cogency of this distinctively ethical
pattern of reasoning, none of our ordinary accounts of sensory perception or introspection or the
framing and confitrning of explanatory hypotheses or inference or logical construction or
conceptual analysis, or any combination of these, will provide a satisfactory answer; "a special
sort of intuition" is a lame answer, but it is the one to which the dear-headed objectivist is
compelled to resort. •
I believe that Mackie states the question of subject matter in an overly metaphysical
way. In neither the case of mathematics nor that of the normative is it a problem that
facts about these domains would be different from facts about the natural world. There
is no reason to believe that the only truths are truths about the natural world. Nor is it
"incompatible with a scientific view of the world" (i.e. that world) to maintain that
there are such truths.
Even if this is so, however, there remain important questions about the subject
matter of mathematics and that of morality and of the normative more generally,
questions that an account of these domains should answer. These are, flrst, questions
about whether these subject matters can be characterized in a way that justifies
confidence that claims about them have determinate truth-values. Second, there is
the question of in what sense truths about these domains are "independent of us."
Third, there is the epistemological question of how it is possible for us to come to
know truths about the subject so characterized. This is not a problem of explaining
how we can be in touch with strange entities, but rather a question of determining
what methods offrnt-order reasoning about that subject matter should be seen as valid.
Finally, in the case of morality there is the question of practical significance-the
problem of explaining the importance that moral truths have for us.
When he speaks of claims about objective values, he may intend to contrast these with claims about
"subjective" values--claims about what a person ought to do, or has reason to do, that, unlike moral dainis,
are claimed to hold only insofar as the agent has certain desires or aims. Mackie rmy have no objectioa to
values, or claims about reasons, of the latter kind. If so, however, his position suffers certain instability. The
claim that a person has reason to do what will promote the satisfaction of his or her desires is itself a normative
claim. Indeed, it is an "objective" norrmtive claim, since it does not itself depend on what people desire, or on
what aims they have. If there is something metaphysically odd about objective oorrmtive truths, then this
supposed truth {that people have reason to do what would satisfy their desires, or promote their aims) is just as
odd as any other. The disagreement between someone who thinks that all reasons for action depend on the
agent's desires and someone who thinks that there are some reasons that do not depend on agents' desires is a
normative disagreement, not a metaphysical one. So Mackie's "argument from queerness," insofar as the
queerness involved is metaphysical, is an argument against irreducibly norrmtive truths of any kind, not just
objective moral values. At least this is how I am going to take his argument, I hope not unEUrly.
Ibid., p. 39.
I argue for these claims in my 2009 John Locke Lectures, Scanlon, forthcoming.
Many of these questions could, I believe, be satisfactorily answered by a sufficiently
clear characterization of the subject matter of a domain in first-order tenns-that is, in
the tenns used by ordinary statements in that domain. The subject matter of arithmetic,
for example, is adequately characterized by saying that it includes zero and all and only
those other numbers reached from zero by repeated applications of the successor
function. This characterization seems evidently true and non-arbitrary. Since the
truth-values of arithmetical statements are determined, ultimately, by facts about the
successor relation, this characterization supports the idea that arithmetical statements
have determinate truth-values (unless this description of an infinite domain is seen as
unintelligible). We are capable of thinking about and comparing particular finite strings
in this sequence, SO, SSO . .. so it is not mysterious how we can arrive at basic truths of
arithmetic "just by thinking about them." Moreover, on the basis of this characteriza-
tion of the domain, we can recognize as true more general axioms (such as Peano's
postulates), which can then be used to establish general theorems about numbers.
Godel's results show that any set of axioms, if consistent, will leave some sentences
undecided, but this does not mean that the sentences that are undecidable in a
particular axiom system have no determinate truth-value.
This is in contrast with the situation in set theory, where there is at present no fully
satisfactory general characterization of the domain of sets. This domain might be
characterized in two ways. The first is through sets of axioms, such as those of
Zermelo-Frankel set theory (ZF). But even if these axioms are plausible and, as far as
we can see, consistent, they may seem arbitrary in the absence of some general
characterization of the realm of sets that they describe, and which supports them in
the way in which our intuitive understanding of the natural numbers supports the
Peano postulates.
One such characterization is provided by what is called the iterative conception of
According to this conception, the universe of sets consists of just those that would
be formed in the following process: Begin, at stage 0, with a finite list of specified
elements (or with the empty set). At stage n + 1 form all sets of the basic elements and
the sets that were created at previous stages. For each limit ordinal A., at stage A. form the
set of all sets formed at stage a for all a<A.. The iterative conception provides a rationale
for most of the standard axioms of accepted set theory. As I have stated it, the account is
vague or incomplete in a number of ways. First, it appeals at various points to the idea
of"all sets" "formed" at previous stages, and these ideas seem to need further specifi-
cation.' Second, it remains unspecified how far the construction extends (through "all"
of the transfmite ordinals?).
• On this conception and the adequacy of the basis it provides for axioms of set theory, see Shoenfield
1977; Boolos 1971; Pa11ons 1983; and Boolos 1989.
' AJ Pa110ns noteS (Parsons 1983), there are also questioru about how the idea of"earlier and later" in the
sequence is to be unde11tood.
The more serious problem, however, is that neither this conception nor the axioms
of ZF seem to give a complete characterization of the realm of sets. Important questions
are left open, such as the status of the continuum hypothesis-the question of whether
there are transfinite sets that are larger than the set of all natural numbers but smaller
than the set of all sets of natural numbers. To resolve this question, and others, we need
a further characterization of the realm of sets. This might be provided by additional
axioms, or by augmenting the description offered by the iterative conception. This
might be done by changing the starting point-beginning with something other than
the empty set--or by redefining the steps through which the hierarchy proceeds.
Ideally, one would want to do both: to find additional axioms that are supported by an
augmented version of this hierarchical process.
These additional axioms, and the characterization supporting them, would be
justified by its ability to unify and explain what seem to be the most evident truths
about sets. As Kurt Godel famously observed:
There might exist axioms so abundant in their verifiable consequences, shedding so much light
on a whole field, and yielding such powerful methods for solving problems ... that, no matter
whether they are intrinsically necessary, they would have to be accepted at least in the same sense
as a well-established physical theory.
It appears that the ultimate form of justification here is a process like what Rawls called
the method of reflective equilibrium: we look for general principles that unify and
explain what seem to us to be the most evident truths about a subject matter, being
ready along the way to change our mind about these "evident truths" when we learn
more about what general principles would be required to explain them.
So, for
example, we might be led to change our mind about the continuum hypothesis
when we discover that it can only be supported by additional axioms that have other
consequences that are very implausible. As Rawls said about the search for principles of
Moral philosophy is Socratic: we may want to change our present considered judgments orice
their regulative principles are brought to light. And we may want to do this even though these
principles are a perfect fit. A knowledge of these principles may suggest further reflections that
lead us to revise our judgments.
We may hope that this process of seeking reflective equilibrium will lead to a set of
axioms and a general characterization of the realm of sets that fit together in a
satisfactory way. This would not be a metaphysical account of the subject matter of
set theory: it would proceed entirely in first-order terms, employing the concept of a
set. Despite this, such an account would support the idea that set theory has a
• GOdel1964, p. 477.
• See Rawls 1971, sec. 9 and "The Independence of Moral Theory" in Rawls 1999.
Rawls 1971, p. 49.
detenninate subject matter, and that statements about sets have detenninate truth-
values, removing the sense that the choice of set theoretic axioms is arbitrary. But it
remains possible that this process will not lead to any single overall account of the realm
of sets but instead to several alternative accounts shaped by different additional axioms.
In this case a kind of pluralism about set theory may be correct: there may be alternative
characterizations of the domain of sets between which it is open to us to choose.
The case of morality is in some important respects analogous to that of set theory in
this regard: we lack a single convincing overall first-order characterization of the
subject matter of morality (analogous to the characterization of the natural numbers).
The search for such a characterization must proceed by the method of reflective
equilibrium, but this method may not yield a single determinate answer. Indeed, the
situation of morality seems worse than in that of set theory in one important respect:
the concepts in question cannot be characterized with sufficient precision to provide a
rigorous account of the methods of reasoning that we employ in carrying out the
search for reflective equilibrium.
I have so far been considering the advantages that could be gained from a compre-
hensive characterization of the subject matter of a domain in purely first-order terms:
by means of general claims about numbers, about sets, or about what is morally right
and wrong. Let me now tum to the idea that such an account might be constructivist.
A constructivist account, in the sense in which I will be using that term, characterizes a
domain in terms of some specified process of construction. This can take many
different forms. The process in question may involve ways of constructing objects in
the dornain of a particular kind-for example, a way of constructing a number a such
that Fa or a general way, given a number, b, of constructing a number a such that Rba.
But since the result of such a construction is always to establish the correctness of some
statement about the domain, such as "Fa," or "3xFx," or "Rba" or "(x)3yRxy," we
can think of the constructions not as ways of constructing objects but as ways of
establishing particular claims about a domain. This will be a constructivist account of
the domain (rather than merely a set of general truths about that domain) only if it is
plausible to think that what makes something a truth about the domain in question is
that it can be arrived at through steps of the kind in question.
A constructivist account will be interesting only if it specifies clearly which se-
quences of steps constitute a construction (or derivation) of the appropriate kind.
Deciding whether a given step is of an appropriate kind may involve an exercise of
judgment. But (and this is the point that justifies this seeming digression) the judgment
that is called for cannot be a judgment directly about the truth of some claim about the
domain in question, or about what seems most likely to be true about this domain.
A procedure that involves judgments of this kind may be a way of arriving at justified
A pot•ibility described in Koellner 2010.
beliefS about a domain. But such a process could not be what detennines the facts about
that domain itsel£
It follows that the process of seeking a set of beliefs about a subject that are in
reflective equilibrium may be a way of arriving at a characterization of the subject matter
of a domain, but it is not an account of that subject matter itself, and, in particular, not a
constructivist account. This is because that process-identifying those statements about
the subject that seem most clearly true, formulating general principles that would
account for those truths, then going through successive steps ofmodifyingjudgments
and principles to fmd ones that fit together properly-requires repeated judgments
about the domain in question.
In the case of mathematics, the term 'constructivist' is generally used in a sense that is
more limiting than the one I am here concerned with. The appeal of accounts that are
constructivist in this sense lies in concern about characterizations of a domain in terms
of a "completed infinite" such as "the set containing 0 and closed under successor." It
may be held that we cannot grasp the idea of such a set, and perhaps even that this
description does not identify a determinate domain. Doubts of this kind lead people to
seek an account according to which the facts about numbers are facts about what has
been or could actually be arrived at by us through certain specified means of construc-
tion. This leads to a different view of the methods of valid reasoning about numbers,
and to the conclusion that statements about numbers may not all have detenninate
truth-values, since for a given A there may be no construction leading to A and yet no
construction leading to not-A.
Things are different in the case of morality, because the source of doubts about this
domain is different. A characterization of the subject matter of morality consisting
simply of very general first-order principles of right and wrong may seem unsatisfactory
for a variety of reasons. Insofar as these principles are claimed to be true independent of
us, one may want some explanation of what kind of facts these are. This problem is
sharpened by the problem of practical significance. If moral truths are truths about a
domain of facts independent of us, why should we take these facts to be significant,
even authoritative, as guides to conduct? If moral truths were true in virtue of faots
about us, such as our preferences, or our wills, then their practical significance fot us
could be readily explained. But if they are objective truths, independent of us, then it
may well be asked why we should care about them. This fundamental tension between
objectivity and practical significance was a central element in Mackie's challenge: what
he expressed skepticism about was the possibility of facts "in the world" that had what
he called "objective prescriptivity."
Some constructivist accounts of morality are appealing because they seem weD
positioned to respond to this problem. To see why, consider first the account offered
by John Rawls, who was the first to introduce the term "constructivism" in this area;
" In Rawls 1980 and Rawls 1993, Lecture III,"Kantian Constructivism in Moral Theory." and Political
Ubrralism, Chapter III, "Political Constructivism."
Rawls was concerned not with morality in general but with justice, that is to say,
principles for the assessment ofbasic social institutions. The function of such principles,
he believed, is to serve as a shared basis for assessing conflicting claims from citizens
about what their institutions should be like. The main conflicts of this kind will be
between individuals in different economic classes and between individuals who have
different religious views or, more generally, different "conceptions of the good." So
principles appropriate for this role will need to be justified on a basis that is neutral with
respect to these forms of disagreement. Rawls suggested that this would true of
principles that would be agreed to by parties who did not know what economic
class they represented or what conception of the good they held. He made this idea
more precise in his idea of an Original Position in which principles of justice are
The resulting view, that the correct principles of justice are ones that would be
agreed to by parties in the Original Position Rawls described, has the form of a
constructivist view: facts about justice are facts about what principles would be arrived
at through a process of a certain kind. Even though Rawls defines his Original Position
quite clearly, deciding what principles of justice would or could be chosen in such a
position requires the exercise of judgment. (This is an important contrast with the case
of set theory.) But the account remains constructivist because the judgments in
question are not judgments about what is just or unjust (or morally right or wrong)
but rather judgments about what individuals who are seeking only to do as well for
themselves as they can would have reason to choose under conditions of limited
The case for accepting this constructivist account of justice lies with a reflective
equilibrium argument. The account has to be defended by claiming that it "fits with"
and provides a satisfying explanation of what are, on reflection, our considered
judgments about justice.
These include judgments "of all levels of generality"-
judgments about the justice or injustice of particular institutions and practices, maxims
stating prima facie just- or unjust-making characteristics, and judgments about the
nature and role of justice itself.
The fact that certain principles would be chosen through a procedure of the kind
Rawls described gives us reason to be concerned with them because it indicates that
they have the kind of impartial justification that makes them fit to play the role that
principles ofjustice are supposed to play. This is only a partial explanation, since it can
always be asked what reason we have to be concerned with impartial justifications of
this kind. The answer is given, I believe, by pointing to the character that our relations
Rawls 1971, Chapter Ill.
•• "On reflection" because the relevant class of judgments is not fixed at the start. Our assessment of which
judgments are "considered" is likely to change as the process of seeking principles that explain these
judgments goes forward. For more discussion, see Scanlon 2002, my "Rawls on Justification."
•• See Rawls 1951; Rawls 1971, sec. 9; and Rawls 1975. For further discussion see Scanlon 2002.
with our fellow citizens have if we are cooperating with them on principles that have
such a justification, compared with the character of these relations when our institu-
tions cannot be justified in this way. The force of this appeal can, of course, be debated
further (every justification has to start somewhere). But the &ct that it can be supported
by this explanation takes the constructivist account beyond what is offered by a first-
order account that takes the form simply of general principles specifying what justice
This constructivist account also offers an explanation of the objectivity of claims
about justice--the sense in which they are correct or incorrect "independent of us."
To assess this explanation we should consider first the various ways in which the idea of
independence can be understood.
The first is the idea that a subject matter is "independent of us," and judgments
about it are objective, if it is possible for us (at least individually) to be mistaken in our
judgments about it. Call this minimal objectivity judgment-independence. Truths of
arithmetic and set theory are objective in this sense. Why do I believe that they are?
Because there are ways of thinking about these questions which anyone who under-
stands the subject can engage in, and which seem to lead clearly to certain conclusions.
These "ways of thinking" need not be algorithms, although in the case of arithmetic
they often are. More informal methods--such as those we use to convince ourselves of
many of the axioms of set theory---can suffice. This minimal objectivity (judgment-
independence) is not just a matter of de facto agreement, but also the tendency of the
judgments of different competent judges to converge, and the stability of our own
judgments, which supports our confidence that they concern judgment-independent
truths. It is important, for example, that there is such a thing as discovering eaors in our
thinking, and that when we have identified something as an eaor this conclusion is
generally stable--we do not generally Bip back the other way.
Judgment-independence is an important property, but it is a quite minimal notion of
objectivity. Even judgments about what is the case according to a make-believe game
can be objective in this sense, as can judgments about what is permitted by certain
social norms. But these things are not "independent of us" in at least one of the senses
that we have in mind in discussing objectivity.
A second, stronger idea is that a subject matter is independent of us, and judgJ;nents
about it are objective, if these judgments are minimally objective and, in addition, the
standards for assessing such judgments do not depend on what we have done, chosen,
or adopted, and would not be different had we done, chosen, or adopted something
dse. I will call this choice-independence. Judgments about what one has reason to do in a
make-believe game and judgments about what one has to do given arbitrary social
norms are judgment-independent but not choice-independent. To be clear: judgments
about what is true within such a game, or what is required by a social norm, can be both
judgment-independent and choice-independent. What are not choice-independent
are judgments about what one has reason to do or believe that presuppose these games
or norms.
Many mathematical judgments, such as propositions of number theory and arithmetic,
are objective in both of these senses. Many axioms of set theory (such as the standard
Zennelo-Frankel axioms) are choice-independent as well as judgment-independent.
But if "pluralism" of the kind mentioned earlier turns out to be correct then there are
further axioms of set theory that are not choice-independent.
I believe that many judgments about reasons for action are both judgment-
independent and choice-independent, although there is disagreement about whether
this is so. But even if judgments about reasons are independent of us in both of these
senses they are not "independent of us" in the further sense of being independent of
what we are like. This might be called independence of human nature. If we were
constituted or situated differently, so as to have different needs, or so as to enjoy
different things, then our reasons would be different. In the case of reasons, lack of
independence in this sense does not seem to be a problem. One would not expect or
want them to be independent of our nature in this way. (Mathematical truths, on the
other hand, might be expected to have this kind of independence, although it is an
interesting question how it would matter if they lacked it.)
What does seem clear, however, is that the significance of mathematical truths does
not depend on their being independent of us in any further, distinctively ontological
sense of having a real existence apart from us. Facts about physical objects do need to
have such an existence. But there is, I believe, no reason to construe mathematical
truths on this model.
According to Rawls's constructivist account, judgments about justice are objective
in the sense ofbeingjudgment-independent if judgments about what parties in Rawls's
Original Position have reason to choose, and judgments about what follows from those
principles, are judgments we can be mistaken about. This seems quite plausible.
Judgments about what the parties in the Original Position would have reason to
choose are nonnative judgments. But their objectivity is less controversial than that
of many nonnative judgments, because they are only hypothetical normative judg-
ments-judgments about what parties have reason to do given certain specified aims
and given certain background information. These judgments are in this respect like the
judgments I mentioned above, such as judgments about games or social norms, which
have judgment-independence but lack choice-independence. This is not, however, a
flaw from the point of view of the aims of a constructivist theory.
As I said earlier, the appeal of a constructivist theory lies in the promise of giving an
account of a subject matter that supports the idea that judgments about it have
detenninate truth-values and that provides or fits with a plausible account of the
practical significance of such judgments. The fact that, on such an account, the
detenninate truth-value of judgments about justice depends on the choice of a
particular way of defining the Original Position is not a problem if the fact that a
principle would be arrived at in an Original Position of that kind is a reason for us to
give that principle the authority claimed for principles of justice.
A claim that being chosen in such an Original Position is a reason to regard a
principle as authoritative is an unconditional nonnative claim about what we have
reason to do in the relevant circumstances, hence more controversial than the condi-
tional claims about what parties in the Original Position would have reason to choose,
given their aims and the infonnation available to them. But this should not undennine
the determinateness ofjudgments about justice on this constructivist account, even for a
person who has doubts about the objectivity of unconditionally nonnative judgments.
What we have, then, is a two-part thesis: an account of the truth-values of judgments
about justice, which depends only on conditional normative claims, and an account of
the significance of such judgments, which depends on claims about reasons whose
objectivity may be more controversial.
Any suggestion of arbitrariness flowing from the dependence of the truth-values of
judgments about justice on the choice of a particular way of defining the Original
Position would be further reduced, or even eliminated, I would say, if a larger
reflective equilibrium argument establishes that the account of justice provided by
this definition is the one that best fits with all of our considered judgments about
justice. (It is exactly the lack of such an argument that could lead to a "pluralist" view of
set theory.)
My own contractualist account of moral right and wrong could also count as a
constructivist account (in this case, an account of individual morality).
According to
this account, in order to detennine whether an action is morally permissible we should
consider a general principle that would permit it. We then consider what objections
individuals in various situations could offer to this principle based on the way in which
they would be affected by it, by living with the consequences of the actions it would
permit and with the possibility that agents may perform such actions, since they would
be permitted to do so. We then compare these reasons with the reasons that individuals
would have to object to a principle that would forbid actions of the kind in question;
based, again, on how they would be affected by such a principle, and consider whether
it would be reasonable for those who have reason to object to the principle permitting
the action to reject it, given the reasons that others have for objecting to the contrary
principle. If it would be reasonable to reject that principle, then the action in question
would be morally wrong. The rightness or wrongness of an action depends on what the
correct outcome of this procedure would be, whether or not anyone has carried it. out.
Like Rawls's constructivist account of judgments about justice, this account of the
subject matter of moral rightness and wrongness makes the truth-values of judgments
about this subject matter depend on facts about what principles individuals in certain
circumstances would have reason to reject. It then explains the practical significance of
judgments about right and wrong on the ground that we have reason to care about
principles that could not be rejected in this way--specifically, that we have reason to
" See Scanlon 1998, Cbapten 4, 5.
care about this kind of justifiability of our actions to others. The case for accepting this
account of rightness and wrongness then depends, as before, on a reflective equilibrium
argument that it provides the best overall account of our considered judgments about
this ·subject matter, including, as before, judgments "of all levels of generality," not just
judgments about the rightness and wrongness of particular actions.
On this account, judgments of right and wrong will have a particular form of
objectivity-will be judgment-independent or choice-independent-just in case
judgments about what individuals in certain circumstances have reason to reject have
these forms of objectivity. In contrast to Rawls's constructivist account of justice, in this
case the judgments on which the truth-values and objectivity of moral judgments are
made to depend are fully normative judgments about reasons for action. They are
judgments about what individuals in specified circumstances who, among other things,
care about fmding principles others could also accept, would have reason to do. But the
judgments that are called for involve judgments about which things individuals have
more reason to want to have, or to avoid. Since someone like Mackie, who has doubts
about the objectivity ofmoraljudgments is likely to have doubts about the objectivity
of judgments about reasons for action in general, this brings us to the general question
of how the determinateness and objectivity of judgments of the latter kind can be
explained and defended. In particular, it raises the question of whether a constructivist
view might provide a satisfactory account of the subject matter of normative judgments
in general.
Such an account would involve a process for "constructing" reasons-that is, for
arriving at conclusions about whether a given consideration is or is not a reason for a
person in certain circumstances to act in a certain way-such that there are determinate
standards for the validity of steps involved in this process. Moreover, for reasons
mentioned above, the validity of these steps cannot depend on the truth of claims
about which things are or are not reasons for action.
If such an account of reasons for action is to have ambitions parallel to those of
constructivist accounts of justice and moral rightness and wrongness, the fact that the
conclusion that P is a reason for a person in C to do A can be arrived at through this
process should help to explain the practical significance of a claim that Pis a reason. In
this case, however, it is not clear what is to be explained. The practical significance of
judgments about justice or about moral right and wrong can be seen as lying in the fact
that when such a judgment is correct then we have reason to be guided by it in deciding
what to do. So in these cases practical significance can be explained in terms of reasons.
But where judgments about reasons for action are concerned this is not an option. It is
nonsensical to ask what reason we have to do what we have reason to do. So if the
practical significance of judgments about reasons is to be explained this must take some
other form.
Finally, if a constructivist account of reasons for action is to be supported
For fuller discussion of this question see Scanlon, forthcoming, Lecture 1.
in reflective equilibrium, it should seem evident that at least those judgments about
reasons that seem most clearly correct are ones that could be arrived at through the
procedure that this account describes.
Although I think that constructivist accounts of justice and morality have some
plausibility, I do not believe that a plausible constructivist account of reasons for action
in general can be given. In the remainder of this chapter I will try to explain why.
The best-known attempt to provide such a view is what has come to be called Kant's
Categorical Imperative procedure.
Kant's Categorical Imperative is a test of the
acceptability of maxims, which I will take to be general policies of taking certain
considerations as reason to act in certain ways. A maxim passes the Categorical
Imperative test if it can be willed to be a universal law or if adopting it is consistent
with regarding rational nature (whether one's own or that of another rational creature)
as an end in itself. This is commonly understood as a test of the moral acceptability of a
maxim, and of acting on such a maxim. So understood, it seems too weak to provide
a general account of reasons for action since, presumably, it can be permissible for a
person to do things that, as it happens, he or she has no reason to do. In Korsgaard's
version of the Kantian account, this gap is filled by the idea of an agent's practical
identities. A practical identity is "a description under which you value yourself and fmd
your life worth living and your actions to be worth undertaking. Conceptions of
practical identity include such things as roles and relationships, citizenship, member-
ships in ethnic or religious groups, causes, vocations, professions, and offices. "
The overall idea is this: insofar as we see ourselves as acting at all, we must see the
Categorical Imperative as constraining our practical thought. So we have the reasons
specified by the maxims that this requires us to adopt. Beyond this, we have reasons to
do those things that are required by the more specific practical identities we have
adopted, provided that these are compatible with the Categorical Imperative. This
view can be seen as constructivist insof.rr as it provides a procedure through which it is
determined whether something is a reason for a person: something is a reason for a
person if denying that it was a reason would violate the Categorical Imperative or be
inconsistent with some practical identity (consistent with the Categorical Imperative)
that that person has adopted.
The fact that valid judgments about reasons arise from
this process is supposed to explain their practical significance: their special authority lies
in the agent's own will-in the fact that they flow from choices the agent has made-or
from an identity that an agent must endorse insof.rr as she sees herself as acting at all.
According to this account, judgments about reasons are objective in the sense of
being judgment-independent: they are the kind of thing that someone can be mistaken
See Kongaard 1996; O'Neill 1989.
•• Korsgaard 2009, p. 20.
This account could also be seen as constructivist insofar as it holds that we construct the domain or
reasons by adopting particular ends and practical identities, in ways consistent with the Categorical Impera-
tive. This seems to me entirely consistent with the version stated in the text so I will not explore it separately.
about. Some judgments about reasons-those following from the Categorical Impera-
tive itself--are also choice-independent. The reasons Bowing from the choices of
practical identity that the agent has made will not be choice-independent. Whether
this is a Baw in the account, or an advantage, is something to be determined by the
process of seeking reflective equilibrium in our overall judgments about reasons. It is
quite plausible to say that some reasons a person has depend on prior choices he or she
has made. The question is when this is true and how this dependence is best explained.
I cannot here give a full examination of the Kantian theory, which is subde and
complex, but will just state briefly my reasons for finding it unsatisfactory.
despite the initial appeal of various forms of Kant's Categorical Imperative as moral
requirements, I am not convinced by any arguments I have seen for the claim that we
must see these requirements as binding on us insofar as we see ourselves as acting at all.
Second, although it seems true that individuals have different reasons depending on the
ends and practical identities they have adopted, these reasons depend on their having
good reasons to adopt those ends or identities in the flrst place, and not to revise or
reject them. And these reasons in tum are not all adequately explained by the
Categorical Imperative test.
If the Kantian constructivist account of reasons is not satisfactory, is there another
constructivist account that would be more satisfactory? Part of the appeal of the
Kantian account is that it not only promises to provide grounding for the correctness
of judgments about reasons but also promises to do this in a way that explains the
practical significance of these judgments. It attempts to do this by grounding facts about
reasons in a conception of rationality. So one question is whether there might be a
different conception of rationality that could play this grounding role. This would have
to be a conception that did not itself involve or depend on substantive claims about
what reasons people have, but which led to conclusions about such claims. I do not
myself see what such a conception could be like. Some things that are referred to as
conceptions of rationality are very general substantive theses about reasons-such as the
idea that it is rational for a person to do what is in his or her self-interest. Such an
account could not serve as an explanation of the practical signiflcance of these reasons.
The only non-substantive alternative that I am aware of is the formal conception of
rationality discussed by John Broome and others, according to which rational require-
ments are simply requirements of consistency among a person's practical attitudes.
These requirements have no substantive implications about the reasons people have. So
no account of either of these kinds would provide a basis for claims about reasons.
It is possible, however, that there might be a constructivist account of reasons of a
less ambitious kind. Such an account would keep the ambition of characterizing the
domain of reasons in a way that supported the idea that judgments about reasons have
determinate truth-values. But it would abandon the further aim of explaining the
For slightly fuller discussion see Scanlon 2011 .
See Broome 19\19, 2005; Kolodny 2005.
practical significance of conclusions about reasons, resting simply with the idea that the
"normative authority" of a reason is simply that-being a reason--and that this cannot
be explained any further way. Such an account would be similar to the "constructivist"
account of sets offered by the Iterative Conception. This account characterizes the
universe of sets by describing the way in which sets are constructed out of sets or non-
set elements. But it is stated in the language of set theory, and does not claim to explain
what a set is. Might there, then, be a constructivist account of reasons of this more
modest kind?
It does seem that some reasons depend on others. Roughly speaking, it seems that if
a person bas good reason to have a certain end, then be or she has good reason to do
what will promote it, and if a person bas good reason to hold a certain value, or to
adopt a particular practical identity, then he or she has good reason to do what is
involved in respecting this value or living in accord with this identity.
One might say,
then, that these relations of dependence between reasons are ways in which the domain
of reasons is determined, by the construction of some reasons on the basis of others.
The problem, however, is that there seem to be too many seemingly independent
reasons that are not constructed from others in this way. To refer again to the analogy
with set theory, the question would be what is to play the role here analogous to the
empty set, or a small set of prior elements from which the domain of sets may be seen as
constructed. One proposal might be that the only underived reasons are reasons to
avoid pain and seek pleasure, and that all other reasons are constructed out of these in
ways like those just described. This does not seem to me a very plausible proposal,
because it does not seem that these are the only underived reasons. And this conclusion
points toward a more general one: the domain of reasons, even reasons for action,
seems too varied and complex to be plausibly analyzed in this way.
How, then, do we come to know particular underived truths about which things are
reasons? My own answer is that we do this simply by thinking carefully about what
seem to us to be reasons, considering what general principles about reasons would
explain them, what implications these would have, considering the plausibility of the
implications of these principles and so on. For example, suppose it seems to me that
someone bas reason to do A because be or she would fmd it pleasant. Pleasure does not
always constitute a reason (pleasure in the suffering of others, for example, does not).
So we need to ask what attitudes this particular pleasure involves, and whether they are
attitudes that one has reason to want, or reason to want in the particular circumstances
in question.
One might characterize this process as one of bringing one's particular judgments
about reasons and one's general principles about when something is a reason into
reflective equilibrium. This seems to me broadly correct, although misleading in some
ways, of which I have space here to mention only a few.
Exactly how this is so is a complicated matter. See Raz 2005 and Kolodny ms.
The fmt is that the distinction between particular judgments and general principles
that explain them is in this case not clear. As I mentioned earlier in discussing pain, the
process I have just described is one of coming to a clearer understanding of the
conditions, c, under which some fact p is a reason. So it is as much a matter of clarifying
what particular judgment we in fact accept, as a matter of finding a separate principle
that "explains" this judgment. Second, the label "equilibrium" is misleading insofar as
it suggests that one's reasons for accepting the judgments one holds at the end of the
process include the fact that one has achieved an "equilibrium" or "coherence" among
one's various beliefS.
The problem with this is that mere coherence is too easily
achieved--one could attain it in many ways, simply by rejecting one or another set of
conflictingjudgments. The justificatory force of the judgments we arrive at lies, rather,
in the details of the process that leads to these judgments. The reason for accepting one
of these judgments is that (as far as one can tell) it entails other judgments that are true,
or is supported by judgments that lead to such conclusions. (Just as, in the case of set
theory, one has reason to accept a new axiom because it supports or is supported by and
unifies other set theoretic claims that appear to be correct.)
These factors being noted, it seems to me that such a process of careful reflection is
the only way we have of arriving at conclusions about reasons for action. Might this
process itself be seen as a constructivist account of the domain of reasons? Sharon Street,
for example, speaks of "a constructivist view according to which the truth of 'X is a
reason for agent A to Y' is a function of whether that judgment would be among A's
evaluative judgments in reflective equilibrium. "
There are several problems with this proposal. First, it is not plausible to claim that if
the judgment that X is a reason for A to Y would be among A's evaluative beliefS in
reflective equilibrium then X is a reason for A to Y however A carried out the process of
reaching this equilibrium. There are many lazy and sloppy ways of reaching equilibrium.
As I just pointed out, the normative status conferred by a judgment's being in a set that
is in reflective equilibrium depends on the quality of the decisions that are made in
arriving at it--decisions about what to count as a considered judgment at the outset and
about what to modify in situations of conflict. So the most that could be said is that X is
a reason for A to Y if the judgment that it is such a reason would be among A's
evaluative judgments in reflective equilibrium if the judgments A made in arriving at this
equilibrium were sound. So understood, however, this is not a constructivist view, since
the steps involved in carrying out the process in question would involve making
judgments about what is or is not a reason.
Leaving aside the question of the applicability of the label "constructivist," however,
it is true for the same reason that the process of seeking reflective equilibrium in one's
beliefS about reasons is not an account of the subject matter of practical reasons at all.
In deciding whether a certain claim is among one's "considered judgments," or in
•• For fuller discussion see Scanlon 2002. •• Street 2006, p. Ito.
deciding whether to modify such a judgment in the light of its conflict with a principle
one has arrived at or whether, instead, to modify or abandon the principle in the light
of this conflict, the question one asks cannot be "Will this judgment be among those
I would arrive at if I reached reflective equilibrium?" but rather "Is this judgment
correct?" The process of seeking reflective equilibrium in one's beliefS about a subject
matter is therefore not a characterization of the truth about that subject matter but
rather a method for arriving at such a characterization.
Barwise, Jon (ed). 1977. Handbook rif Mathematical Logic (Amsterdam: North Holland).
Benacerraf, Paul and Hilary Putnam (eds). 1983. Philosophy of Mathematics: Selected Readings,
Second edition (Cambridge: Cambridge University Press).
Boolos, George. 1971. "The Iterative Conception ofSet," in Benacerrafand Putnam, 1983,
pp. 486--502.
Boolos, George. 1989. "Iteration Again," in Philosophical Topics 42, pp. 5-21.
Broome, John. 1999. ""Normative Requirements," in Ratio 12, pp. 398-419.
Broome, John. 2005. "Does Rationality Give Us Reasons?" in Philosophical Issues 15,
pp. 321-37.
Buneo, Otavio and Oystein Linnebo (eds). 2009. New Waves in Philosophy rif Mathematics
(Houndmills: Macmillan Publishing).
Freeman, Samuel (ed.). 2002. The Cambridge Companion to Rawls (Cambridge: Cambridge
University Press).
Godel, Kurt. 1964 "What Is Cantor's Continuum Problem?" in Benacerrafand Putnam, 1983,
pp. 47o-85.
Koellner, Peter. 2010. "Truth in Mathematics: The Question of Pluralism," in Buneo and
Linnebo, 2010, pp. 8o-117.
Kolodny, Niko. 2005. "Why Be Rational?" in Mind 114, pp. 509-63.
Kolodny, Niko. Unpublished ms. "Toward a Principle oflnstrumental Transmission".
Korsgaard, Chrisrine. 1996. The Sources rifNormativity (Cambridge: Cambridge University Press).
Korsgaard, Christine. 2009. Self-Constitution: Agency, Identity, and Integrity (Oxford: Oxford
University Press).
Mackie,). L. 1977. Ethics: Inventing Right and Wrong (Harmondsworth: Penguin).
O'Neill, Onora. 1989. Constructions rif Reason (Cambridge: Cambridge University Press).
Parfit, Derek. 2011. On Hlhat Matters (Oxford: Oxford University Press).
Parsons, Charles. 1983. "What Is the Iterative Conception of Set?" in Benaceraf and Putnam,
1983, pp. 503-29.
Rawls, John. 1951. "Outline of a Decision Procedure for Ethics," in Rawls, 1999, pp. 1-19.
Rawls, John. 1971. A Theory rifjustice (Cambridge, MA: Harvard University Press).
Rawls, John. 1980. "Kantian Constructivism in Moral Theory," in Rawls 1999, pp. 303-58.
Rawls, John. 1993. Political Liberalism (New York: Columbia University Press).
Rawls, John. 1999. Samuel Freeman, ed., Collected Papers (Cambridge, MA: Harvard University
Raz, Joseph. 2005. "The Myth of Instrumental Rationality," in o u r t u ~ l of Ethics and Social
Philosophy 1, pp. 1-28.
Scanlon, T. M. 1982. "Conttactualism and Utiliarianism," in Sen and Williams 1982,
pp. 103-129.
Scanlon, T. M. 1998. What We Owe to Each Other (Cambridge, MA: Harvard Univenity Press).
Scanlon, T. M. 2002. "Rawls on justification," in Freeman, 2002, pp. 139-67.
Scanlon, T. M. 2011. "How I Am Not a Kantian," in Parfit, 2011, volume 2.
Scanlon, T. M. Forthcoming. Being Realistic about Reasons.
Sen, Amartya and Williams, Bernard (eds). 1982, Utilitarianism and Beyond (Cambridge: Cam-
bridge Univemty Press).
Shoenfield, J. R. "Axioms of Set Theory," in Batwise, 1977, pp. 321-44.
Street, Sharon. 2006. "A Darwinian Dilemma for Realist Theories of Value," in Philosophical
Smdies 127. pp. 109-66.
(This bibliography is limited to contemporary philosophy written in English.)
1. Bruce Ackerman: Social]ustice in the Liberal St4te (New Haven: Yale Univenity Press, 1981).
2. Carla Bagnoli: "Rawls on the Objectivity of Practical Reason," in Croatian journal of
Philosophy 3, 2001, pp. 307-31.
3. Carla Bagnoli: "Truth, Invention, and Construction," in Notizie di Politeia 61, 2001,
pp. 4 ~ 7 .
4. Carla Bagnoli: ''Realism as a Moral Achievement," in Notizie di Politeia 66, 2002, pp. 51-63.
5. Carla Bagnoli: "Moral Constructivism: A Phenomenological Argument," in Topoi 21,2002,
pp. 125-38.
6. Carla Bagnoli: "The Authority of Reflection," in Theoria 58, 2007, pp. 43-52.
7. Carla Bagnoli: "Respect and Membenhip in the Moral Community," in Ethical Theory and
Moral Praaice 10, 2007, pp. 113-28.
8. Brian Barry: Theories of justice (Berkeley; Univenity of California Press, 1989).
9. Brian Barry: justice as Impartiality (Oxford: Clarendon Press, 1995).
10. Kenneth Baynes: "Constructivism and Practical Reason in Rawls," in Analyse et Critique 14,
1992, pp. 18-32.
11. Lisa Bortolotti (ed.): Philosophy and Happiness (London: Palgrave, 2009).
12. Michael Bratman: "Practical Reason and Acceptance in a Context," in Mind 101, 1992,
pp. 1-14 and in Faces of Intention, pp. 15-34.
13. Michael Bratman: "Identification, Decision and Treating as a Reason," in Philosophical Topics
24, 1996, pp. 1-18 and in Faces of Intention, pp. 185-206.
14. Michael Bratman: "The Sources ofNormativity," in Philosophy and Phenomerwlogical Research
58, 1998, pp. 699-702, and in Faces of Intention, pp. 265-78.
15. Michael Bratman: Faces of Intention: Selected Essays on Intention and Agent}' (New York:
Cambridge Univenity Press, 1999).
16. Michael Bratman: "Reflection, Planning and Temporally Extended Agency," in Philosophi-
cal Review 109, 2000, pp. 35-61.
17. Michael Bratman: "Valuing and the Will," in Philosophical Perspectives 14, 2000, pp. 249-65.
18. Michael Bratman: "Hierarchy, Circularity and Double Reduction," in Buss and Overton,
Contours of Agent)', pp. 65-85.
19. Michael Bratman: "A Desire of One's Own," in]oumal of Philosophy 100,2003, pp. 221-42.
20. Michael Bratman: "A Thoughtful and Reasonable Stability," in Harry Frankfurt, Taking
Ourselves Seriously and Getting it Right, pp. 77-90.
21. David 0. Brink: "Rawlsian Constructivism in Moral Theory," in Canadian Journal of
Philosophy 17, 1987, pp. 71-90.
22. David 0. Brink: Moral Realism and the Foundations of Ethics (Cambridge: Cambridge Univer-
siry Press, 1989).
23. Sarah Buss and Lee Overton {eds): Contours of Agency: Essays on Themes ..from Harry Frankfort
(Cambridge, MA: MIT Press, 2002).
24. Matthew Chrisman: "Constructivism, Expressivism and Ethical Knowledge," in International
journal of Philosophical Studies 18, 2010, pp. 331-53.
25. G. A. Cohen: Rescuing Justice and Equality (Cambridge, MA: Harvard University Press,
26. Joshua Cohen: "Moral Pluralism and Political Consensus," in Copp, Hampton and Roe-
mer, The Idea of Democracy, pp. 27Q-91.
27. Mary Clayton Coleman: "Public Reasons and Practical Solipsism," in Southern journal of
Philosophy 43, 2005, pp.: 317-36.
28. Mary Clayton Coleman: "Korsgaard on Kant on the Value ofHumanity," in journal of Value
Inquiry 40, 2006, pp. 475-8.
29. David Copp: "Contractarianism and Moral Skepticism," in Vallentyne, Contractarianism and
Rlltional Choice, pp. 196-228.
30. David Copp, Jean Hampton and John Roemer (eds): The Idea of Democracy (Cambridge:
Cambridge University Press, 1993).
31. D. Copp and D. Zimmerman (eds): Morality, Reason and Truth (Totowa: Rowman and
Allanheld, 1985).
32. Norman Daniels (ed.): Reading Rllwls (Oxford: Blackwell, 1975).
33. Norman Daniels: "Wide Reflective Equilibrium and Theory Acceptance in Ethics," in
Journal of Philosophy 76, 1979, pp. 256-82.
34. Norman Daniels: justice and Justification: Reflective Equilibrium in Theory and Practice (New
York: Cambridge University Press, 1996).
35. Stephen Darwall: Impartial Reason (London: Cornell University Press, 1983).
36. Stephen Darwall: The Second-Person Standpoint (Cambridge MA: Harvard University Press,
37. Stephen Darwall, Allan Gibbard, and Peter Railton, "Toward Fin de siecle Ethics: Some Trends,"
in Philosophit41 Review 101, 1992, pp. 115-89, reprinted in Darwall, Gibbard and Railton, Moral
Discourse and PractU:e, pp. 3-47.
38. Stephen Darwall, Allan Gibbard, and Peter Railton: Moral Discourse and Practice (Oxford:
Oxford University Press, 1997), pp. 3-47.
39. Gerald Dworkin: "Contractualism and the Normativity ofPrinciples," in Ethics 112, 2002,
pp. 471-82.
40. Ronald Dworkin, "The Original Position," in Daniels, Reading &wls, pp. 16-53.
41. David Enoch: "Agency, Shmagency: Why Normativity Won't Come from What Is
Constitutive of Action," in Philosophical Review 115, 2006, pp. 169-98.
42. David Enoch: "Can There Be a Global, Interesting, Coherent Constructivism about
Practical Reason?" in Philosophical Explorations 12, 2009, pp. 319-39.
43. Eckart Forster (ed.): Kant's Transcendental Deductions (Stanford: Stanford University Press,
44. Harry Frankfurt: "Freedom of the Will and the Concept of a Person," in Journal of Philosophy
68, 1971, pp. 5-20 and in The Importance of What We Care About, pp. 11-25.
45. Harry Frankfurt: "Identification and Wholeheartedness," in Schoeman, Responsibility, Char-
acter and the Emotions, pp. 27-45 and in The Importance of W'hat We Care About, pp. 159-76.
46. Harry Frankfurt: The Importance of Hlhat We Care About: Philosophical Essays (Cambridge:
Cambridge University Press, 1988).
lllliLIOGRAPHY 245
47. Harry Frankfurt: "The Faintest Passion," in Proceedings and Addresses of the American Philo-
sophical Association 66, 1992, 5-16 and Necessity, Volition and Love, pp. 95-107.
48. Harry Frankfurt: "On the Necessity of Ideals," in Noam and Wren, The Moral Self,
pp. 16--27 and Necessity, Volition and Love, pp. 108-116.
49. Harry Frankfurt: Necessity, Volition and Love (Cambridge: Cambridge University Press,
50. Harry Frankfurt: Taking Ourselves Seriously and Getting it JQght {Stanford: Stanford University
Press, 2006).
51. Samuel Freeman: "The Burdens of Public Justification: Constructivism, Contractualism and
Publicity," in Politics, Philosophy and Economics 4, 2007, pp. 5-43.
52. Samuel Freeman: RAwls (London: Routledge, 2007).
53. Samuel Freeman (ed.): The Cambridge Companion to Rawls (Cambridge: Cambridge Univer-
sity Press, 2003).
54. Richard Galvin: "Rounding Up The Usual Suspects: Varieties ofKantian Constructivism in
Ethics," in Philosophical Quarterly 61, 2011, pp. 16--36.
55. David Gauthier: "Justice as Social Choice," in Copp and Zimmerman, Morality, Reason and
Truth, pp. 251--{)9.
56. David Gauthier: Morals by Agreement (Oxford: Clarendon Press, 1986).
57. David Gauthier: "Morality, Rational Choice, and Semantic Representation," in Social
Philosophy and Policy 5, 1988, pp. 173-221.
58. David Gauthier: Moral Dealing (Ithaca: Cornell University Press, 1990).
59. David Gauthier: "Why Contractarianism?" in Vallentyne, Contractarianism and Rational Choice.
60. David Gauthier: "Political Contractarianism," in ]oumal of Political Philosophy 5, 1997, pp.132-48.
61. David Gauthier: 'Uniting Separate Persons', in Gauthier and Sugden, RAtionality, justice and
the Social Contract, pp. 176--92.
62. David Gauthier and Robert Sugden {eds): RAtionality, Justice and the Social Contract: Themes
from "Morals by Agreement," (London: Harvester Wheatsheaf, 1993).
63. Allan Gibbard: Wise Choices, Apt Feelings (Cambridge MA: Harvard University Press, 1990).
64. Allan Gibbard, "Morality as Consistency in Living: Korsgaard's Kantian Lectures," in Ethics
110, 1999,pp. 140--{)4.
65. Allan Gibbard: Reconciling Our Aims (Oxford: Oxford University Press, 2008).
66. Jean Hampton: "Equalizing Concessions in the Pursuit of Justice: A Discussion of Gauthier's
Bargaining Solution," in Vallentye, Contractarianism and RAtional Choice, pp. 149--{)1.
67. Barbara Herman, 'Justification and Objectivity: Comments on Rawls and Allison," in
Forster, Kant's Transcendental Deductions, pp. 131-41.
68. Barbara Herman: "Pluralism and the Community of Moral Judgment," in Heyd, Toleration,
pp. 60--80.
69. David Heyd: Tokration: An Elusive Virtue (Princeton: Princeton Univenity Press, 1996).
70. Thomas E. Hill Jr.: "Kantian Constructivism in Ethics," in Ethics 99, 1989, pp. 752-70).
71. Susan Hurley: Natural Reasons: Personality and Polity (Oxford: Oxford University Press,
72. NadeemJ. Z. Hussain and Nishi Shah: "Misunderstanding Metaethics," in Oxford Studies in
Meatethics 1, 2006, pp. 265-94.
73. Aaron James: "Rights and Circularity in Scanlon's Contractualism," in journal of o r r ~ l
Philosophy 1, 2004, pp. 369-77.
74. Aaron James: "Constructing Justice for Existing Practice: Rawls and the Status Quo," in
Philosophy and Public Affairs 33, 2005, pp. 281-316.
75. Aaron James: "Power in Social Organization as the Subject of Justice," Pacific Philosophical
Quarterly 86, 2005, pp. 25-49.
76. Aaron James: "Equality in a Realistic Utopia," in Sodal Theory and Praaice 32, 2006,
pp. 699-724.
77. Aaron James: "Legal and Other Governance In Second Person Perspective," in Loyola Law
Review 40, 2006, pp. 911-37.
78. Aaron James: "The Objectivity of Values: Invariance without Explanation," in Southern
journal of Philosophy 44,2006, pp. 581-605.
79. Aaron James: "Constructivism about Practical Reasons," Philosophy and Phenomenological
Research 74, 2007, pp. 302-25.
80. Robert Johnson: "Value and Autonomy in Kant's Ethics," in Oxford Studies in Metaethics 2,
2007. pp. 133-48.
81. Erin Kelly: "Doing without Desert," in Pacific Philosophical Quarterly 83, 2002, pp. 18G-205.
82. Erin Kelly and Lionel McPherson: "On Tolerating the Unreasonable," in journal of Political
Philosophy 9, 2001, pp. 38-55.
83. George Klosko: "Political Constructivism in Rawls' Political Liberalism," in American Political
Science Review 91, 1997, pp. 635-46.
84. Christine Korsgaard: "Aristotle and Kant on the Source of Value," in Ethics 96, 1986,
pp. 486-505 and Creating the Kingdom of Ends, pp. 225-48.
85. Christine Korsgaard: "Personal Identity and the Unity of Agency: A Kantian Response to
Parfit," in Philosophy and Public Affairs 18, 1989 pp. 101-32 and in Creating the Kingdom of
Ends, pp. 363-97.
86. Christine Korsgaard: "Creating the Kingdom of Ends: Reciprocity and Responsibility in
Personal Relations," in Philosophical Perspectives 6, 1992, pp. 305-32 and in Creating the
J(jngdom of Ends, pp. 188-221.
87. Christine Korsgaard: "The Reasons We Can Share: An Attack on the Distinction
between Agent-Relative and Agent-Neutral Values," in Sodal Philosophy & Policy 10,
1993, pp. 24-51 and in Creating the J(jngdom of Ends, pp. 275-310.
88. Christine M. Korsgaard: Creating the Kingdom of Ends (Cambridge: Cambridge University
Press, 1996).
89. Christine M. Korsgaard: The Sources ofNormativity (Cambridge: Cambridge University Press,
90. Christine Korsgaard: "Self-Constitution in the Ethics ofPlato and Kant," in The journal of
Ethics 3, 1999, pp. 1-29 and in The Constitution of Agency, pp. 10G-28.
91. Christine Korsgaard: "The Dependence of Value on Humanity," in Raz, The Practice of
Value, pp. 63-85.
92. Christine Korsgaard: "Realism and Constructivism in Twentieth-Century Moral Philoso-
phy," in]oumal of Philosophical Research APA Centennial Supplement, 2003, pp. 99-122 and
in The Constitution of Agency, pp. 302-326.
93. Christine Korsgaard: "Morality and the Logic of Caring: A Commentary on Harry Frank-
furt," in Frankfurt, Taking Ourselves Seriously & Getting it Right, pp. 55-76.
94. Christine Korsgaard: The Constitution of Agency (Oxford: Clarendon Press, 2008).
95. Christine Korsgaard: Self-Constitution: Agency, Identity, and Integrity (Oxford: Clarendon
Press, 2009).
96. Larry Krasnoff: "How Kantian is Constructivism?" in Kant-Studien 90, 1999, pp. 385-409.
97. Christine Lafont: "Moral Objectivity and Reasonable Agreement: Can Realism be Re-
conciled with Kantian Constructivism?" in Ratio Juris 17, 2004, pp. 27-51.
98. Charles Larmore: The Morals of Modernity (Cambridge: Cambridge University Press, 1996).
99. Charles Larmore: The Autonomy of Morality (Cambridge: Cambridge University Press, 2008).
100. Mark LeBar: "Aristotelian Constructivism," in Social Philosophy and Policy 15, 2008,
pp. 182-213.
101. James Lenman: "Michael Smith and the Daleks: Reason, Morality and Contingency," in
Utilitas 11, 1999, pp. 164-77.
102. James Lenman: "Against Moral Fictionalism," in Philosophical Books 49, 2008, pp. 23-32.
I 03. James Lenman: "What is Moral Inquiry?" Proceedings of the Aristotelian Society Supple-
mentary Volume 81, 2007, pp. 63-81.
104. James Lenman: "Achieving Objectivity," in Philosophical Perspectives 23,2009, pp. 293-304.
105. James Lenman: "The Politics of the Self Stability, Normativity and the Lives We Can Live
With Living," in Bertolotti, Philosophy and Happiness, pp. 182-99.
106. James Lenman: "Naturalism Without Tears," in Ratio 22, 2009, pp. 1-18.
107. James Lenman: "Humean Constructivism in Moral Theory," in Oxford Studies in Metaethics
5, 2010,pp. 175-93.
108. Lionel McPherson: "Normativity and the Rejection of Rationalism," inJoumal of Philoso-
phy 104 (2007): 55-70.
I 09. Tito Magri: "Frires Ennemis: The Common Root of Expressivism and Constructivism," in
Topoi 21, 2002. pp. 153--64.
110. Ronald Milo: "Contractarian Constructivism," in Journal of Philosophy 122, 1995, pp. 181-204.
111. Christopher W. Morris: "A Contractarian Account of Moral Justification in Sinnott-
Armstring and Timmons," in Moral Knowledge?, pp. 215-42.
112. G. C. Noam and T. Wren (eds): The Moral Self (Cambridge MA: MIT Press, 1993).
113. Onora O'Neill: Constructions of Reason: Explorations of KAnt's Practical Philosophy (Cam-
bridge: Cambridge University Press, 1989).
114. Onora O'Neill: Towards Justice and Virtue: A Constructive Account of Practical Reasoning
(Cambridge: Cambridge University Press, 1996).
115. Onora O'Neill: "Constructivism vs. Contractualism," in Ratio 16, 2003, 319-31.
116. Onora O'Neill: "Constructivism in Rawls and Kant," in Freeman, The Cambridge Com-
panion to Rawls, pp. 347-67.
117. Japa Pallikkathayil: "Deriving Morality from Politics: Rethinking the Formula of Human-
ity," in Ethics 121, 2010, pp. 116-47.
118. John Rawls: "Oudine of a Decision Procedure for Ethics," in Philosophical Review60, 1951,
pp. 177-97 and Collected Papers, pp. 1-19.
119. John Rawls: A Theory of Justice (Cambridge, MA: Harvard University Press, 1971 and
Oxford: Clarendon Press, 1971).
120. John Rawls: '1ustice as Fairness: Political Not Metaphysical," in Philosophy and Public a i r s
14, 1985 pp. 223-52 and Collected Papers, pp. 388-414.
121. John Rawls: "The Independence of Moral Theory," in Proceedings of the American Philo-
sophical Association 48 1975, pp. 5-22 and Collected Papers, pp. 286-302.
122. Jolm Rawls: "Kantian Constructivism in Moral Theory: The Dewey Lectures 1980," in
]oumal of Philosophy 77, 1980, pp. 515-72 and Collected Papers, pp. 303-358.
123. Jolm Rawls: Political Liberalism (New York: Columbia University Press, 1993).
124. Jolm Rawls: "The Idea ofPublic Reason Revisited," in University of Chicago Law Review 64,
1997, pp. 764-807 and Collected Papers, pp. 573-615.
125. Jolm Rawls: Colleaed Papers (Cambridge, MA: Harvard University Press, 1999).
126. Jolm Rawls: A Theory rifjustice, revised edition (Cambridge, MA: Harvard University Press,
127. Jolm Rawls: Lectures on the History of Moral Philosophy (Cambridge, MA: Harvard University
Press, 2000).
128. Joseph Raz: The Practice rif Value (Oxford: Clarendon Press, 2003).
129. Michael Ridge: "Saving Scanlon: Contractualism and Agent-Relativity," in journal rif
Political Philosophy 9, 2001, pp. 472-81.
130. Miriam Ronzoni: "Constructivism and Practical Reason: On lntersubjectivity, Abstrac-
tion, and Judgment," in]oumal rif Moral Philosophy 7, 2010, pp. 74-104.
131. Miriam Ronzoni and Laura Valentini: "On the Meta-Ethical Status of Constructivism:
Reflections on G.A. Cohen's 'Facts and Principles'," in Politics, Philosophy and Economics 7,
2008. pp. 403-22.
132. Mark van Roojen: "Rationalist Realism and Constructivist Accounts of Morality," in
Philosophical Studies 126, 2005, pp. 285-95.
133. T. M. Scanlon: "Rawls' Theory of Justice," in Daniels, Reading Rawls.
134. T. M. Scanlon: "Contractualism and Utilitarianism," in Sen and Williams, Utilitarianism and
135. T. M. Scanlon: "The Aims and Authority of Moral Theory," in Oxford journal rif Legal
Studies 12, 1992, pp. 1-23.
136. T. M. Scanlon: H'hat We Owe to Each Other (Cambridge, MA: Harvard University Press,
137. T. M. Scanlon: "Reasons and Passions," in Buss and Overton, Contours of Agency.
138. T. M. Scanlon: "Metaphysics and Morals," in Proceedings of the American Philosophical
Association 77, 2003, pp. 7-22.
139. T. M. Scanlon: "Rawls on Justification," in Freeman, The Cambridge Companion to Rawls.
140. Tamar Schapiro: "Three Conceptions of Action in Moral Theory," in Nous 35, 2001,
pp. 93-117.
141. Ferdinand Schoeman (ed.): Responsibility, Character, and the Emotions: New Essays in Moral
Psychology (New York: Cambridge University Press, 1987).
142. Amartya Sen and Bernard Williams (eds): Utilitarianism and Beyond (Cambridge: Cambridge
University Press, 1982).
143. Russ Shafer-Landau: Moral Realism: A Defense (Oxford: Oxford University Press, 2003),
Chapter 2.
144. Yonatan Shernmer: "Desires as Reasons," in Philosophy and Phenomenological Research 75,
2007, pp. 326-48.
145. Yonatan Shemmer: "Full Information, Well Being, and Reasonable Desires," in Utilitas 23,
2011, pp. 206-27.
146. Peter Singer: "Sidgwick and Reflective Equilibrium," in The Monist 58, 1974, pp. 490-517.
147. Walter Sinnott-Annstrong and Mark Timmons (eds): Moral Knowledge?: New Readings in
Moral Epistemology (Oxford: Oxford University Press, 1996).
148. Walter Sinnott-Annstrong: Moral Skeptidsms (Oxford: Oxford University Press, 2006).
149. John Skorupski: "The Ontology of Reasons," in Topoi 21, 200, pp. 113-24.
150. Holly Smith: "Deriving Morality From Rationality," in Vallentyne, Contractarianism and
Rational Choice, pp. 229-53.
151. Michael Smith: The Moral Problem (Oxford: Blackwell, 1994).
152. Cynthia Stark: "Hypothetical Consent and Justification," in Journal of Philosophy 97, 2000,
pp. 313-34.
153. Sharon Street: "Constructivism About Reasons," in Oxford Studies in Metaethics 3, 2008,
pp. 207-45.
154. Sharon Street: "Reply to Copp: Naturalism, Normativity, and the Varieties of Realism
Worth Wonying About," in Philosophical Issues 18, 2008, pp. 207-28.
155. Sharon Street: "In Defense of Future Tuesday Indifference: Ideally Coherent Eccentrics
and the Contingency ofWhat Matters," in Philosophical Issues 19,2009, pp. 273-98.
156. Sharon Street: "What is Constructivism in Ethics and Metaethics?" in Philosophy Compass 5,
2010, pp. 363-84.
157. Valerie Tiberius: "Full Information and Ideal Deliberation," in journal of Value Inquiry 3,
1997, pp. 329-38.
158. Valerie Tiberius: "Justifying Reasons for Valuing: An Argument Against the Social
Account," in Southern journal of Philosophy 37, 1999, pp. 141-58.
159. Valerie Tiberius: "Humean Heroism: Value Commitments and the Source ofNormativ-
ity," in Pad.fic Philosophical Quarterly 81, 2000, pp. 426-46.
160. Valerie Tiberius: "Maintaining Conviction and the Humean Account ofNormativity," in
Topoi21, 2002,pp. 165-73.
161. Valerie Tiberius: "Practical Reason and the Stability Standard," in Ethical Theory and Moral
Practice 5, 2002, pp. 339-53.
162. Valerie Tiberius: "Value and Practical Deliberation," in Philosophical Studies 111, 2002,
pp. 146-72.
163. Valerie Tiberius: The Rt:fiective Life: Living Wisely Within Our Limits (Oxford: Oxford
University Press, 2008).
164. Mark Timmons: "The Limits of Moral Constructivism," in Ratio 16, 2003, pp. 391-423.
165. Peter Vallentyne (ed.): Contractarianism and Rational Choice: Essays on David Gauthier's
"Morals by Agreement" (Cambridge: Cambridge University Press, 1991).
166. Peter Vallentyne: "Contractarianism and the Assumption of Mutual Unconcern," in
Vallentyne, Contractarianism and Rational Choice, pp. 71-5.
16 7. Peter Vallentyne: "Gauthier's Three Projects," in Vallentyne, Contractarianism and Rational
Choice, pp. 1-11.
168. David Velleman: Practical Rtfiection (Princeton: Princeton University Press, 1989).
169. David Velleman: The Possibility of Practical Reason (Oxford: Oxford University Press, 2000).
170. David Velleman: How We Get Along (Cambridge: Cambridge University Press, 2009).
171. Ralph W edgwood, "Practical Reasoning as Figuring out What is Best: Against Construc-
tivism," in Topoi 21, 2002, pp. 139-52.
172. R. Jay Wallace: "Nonnativity and the Will," in Royal Institute o/ Philosophy Supplement 55,
2004, pp. 195-216.
173. R.Jay Wallace: "Constructing Nonnativity," in Topics 32,2004, pp. 451-76.
174. Andrew Williams: "Justice, Incentives and Constructivism," in Ratio 21, 2008, pp. 476-93.
175. Crispin Wright: Truth 11nd Objectivity (Cambridge, MA: Harvard University Press, 1992).
action, philosophy of 81, 95
alethic pluralism 109, 112-16
Alston, William P. 101
Altham,J. E.J. 149
Ameriks, Karl 122, 134
Anscombe, G. E. M. 129-32, 148
anti-realism see realism
Aristode 196, 206
arithmetic 132, 228, 233-4
attitude-dependence see mind-dependence/
autonomy 119-35,140, 143,152-4,176
Azzouni, Jody 116
Bagnoli, Carla 41
Blackburn, Simon 4, 23, 64, 66,217,223-4
Boolos, George 228
bootstrapping 18, 28-32, 36
Bratman, Michael 165, 222
Brink, David 105
Broome,John 19,23, 77, 89,161,238
categorical imperative 1, 139, 237--8
choice-independence see mind dependence/
Chrisman, Matthew 5
circularity 48-9, 87, 102-3, 106-8, 110,
112-13, 116, 145, 164, 170, 180, 184, 187,
Clarke, Samuel 131-3
coherence 108-15,160,164,220-2,240
see also reflective equilibrium
principle of 160, 164
conunitments, practical 22-37, 30, 45, 50-2,
56-7,81,84, 86,92-7,202,206,
208, 217
Confucius 215
Consistency 20, 71, 1594>0, 163, 164, 167,
principle of 159-60, 164, 167,169-70, 175
about justice 1-2, 232-7, 213-14
about mathematics 226-31
about morality 213-14, 222-4, 231-7
about 40-58, 99-118, 139-57,
216-22, 236-41
andnaturalism 19-20,119-20,150,153,157,
215.217. 184
dependent 161, 183
formalist /procedural 31, 33, 35-7, 88, 208
foundational 162
Humean 4, 41-5, 49, 51-8, 143-57,
independent 161
local 3, 139,161,183,206
Kantian 4,41-2,45-58,138-43,213-14,
metaethical, metanormative 139, 143-57,
180, 186, 216-22 see also constructivism,
Neuradllan 83, 162,206,217
political 2-3, 138, 140
restricted see constructivism, local
substantive 88, 208
thoroughgoing see constructivism,
wise judgment 195-212
contingency 42, 55--8
continuum hypothesis 229
contractualism 21, 82, 132, 209
Crisp, Roger 132
Cullity, Garrett 121
Cuneo, Terence 213
Darwall, Stephen 2, 61, 119, 140-3,
Darwin, Charles 63
Dean, Richard 125
decision making 73-4, 126, 159-60, 165-6,
206, 210-11, 240
desire 3, 6, 42-4, 46, 48, 55, 147-57,
Dewey, John 126-7
Dorsey, Dale 5, 41, 220
Dworkin, Ronald 63-4, 66, 100-1
direction of fit 148-5 7, 220
Enoch, David 2, 5, 6, 216
epistemology see rnoraVnormative epistemology
error theory see scepticism
Euthyphro Dilemma 64, 130
expressivism 4-5, 20-1, 23, 32, 68, 140-1,
Firth, John 203
Fisher, Andrew 213
Foucault, Michel 63
Frankfurt, Harry 4, 37, 43, 81, ~ . 90,92-4,97
Freeman, Samuel 4
Frege, Gortlob 192
Gaut, Berys 121
Gauthier, David 1-2
Gibbard,Allan 2,5,23,61,64,66,68,84,87,94,
Godel, Kurt 228-9
Guyer, Paul 125
Hare, John E. 120, 122, 129, 134
heteronomy 123-5, 129, 133
Hill, Thomas E. 2, 205
Hills, Alison 121, 122, 125
Hobbes, Thomas 1-2, 162
Humberstone, lloyd 148
Hume, David 42, 63, 127
Hussain, NadeemJ. Z. 5, 61, 78, 81, 91, 141,
220-1, 224
Irwin, Terence 134
James, Aaron 2
Johnson, Robert 122-3
judgement-independence see mind-
justice 1, 3, 213-14, 229, 232-7
justification see moral/normative epistemology
Kain, Patrick 122, 134
Kant, Immanuel 1, 6, 22, 41-2, 46, 48, 50,
52, 55,58,68,120-8,131,133-4,139,
Kirchin, Simon 213
Kirkham, Richard 109, 111-12,190
knowledge see moral/nonnative epistemology
Koellner, Peter 230
Kolodny, Niko 89, 161, 238-9
Korsgaard, Christine M. 2-5, 18, 22, 24, 26-9,
Kripke, Saul 103
Lafont, Christine 121, 124
Langton, Rae 122, 125
Larmore, Charles 119, 128, 130
Leibniz, G. W. 121
Lenman,James 41, 57, 68, 162-3, 200,
Lewis, David 42
Lynch, Michael 109, 114-15, 117
Mackie, J. L. 111, 226-7, 236
Magri, Tito 224
mathematics 3, 72, 132, 226-31, 234
McDowell,John 62, 78, 119, 127-8, 151-2, 196
Millikan, R. G. 151-3
mind-dependence/independence 42-4,
Moore, G. E. 140
moral/nonnative epistemology 2, 4, 5, 217,
moral obligation 123, 129, 131
morality 1-3, 6, 42, 45, 52, 55-7, 103, 109,
ideal observer theory 203-4
Platonism 60, 65-7, 73
theological voluntarism 60, 6 ~
moral obligation 123, 129, 131
Nagel, Thomas 22, 61, 63-4,
Neiman, Susan 121
Nietzsche, Friedrich 63, 126
authority 19, 27, 33, 35, 76, 130-1, 171,
constitutive account of normative judgments 87
judgments 70, 85, 87, 91, 94, 184, 198
normativity 4-5,25-7,41-58,99-118, 143-57,
see also reason, reasons
constitutive account of 19, 22, 25, 27-8, 31,
33, 40,43-4,46, 51,55-7, 62,64, 69,
70-1, 73
conventionalism 60,66-7,73
justification of 163
voluntarism 60, 65, 76, 77, 122
objectivity 42,45,214,220,226-7,231,233-8
O'Neill, Onora 2, 3, 237
Parfit, Derek 27
Parsons, Charles 228
Peano's postulates 228
Peirce, C. S. 129
Pippin, Robert 130
practical identity 45-51, 237-9
practical point of view I standpoint of the agent
3,5-6, 24,32,40-2,45,52-3,68-9,76,
83-5, 91, 93, 95-7, 106, 139, 205
procedure, constructive 2-3, 35, 125, 138-40,
authority of 171
constitutive account of 130-1
psychologism 24-8
Quine, W. V. 0. 109
quietism, metaethical 61, 63, 68, 78, 186, 188
Railton , Peter 2, 61, 140-3, 184, 213
rationality, principles of 19, 160
recalcitrant irrationality 36, 177-8
Rawls, John 1-4, 21, 41, 60, 68, 69, 75, 82,
120, 121, 138, 140, 143, 183, 213-16,
Raz, Joseph 18, 63, 89, 239
realism 3, 41,47-51,56,78, 100, 106, 109, 111,
119, 123-35, 140-1, 184, 206
authoritarianism of 126
reason, reasons 18-21, 24--6, 29, 32, 41-58,
61,66, 70,76,81-97,99-126,143-57,
159--61, 165-70, 175-8, 18()-7, 196-7,
constitutional model 17 4
diachronic stability 173
intelligibility of 73
theoretical 185
Reath, Andrews 120
reflective equilibrium 199, 207, 217-19,
229-30, 232, 235-41
relativism 193
Ridge,Ntichael 5--6,215,218,219-20,224
Rorty, Richard 125-9
Rousseau, Jean-Jacques 1, 176
Russel, Bertrand 189
Sartre,Jean-Paul 93
Scanlon, T. M. 2, 3, 18, 21, 26, 60, 63, 64, 77, 81,
scepticism/error theory 2, 4, 20, 27, 33, 42, 62,
67, 111, 117, 128, 148, 15()-2, 220, 232
Schapiro, Tamar 2
Schneewind, J. B. 65, 120,
Schoenfield, J. R. 228
Schroeder, Mark 42
self-legislation 130, 132-3
Sensen, Oliver 125
set theory 228-30, 233-4, 239-40
Shafer-Landau, Russ 61, 63, 100, 210, 213
Shah, Nishi 5, 61, 141, 22Q-1, 224
Shemmer, Y onatan 206
Smith, Michael 28, 152, 184
Sterelny, Kim 151-2
Street, Sharon 2, 3, 5, 6, 21, 22, 26, 29, 31-3,
35,37, 76,81-97,100,103-4,106,109,
12()-1, 127,138-9,143-57,162,206,
Tappolet, Christine 113-14
Tarski, Alfred 101
Taylor, Charles 52
Thrasymachus 63
Tiberius, Valerie 2, 5, 41
Trivers, R. 53
truth 3, 4, 42, 99-118, 144, 150, 220, 227,
231, 234, 238
semantic theory of 101--6, 108-13,
coherence theory of 108-13, 117-18, 189
value/valuing 5, 19-20, 4()-58, 6()-78, 85--6,
92-3, 119, 121-5, 132, 141-4, 146, 154,
198,20()-2,208, 22()-1,223-4,226-7,
Protagorean objectivity 6<>-1, 65
Walker, Ralph 189
Wallace, R.Jay 4, 5, 90, 164, 171,207,
Watson, Gary 43, 86, 221-2
Wedgwood, Ralph 103
Williams, Bernard 6, 42, 81, 84, 93, 156
wisdom 195-211
Wood, Allen 125,134
Wright, Crispin 43, 109, 114, 190, 215
Zermelo-Frankel set theory 228, 234

You're Reading a Free Preview