Paul Faulkner, Thomas Simpson - The Philosophy of Trust-Oxford University Press (2017)

OUP CORRECTED PROOF – FINAL, 01/11/2017, SPi
The Philosophy of Trust

The Philosophy
of Trust
edited by
Paul Faulkner and Thomas Simpson
1
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© the several contributors 2017
The moral rights of the authors have been asserted
First Edition published in 2017
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2016955342
ISBN 978–0–19–873254–9
Printed in Great Britain by
Clays Ltd, St Ives plc
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Contents
List of Illustrations vii

List of Contributors ix
1. Introduction 1
2. The Empowering Theory of Trust 14
Victoria McGeer and Philip Pettit
3. Trust as a Second-Personal Attitude (of the Heart) 35
Stephen Darwall
4. On the Risks of Resting Assured: An Assurance Theory of Trust 51
Edward S. Hinchman
5. Betraying Trust 70
Collin O’Neil
6. ‘But I Was Counting On You!’ 90
Karen Jones
7. The Problem of Trust 109
Paul Faulkner
8. Trust and Collective Agency 129
Bernd Lahno
9. Trust as a Two-Place Relation 149
Jacopo Domenicucci and Richard Holton
10. Deciding to Trust 161
Benjamin McMyler
11. Trust and Evidence 177
Thomas Simpson
12. Being Pragmatic about Trust 195
Philip J. Nickel
13. Trusting a Promise and Other Things 214
David Owens
14. Trustworthy Groups and Organizations 230
Katherine Hawley
vi Contents
15. Faith in Kant 251

Guy Longworth
16. ‘Trust is Basic’: Løgstrup on the Priority of Trust 272
Robert Stern
Index 295
List of Illustrations
2.1 Responsiveness to evidence of trust 26

2.2 Trust-responsive dependability 27
8.1 Hi-Lo game 133
12.1 Trust and cooperation 199
List of Contributors
Stephen Darwall, Yale University

Jacopo Domenicucci, École Normale Supérieure
Paul Faulkner, University of Sheffield
Katherine Hawley, University of St Andrews
Edward S. Hinchman, University of Wisconsin, Milwaukee
Richard Holton, University of Cambridge
Karen Jones, University of Melbourne
Bernd Lahno, Frankfurt School of Finance and Management
Guy Longworth, University of Warwick
Victoria McGeer, Princeton University
Benjamin McMyler, Texas A&M University
Philip J. Nickel, Eindhoven University of Technology
Collin O’Neil, Lehman College, New York
David Owens, King’s College London
Philip Pettit, Princeton University
Thomas Simpson, University of Oxford
Robert Stern, University of Sheffield

1
Introduction
Trust is central to our social lives. We know by trusting what others tell us. We act on
that basis, and on the basis of trust in their promises and implicit commitments. Trust
is central to our having a shared life together. Further, and non-instrumentally, trust-
ing relations are themselves of great value. In trusting others, we realize distinctive
forms of value.
This volume collects new philosophical essays on trust. By doing so, we help remedy
the relative neglect that the topic has suffered in Anglophone philosophy. This neglect
is especially striking when compared with the quantity of work on concepts of similar
significance, such as knowledge, justice, or truth. The neglect is worth remedying
because of both the importance of trust and its intrinsic interest. The task of this
Introduction is to illustrate the importance of the topic, by tracing out three broad
areas of enquiry in which trust is significant. As the essays predominantly address the
core analytical question of identifying and understanding the attitude of trust, this also
helps to set the volume in intellectual context. Regarding the intrinsic interest of the
topic, we let the essays speak for themselves. Summaries of each chapter then orient
readers to the volume’s contents.
1. Trust and Cooperation

Annette Baier’s ‘Trust and Antitrust’ (1986) is the starting point for contemporary
philosophical reflection on trust. Her paper is cited invariably for its account of trust,
on which trust is reliance on another person’s goodwill towards oneself. Her reasons
for addressing the topic, however, are pertinent here. She starts by noting a ‘strange
silence’ on trust in the moral philosophical canon, and specifically on the question of
whom I should trust, in what way, and why (p. 232). In answering that, we should find
a test for judging trust relationships from a moral point of view. The importance of
doing so, and her explanation for the strange silence, derive from the same claim.
Her central claim is that contract between free persons of roughly equal power and
capacities is a poor model for swathes of the moral life, in which we cooperate with and
2 Paul Faulkner and Thomas Simpson
care for each other. Baier justifies this by observing the diversity of relationships where
inequality of power is a fundamental fact. This is certainly the case descriptively, where
many relationships are marked with inequalities of power—think of such relations as
parent–child, husband–wife, disabled person–carer, or citizen–government official. In
some cases, though of course not all, this is normatively unproblematic. But this nor-
mativity is not well evaluated in terms of when someone is obliged to keep a contract.
Not only is this because the terms of these relationships are not actually structured by
contract, but more fundamentally, these relationships are not chosen—often and
sometimes usually un-chosen, and by at least one of the parties. In contrast, contracts
are binding only when freely entered into. The same is true of promises.
Baier takes the observation to undermine a broader liberal vision, on which the
voluntariness of relationships is a condition on their respectability. The contractarian
is hoist with the same petard. Their central device, of what people would agree to
under appropriately specified conditions, similarly fails to address the morality of
non-voluntary relationships. In contrast, trust can well exist in cooperative and caring
relationships which are not chosen. So Baier’s central aim is to provide a test for evalu-
ating when trust relationships are morally decent. The importance of addressing trust,
then, derives from the incompleteness of inherited moral theory.
Her explanation for the strange silence relates her central claim to the position of the
great dead moral philosophers. They were men, mostly of independent means, many
of whom were bachelors or functional bachelors, in a context where women were
counted on to serve men. She remarks that it should not surprise us that
they managed to relegate to the mental background the web of trust tying most moral agents to
one another, and to focus their philosophical attention so single-mindedly on cool, distanced
relations between more or less free and equal adult strangers, say, the members of an all male
club. . . . Relations between equals and nonintimates will be the moral norm for adult males
whose dealings with others are mainly business or restrained social dealings with similarly
placed males. But for lovers, husbands, fathers, the ill, the very young, and the elderly, other
relationships with their moral potential and perils will loom larger. (248)
In juxtaposition to the equality presumed by members of a club, Baier proposes infant–

parent relations as an alternative paradigm of relationship that moral theory must
address.
Baier’s claims against the contractarian and liberal are contestable. John Locke, for
instance, explicitly addresses the descriptive fact of the inequality of power, arguing
that it constitutes no objection to his contractualist doctrine (Two Treatises, II.8). But
she is surely right that there is an omission here and the incompleteness of contract for
patterns of actual cooperation has also been the starting point for a large-scale enquiry
into trust outside philosophy. After observing that much cooperation takes place out-
side of contract, social scientists from a variety of disciplines have sought to explain
how trust can sustain cooperation. Contemporary work on trust was galvanized by the
landmark publication of Trust: Making and Breaking Cooperative Relations (1988),
introduction 3
edited by Diego Gambetta. That volume brought together scholars from philosophy,
economics, history, zoology, political theory, psychology, anthropology, and sociology.
Although the contributors have no unifying intellectual vision, Anthony Pagden’s
study of some political economists from eighteenth-century Naples is illustrative of
the kind of concern animating their work. He reports their explanation of how new
Spanish rulers, the Habsburgs, reduced a once flourishing trading state to economic
ruin. They did so through instituting an aristocratic honour code that dissolved
pre-existing bonds of trust which were based on civic virtue. Similarly influential
subsequent work has picked up this theme. Francis Fukuyama (1995), for instance,
addressed the interrelation between interpersonal ties and national economic pros-
pects; and Robert Putnam’s work on social capital (1995, 2000) investigated the
political consequences of those ties.
While this emphasis on how pre-existent social ties sustain cooperation is not new,
work on trust in the context of understanding cooperation has significantly multiplied
since. Trust as a basis for social order is an established theme in sociology, with Emile
Durkheim (2013 [1893]) and Georg Simmel (1990 [1900], 1950) the early pioneers,
continued by Niklas Luhmann (1979 [1968]; for surveys of the classical sociological
work, see Mistzal 1996 and Möllering 2006). Individual variation in trusting disposi-
tions and its effects was an early theme of psychology (e.g. Rotter 1967, 1971, 1980).
But recent work has significantly expanded beyond these disciplines. Perhaps the most
impressive fruit is the Russell Sage Foundation series on trust, with sixteen volumes in
print, including contributions from both theoretical and empirical perspectives. One
area of work that has focused especially heavily on trust is Organizational and
Management Studies (see Kramer and Tyler 1996; Bachmann and Zaheer 2006, 2008).
Accessible general introductions to the range of work on trust include Kohn (2008)
and Hawley (2014); Karen Cook’s excellent annotated bibliography fulfils the same
function (2011). One of the few philosophers to be directly inspired by this line of
work is Martin Hollis (1998). A first source of interest in the topic of trust, then, are the
fundamental questions around the conditions for the possibility of social order, and
the roles of morality, sympathy, and rationality in sustaining it. These questions are
both philosophical and empirical. It is clear that trust has a more significant role to play
in philosophical attempts at answering them than has been appreciated hitherto.
2. Trust and Knowledge

Trust enables knowledge via testimony. This is a second source of interest.
C. A. J. Coady’s Testimony: A Philosophical Study (1992) sparked current philosophical
interest in the epistemology of testimony. Taking inspiration from Thomas Reid, he
defended anti-reductionism about testimony, the conjunctive view that the justification
for believing testimonial reports is sui generis in nature, akin to other fundamental
sources of belief such as sense perception or memory; and that hearers have a default
entitlement to believe what they are told. This is in opposition to David Hume’s
reductionist view, which denies both conjuncts: the justification for testimony-based
belief is inductive, based on past observed correlation between reports that p and it
being the case that p or ~p; evidence of such observed correlation is therefore
required for justified belief. The basic strategy for classical anti-reductionism is
made explicit by the title of the earlier of Reid’s two discussions: ‘Of the analogy
between perception, and the credit we give to human testimony’ (Inquiry into the
Human Mind, 6.24. Formulated this way, reductionism and anti-reductionism are
contraries, not contradictories).
There is something right and something wrong about Reid’s anti-reductionism.
‘She told me so’ is very often a satisfactory defence for why someone believes that p.
The default entitlement to believe a report defended by anti-reductionists explains
the success of that defence nicely. (A qualified reductionism, which endorses only the
inductive justification of testimony-based belief, can endorse the default entitlement;
e.g. Adler 2002.) In contrast, and while not certainly false, many epistemologists have
nonetheless been unpersuaded by the claimed analogy between testimony and sense
perception. As a source of knowledge, testimony constitutively relies on the will of
other people, namely testifiers, in a way that sense perception does not. This makes
hearers vulnerable to deception. Moreover, there is no shortage of reasons to deceive.
So the analogy breaks down at a crucial point. The purely epistemic resources of clas-
sical anti-reductionism are ill equipped to explain why speakers choose—as a matter
of practical reason—to testify truthfully. As Bernard Williams wryly remarked of a
related idea, the broadly transcendental argument Coady offers ‘does not do much to
help us understand how these linguistic practices [of assertion] survive the level of
abuse which year in and year out they regularly receive’ (2002: 86).
An alternative proposal justifies the acceptance of testimony on the basis of the
commitment that the testifier undertakes in asserting p. J. L. Austin provides an early
elaboration of the parallel between testifying to one’s knowledge that p and promising.
In both, I have ‘bound myself to others. . . . When I say “I know,” I give others my word:
I give others my authority for saying that “S is P” ’ (1961 [1946]: 368). More recently,
Richard Moran has made the parallel central to his ‘Assurance view’ of testimony, on
which giving testimony—telling someone something—is ‘an overt assumption of spe-
cific responsibilities on the part of the speaker. This is no more (or less) mysterious
than how an explicit agreement or contract alters one’s responsibilities’ (2006: 288–9).
Moran’s Assurance view is an exemplar of a family of related views (see also Ross 1986;
Elgin 2002; Hinchman 2005). These are clearly not reductionist. But they differ from
classical anti-reductionism in basing the epistemic significance of testimony on practical
attitudes that ground and constitute interpersonal relationships. A principal attraction
of assurance accounts is that they make sense of the moral and interpersonal phenom-
enology of testimony, of how it is that the ground for belief is the person testifying and
not merely what they have asserted. The phenomenology is well illustrated by contrast
with how it feels to be disbelieved. As Elizabeth Anscombe remarks, ‘It is an insult and
introduction 5
it may be an injury not to be believed’ (1979: 150). In believing someone, I show my

respect for them.
Trust is then arguably required as part of an account of how testimony can function
to assure. This is so just if—as many find attractive—in felicitous cases of promissory
exchange, the appropriate response to a promise made is then trust that the promise
will be kept, on the basis of trust in the promise-maker. Insofar as, in the normal case,
felicitous testimony relies on the same central social fact, of commitment, so too the
appropriate response to testimony given is also trust, manifest in believing a speaker.
One account of how this can be rational—of how trust can ground reasonable testimo-
nial uptake—is offered by Paul Faulkner (2011). He proposes that A affectively trusts S
if and only if A depends on S φ-ing, and expects his dependence on S to motivate S to
φ—for A’s dependence on S to be the reason for which S φs (2011: 143–50). As a result,
affective trust is a bootstrapping attitude: I can choose to trust someone affectively and
my doing so creates the reasons which justify the attitude. In believing that you recog-
nize my dependence on you φ-ing and presuming that you will take my dependence on
your φ-ing as a reason to φ, so I come to presume that you will φ, which in the case of
testimony is tell me the truth because I need it. Regardless of the specifics of the
account, given the connections between assertion, commitment, and trust, the last
must arguably play some role in explaining how we believe speakers and so how we get
to know what speakers know. This is unsurprising given that the interrelation—even
entanglement—of practical and epistemic reasons found in testimony is found also in
the attitude of trust.
3. Trust and Social Philosophy

A third source of interest in trust derives from the disparate areas of philosophical
enquiry that are concerned with relations between people. These are predominantly in
the domain of practical philosophy. Onora O’Neill has argued prominently for the
importance of trusting relations as a valuable state of affairs that should form a basis
and goal of public affairs. In bioethics, she proposes trust as an ideal for patient–physician
relations, in contrast to that of promoting individual autonomy, and argues that efforts
to achieve the latter have undermined the former (2002a). Trust should not be traded
off in this way. Her concern that unreflective pursuit of other ideals may have a similar
undermining effect is also evident in her Reith Lectures on trust and public life, where
she surveys the loss of trust in professions and institutions in the UK (2002b). The
paradox is that regimes of accountability that aim at increasing trustworthiness have
the effect of undermining trust. O’Neill’s work well illustrates a general question,
which likely permits only a series of piecemeal answers: for any area of applied ethics,
to what extent is trust a valuable ideal or goal? A smorgasbord of contemporary areas
where the question iterates includes banking; policing and the law; politics and politi-
cians; technologies such as the Internet, robotics, genetic modification, and artificial
intelligence; international relations; government and corporate surveillance; and

other undeclared security-related operations.
Many of these are specific, modern instances of established concerns from early
modern political philosophy, on which trust is a condition for and consequence of
legitimate government. The absoluteness of Hobbes’ sovereign invited the Lockean
question: why obey it? Hobbes’ proposal is subtler than the sovereign’s bald threat of
massive coercive violence. The threat is authorized by the people as a device for escap-
ing the state of nature, by which a plurality of wills are united to ‘a real Unity’ (Leviathan,
§17). The subject should obey the sovereign because, in being so authorized, the sover-
eign’s will is her own. Locke’s dissatisfaction with this answer and his alternative have
been the more influential. Whereas Hobbes merely predicts, as a descriptive matter,
that the sovereign will act in the interests of its subjects, for Locke this is a normative
requirement. The legislative should design laws for the good of the people. It holds
power in ‘fiduciary trust’. Success in protecting citizens’ natural rights, especially pres-
ervation of property, is thus a condition of legitimate government. So fulfilment of
trust is a political requirement for legitimacy; it is a consequence that governments
should have the powers needed to exercise that trust. Locke’s account of the relation-
ship between the governed and government as one of trust makes clear the basic own-
ership and purpose of power—it is exercised on behalf of those governed. A different
part of the liberal tradition focuses on the appropriate attitude of the governed towards
those in power. The frailties of human nature, the risks of abuse of power, and the
temptations and opportunities to do so mean that the default attitude should be dis-
trust (e.g. Hardin 2002). How these claims are to be reconciled, or indeed their sound-
ness, is of course a further matter.
Trust is politically significant not only in relation to legitimacy. It is significant also
for stability. The object of Rawls’ concern for stability was conceptions of justice.
Specifically, a conception of justice must generate its own support among those who
live in a society whose basic structure is so marked, so that they desire to live in accord-
ance with its principles (Theory of Justice, §24). The condition is a demanding one, rul-
ing out utopian conceptions of justice that cannot command support from people as
they actually are. A consequence of the demand is that, insofar as cooperation outside
the shadow of violence is sustained by trust, so those factors that undermine or sup-
port trust empirically become normatively significant. This is one of the foci in the
debate on the ethics of immigration (Miller 1995; Pevnick 2009). But the issue reiter-
ates more widely. Both conservative and socialist critics of neo-liberal capitalism have
argued that the dominance of the market as means of exchange undermines the condi-
tions that make market exchange possible in the first place. It does so by weakening or
eliminating the sympathetic bonds of trust fostered by non-transactional interaction
and which undergird the market. (Bowles 2011 provides critical discussion of the
objection; Hirsch 1977 is an example.)
Trust is also significant in non-practical areas of philosophy. As examples, consider
the philosophy of religion and the philosophy of action. Two central and related
introduction 7
c oncerns in the philosophy of religion are evaluating the justificatory status of belief
in God and identifying the attitude of faith. It is plausible that trust answers both.
Belief in God may be based on acceptance of testimony. The relation between believer
and God, at least for the first, is characteristically one of trust. (Discussion of the for-
mer possibility is young, but see e.g. Lamont 2009; Dougherty 2014; Simpson 2014;
and on the latter, Swinburne 2005; Godfrey 2012; Kvanvig 2016). In the philosophy of
action, trust is plausibly a constitutive requirement on shared agency. Further, a
precondition by which group agents are characteristically formed may likewise be
trust. Moreover, this survey of areas of philosophy where trust is important is illustra-
tive, not exhaustive.
4. Chapter Summary
Victoria McGeer and Philip Pettit begin the papers in Chapter 2 with ‘The Empowering
Theory of Trust’. It is commonly held that both trust and reliance involve dependence.
Trust differs from reliance, according to McGeer and Pettit, because it involves the
expectation that the trusted will see a trusting party’s dependence as a reason to do
what the trusting party expects. Others in this volume also emphasize that trust
involves this expectation, particularly Darwall, Jones, Faulkner, and Stern. McGeer
and Pettit argue that by displaying the expectation that a trusted party will be sensitive
to their reliance, a trusting party can increase that very sensitivity in two distinct ways.
They can make it the case in a given context that the trusted party is more reliably sen-
sitive to their dependence: that is, disposed to respond appropriately under more pos-
sible variations on the situation. And even if that trusted party’s sensitivity is already
totally reliable, they can make it the case that it is more resiliently in place: it is less
subject to being inhibited or undermined by disrupting influences. That is to say, trust
can empower the trusted. In a variety of contexts and relationships, trusting someone
can thereby enable them to live up to the expectations we invest in them. It can elicit
the very trait that makes it sensible for us to give them our trust.
The idea that trust involves some expectation that the trusted will respond positively
to one’s reliance then suggests to Stephen Darwall in Chapter 3 that trust is a second-
personal attitude. However, paradigm second-personal attitudes are deontic: they pre-
suppose authority and accountability relations. Consider how one might blame or
resent someone who broke a promise: the promise relation gives the promisee the
authority to hold the promisor accountable for doing what they promised to do; and it
is this authority and the associated accountability relation that is presupposed by the
attitudes of blame and resentment that the promisee would be susceptible to were the
promisor not to do this. By contrast, trusting someone to do something involves laying
oneself open. One doesn’t have any authority over the trusted and cannot hold them
accountable for doing what one trusts them to do. So trust is non-deontic. But it is
second-personal in that it is similarly reciprocating: it invites the other to take the same
view of the situation. This is not an invitation to recognize authority, but an invitation
to recognize the trust and respond in kind, trustworthily. So trust, Darwall argues, is
comparable to love: it is a second-personal ‘attitude of the heart’.
Edward Hinchman also takes up the idea that trust involves some expectation that
the trusted will respond positively to one’s dependence in Chapter 4. Trust, he sug-
gests, involves a Gricean form of mutual recognition: in inviting trust, the trusted can
give the trusting a reason for action or belief grounded, in part, in the trusting’s recog-
nition of the trusted’s intention to give this reason. What Hinchman emphasizes are
two things: that the trust relation is thereby, at heart, a rational relation; and that trust
can thereby both be betrayed without being disappointed and be disappointed without
being betrayed. These possibilities reflect how the reasons made available through a
trust relation are grounded not merely in the trusted’s reliability but also in the trusted’s
responsiveness to relevant need.
The betrayal of trust is also central to Collin O’Neil’s paper in Chapter 5, where he
gives an analysis of betrayal and the form of trust that can be betrayed. On O’Neil’s
account, a betrayal of trust should be understood as the violation of a certain trust-
based obligation, just as a betrayal of loyalty is naturally understood as the violation of
a certain relationship-based obligation. Trust itself can give the trusted an obligation
to do what they are trusted to do because it is a form of honour, and here O’Neil
echoes McGeer and Pettit. Once there is such an obligation there can be a failure to
act on it; so trust can be betrayed, where this then constitutes a unique wronging of the
trusted party.
Supposing trust were betrayed, one form of complaint would be ‘But I was
counting on you!’ The aim of Karen Jones’ paper in Chapter 6 is to understand the
normative status of this complaint. Jones offers a ‘counting on’ theory of trust, so
the complaint she investigates might also be voiced as ‘But I was trusting you!’
According to Jones’ theory, the constitutive expectation of trust is that the trusted
will respond positively to the fact that the trusting party is counting on them—or
depending on them in some way. In expecting a trusted party to act for this reason,
trust is a way of ‘extending our agency’, which is to say achieving the goods of
cooperative agency. It follows that the normative force of the complaint is not
moral—we can cooperate in doing immoral things—but that the norms specific to
trust have been flouted. Our commitment to these norms is then grounded in our
interest in living cooperative social lives.
It is the problem of the rationality of cooperation that is at the centre of Paul
Faulkner’s chapter. Cooperation threatens to become rationally problematic insofar as
the following conditions hold: reliance has a worst outcome—we rely and the other
proves unreliable; the interaction is one-off; and we are ignorant of the other’s particu-
lar motivations but recognize a general motivation to be unreliable. The problem,
Faulkner argues in Chapter 7, is that the satisfaction of these conditions is common-
place. Thus cooperation should be much less common than it in fact is. So what
explains it? Faulkner considers and rejects various game-theoretical solutions before
introduction 9
canvassing a ‘trust-based’ solution. According to this solution the problem is dissolved

once one recognizes how trust itself can give reasons for cooperating.
While a simple trust game—the Trust game—is central to Faulkner’s chapter,
another—the Hi-Lo game—is central to Bernd Lahno’s. In Chapter 8 we are to imagine
two canoeists being swept towards a rock: it is imperative that they both steer either left
or right, with the right course looking marginally safer. For Lahno this coordination
problem illustrates the paucity of individual game-theoretical reasoning since each
will only have a conditional preference: to steer right if the other does, and otherwise to
steer left. And on this basis it is impossible to explain coordination. This theoretical
problem is easily solved if teams are added as possible agents and the canoeists are
taken to be a team since the team will unconditionally favour the right path. However,
making sense of team reasoning, Lahno argues, presupposes that trusting relations
structure the team, which in turn presupposes a background of trust. Thus it turns out
that the resolution of a basic coordination problem requires a background of trust. So
Lahno and Faulkner follow parallel argumentative trajectories to similar conclusions.
The background attitude of trust that Faulkner takes to explain cooperation is two-
place (‘A trusts B’) rather than three-place (‘A trusts B to φ’). It is the two-place form,
Jacapo Domenicucci and Richard Holton argue in Chapter 9, that is the basic form of
trust; and in arguing this they break with the tradition, which concentrates exclusively
on its three-place form. In this respect, trust is like love or friendship; one would not,
they observe, understand Antony’s love for Cleopatra in terms of the three-place
‘Antony loves Cleopatra for her ϕ’. Ditto trust. The two-place relation—our trusting
someone—ordinarily grounds instances of relying on them in some way, so it ordinar-
ily grounds three-place trust. But there can be trust when there is no opportunity of
reliance. Trust is thus essentially a readiness to hand over control or power to another,
which is associated with the reactive attitudes.
Handing over control, like handing over a set of car keys, is, one would think, some-
thing you can decide to do. But that one can decide to trust is something Benjamin
McMyler takes issue with in Chapter 10. Action can be willed—one can decide to act—
just because one can act for any reason, be it good or bad. So one can decide to act in a
trusting way—to act as if one trusted. But one cannot decide to trust in the way that one
can decide to act because one can only trust for certain kinds of reason. In this respect,
McMyler argues, trust is like belief. One cannot believe that p for any reason. A bribe
and a threat can be reasons for acting, so each can be a reason for pretending one
believes, but neither can be a reason for belief. Only certain things—paradigmatically
evidence—can be a reason for belief, and for this reason one cannot decide to believe.
And the same is true, McMyler claims, for trust. One can trust only if one has reasons
for thinking that the trusted is worthy of trust. So one cannot trust at will.
McMyler doesn’t commit an answer to the question, what is a reason for thinking
the trusted is worthy of trust? With a significant qualification, Thomas Simpson does
in Chapter 11. Thus he argues for the following ‘evidentialist constraint’ on trust: it is
rational for A to trust B to φ only if, on A’s total evidence, it is likely that B will φ. For
instance, on an expedition to the Antarctic whose success required a supply drop, it

would be rational to trust the supplier only if one’s total evidence made it likely they
would successfully execute the drop. The significant qualifier is that while trust in this
case is cognitive—amounting to the belief that the supplier is trustworthy—not all
instances of trust are cognitive. ‘Trust’ is equivocal, and there are, Simpson credits,
richer notions of trust like those in play in the previously considered chapters. On
those occasions when the evidential constraint then applies, agents can or should
refuse moral demands to trust which are unsupported by the evidence.
Whether such richer notions should be recognized is a debate opened by the next
two chapters. A seeming advantage of aligning trust with the belief that the trusted is
trustworthy—or at least will prove so on this occasion—is that trust can thereby easily
explain cooperation. That trust should be able to do this, Philip Nickel proposes as a
methodological constraint on theorizing about trust in Chapter 12. The satisfaction
of this constraint can then be used to differentiate theories of trust with the better
theories being able to explain cooperation more generally and effectively. Reasoning
thus, Nickel argues for the superiority of what he terms unrestricted views of trust,
which take trust to be no more than the disposition to rely on others, over restrictive
views, which require the trusting person to have some further attitude in addition to
this disposition. The same criterion also favours some restrictive views over others.
David Owens goes one step further than Nickel in Chapter 13: it is not merely that
restrictive views of trust are to be rejected for methodological reasons; rather they
should be rejected because there is no unique attitude of trust. What gets counted as
trusting depends on the nature of the object of trust. Owens illustrates this by reference
to trust in a promise. Conceive promises as a vehicle of social coordination and acts of
reliance count as trusting. (As Nickel observes, it is the explanation of such acts that
matters.) But conceive promises as a vehicle for cultivating normative relationships and
acts of acceptance can be sufficient and so can count as trusting, and actual reliance is
not necessary. What counts as trusting a promise, Owens argues, hinges on what one
takes the distinctive value of a promise to be, and a ‘trusting’ action or attitude is one
that realizes this distinctive value. Thus trusting someone’s promises, their assertions,
their judgement, and so on may well involve quite different actions or attitudes.
We can expect others to be trustworthy and value their being so. This is true even
if there is no single thing referred to by ‘trust’ and ‘trustworthiness’. But what about
groups? Can groups be trustworthy in the way that individuals can? Implicit in
Lahno’s talk of team reasoning is the idea that a team can be trustworthy. Katherine
Hawley takes issue with this in Chapter 14. Her focus is group testimony. It is possible to
distinguish mere reliability from trustworthiness with respect to individual testifiers—
a liar could, despite their intentions, be reliable. And it is equally possible to make
this distinction at the group level. The question Hawley pursues is then whether
we should make this distinction at the level of group testimony. Hawley proposes
not: when we consider group testimony, all that matters is a group’s reliability. She
then considers, and counters, several arguments, both epistemological and ethical,
to this view.
introduction 11
The last two chapters take up philosophical issues of trust from a historical perspec-
tive, investigating what Kant and Løgstrup had to say about trust. Can it ever be
reasonable to trust someone without evidence of their trustworthiness? In Chapter 15,
Guy Longworth defends a positive answer to this question based on Kant’s idea that
we can have practical reasons for holding things to be true. So we can have practical
reasons for holding true that someone is trustworthy, and so for trusting them. The
obvious worry here is that rationally holding something to be true would seem to
require rationally believing this thing to be true. However, Longworth rejects this
requirement on the basis of an account of deciding to do something. In deciding to φ,
one holds true that one will φ but the rationality of this is not that of belief: it is one’s
will that constrains one’s holding true and not the evidence. So it is the practical ration-
ality of operating under this self-imposed constraint that determines the rationality of
holding true that one will φ. Thus the way is clear to allowing that one can trust and be
reasonable solely on the basis of practical reasons. In drawing this conclusion
Longworth takes issue with McMyler’s claim that trust cannot be voluntary.
In Chapter 16, Robert Stern investigates Løgstrup’s claim that trust is basic. There are,
Stern suggests, four plausible interpretations of this: (1) a psychological interpretation
(children start out with a trusting attitude and learn to distrust); (2) a transcendental
interpretation (trust is a necessary condition of language and society, so must be the
default attitude); (3) an axiological interpretation (trust is an intrinsic good while
distrust is a privation); (4) an ontological interpretation (trust is essential to the proper
functioning of human life itself, and as such cannot be a norm or practice we have
instituted for ourselves). All four claims can be found in Løgstrup but it is the last two,
Stern claims, which best represent his view. On this interpretation, trust is to distrust
as health is to sickness. Being healthy is a basic human good, one that needs no justifi-
cation; so too trust. As a human good it is realized in and by an openness to others. By
contrast, to distrust someone is to conceive of them in a way that limits them and closes
one’s relation to them. In this respect, trust is comparable to love and sympathy and is
what Løgstrup calls a ‘sovereign expression of life’.1
References
Adler, Jonathan E. 2002. Belief ’s Own Ethics. Cambridge, MA: MIT Press.
Anscombe, Elizabeth. 1979. What Is It to Believe Someone? In Rationality and Religious Belief,
ed. C. F. D. Delaney, pp. 141–51. Notre Dame, IN: University of Notre Dame Press.
Austin, J. L. 1961 [1946]. Other Minds. In his Philosophical Papers, 44–84. Oxford: Clarendon Press.
Bachmann, Reinhard and Akbar Zaheer (eds.). 2006. Handbook of Trust Research. Cheltenham:
Edward Elgar.
Bachmann, Reinhard, and Akbar Zaheer (eds.). 2008. Landmark Papers on Trust, 2 vols.
Cheltenham: Edward Elgar.
Baier, Annette C. 1986. Trust and Antitrust. Ethics 96: 231–60.
1
We are grateful to the Mind Association, for a grant which helped to support the workshop at which a
number of these essays were presented while in early draft; and to Peter Momtchiloff, OUP, for his support
for the volume
Bowles, Samuel. 2011. Is Liberal Society a Parasite on Tradition? Philosophy and Public Affairs
39(1): 46–81.
Coady, C. A. J. 1992. Testimony: A Philosophical Study. Oxford: Clarendon Press.
Cook, Karen. 2011. Trust. Oxford Bibliographies. Available at <http://www.oxfordbibliographies.
com/view/document/obo-9780199756384/obo-9780199756384-0062.xml> accessed 8 Sept
2015.
Dougherty, Trent. 2014. Faith, Trust, and Testimony: An Evidentialist Account. In Religious
Faith and Intellectual Virtue, ed. Laura Frances Callahan and Timothy O’Connor, pp. 97–123.
Oxford: Oxford University Press.
Durkheim, Emile. 2013 [1893]. The Division of Labour in Society, 2nd edn., ed. Steven Lukes.
Basingstoke: Palgrave Macmillan.
Elgin, Catherine Z. 2002. Take It From Me: The Epistemological Status of Testimony. Philosophy
and Phenomenological Research 65(2): 291–308.
Faulkner, Paul. 2011. Knowledge on Trust. Oxford: Oxford University Press.
Fukuyama, Francis. 1995. Trust: The Social Virtues and the Creation of Prosperity. London:
Hamish Hamilton.
Gambetta, Diego (ed.). 1988. Trust: Making and Breaking Cooperative Relations. Oxford: Basil
Blackwell.
Godfrey, Joseph J. 2012. Trust of People, Words and God: A Route for Philosophy of Religion.
Notre Dame, IN: University of Notre Dame Press.
Hardin, Russell. 2002. Liberal Distrust. European Review 10(1): 73–89.
Hawley, Katherine. 2014. Trust: A Very Short Introduction. Oxford: Oxford University Press.
Hinchman, Edward S. 2005. Telling as Inviting to Trust. Philosophy and Phenomenological
Research 70(3): 562–87.
Hirsch, Fred. 1977. The Social Limits to Growth. London: Routledge & Kegan Paul.
Hollis, Martin. 1998. Trust within Reason. Cambridge: Cambridge University Press.
Kohn, Marek. 2008. Trust: Self-Interest and the Common Good. Oxford: Oxford University
Press.
Kramer, Roderick M. and Tom R. Tyler (eds.). 1996. Trust in Organizations: Frontiers of Theory
and Research. Thousand Oaks, CA: Sage.
Kvanvig, Jonathan. 2016. The Idea of Faith as Trust: Lessons in Noncognitivist Approaches to
Faith. In Reason and Faith: Themes from Richard Swinburne, ed. Michael Bergmann and
Jeffrey Brower, pp. 4–25. Oxford: Oxford University Press.
Lamont, John. 2009. A Conception of Faith in the Greek Fathers. In Analytic Theology: New
Essays in the Philosophy of Theology, ed. Oliver D. Crisp and Michael C. Rea, pp. 87–106.
Luhmann, Niklas. 1979 [1968]. Trust and Power. Trans. Howard Davis, John Raffan, and
Kathryn Rooney. Chichester: John Wiley.
Miller, David. 1995. On Nationality. Oxford: Oxford University Press.
Mistzal, Barbara A. 1996. Trust in Modern Societies. Cambridge: Polity.
Möllering, Guido. 2006. Trust: Reason, Routine, Reflexivity. Oxford: Elsevier.
Moran, Richard. 2006. Getting Told and Being Believed. In The Epistemology of Testimony, ed.
Jennifer Lackey and Ernest Sosa, pp. 272–306. Oxford: Oxford University Press.
O’Neill, Onora. 2002a. Autonomy and Trust in Bioethics. Cambridge: Cambridge University Press.
O’Neill, Onora. 2002b. A Question of Trust. Cambridge: Cambridge University Press.
introduction 13
Pevnick, Ryan. 2009. Social Trust and the Ethics of Immigration Policy. Journal of Political
Philosophy 17(2): 146–67.
Putnam, Robert D. 1995. Bowling Alone: America’s Declining Social Capital. Journal of
Democracy 6(1): 65–78.
Putnam, Robert D. 2000. Bowling Alone: The Collapse and Revival of American Community.
New York, NY: Simon Schuster.
Ross, Angus. 1986. Why Do We Believe What We Are Told? Ratio 28: 69–88.
Rotter, Julian B. 1967. A New Scale for the Measurement of Interpersonal Trust. Journal of
Personality 35(4): 651–65.
Rotter, Julian B. 1971. Generalized Expectancies for Interpersonal Trust. American Psychologist
26(5): 443–52.
Rotter, Julian B. 1980. Interpersonal Trust, Trustworthiness, and Gullibility. American
Psychologist 35(1): 1–7.
Simmel, Georg. 1950. The Sociology of Georg Simmel, ed. Kurt H. Wolff. New York, NY: Free
Press.
Simmel, Georg. 1990 [1900]. The Philosophy of Money, 2nd edn. Trans. Tom Bottomore and
David Frisby. London: Routledge.
Simpson, Thomas W. 2014. Testimony in John’s Gospel: The Puzzle of 5:31 and 8:14. Tyndale
Bulletin 65(2): 201–18.
Swinburne, Richard. 2005. Faith and Reason, 2nd edn. Oxford: Oxford University Press.
Williams, Bernard. 2002. Truth and Truthfulness: An Essay in Genealogy. Princeton, NJ:
Princeton University Press.
2
The Empowering Theory of Trust
Victoria McGeer and Philip Pettit
We have argued independently in earlier papers that if I trust you in a certain domain,
then that may help to improve your capacity to prove reliable and deserving of
trust; it may have a positive, empowering impact on your psychology (Pettit 1995;
McGeer 2008b). It is a matter of common assumption that any capacity may come in
degrees or levels; more on this later. But how can my relying on you in a trusting man-
ner raise the degree to which you are reliable and deserving of trust? We sketch a
defence of the possibility in this chapter and elaborate on some of its implications.
The chapter itself is in three sections. The first provides a background account of
what it means for me to rely on you and trust you, and what it means for you to have the
capacity to respond: to be trust-responsive. The second explores the case for the theory
that my trusting you can increase your trust-responsiveness. The third looks at some
implications of the fact that your trust-responsiveness is sensitive in that way to my
trusting you.
1. Trust and Reliance

The most basic idea is that of reliance. If I am to count as relying on you, in our usage,
then I must assume that you have a disposition to act in a certain manner and I must act
on that assumption. Thus I may assume that you are disposed to put out your garbage
for collection on the appropriate day and treat your putting it out as a signal that today is
the day. Or I may assume that you have a disposition to check the weather report before
you go out and treat your taking an umbrella as a prompt to prepare for rain. Or I may
assume that you have a disposition to recognize what is best in the long-term for your
children and be ready to follow the same line in raising my own. In any such case I dis-
play an attitude of reliance insofar as I assume the presence of a disposition in you that I
treat as a potential guide for my own behaviour. And I perform an act of reliance in let-
ting that assumption dictate how I do indeed behave on this or that occasion.1
1
In all of these cases you act in a way appropriate to the purpose on hand: to getting the garbage col-
lected, or to guarding against rain, or to raising your children. But I might rely on you in a certain sense so
the empowering theory of trust 15
If this is what relying on you involves, what extra is needed for it to be the case that I
trust you? In ordinary usage, the word ‘trust’ does not always come apart from the
word ‘rely’. But in the terminology we use in this chapter, and in keeping with a well-
established philosophical tradition, it does come apart (Baier 1986; Holton 1994;
Jones 1996). ‘Trust’ in that tradition of thinking is a term of art.
In this usage, I will count as trusting you to do something X just insofar as three
conditions are met. First, I manifestly rely on you to do X: I make clear to you my
assumption that you will prove reliable in doing X.2 Second, I assume that the manifest
fact of my reliance will weigh with you as a reason for choosing voluntarily to X. And
third, this assumption helps to explain or reinforce my relying on you.
What does it mean to say that the manifest fact of my reliance weighs with you as a
reason for choosing voluntarily to X: that is, to do X without being forced to do it? That
fact must provide you with a reason, by your view of things, for proving reliable,
whether by making X-ing as such attractive or by making it attractive as a means to
some other end. In particular, it must provide you with a reason to X without making
X-ing into a more or less involuntary choice: that is, without making not X-ing into an
intuitively unacceptable alternative, as it would do if it meant, for example, that you
would be punished for not X-ing (Olsaretti 2004). The reason for X-ing which
the manifest fact of my reliance on you provides may actually motivate you; the
motivation may be sufficient to get you to X or a necessary part of a condition sufficient
to get you to X. Or, to take a further possibility, it may reinforce some independently
sufficient motive you may have for X-ing: it will be there, if needed, to provide back-up
assistance.
As reliance involves both an attitude and an action so too, on this account, does
trust. I will display an attitude of trust just insofar as I assume that the manifest fact that
I am relying on you will weigh with you as a reason to prove reliable; henceforth we
take the requirement of voluntariness to be understood. And I will exercise an act of
trust just insofar as I act on that assumption or belief.
On this usage, I may often display or exercise reliance without displaying or exercis-
ing trust. I regularly rely without trust on inanimate objects and non-human animals,
as when I act as if the bridge I am crossing will not collapse, as if the weather tomorrow
will be fine, or as if the dog will not soil the carpet. And I may even rely on people in a
way that does not involve trust. When I go on a political demonstration, I rely on o thers
to join me but I do not display or exercise trust. I do not expect that they will have been
aware of my intention to participate—it may not have been manifest to them—let
far as you act in some regular way, even perhaps inappropriately by my lights to the situation at hand.
Believing you always put out the garbage a day too early, I might rely on your doing so and put out my
garbage the day after. In the first sort of case, the reliance is straight, as we may say, in the second bent.
When we speak of reliance in the text, it is always straight reliance we have in mind.
2
This condition may be weakened to agree with a developmental view of trust, according to which I
may sensibly manifest reliance without having much confidence that you will actually prove reliable. In
that sort of case, mentioned at the end of this section, I manifest the belief in your reliability with the aim
of enhancing your capacity to prove reliable over time.
16 Victoria McGeer and Philip Pettit
alone aware of my relying on them to take part as well. And when I rely on some rough-
necks not to assault me in a public place, this reliance will not typically involve trust
either. While I may take them to recognize that I am relying on them in this way, I am
unlikely to think that that will weigh with them as a reason to prove reliable. I will
probably think that the only consideration that moves them is the fear that an assault
would prompt third parties to come to my aid.
Suppose, by contrast with such cases, that I have been sitting near you in a restaurant
and ask you to keep an eye on my computer when I go to the toilet. I am manifestly
relying on you not to make off with the computer, and not to let anyone else do so
either. And to the extent that I rely on you in this way, taking you to be significantly
responsive to that reliance—taking you to have the capacity to be moved by my relying
on you—I count as putting my trust in you. I have and reveal an attitude of trust in
putting myself in your hands—this may be based on how you look or on how you
seemed in an earlier exchange of pleasantries—and in acting on that attitude I exercise
trust: I perform an act of trust.
In offering this account of trust, we take your responsiveness to my manifest
reliance—your trust-responsiveness, for short—as a capacity: specifically, as the capacity
to be moved voluntarily to action by the manifest fact of my reliance. You will have the
capacity involved just insofar as you satisfy an important modal condition. For an
open range of variations on the actual situation that keep in place the fact that I am
manifestly relying on you, you would be moved to action in most of them; or at least
you would be moved to action when those variations did not introduce countervailing
reasons, impeding passions, or anything of a recognizably disruptive kind.3 In short,
the capacity consists in a disposition to be robustly, if not inevitably, moved to action
by the fact that I am manifestly relying on you.4
By all accounts a capacity like trust-responsiveness may come in degrees and this
will be important in later discussion. Taking the capacity as a disposition of the kind
described, it turns out that there are two ways in which trust-responsiveness may vary
in degree; it may vary in dependability, as we put it, or in durability. The capacity will be
present with a higher rather than a lower degree of dependability to the extent that the
disposition is relatively sensitive to appropriate triggers: it is fit to be activated over a
greater rather than a lesser number of scenarios in which the relevant reasons for trust-
responsiveness are in place. By contrast, the capacity will be present with a higher
rather than a lower degree of durability to the extent that it is relatively resistant to
extraneous features that inhibit its manifestation: it is fit to display itself over a greater
rather than a lesser number of scenarios where potential disrupters are present (Pettit
2015, ch. 2).
3
The idea that you would be moved to action in most of the scenarios is not essential; it might be
replaced, for example, by the idea that you would have a significant chance of being moved to action in all
of them. See Pettit (2015, Appendix II).
4
For a characterization of capacity on these lines see Smith (2003) and McGeer and Pettit (2015).
To illustrate the distinction between dependability and durability in our usage,

consider the mechanical disposition of a sophisticated thermostat designed to maintain
ambient room temperature at a set degree. The system will be more or less dependably
disposed to do this job just to the extent that it can cope with various, limited increases
in cold or heat: they trigger it, as appropriate, to turn on the central heating or the air
conditioner. But, however dependable, the system will not be very durably disposed
to do its job if the power supply is erratic. And its durability will be increased,
independently of any impact on its dependability, if the supply becomes less erratic or
a back-up source of energy is put in place.
The same distinction between durability and dependability will apply with any
human disposition or capacity that we ascribe, including that of trust-responsiveness,
insofar as the following conditions obtain. First, the disposition is interpreted as a
disposition to respond dependably to a certain range of triggering scenarios: this, in
the way the heating/cooling system is interpreted as a mechanism for restoring the
pre-set temperature in response to a limited range of increases in cold or heat. Second,
there are some scenarios where the disposition fails to generate the expected response,
by analogy with the scenarios where the heating system does not work. Third, the
intuitively best explanation of this failure is that in those scenarios there are factors
present that disrupt or mask the operation of the disposition, as the failure of the
electric supply disrupts the heating system.
There are many cases where such conditions are satisfied, allowing us to distinguish
quite naturally between the dependability and durability of various ordinary human
capacities. We may be happy to ascribe an outstanding mathematical capacity to you,
treating it as highly dependable—you can solve a great range of tricky problems—
while recognizing that the ability is disrupted or masked by inhibition in the company
of hostile or assertive competitors. We may be happy to ascribe a highly dependable
capacity to play very difficult pieces on the piano—a capacity demonstrated time and
time again—while admitting that, unfortunately, it is disrupted or masked by stage
fright in the presence of a large audience. And we may be confident in treating you as a
dependably kind person—this shows up in how you routinely treat your family and
friends, colleagues, and strangers—while acknowledging that this capacity is dis-
rupted or masked when you imbibe a few too many alcoholic drinks. By analogy with
these cases, we hold that the trust-responsiveness that I ascribe in investing my trust in
you may be suitably dependable—it may mean that you score quite highly among your
peers—without always being very durable.5
5
While this is the possibility of particular interest in this chapter, it is worth noting more generally that
there is a sense in which dependability and durability may vary independently of each other, so that the
disposition or capacity may not only be dependable without being durable, for example, but also durable
without being dependable. If we map dependability on the horizontal axis of a graph, and durability on
the vertical, it is possible for a disposition to be situated at any set of coordinates, disappearing from view
only at 0,0.
Consider, for instance, the scenario that Plato makes vivid in the Republic, invoking
the ring of Gyges. Though it may be true that your trust-responsiveness is highly
dependable, it is reasonable to ask whether you would retain that disposition—or any
other of a range of virtues—if you were provided with a ring that made your actions
undetectable and that put you effectively beyond rebuke or penalty. Assuming you
think that you probably would not retain the disposition in that scenario—that you
would probably be corrupted by the ring—should you conclude that you are not as
dependably trust-responsive or kind or honest now as you thought you were? Not if, as
seems reasonable, you think that the dependability of those dispositions is measured
only over scenarios where no magical intervention occurs and you live in the presence,
and under the potential regard, of others. Not, in other words, if you interpret the
disposition as a disposition to respond dependably to just those scenarios. What the
ring of Gyges shows, on this interpretation, is that virtuous dispositions that we prize
in ourselves and others, however dependable they are, may not be durable in the meas-
ure to which we thought them.6
The analysis of trust outlined in this section of our chapter does not yet provide sup-
port for the empowering view of trust. It is quite consistent with the analysis that when
the manifestness of my reliance motivates you to prove reliable, that is entirely because
of a pre-existing disposition on your part not to let people down when they are manifestly
relying on you; that disposition may not be affected in any way by my trusting you. The
act of manifesting reliance may serve as a trigger that activates that disposition, in
other words, without giving it a boost. It may prompt you to reveal your established
capacity to be motivated to act as others manifestly depend on you to act but not serve
in any way to improve that capacity. The empowering theory goes further in suggesting
that by manifestly relying on you—by exercising trust—I may not only cause you to
exercise your existing capacity for trust-responsiveness; I may also cause you to
develop that capacity, achieving a higher degree of dependability or durability.
Although the empowering theory of trust is more encompassing than the view
implied in the bare analysis of the phenomenon, we take it here in a way that makes it
less encompassing than a third approach. Our focus is on straightforward cases where
I expect that the manifestation of my relying on you will generally ensure that you
actually prove reliable; it will help to prompt or reinforce your acting as I rely on you to
act. But the considerations that support the empowering theory, so understood, sug-
gest that there will also be cases where it makes developmental sense for me to manifest
reliance on you, even though I do not expect you actually to prove reliable: certainly
I do not expect this with any degree of confidence. Manifesting reliance will make
developmental sense to the extent that I believe that you can be brought to develop
trust-responsiveness over time, as a result of my efforts; you may be an adolescent, to
6
Pettit (2015, ch. 2) argues that the situationist claims in recent psychology do not establish that virtues
are undependable and in that sense rare—this is the standard interpretation—but only that they are not as
durable as might have been thought; they are subject to unexpected disrupters, as various experiments in
social psychology have shown. See also Merritt (2000).
take a salient case, and I may be a parent who is anxious to make you trust-responsive.
The empowering theory might be taken as a theory governing a spectrum of cases that
includes these examples as well as the straightforward cases, and that also includes
examples that do not fall neatly into either category. But for simplicity we restrict atten-
tion here to the theory as it applies in the straightforward cases.
We try to sketch some considerations that support the theory in section two. And
then in the third section we elaborate some implications of the theory for how we
should think about trust-responsiveness and, more generally, about capacities of the
kind it exemplifies. The considerations we will be introducing in support of the
empowering theory are broadly empirical. But they are rooted in matters of common
experience and assumption and we think of the case they make for the theory as an
elaboration of common-sense tenets. Still, the case we present is certainly subject to
testing and might be usefully brought into contact with the empirical literature on
trust. Unfortunately we are not in a position to do that ourselves, and certainly not
within the confines of this chapter.
2. The Case for the Empowering Theory

2.1 Two theories
When I display or exercise trust in dealing with you, as we saw, then I act on the belief
that you are responsive to my manifestly relying on you to act in a certain way. In
taking this line I make myself vulnerable to you. Depending on our relationship with
one another, and indeed on what is at stake, this may involve exposing myself to any of
a variety of ways in which you may let me down (Baier 1986; Holton 1994; Jones 1996).
It is one thing for me to trust you, a complete stranger, with my computer; it is quite
another to trust you, a friend, with a secret. But whatever the mode of vulnerability
involved, I trust you just insofar as I take my manifest reliance and my associated vul-
nerability as a reason that weighs with you to prove trust-responsive, whether that
reason serves in an active or back-up role. If I did not believe that you were responsive
to such considerations of reliance—for short, to reasons of trust—then I would not
count as putting my trust in you.
But to be responsive to reasons of trust—to have the capacity to be suitably m
otivated
by the manifest reliance of others—is an ambiguous notion. It may mean either or both
of two different things.
First, it may mean that you have a standing disposition to be responsive to
reasons of trust, where the reasons present may vary depending on your relation-
ship to the trustee and even to what is at stake; you may have different reasons for
being responsive to the trust of a friend from the reasons you have to be responsive
to the trust of a stranger. But within a given category—say, in your relationship to a
given friend—your sensitivity to the reasons themselves is a standing feature of
your psychology.
Alternatively, to be responsive to reasons of trust may mean that you are ready or
predisposed, on being presented with evidence of trust, to form a disposition there
and then, in the situation at hand, to respond to reasons of trust. Where in the first
example your responsiveness is a standing, independent disposition, in the second it
is a disposition that is actually formed or developed in response to another’s display
of trust.
An analogy may help make this difference clear. Consider what it means to have an
adaptive immune system. Creatures with such a system are disposed to fight off infec-
tions but they may be disposed to do this in either of two different senses. They may
have been exposed to a non-lethal form of the disease, in which case they will have
developed the relevant antibodies and will be actively disposed—disposed in a once-
for-all-inputs fashion—to fight it off: the antibodies will be there, waiting to get to work
when suitably triggered. But even if they have not been exposed to the disease and
not developed this active immunity, the creatures will be disposed in a more remote,
input-by-input sense to fight it off; by hypothesis, their adaptive immune system will start
manufacturing antibodies in response to exposure. This passive immunity is not quite as
effective as the active sort but it can provide a perfectly adequate level of protection.
We may all agree that when I trust you I must believe that you are responsive to
reasons of trust: that is, to considerations to the effect that someone—say, someone in a
particular relationship—is manifestly relying on you in a certain way. But you may be
trust-responsive in either or both of two ways, just as your body may fight off disease in
either of two ways. Setting aside the complication raised by the relativization of trust to
different relationships, you may have a standing disposition to be motivated by reasons
of trust across all instances of reliance. Or you may have an input-sensitive predisposition
or readiness to develop the disposition to respond appropriately to reasons of trust in
specific instances of reliance. Where the disposition you act on in the first case has
a standing character, the disposition you act on in the second has a case-specific
character: it is formed anew in each situation.7
Though not all trust-theorists subscribe to this view, the standard or received theory
of trust, as we shall characterize it, holds that trust-responsiveness consists in a once-
for-all-cases type of responsiveness. On this picture, trust-responsive people have a
standing disposition, more or less constant across different situations, to respond to
reasons of trust. If standing trust-responsiveness is equated with trustworthiness, in
the ordinary sense of that term, then the claim of the received theory is that people’s
trust-responsiveness is nothing more or less than their (standing) trustworthiness. It is
7
Consistently with this distinction, it is possible that as I and others display trust in you and help to
elicit a situational disposition to prove reliable, this will have a boosting effect on your standing disposition
over time. The idea is that the more you are situationally activated to prove responsive to trust, the more
dependable your standing disposition to prove responsive will become. It may even be that situational
activation remains essential to keeping the standing disposition in place at a minimal level of dependabil-
ity. This would be analogous to the immunity case, where it may be that while past exposure to disease has
elicited antibodies that provide you with active rather than just passive immunity, continuing exposure—
regular booster shots of a vaccine, for example—may be necessary to maintain that active immunity.
a virtue that they bring to every situation and provides a solid, unchanging base on
which we can predict what they will do, or be inclined to do, now in this scenario,
now in that.8
The empowering theory of trust does not deny that people may be trustworthy in
this standing sense and that this disposition may play a role in helping to explain why
they do not let down their trustors. But it adds that how trust-responsive such trustees
are in dealing with another may also be a function of their situational sensitivity to the
presence of that person in the role of a salient trustor: someone who is manifestly rely-
ing on them to behave in a certain manner. The idea is not that the trustor may have a
special hold on them, although that may be true too. It is rather that any trustor is liable
to elicit a novel degree of trust-responsiveness in a trustee—not necessarily the same
degree for every trustor—as well as triggering the trustee’s existing trustworthiness.
On the empowering theory, then, the net trust-responsiveness that you display, now
in this scenario, now in that, may be a function of two variables. The first variable is
how trustworthy you are in general: how much sensitivity you have to reasons of
trust—reasons of manifest reliance—in themselves. The second is how far you are
moved by the presence of those who manifestly rely on you in certain ways: how sensi-
tive you are to the fact that here and now others are putting their trust in you. We will
look shortly at how to model the contribution of each of these factors in representing
your degree of trust-responsiveness. But first we must look at the precise claims of the
empowering theory and the case for accepting them.
Suppose for a given scenario of trust—a case where someone manifestly relies on
you—that you are trust-responsive with a certain degree of dependability and a certain
degree of durability. You are disposed to respond to those reasons across a range of
possibilities that measures the degree of dependability; and your disposition is proof
against a range of disrupters that measures the degree of durability. It may be in such a
case that your responsiveness reflects just a standing disposition, as in the received
theory. Or it may be that in each relevant case it also reflects the effect of a case-specific
disposition. Thus it may be that your final disposition is the resultant of standing and
situational components.
The empowering theory is committed to two theses. First, that there are grounds for
thinking that even if you bring a standing capacity of trust-responsiveness to a given
situation of trust, you are likely to be influenced within that situation in a way that
enhances the capacity; you are likely to develop a disposition with a higher degree of
dependability in reaction to situational inputs. And second, that there are grounds, not
8
This take on trustworthiness is motivated only by considerations of convenience and clarity. The
t heory outlined in the chapter might also be construed to show that trustworthiness is not a standing trait
of the kind presented here; it might identify trustworthiness with trust-responsiveness. As we use the
term here, ‘trustworthiness’ measures your standing responsiveness to reasons of trust, where as we have
noted, the requirements of those reasons may vary in keeping with your relationship to the trustor (stran-
ger, colleague, friend) and in keeping with what is at stake in the context (their computer, their secret, or
whatever).
just to think that you can be made trust-responsively more dependable by situational
influences, but also to hold that situational influences can make you trust-responsively
more durable. The empowering theory defends both the situational enhancement of
dependability and the situational reinforcement of durability.
2.2 The situational enhancement of dependability

There are three grounds for postulating the likelihood of situational enhancement
of dependability. The first is that when I trust you, I display and communicate a belief
in your capacity to be motivated by my manifest reliance, thereby encouraging you
to prove reliable. The second is that when I trust you to do something, I often make a
request, explicit or implicit, that you do it. And the third is that when I trust you I dis-
play a good opinion of your dependability, thereby giving you an extra esteem-based
motive for not letting me down. In any situation, these effects are liable to make you
trust-responsively dependable in a higher degree than would be ensured by the stand-
ing capacity of trust-responsiveness that you bring to the situation.
2.2.1 tHE ENCOURAGEMENT ARGUMENT

The first argument for the situational enhancement thesis claims that insofar as
I trust you to do something, I encourage you to do all you can to prove dependable
(McGeer 2008b). This turns on three plausible psychological observations: first, that in
trusting you I display and communicate a belief in your capacity to be motivated by my
manifest reliance on you; second, that in communicating the belief that you have such
a capacity, I may boost your own belief that you have that capacity; and third, that in
doing this I may improve the capacity itself, making it more dependable: I may give
you the courage to test and thereby develop your own powers of agency.
The first of these observations simply falls out of our account of trust, as distin-
guished from mere reliance. When I trust you, the manifestness of my reliance means
that it must be a matter of common awareness that I am doing so: namely, that I am
acting on the belief that you will recognize that I am relying on you and be motivated
by that very fact to respond to my reliance. This will be a matter of common awareness
insofar as the evidence that supports the claim that I am trusting you is clearly available
to each of us, ensuring that each will be aware that I am acting out of trust, aware that
each is aware of this, and so on (Lewis 1969). Given that trust requires a common
awareness of this fact, I will not only display my belief that you are trust-responsive in
the involuntary manner in which my yawning might display my boredom. I will dis-
play my belief in an intentional manner, communicating that I believe in your capacity
(Grice 1957).
The second observation in support of the encouragement argument turns on the
commonsense assumption that the fact that I communicate a belief in a capacity like
your trust-responsiveness is likely to increase your own confidence in having that
capacity. Short of my being a doting parent, or something of that kind, I credibly
testify to signs of your having that capacity when I communicate my belief in your
responsiveness, and stake my welfare on its being true. In effect I say: ‘Have no doubts;
you possess the capacity to live up to my trust, to prove reliable in the ways that I
anticipate!’ And in saying this—indeed, in staking my own welfare on its being true—
I provide good ground for your being more confident of having the responsiveness
that I ascribe to you. After all, the signs to which I testify may not be accessible to you
yourself; your diffidence or fear may lead you not to credit them; or your experience
(or inexperience) may lead you to underestimate what they imply about your own
powers of agency.
The third observation may seem more problematic, but it is also pretty plausible and
follows quite naturally from the last. If you gain in confidence about having a certain
sort of capacity—in particular, the capacity to be moved to action by a consideration
like my relying on you to do something—then that is likely to improve the capacity
itself, increasing the degree to which you dependably exercise it. It is likely to increase
the range in which you are trust-responsive beyond the range ensured by your stand-
ing responsiveness to reasons of trust: what we earlier called ‘trustworthiness’. This
observation is grounded in the fact, borne out in everyday experience, that becoming
more confident about being able to do something X in suitable circumstances—say,
about being able to prove reliable in response to someone’s manifest reliance on you—
means improving that very capacity; it means that you will do X more dependably
across relevant circumstances.9
Thus, suppose you have a low opinion of your own moral quality, believing that you
are weak and give in to self-interest more than you ought to do; believing in particular
that you are less responsive than you should be to others who manifestly rely on you
not to let them down. Your self-confidence in your own capacity to do as you ought is
likely to improve as others display faith in your capacity to prove trust-responsive. And
that in turn is likely to encourage you in your display of such responsiveness.
2.2.2 the request-based argument

The second, request-based argument for the situational enhancement thesis turns
on the intuition that when I ask you to do something, thereby presupposing that
you will be motivated to comply by my manifest reliance on you—that is, that you
will respond in an active or back-up way to the reasons of trust relevant in our
relationship—the fact that I am making a request may have a force over and beyond the
persuasive power of the reasons themselves. Moving away from trust for a moment,
suppose I believe that you are fully aware of all the reasons there may be for helping me in
some manner. Even in the presence of such a belief it may make sense to ask for the help
I need. Doing so will amount to recognizing the burden that helping me is likely to
9
Here as elsewhere there is some interpretive latitude about how to understand the disposition and
about where to draw the line between dependability and durability. Thus the text assumes that having the
capacity to do X at any level of dependability is responsive to your confidence about being able to do X. If
it were unresponsive, then the absence of confidence, assuming it affects your performance, would count
as a disrupter of the capacity, affecting the level of its durability rather than its dependability.
impose, acknowledging that you have no obligation to bear that burden, and asking
your permission nevertheless to impose it. If you do not then decline the burden, but
implicitly accept it, you will effectively have promised to provide the help I require.
And so you will have an additional normative reason, and presumably an additional
motive, for actually providing that help.
When I manifestly rely on you to act in a certain way, I may often make an explicit
request to do what I am relying on you to do, as when I ask you to keep an eye on my
computer. And even where I do not make an explicit request, reliance may be so mani-
fest as to constitute an implicit request. By analogy with the help case, that means that
in either of these cases I effectively ask your permission to impose on you the burden of
proving reliable, acknowledging that it will indeed be a burden and that you have no
obligation to bear it. And if you do not decline to bear the burden—if in effect you
accept it—then that means that there is further reason to bear it: that otherwise you
will breach an implicit promise. However morally binding you consider that reason, it
is very likely that you will be motivated by it; no one wants to be cast as unmoved by
their promises. And the motivational effect of the promise will be to make you more
responsive to the reasons or requirements of trust.10
2.2.3 the esteem-based argument

We have looked at two arguments for believing in the situational enhancement
thesis, according to which I may increase the dependability of your trust-responsive
disposition by the very fact of displaying trust in you. Both of those arguments
suggest that I may increase your dependability by introducing factors that we can each
acknowledge openly (Baier 1986; Jones 2004). I do not undermine my impact by mak-
ing clear that I am hoping to have an impact, whether in expressing my faith in the
capacity that I take you to have or in making a more or less explicit request to you. The
factor invoked in each case is publicly avowable between us. Consistently with operat-
ing as described, it may be something of which we are each aware, aware that each is
aware, and so on in the usual regress of common awareness.
The factor introduced in the third argument for situational enhancement is different
in this regard from the other two (McGeer 2008b). It consists in the fact that when
I display trust in you to act in a certain manner I give evidence, by regular accepted
criteria of attitude ascription, of having a good opinion of you. And I thereby provide
you with a new motive for behaving as I am manifestly relying on you to behave; that by
doing so, you will reinforce the good opinion I exhibit and secure an intuitively attractive
reward in the economy of esteem (Brennan and Pettit 2004). This reward will be all the
10
A distinct argument might appeal to the thought that second-person claims constitute a unique
source of moral demand, distinct from any impersonal demand to improve the world or indeed any per-
sonal demand to act in a morally righteous way (Darwall 2006). If this is correct, then the request-based
argument need only assume that in asking for your trust-responsiveness, I create a second-personal reason
why you should prove reliable.
greater of course, and the new motive will be all the more powerful, if I trust in you as a
public act; in this case, you will be able to secure the good opinion of witnesses, and not
my good opinion alone, by proving reliable.
This argument suggests that there is a certain cunning in trust (Pettit 1995). Our
accepted criteria of attitude ascription support the view that when I trust you to act in a
certain manner, that must be because I hold a good opinion of you. Those criteria
embody ‘the fundamental attribution bias’ (Jones 1990): the tendency to explain
responses like my trusting you in terms of some manifest, standing disposition such as
the belief that you are trustworthy. As a matter of fact, however, I may not be acting on
such a belief in trusting you. Rather I may be grounding my trust in the belief, first, that
you will care for the effects on your reputation with me and with any witnesses of your
proving to be reliable; and, second, that this concern will almost certainly lead you to
act as I am purportedly trusting you to act.
There is more to be said about the dynamics of esteem in supporting trust and,
indeed, other such overtures between people. But these remarks should be suffi-
cient to identify a third factor that supports the idea that by trusting you I may do
more than activate a pre-existing disposition of trustworthiness to be moved by the
manifest reliance of others; I may induce a higher level of dependability in your
trust-responsiveness. My display of trust may not only constitute a motivating
request for help and an intentional form of encouragement to provide that help.
It may also link your provision of help—your acting as I rely on you to act—with
an attractive prospect of securing esteem and standing in my eyes and in the eyes
of witnesses.
2.2.4 a question
We describe the three factors surveyed here as enhancing the dependability of your
standing disposition to respond to reasons of trust: that is, as making it more depend-
able. But why not take them to complement your standing disposition by putting inde-
pendent forces in place that have a parallel, if convergent effect? Why not take them to
do for your trustworthiness what your self-interest does for your honesty, when it is
clearly in your interest to tell the truth?
The answer is that the three factors in our examples each serve to make you more
responsive to reasons of trust as such, not just to make you more likely to act in the way
that, as it happens, reasons of trust support. The encouragement I provide in trusting
you boosts your existing capacity, whatever that involves, to be motivated by reasons of
trust. The request I make in doing this provides you with a further, presumptively
motivating reason—that it will enable you to keep an implicit promise—to act in what-
ever manner those reasons require. And equally the prospect of esteem that I put in
place when I trust you provides you with a further, motivating reason to be sensitive to
the reasons of trust as such: after all, you can only expect to win credit from me by
proving yourself trust-responsive in this sense.
2.3 Modelling situational enhancement

The upshot of this picture is that your responsiveness to evidence of trust in a given
instance, your capacity to be moved to action by my manifestly relying on you, may be
a function of two factors.11 First, the standing sensitivity to reasons of trust that you
bring to that encounter: that is, your trustworthiness. And, second, the situational sen-
sitivity to my presence that enhances your trustworthiness, increasing the dependabil-
ity of your disposition to act as the reasons of trust require. We might picture the play
of these factors as shown in Figure 2.1.
If your trust-responsive dependability is a function of two forces, of course, then it
becomes possible that in a given case it may result from different combinations of
those forces. The two sensitivities may combine in different measures to produce
responsiveness and any degree of responsiveness may be realized via any of a range of
equivalent combinations. In some combinations sensitivity to reasons of trust will be
high and sensitivity to the presence of a trustor low, in others the reverse will hold, and
in still others the factors will be more or less equal.
Thus we might represent your trust-responsive dependability in a given instance on
a graph in which the vertical axis depicts sensitivity to reasons of trust, the horizontal
sensitivity to the presence of a trustor (see Figure 2.2). Take a given combination as a
point on the graph and now connect up those points that yield the same degree of
trust-responsiveness. Such points will connect in an equivalence curve—one of the
diagonal lines in the diagram—in the space of trust-responsive dependability. We may
characterize any of a range of curves, each corresponding to a different degree of
dependability: the lines to the upper right in the diagram will represent higher degrees,
those to the lower left lower degrees, in analogy with the familiar picture of indiffer-
ence curves in the space of utility. One of those curves will presumably characterize
your degree of trust-responsiveness in the case in question. And on the curve that
identifies you, there will be a point that represents your particular combination of
sensitivities in that case.
Trust-responsiveness
Sensitivity to reasons of + Sensitivity to trustor’s

trust presence
Figure 2.1 Responsiveness to evidence of trust
11
We borrow in discussion here from McGeer and Pettit (2015).
Sensitivity
to reasons of
trust
Sensitivity to trustor’s
presence
Figure 2.2 Trust-responsive dependability
2.4 The situational reinforcement of durability

If the considerations rehearsed so far are sound, they suggest that situational influ-
ences are likely to make you trust-responsively more dependable than could be
explained by the standing capacity of trust-responsiveness that you bring to the situ-
ation. But the empowering theory of trust, as we understand it, holds also by a second
thesis: that situational influences are likely not only to make your disposition to
respond to reasons of trust more dependable, but also to make it more durable; they
are likely not just to enhance the disposition, but to reinforce it. The idea is that just as
situational influences can increase the range of scenarios over which that disposition is
dependably triggered, so they can increase the range of scenarios in which it is durable
enough to survive potential disrupters.
The easiest way to support the reinforcement thesis is to consider the case
where you have developed a more or less dependable, standing disposition of trust-
responsiveness, perhaps under the long-term influence of situational sensitization. In
our terms, you have become a more or less trustworthy person, not just someone
who is trust-responsive. The considerations rehearsed in support of the enhancement
thesis argue that situational influences can also reinforce any standing disposition
of trust-responsiveness that already exists.
One way in which those influences may reinforce your trust-responsiveness, of
course, is by habitually reminding you of the lessons they teach: that if you do not
prove reliable, you must often count as breaking a promise to a trustor, you must disap-
point the faith in you that they display in encouraging you, and you must expect to lose
the good opinion and reputation that they give you the chance to earn. The causal
effect of being constantly under such tutelage will surely be to maintain and reinforce
in you the disposition to respond to reasons of trust. If, for example, your standing
trust-responsiveness has begun to decline, say as a result of losing faith in yourself, an
exposure to such influences from others—in particular, perhaps, their influence in
encouraging you—ought to help to restore the durability of that disposition. This
exposure will help to maintain your standing trust-responsiveness in the way in which
booster shots of a vaccine may help to maintain your active immunity to a disease.
But apart from the causal impact of situational influences in maintaining or rein-
forcing your trust-responsiveness, there is also a further consideration that argues for
the reinforcement thesis. Suppose that your standing disposition to respond to reasons
of trust is so dependable that there is no room for further enhancement by situational
influences. Even in that case, there is a role that those influences can continue to play
in making the disposition more durable.
Once again, the ring of Gyges story brings out that role nicely. Suppose, as seems
reasonable in many cases, that you would cease to be trustworthy—you would lose
any standing capacity to respond to reasons of trust—if you enjoyed the total
undetectability and impunity that the ring would confer; those factors would disrupt
your trustworthiness, suspending it or even obliterating it. What does that teach us
about the factors that play a role in keeping the disposition in place in the ordinary
world? Presumably, that your exposure to others is essential to your continuing to
have the disposition, to its remaining a durable part of your psychology, and plaus-
ibly, that your exposure to the sorts of situational influences listed has a major part
in securing that result.
How are we to conceive of the role that those factors play, assuming that you are
already fully trustworthy? You may not be focused on the goal of proving yourself a
promise-keeper, living up to the encouragement of others, or winning esteem in their
eyes. And you may not even be aware of such influences, let alone attentive to them. All
that need be true is that the influences have a virtual impact on your thinking and
behaviour. They are there to put you back on course—there to maintain and reinforce
your trust-responsiveness—should you fall away. The ordinary world is one where any
tendency to give into temptation, or any step in that direction, is likely to trigger those
influences and restore the disposition in you. And that shows up in the fact, as we pre-
sume it to be, that in the Gyges world there is nothing to block any such fall from grace
and much indeed to prompt a fall: namely, the salient fact that you can do as you wish
with total impunity.
On this picture, the situational influences may not have any causal effect in ensuring
that your trustworthiness remains durably present. What they have rather is a virtual
effect. They may play no causal role in keeping your trustworthiness in place—it sur-
vives as a matter of deliberative habit—but if they were absent then, sooner or later,
that standing responsiveness to reasons of trust would fall away. Their absence would
make self-seeking deliberation a salient option and would almost certainly undermine
the disposition.
This picture implies that the situational influences to which you are subject control
for the durable presence of the disposition. Think of how the cowboy controls his
cattle in the classic western movie. He lets them follow their head—he does nothing
causally to direct them—but is ready to intervene should any head of cattle wander off
track. The situational influences that surround you in your role as a potential trustee
have the same sort of non-causal control over your trust-responsiveness. They give
that disposition a durable presence that it would not enjoy in their absence. And they
do this just by being there, ready to play a causal role if they are needed to maintain
the disposition.
3. Trust-Responsiveness on the Empowering Theory

3.1 The evocative ascription of trust-responsiveness
When I trust you I display, and indeed communicate, the belief that you are trust-
responsive, as we saw in section two. But that means that trusting you has the force of
an ascription of trust-responsiveness. It amounts to saying in an ascriptive mode: ‘you
can respond to the fact that I am relying on you; you have the capacity to be moved by
that consideration; you can prove yourself reliable’.
On the empowering view of trust, however, this ascription of capacity has a very
unusual character. That you have the capacity of trust-responsiveness in a given situation
means that you are disposed, however durably, to respond more or less dependably to
reasons of trust. Whether or not you actually fail to act on the capacity, it must be true
that in a range of variations on actual circumstances—in particular, ones where
reasons of trust remain in place, where no outweighing considerations are introduced,
and where no novel perturbation affects you—you would respond as appropriate. But
if the empowering view is sound, then when I ascribe such a capacity to you in displaying
trust, I am likely to have an effect on the very capacity I ascribe.
The likelihood of having such an effect is surely going to be salient in the culture.12
And so I cannot ascribe the capacity in a purely descriptive spirit, as if I were recording
a pre-existing state of affairs. In asking you to take down a bottle from a shelf I cannot
reach, I am implicitly ascribing a height to you that I do not possess myself and in
doing so I am recording what I take to be a pre-existing state of affairs and making the
ascription in a purely descriptive spirit. The ascription of trust-responsiveness that
I implicitly make in trusting you is clearly not like that. It is akin to the ascription of a
capacity—say, a capacity to reach the top of a hill on your bicycle—when I call out in
encouragement: ‘You can do it’. The ascription is inevitably designed to elicit the very
capacity that it attributes.
Descriptive reports are often contrasted with performative reports. In the case of a
performative, the utterance brings about the very state of affairs on which it reports, as
in examples like ‘I resign my post’ or ‘I beg your favor’ (Lewis 1983, ch. 12). But as the
ascription of a capacity that I implicitly make in trusting you is unlike a descriptive
report, so it is also unlike a performative utterance: I do not make you responsive to
evidence of trust just by ascribing such responsiveness to you.
The descriptive and performative alternatives, however, do not exhaust the options.
There is a third possibility that our discussion makes salient. This is that the ascription
of trust-responsiveness implicit in my displaying trust in you is an evocative ascription,
as we may call it: an ascription that is designed to help bring about the very state of
affairs on which it reports. The idea is that in communicating the message that you are
12
This may not mean that we regularly pay attention to the effect but only that the frequency with which
we resorted to ascribing trust-responsiveness would be affected if it became clear that the effect did not
generally materialize.
trust-responsive and can prove reliable, I do so with complex intent. I mean to try to
elicit your capacity to prove reliable, or at least to support it, in the very act of recording
its presence. Whether or not I make this explicit, I am effectively exhorting you to
prove reliable. I am speaking in a way that I may reasonably expect to have a direct
effect in enhancing the dependability of your trust-responsiveness—and an indirect
effect perhaps in reinforcing its durability.
How are we to think of a capacity that has such a plastic form that I can ascribe it to
you in this evocative fashion? The question is worth exploring as it turns out that there
are three significant lessons that the argument teaches us about how to think about the
capacity. The first is that the capacity is context dependent; the second that it is a work
in progress; and the third that assigning the capacity is subject to moral as well as
epistemic norms.
3.2 Trust-responsiveness is context dependent

The first clear implication of the empowering theory is that your capacity to prove reli-
able in dealing with a trustor like me is not fixed solely on the basis of how you are in
yourself. It may be fixed not just by your standing sensitivity to the reasons of trust—
that is, by your pre-established trustworthiness—but also by your situational sensitivity
to me, your trustor. And even if it is wholly fixed by your standing sensitivity to reasons
of trust, that sensitivity is itself kept in place by your situational sensitivity to me, and to
other trustors. The capacity of trust-responsiveness, on this picture, is the product or
resultant both of how you are in yourself and of how things are in the world you occupy.
It is common to assume that any disposition you possess—and so any capacity
linked with that disposition—must be grounded in categorical facts; it cannot be a bare
disposition, as it is sometimes put (Dummett 1973). The basis of the assumption is that
it is hard to imagine that two otherwise identical subjects should be the bearers of
different dispositions. What could explain the difference in disposition other than a
difference of a more categorical kind?
But when it comes to dispositions like trust-responsiveness, and human capacities
more generally, this common assumption takes a particular form. It is taken to imply
that as between two subjects who differ in such a disposition, the categorical difference
that explains the divergence must be a difference in their inherent make-up. If our
observations about trust-responsiveness are correct, however, this is not so. How
trust-responsive you are in dealing with me may be a function of many factors: my
nature, our relationship, and the company that we share with others, such as others
who serve as witnesses to my relying on you in a certain way. In short, it may be a func-
tion of the context in which you operate as well as a function of your inherent charac-
ter. It may be a context-dependent or ecological capacity, not a capacity that is fixed
only by how you are in yourself.13
13
This notion of an ecological capacity resembles one defended in another context by Manuel Vargas
(2013); we borrow the term ‘ecological’ from his book. See also McGeer and Pettit (2015). The fact that the
3.3 Trust-responsiveness is a work in progress

The standard view of trust reifies trust-responsiveness and suggests that its presence or
absence is an on or off matter, so that there is a clear dividing line between those who
are worthy of trust and those who are not. The capacity is equated with a disposition to
appreciate and respond to reasons of trust and like any such disposition it may come in
various degrees of dependability and durability; it may dispose the agent more or less
durably to respond to reasons of trust more or less dependably. But the idea in the
standard view is that there is a certain threshold of durability and dependability at
which this capacity materializes and that every agent reaches this or fails to reach it;
there is no middle ground. It may take educational and moral development for an
agent to pass the threshold. But given that development, the capacity is fixed in place,
with a suitable degree of dependability and durability, and the agent joins the ranks of
the trustworthy.
There may well be a certain threshold of responsiveness to reasons of trust, both in
durability and dependability, which we require those who count as trustworthy to
pass. But that should not lead us to adopt the standard view. For on our picture, the
responsiveness may enjoy that degree of dependability under equivalent mixtures of
two factors: one, a standing sensitivity to reasons of trust as such; and two, a situational
sensitivity to the presence of a trustor. And on our picture, the responsiveness may
be more or less durable, depending on the effectiveness of situational influences on
the trustee.
Once we recognize that trust-responsiveness has this ecological character, it
becomes difficult to think of the capacity involved as something that is there or not
there in an agent, depending on the threshold reached. By our account, the responsive-
ness or capacity required may materialize at a given level of dependability on the basis
of any mix of standing and situational sensitivity, and it may materialize with one or
another degree of durability. It would make sense to posit the sort of threshold envis-
aged in the standard theory if trust-responsiveness were something established in the
make-up of a person. But this scarcely makes sense on the basis of the assumptions
defended here.
The image suggested by our approach is that the capacity that makes you fit to be
trusted is as likely to be a work in progress—a work in which others play a role and not
just you alone—as it is to be a finished achievement: something that is fixed in place,
once and for all. The moral life promises to be agonistic in nature, involving a continu-
ous struggle to maintain an appreciation of the reasons relevant in trust and to sustain
your responsiveness to those reasons (McGeer 2008a). It is likely to require you to call
on all the reserves at your disposal, including reserves that only suitable social inter-
actions can provide, in order to maintain your status as someone it makes sense to trust.
capacity is context-dependent might lead us to treat ascriptions as indexical, so that what I ascribe is always
responsiveness-to-me, but it would also be consistent with treating them as ascriptions of responsiveness,
now to this sort of audience, now to that. See Hawthorne and Pettit (1996).
There may be an attraction, of course, in the prospect of achieving an increasing level

of sensitivity to reasons of trust and a decreasing dependence on sensitivity to the pres-
ence of one or another trustor. This is the appeal of becoming a virtuous person—specif-
ically, a trustworthy person—in the Aristotelian sense of virtue: someone with a potential
to act appropriately that is not hostage to the fortune of circumstance; someone who not
only enjoys excellence or areteia but enjoys it with a certain self-sufficiency.
But, however appealing in other ways, the Aristotelian ideal may be a myth in at
least one respect. Even if you have a standing disposition to respond dependably to
reasons of trust—even if the dependability of your trust-responsiveness is grounded in
your nature—you may depend on context, and not just on your inherent nature, for the
durable presence of that disposition. This is a lesson that the ring of Gyges teaches. Or a
lesson that it teaches, at any rate, on the assumption that many of us would not main-
tain even a hard-won degree of trustworthiness—a standing capacity to respond to
reasons of trust—if we could manage with impunity to break promises, flout the
encouragement of others, and perform manifestly shameful acts. Even if some of us
have achieved the context-independent dependability associated with the Aristotelian
ideal, so the message goes, we are unlikely to have achieved a corresponding degree of
context-independent durability.
3.4 Ascribing trust-responsiveness is subject to moral norms

Since the ascription of a capacity like trust-responsiveness is a judgemental exercise,
bound to the requirements of truth, it is certainly subject to epistemic or cognitive
norms. I ought not to ascribe a capacity to appreciate and act on the manifest reliance
of others in the absence of all evidence that you have a suitable disposition to respond
to reasons of trust. Certainly, for example, I ought not to ascribe such a capacity if
experience shows that you routinely let others down, even when they go out of their
way to make it clear that they are relying on you.14
But to ascribe trust-responsiveness, by the argument of this chapter, is often to
enhance or reinforce the capacity, helping to make it more dependable or durable. And
that means that it could not be right to let the ascription of trust-responsiveness be
governed by epistemic or cognitive norms alone. That would be appropriate only if
the ascription amounted to nothing more than the report of an unobtrusive observer.
But in interacting with you as trustor to trustee—or would-be trustor to hoped-for
trustee—I am never an observer of that merely spectatorial kind.
The lesson of this observation is that while I should certainly not breach cognitive
norms in crediting you with trust-responsiveness, whether by explicit ascription or by
implied assumption, I should be guided in the slack that those norms leave by norms of
a moral or ethical kind. I should let myself be led to attribute this capacity to you on the
14
Of course even in this case I might have a developmental rationale for manifesting reliance, and at
least purporting to ascribe trust-responsiveness to you: I might plan to nudge you over time towards being
trust-responsive; see the discussion at the end of section one.
basis of how far this is likely to be productive, helping to enhance or reinforce the cap-
acity I posit. I should take into account the extent to which you are likely to be scaffolded
by my attribution and to perform in a manner that is immediately beneficial to me and
that may be more generally beneficial in reinforcing your trust-responsiveness.
Ascribing trust-responsiveness to you, at least in the way in which this is encoded in a
display of trust, means investing in you, as it is said, holding out the hope that you will live
up to the expectations associated with that ascription (McGeer 2004; Pettit 2004). When
I invest this hope in you, I have reason to think that the relationship I thereby announce
will motivate you to live up to my expectations, proving worthy of my faith. But that
means that there are moral norms that should guide me in displaying trust in you and
ascribing a suitably dependable and durable degree of trust-responsiveness. The ascrip-
tion of such responsiveness may be subject to constraints of evidence and a fit topic for
epistemology. But it is also subject to moral standards and is a suitable topic for ethics.
We end with an illustration of this last lesson (McGeer 2008b). In George Eliot’s
novel Middlemarch, the local doctor, Tertius Lydgate, has fallen on bad days, guilty by
association with the villainous Nicholas Bulstrode and is abandoned in disgrace, left to
despair, by the suspicious and gossip-mongering townsfolk. The high schemes he had
entertained of ushering in new medical discoveries and practices are certainly in tat-
ters, and his prospects otherwise look dim. But at this crucial juncture Eliot’s heroine,
Dorothea, calls his coterie of friends to account, reminding them that true friends have
a duty to support one another, so that Lydgate might rise to the challenges he faces
(Baker 1987). Admonishing them, she declares:
‘And would you not like to be the one person who believed in that man’s innocence, if the rest
of the world belied him? . . . I should not be afraid of asking Mr Lydgate to tell me the truth, that
I might help him. Why should I be afraid? I . . . [could] ask for his confidence; and he would be
able to tell me things that might make all the circumstances clear. Then we would all stand by
him and bring him out of his trouble. People glorify all sorts of bravery except the bravery they
might show on behalf of their nearest neighbours.’
True to her intentions, Dorothea seeks Lydgate out to discover the full story behind
his disgrace, and while his tale is far from completely self-exonerating, he is able to
confess to, and take responsibility for, his own follies and weaknesses thanks to
Dorothea’s continuing display of trust in him. Better, he is able to leave her with a
renewed faith in his own capacity to overcome his difficulties, thereby living up to the
faith she expresses in him. As Lydgate himself remarks, ‘you have made a great
difference to my courage by believing in me. Everything seems more bearable since I
have talked to you’ (Eliot 1994: 768). These empowering effects of Dorothea’s trust
make perfect sense in light of the themes we have been charting in this chapter.15
15
We were helped enormously in revising this chapter through the discussion of participants and a
commentary provided by Wlodek Rabinowicz at a moral philosophy conference in Beaune, France in June
2015; by delegate input at the Kioloa Moral Philosophy Workshop in Jan. 2016; and by copious comments
from Paul Faulkner and Thomas Simpson.
References
Alfano, M. (2014). What Are the Bearers of Virtues? In Advances in Experimental Moral
Psychology, ed. H. Sarkissian and J. C. Wright. London, Bloomsbury Academic: 36–54.
Baier, A. (1986). Trust & Antitrust. Ethics 96: 231–60.
Baker, J. (1987). Trust and Rationality. Pacific Philosophical Quarterly 68.
Brennan, G. and P. Pettit (2004). The Economy of Esteem: An Essay on Civil and Political Society.
Oxford, Oxford University Press.
Darwall, S. (2006). The Second-Person Standpoint: Morality, Respect, and Accountability.
Cambridge, MA, Harvard University Press.
Dummett, M. (1973). Frege: Philosophy of Language. London, Duckworth.
Eliot, G. (1994). Middlemarch. New York, Penguin Books.
Grice, P. (1957). Meaning. Philosophical Review 66: 377–88.
Hawthorne, J. and P. Pettit (1996). Strategies for Free Will Compatibilism. Analysis 56: 191–201.
Holton, R. (1994). Deciding to Trust, Coming to Believe. Australasian Journal of Philosophy
72: 63–76.
Jones, E. E. (1990). Interpersonal Perception. New York, Freeman.
Jones, K. (1996). Trust as an Affective Attitude. Ethics 107: 4–25.
Jones, K. (2004). Trust and Terror. In Moral Psychology: Feminist Ethics and Social Theory, ed.
P. DesAutels and M. U. Walker. Lanham, MA, Rowman and Littlefield: 3–17.
Kaplan, D. (1978). On the Logic of Demonstratives. Journal of Philosophical Logic 8: 81–98.
Lewis, D. (1969). Convention. Cambridge, MA, Harvard University Press.
Lewis, D. (1983). Philosophical Papers: Volume I. Oxford, Oxford University Press.
Merritt, M. (2000). Ethics and Situationist Personality Psychology. Ethical Theory and Moral
Practice 3: 365–383.
McGeer, V. (2004). The Art of Good Hope. Annals of the American Academy of Political and
Social Science 592: 100–27.
McGeer, V. (2008a). The Moral Development of First-Person Authority. European Journal of
Philosophy 16: 81–108.
McGeer, V. (2008b). Trust, Hope and Empowerment. Australasian Journal of Philosophy 86:
237–54.
McGeer, V. and P. Pettit (2015). The Hard Problem of Responsibility. In Oxford Studies in
Agency and Responsibility, ed. D. Shoemaker. Oxford, Oxford University Press. Vol. 3:
160–88.
Olsaretti, S. (2004). Liberty, Desert and the Market. Cambridge, Cambridge University Press.
Pettit, P. (1995). The Cunning of Trust. Philosophy and Public Affairs 24: 202–25; reprinted in
P. Pettit (2002). Rules, Reasons, and Norms. Oxford, Oxford University Press.
Pettit, P. (2004). Hope and its Place in Mind. Annals of the American Academy of Political and
Social Science 592: 152–65.
Pettit, P. (2015). The Robust Demands of the Good: Ethics with Attachment, Virtue, and Respect.
Oxford, Oxford University Press.
Smith, M. (2003). Rational Capacities, or: How to Distinguish Recklessness, Weakness and
Compulsion. In Weakness of Will and Practical Irrationality, ed. S. Stroud and C. Tappolet.
Oxford, Oxford University Press: 17–38.
Vargas, M. (2013). Building Better Beings: A Theory of Moral Responsibility. Oxford, Oxford
University Press.
3
Trust as a Second-Personal
Attitude (of the Heart)
Stephen Darwall
Used sufficiently broadly, ‘trust’ need not contrast with ‘rely’ or ‘reliance’. When we
trust our cars or eyes to work properly, or perhaps even when we trust other motorists,
no more need be going on than our relying on them. We simply proceed on the
assumption that these will function or conduct themselves correctly or appropriately.
Philosophers who write about trust, however, have mostly been concerned to distin-
guish trust from mere reliance of these kinds.
In ‘Trust and Antitrust’, from which all subsequent discussions derive, Annette
Baier attempts to identify ‘the difference between trusting others and merely relying
on them’ (Baier 1986: 233). When Kant’s neighbours relied on his regular walks to tell
them the time of day, they did not really trust him in the relevant sense. One difference
that Baier and those following her have found telling is that trust, but not reliance, can
be betrayed and appropriately give rise to distinctive attitudes or feelings, such as feel-
ing letting down, personal disappointment, or personal hurt, that are not really intelli-
gible when mere reliance is frustrated.1
1. Baier, Jones, and Holton on Trust: The Participant

Stance
Even so, Baier ties trust closely to reliance. Even if there are instances of reliance that
do not amount to trust, Baier nonetheless holds that trust is a kind of reliance. Trust is
‘reliance on [others’] good will toward one’, whereas we merely rely on, and so do not
trust, others when our governing assumptions about them stem from ‘their dependable
1
Baier says that Kant’s neighbours might intelligibly have been ‘disappointed with him’ (though not
‘let down’ or that their trust had been ‘betrayed’) had he diverged from his regular patterns (Baier 1986:
235). This sounds odd to my ear, though it seems perfectly intelligible to express disappointment by
the divergences from what they had relied on. Disappointment with or in a person, that is interpersonal
disappointment, as opposed to disappointment by something involving them, even their actions, seems
too close to being letting down by the person to be justified by mere frustrated reliance.
36 Stephen Darwall
habits’ or from motives that ‘are compatible with ill will’ toward us (Baier 1986: 234).
Trust for Baier is thus a species of reliance: reliance that is grounded in what we take to
be others’ goodwill, at least insofar as it concerns us.
Karen Jones strikes a similar theme when she writes, ‘trust is an attitude of optimism
that the goodwill and competence of another will extend to cover the domain of our
interaction with her’ (Jones 1996: 4). This is similar to what Baier says, so far, anyway,
although Jones holds optimism (and so trust) to be a distinctive affective attitude that
differs from mere reliance, as well as from belief. Unlike both of these latter attitudes,
optimism involves an affectively loaded way we have of seeing the object of our trust
(Jones 1996: 4).
There is, however, another element to Jones’ account, the need for which can be seen
by noting that the elements from Baier and Jones so far considered cannot explain
trust’s distinctive connection to feelings of betrayal, being let down, personal disap-
pointment, and personal hurt. Plainly I can be optimistic about (and, indeed rely on)
someone’s goodwill, and reason on that assumption, without taking it that I have any
reason to feel betrayed or personally hurt or let down if my assumption turns out to be
mistaken. Maybe I could see all along that the person whose goodwill and competence
I am optimistic about had done nothing to encourage or welcome my seeing him as
well disposed towards me. My optimism need not have been misplaced. Maybe, as it
happens, he was well disposed towards me and maybe it was not unreasonable for me
to rely on his continuing to be so. But suppose he neither encouraged nor welcomed
my optimistic view, or, even more, suppose he neither saw, nor had any reason to see,
that I was relying on him. In such a situation, it is difficult to understand where
thoughts or feelings of betrayal, being let down, or personal hurt could get a foothold.
Perhaps for these reasons, Jones adds that in addition to optimism about the trusted
person’s goodwill, trust also involves ‘the expectation that the one trusted will be directly
and favorably moved by the thought that we are counting on her’ (Jones 1996: 4). There
are several elements here. First, although optimism about someone’s goodwill does not
itself entail relying on it, Jones is obviously assuming that reliance is present when she
adds the expectation that the other will be appropriately moved by ‘the thought that we
are counting on her’. So, first, in addition to optimism about another’s goodwill, trust
includes relying on this. But, second, it also includes a state of mind whose object is the
other’s being appropriately moved by the thought that one is relying on her. And, third,
this state of mind is one of expectation.
But an expectation of what sort? We can expect that something will occur or expect
something from or of someone. Expectations that take propositions or possible states of
affairs as objects, whereas the object of an expectation of is a person (or group or
collective of persons).2 Expectations of both sorts impose standards. But where an
expectation that says how things will be (or how we have reason to believe they will be),
an expectation of says how someone should act or be.3
2
For simplicity, I will drop these qualifiers in what follows.
3
‘I expect S to A’ is ambiguous between ‘I expect that S will A’ and ‘I expect of S that S will A’. Note that
it is more natural to hear ‘I expect you to come’ as an expectation of rather than expectation that, whereas
trust as a second-personal attitude (of the heart) 37
Jones’ formulation of trust’s expectational element is: ‘the expectation that the one
trusted will be . . . moved by the thought that we are counting on her’. This might sug-
gest the view that trust partly consists in a special kind of expectation that, one whose
object is the proposition that the other will be appropriately moved by our reliance on
her. However, other things Jones says do not fit with this.
Jones makes the important point that trust can sometimes be unwelcome and felt
as a burden, a point to which we shall return. In such situations, the one trusted does
not welcome being in the position of having to ‘live up to’ the other’s expectations
(Jones 1996: 9). However, the only kind of expectation to which it is possible to live up
is an expectation of that cannot be reduced to an expectation that. Not living up
to someone’s expectation is not to confound what she thinks will happen; it is to fail to
meet her standard of how one should act or be.
Consider the way expectations enter into Hume’s theory of causation. Hume fam-
ously holds that reason cannot supply the idea of necessary connection essential to
our notion of cause and effect, that we only acquire knowledge of causes and effects by
experience of their constant conjunction, and that we form the idea of their necessary
connection by a kind of external projection of our internal expectation, acquired by
habit, that the ‘effect’ will take place once we have experienced the ‘cause’ (Hume
1985: 60–7).
Suppose that a neighbour of Kant’s is aware of Kant’s awareness of his neighbours’
reliance. So far we just have the neighbours’ reliance together with Kant’s awareness of
this and the neighbours’ awareness of his awareness. Suppose also that in other cases in
which his neighbours have experienced Kant’s awareness of others’ reliance on his
actions they have also experienced Kant subsequently acting in conformity with this
reliance, and that, in Humean fashion, they are disposed, when they are aware of Kant’s
awareness of some new form of reliance on his regular habits, to expect him to be
moved to conform to this new form of reliance also.
All of this might be, as it were, on the epistemic side. The neighbours might have
no view whatsoever about whether Kant should conform to others’ reliance. They may
simply have experienced the constant conjunction of his past conformance with his
awareness of others’ reliance and expect, for Humean reasons, that he will do so in new
and future cases. Indeed they might, consistently with all that, think that Kant was
mistaken to conform to others’ reliance in the past and, as well, that it would be a
mistake for him to conform to his neighbours’ reliance on him now.
For the neighbours to have expectations to which it is possible for Kant to fail to live
up, these must be expectations of him. We can see the same point from a different
direction if we notice that any expectational element capable of accounting for trust’s
distinctive connection to the intelligibility of feelings of betrayal, personal disappoint-
ment (disappointment in someone), or personal hurt must also be an expectation of a
this is not so with, say, ‘I expect Jones to come’. I will be arguing below that expectations of have a second-
personal aspect. Note also that expectations of can be expressed by ‘I expect that’, especially when they are
(if only implicitly) second-personal, as when a parent says to her children, ‘I expect that these dishes will
be done when I return’. I am indebted here to Juan Piñeros Sanchez.
38 Stephen Darwall
person. If, consequently, the expectational element in Jones’ account is to capture

trust’s connection to these distinctive attitudes and feelings, it must be an expectation
of rather than a (mere) expectation that. The fact that we expect that someone will be
moved by our dependence on him is, as we have just seen, insufficient to that task.
Interpreting the expectational element of Jones’ account as an expectation of brings
it very close to the kind of account that Richard Holton gives of trust as an essentially
‘participant’ attitude that includes susceptibility to ‘reactive attitudes’ in the senses
P. F. Strawson made famous in ‘Freedom and Resentment’ (Holton 1994; Strawson
1968). Here is Strawson:
The personal reactive attitudes rest on, or reflect, an expectation of and demand for, the mani-
festation of a certain degree of good will or regard on the part of other human beings toward
ourselves; or at least on the expectation of, and demand for, an absence of the manifestation of
active ill will or indifferent disregard. (Strawson 1968: 85)
Strawson calls these ‘participant attitudes’ because ‘they belong to involvement or

participation with others in inter-personal human relationship’ (Strawson 1968: 79).
He contrasts them with third-personal, ‘objective’ attitudes that are not held from the
perspective of relating to persons second personally.4
The core of Holton’s account is that trust is a participant attitude, held from the ‘par-
ticipant stance’ (Holton 1994: 66). This means that trust is an attitude we have from the
perspective of implied relationship with the person we trust (perhaps one that trust
itself initiates and, indeed, that the one trusted might find unwelcome or resist).
Holton’s diagnosis of the difference between trust and reliance, then, is that trust is
reliance from the participant stance. Feeling betrayed and hurt feelings are Strawsonian
reactive attitudes, and the disposition to have these responses when our reliance is
frustrated is a marker of the participant stance from which these responses, and there-
fore our trust, is held (Holton 1994: 66; see also Faulkner 2007).
We can now see how once we interpret the expectational element of Jones’ account
as an expectation of, we arrive at essentially the same result. On a Strawsonian analysis,
expectations of implicate dispositions to reactive attitudes and, therefore, the partici-
pant stance. Expectations of someone on whom we rely that he will be moved by the
fact of our reliance is reliance on his being so moved from the participant stance.
For Strawson, again, the participant stance is the standpoint ‘of involvement or
participation with others in inter-personal human relationship’ (Strawson 1968: 79).
Interpersonal relations do not merely have persons as logical relata. They are consti-
tuted by people relating to one another from what I call the ‘second-person standpoint’
(Darwall 2006). As I have argued in a variety of places, the objects of reactive attitudes
are therefore implicit addressees; the attitudes implicitly address their objects and
invite or demand reciprocation in some way; they come with an implicit RSVP
(Strawson 1968: 85; Darwall 2006, 2013a, 2013b).
4
For an extended analysis, see Darwall (2006).
It follows that if Holton is right, as I believe he is, that trust is a participant attitude,
then trust is a second-personal attitude we hold from the perspective of implied
relationship to him. My object in this chapter is to elaborate trust’s distinctive features
as a second-personal attitude and to consider how trust differs from second-personal
attitudes of other kinds.
We have already seen how the second-personal character of trust is implicit in
Holton’s and Jones’ accounts, at least on the most promising interpretation of the lat-
ter’s expectational element. It is implicit also in a central formulation of Baier’s: ‘letting
other persons (natural or artificial) take care of something the truster cares about,
where such “caring for” involves some exercise of discretionary powers’ (Baier 1986:
240). ‘Letting’ and ‘discretionary powers’ both implicate what moral philosophers call
‘normative powers’, that is, powers persons can exercise to change obligations they
have to and rights they hold against one another: their respective ‘directive’ or ‘bipolar’
obligations, on the one hand, and ‘claim rights’, on the other (Raz 1972; Hohfeld 1923).
For example, promises create bipolar obligations of the promiser to the promisee and
correlative claim rights that the promisee holds against the promiser. The power to
create these rights and obligations by promising is thus a normative power.
Similarly, to let someone take care of something is, inter alia, to consent to his doing
so. Consent, like promise, is a normative power. And consent of this kind can only
make sense against the background of assumed bipolar obligations and claim rights.
By consent, I give up a claim right I would otherwise have had and thereby make
others’ actions that would otherwise have violated that right permissible. I give others
a discretionary power to do things and to exercise their own judgement in doing so. If
trust involves letting and the giving of discretionary power, therefore, it requires a
background of relationships governed by normative powers, therefore of obligations
owed to and claims against one another.
In Darwall 2011 and 2012, I argue that the ideas of normative power, bipolar obliga-
tion, and claim rights are all second-personal concepts since they all implicate the
notion of justified reactive attitudes (and other forms of powers of address) that involve
the second-person standpoint. For example, a promisee has an authority or standing
others do not have, as the individual to whom the promise was made, to address the
promiser in certain ways: to release her from the promise, to hold her personally
accountable if the promise is violated, for example, by objecting or seeking compensa-
tion, by resenting the violation, by forgiving it, and so on (Darwall 2011). Similarly, if
one who is trusted acquires the discretionary power to use his own judgement in car-
ing for something as the truster has let him do, he also thereby gains standing to relate
to the truster in distinctive ways—to object and/or respond with reactive attitudes—
should the truster fail to respect this power by not letting him exercise it in fact
(Darwall 2006, 2013a). Also, we might think, if the truster has not merely divested her-
self of the power and transferred it to the trusted, but trusted him to ‘take care’ in exer-
cising it, then she will take herself to have warrant for reactive attitudes, from a
reciprocal participant stance, if the power is not exercised carefully. On this analysis,
40 Stephen Darwall
then, Baier’s account ends up rendering trust as a participant attitude also. So all
three of Holton’s, Jones’, and Baier’s accounts of trust, therefore, either explicitly or
implicitly bring trust within the family of Strawsonian participant attitudes.
2. Participant, Reciprocating Attitudes: Deontic

and Non-Deontic
I take it as common ground among these three major accounts, consequently, that trust
is a participant, and therefore a second-personal, relationship-implying attitude. In
what follows my aim will be to try to say how trust differs from other second-personal
attitudes. To anticipate, the second-personal attitudes with which I have been mostly
concerned in my recent work in moral and legal theory, such as resentment, blame, and
guilt, are all implicitly deontic or juridical attitudes (Darwall 2006, 2013a, 2013b). They
implicitly make claims or demands, presuppose the authority to do so (in relation to
assumed deontic or juridical norms), and hold their objects (implicit addressees)
accountable for complying with these. I will be arguing, however, that trust, or at least
the form trust takes in personal relationships (personal trust), is not a deontic attitude.5
Unlike deontic attitudes trust neither presupposes nor entails any authority to demand
that the trusted act as she is being trusted to do, nor consequently any authority to hold
her accountable for doing so. In this way it is like love, to which it is intimately related.
When we give others our love or friendship, or more generally, when we place our trust
in them, we lay ourselves open to them in distinctive second-personal ways that do not
thereby involve the authority to make claims and demands of them.
To be clear, I am not arguing that there are not cases, for example, genuine betrayals
of trust, that can justify deontic reactive attitudes, like resentment.6 My claim is rather
that unlike deontic attitudes like blame, the attitude of trust does not itself presuppose
any standing to make demands or holds its object accountable. In cases of betrayal
where resentment and blame are warranted, that will be owing to some other element
that triggers some obligation of relationship, say, that someone actively encouraged
our trust. Although trust does presuppose some second-personal standing, and can
therefore be felt as an imposition when it is not welcomed, since it puts the trusted in
the position of failing to ‘live up’ to the truster’s aspirational expectations, such failures
do not, in themselves, justify resentment and blame, so much as other more personal
responses like being ‘let down’ or some other form of personal hurt. Trust, I shall argue,
is a ‘second-personal attitude of the heart’.
5
Faulkner (2014) argues similarly that trust is not second-personal in the sense of Darwall (2006) as
entailing mutual accountability. In effect, I am arguing here that there is a more extensive sense of second
personality as involving mutual reciprocation and that trust is second-personal in this more extensive
sense.
6
See, e.g., Hawley (2014). I am indebted here to an anonymous referee for the Press.
In this I am following the Danish theologian and philosopher Knud Ejler Løgstrup
who made trust a central feature of his anti-Kantian ethics. In Løgstrup’s pregnant
phrase, we ‘lay ourselves open’ with trust and put ourselves in one another’s ‘hands’
(Løgstrup 1997: 9).7 Like Løgstrup, I shall argue that trust involves a form of vulner-
ability that shows itself in distinctively personal responses like personal hurt rather
than putatively impartial responses like blame that presuppose violation of some
deontic norm.
Deontic or juridical attitudes are related to concepts like moral duty or obligation
(whether bipolar or not), wrong, right, rights, and so on. The concepts of moral obliga-
tion and wrong, I argue, are tied to the reactive attitude of moral blame (Darwall 2006).
It is not a substantive, but rather a conceptual, truth that if an action violates a moral
obligation (and is therefore morally wrong), then it is an action of a kind that would be
blameworthy were it to be done without excuse, where blame is understood as a
Strawsonian reactive attitude.
Deontic reactive attitudes like resentment and moral blame implicitly hold
their objects accountable for violations of a deontic standard (obligation or duty).
Accountability is a second-personal relation; it is always to someone or other. As
Strawson argues, when we hold someone accountable through what he calls ‘imper-
sonal’ reactive attitudes like moral blame, we implicitly make a demand of him as
representative persons or on behalf of the moral community (Strawson 1968: 85;
Darwall 2006). Similarly, ‘personal’ deontic reactive attitudes like resentment view
their objects as having violated some bipolar obligation owed to the resenter, correla-
tively, some claim right held against the resented, and hold the resented personally
accountable to the resenter as the very individual to whom the obligation was owed
(Strawson 1968: 85; Darwall 2012).
I argue that deontic accountability-seeking reactive attitudes come with an implicit
RSVP. Unlike third-personal critical attitudes, like contempt, they implicitly call their
objects to hold themselves accountable and take responsibility for their wrongful
actions, for example, with the reciprocal and reciprocating attitude of guilt. That is why
second-personal reactive attitudes like resentment and blame, but not third-personal
attitudes like contempt, carry the presuppositions of moral capacity and will (what
I call second-personal competence) that are the hallmarks of Strawson’s argument.8
It is also why sincere apologies can lead to forgiveness.9
Because the idea that second-personal attitudes are reciprocating attitudes will be
important also in our discussion of trust, it is worth pausing to develop the point
7
I am indebted here to Paul Faulkner. See the essays by Faulkner (ch. 7) and Robert Stern (ch. 16) in this
volume. See also Darwall (2016c).
8
On this point, see Watson (1987: 263, 264) on the presuppositions or ‘constraints of moral address’
that are implicit in reactive attitudes and also McKenna (2014) on the implicitly ‘conversational’ character
of reactive attitudes.
9
See Dill and Darwall (2014: 51).
42 Stephen Darwall
f urther here.10 There is a sense in which shame is a reciprocal attitude of contempt as

guilt is of blame. We can call two attitudes, A and B, reciprocals, in this sense, if, and
only if, for any persons X and Y, if Y is (part of) a fitting object of X’s attitude A, then B
is an attitude it would be fitting for Y to have.11 To feel shame is to feel as one would
seem to be to someone who views one with disdain or contempt. The view from
contempt is from above, looking down on someone or something seen as low, base, or
beneath one. Reciprocally, the view from shame is the view from below, but on oneself
as justifiably seen (from above) as low, base, or unpresentable. X’s contempt for Y is
fitting if, and only if, it would be fitting for Y to feel shame. So contempt and shame are
reciprocals. Similarly, guilt and moral blame are reciprocal attitudes also. An action is
blameworthy if, and only if, it would be fitting for the agent to feel guilt. So X’s blame
of Y is fitting, if, and only if, it would be fitting for Y to feel guilt.
But whereas contempt/shame and blame/guilt are both reciprocal pairs, there is a
different sense in which although blame calls for second-personal reciprocation,
whether by apology or feelings of guilt, and can be fitting only if its object is capable of
reciprocating by internalizing the blame as guilt, nothing parallel is true of contempt
and shame. Contempt does not implicitly call for second-personal reciprocation by its
object; neither does its fittingness depend on its object being able to internalize the
contempt as shame.
Guilt and blame are in this sense not just reciprocals; they (second-personally)
reciprocate one another. But shame does not reciprocate contempt in this sense. The
natural expression of guilt is confession and apology, whereas the natural expression of
shame is nothing second-personal; it is withdrawal. Contempt seeks implicitly not to
engage, but to exclude, its object (Mason 2003; Bell 2013). If, consequently, shame is
the feeling that contempt for one is fitting, it follows that the fitting response is to
exclude oneself, withdraw and hide one’s face from social view. Contempt does not call
for reciprocation; neither is it reciprocated by shame.
Now I shall presently be stressing ways in which trust, though a second-personal
attitude, differs from attitudes like blame, resentment, and guilt in not being a deontic
attitude. But I shall nonetheless be arguing that trust shares deontic attitudes’ recipro-
cating character even so. Like deontic attitudes, trust calls for a kind for reciprocation
by the trusted. The point is not so much that trust calls for the trusted to trust the
truster in ways that mirror the initial trust. If I trust you with my baseball card collec-
tion, I do not necessarily implicitly invite you to trust me with things that are comparably
dear to you, although in many cases that would be a natural progression of relation-
ship. But it does seem essential to my trusting you that I invite you to accept my trust
and, indeed, that I invite you to trust that I am indeed trusting you, to trust in my trust
and in me, trusting you. It will turn out that trust is a reciprocating attitude to itself.
Trust always necessarily invites trust in return.
10
For further discussion, see Darwall (2016a).
11
I used ‘fitting attitude’ in the sense of D’Arms and Jacobson (2000).
3. Trust as a Non-Deontic Attitude: A Contrast

with Promising
We turn now to investigating ways in which trust differs as a second-personal attitude
from deontic attitudes. A useful way to begin is to consider a recent proposal that links
trust to a deontic normative power that, at least on my analysis, is conceptually tied to
deontic second-personal attitudes and moral accountability, namely Daniel Friedrich
and Nicholas Southwood’s account of promising as an invitation to trust (Friedrich
and Southwood 2011). Friedrich and Southwood call this the ‘trust view of promissory
obligation’.
There are many ways in which promise and trust are intimately related that make
trying to understand the notoriously vexing phenomenon of promising in terms of
trust attractive.12 Both promise and trust call for a kind of uptake; both call for accept-
ance and, if this is not forthcoming, the relation is either cancelled or rendered ques-
tionable. And frequently, and perhaps normally, promisees trust promisers to keep
promises that have been made to them. Nevertheless, there are also ways in which
the relationship between promiser and promisee is importantly different from that
between trusted and truster.
The promiser/promisee relation is a deontic relation; more specifically, a relation of
right. Promisers give their promisees a claim right to what they have promised. And if
the promise is broken, the promisee acquires the authority to hold the promiser per-
sonally accountable, to resent the injury, to seek compensation, and so on. Trust, how-
ever, is not a deontic relation in this sense. Those who trust, even when their trust is
accepted, do not normally acquire similar rights.
If you promise me to call the doctor when you are sick, to take care of yourself, and
get good rest, then I acquire a promisee’s rights to your doing so and I can hold you
accountable if you do not. But if I simply trust you to do these things, I do not seem to
acquire similar rights, even if you do not reject my trust. If you allow me to trust you,
I no doubt have some justification for such responses as feeling let down or hurt
personally if you do not take good care of yourself. In trusting you, I make myself
vulnerable to you personally in a way that justifies responses like these.
To make the difference vivid, suppose I ask you to promise that you will take care of
yourself and that you respond that you would prefer not to be obligated to me in this
way, but say also that I am welcome to trust you to take good care. Since you will have
explicitly rejected my attempt to obligate yourself, it seems clear that if I do trust you
and you do not take care of yourself, I will have no justification for attempting to hold
you accountable with an attitude like resentment or moral blame.13 Still, in permitting
me to ‘lay myself open’ to you through trust, you will have made any failure on your
12
In Darwall (2011), I provide an account of promising as an instance of a second-personal account
of normative powers.
13
I am indebted here to Juan Piñeros Sanchez and to an anonymous reviewer for the Press.
44 Stephen Darwall
part to take good care to be an occasion, not just for disappointment that you so failed,
but the more personal disappointment of being ‘let down’.
What I have said so far about the difference in the relations between truster and
trusted, on the one hand, and promisee and promiser, on the other, is not, however, an
argument against the trust view of promising, since it holds not that being a promisee in
relation to a promiser is the same thing as trusting her, but that being an accepter of an
invitation to trust amounts to being promised something. Consistently with the differ-
ences between promise and trust we have noted, it might nonetheless be the case that
the obligation that a promiser has to a promisee can be accounted for by the fact she has
issued an invitation to trust that the promisee has accepted (or at least not rejected).
But what sort of obligation is generated by an invitation to trust, and how is it related
to a promissory obligation? Inviting is obviously a second-personal transaction
(like promising); it is issued to someone who has the standing to accept or reject it
(Darwall 2011). And promising and inviting to trust also have the additional element
that accepting them puts one into a second-personal relation to the person who issues
them. Promisees have a distinctive second-personal standing in relation to promisers,
and if trust is, as we have seen, a second-personal attitude, it follows that accepting an
invitation to trust puts the invitee in a distinctive second-personal relation to the
person issuing the invitation also.
But is it the same second-personal relation? We have already seen some reason to
think that although accepting a promise gives the promisee a claim right against the
promiser, it is not the case that trusting is itself a relation of right. If I trust you to care of
yourself when you are sick, I do not thereby acquire a right to your doing so of the kind
that would have been generated if you had promised, as, again, you could make clear
by refusing to promise but saying you would not reject my trust.
But again, this does not refute the trust theory of promise, since even if trusting
does not generate claim rights, perhaps accepting an invitation to trust does. But how
would it? In general, the only rights that accepted invitations generate are to engage in
whatever form of activity or interaction to which one has been invited. If you invite me
to dance with you, then perhaps you give me some right to do so. So far, this would
mean that an invitation to trust would generate a right of the invitee to trust the inviter.
Now this is not nothing. As we have seen, trust always implies or implicitly initiates
some relation with the person trusted. And because this is so, trust can be unwelcome
or even rejected by someone who does not want to be in that relation. It is hard to see
how Kant could have been in a position to object to his neighbours’ relying on his regu-
lar habits to set their watches, but he might well have been to object to their trusting
him to maintain them. We can well imagine him objecting that they had no right to
trust him, thereby putting him in the position of letting them down and giving them
reason to have towards him the participant reactive attitudes that are distinctive of
trust. He could complain that their trust was presumptuous.
If, however, Kant had invited his neighbours’ trust, then he could not make this
complaint. It would not strain language, too much anyway, to say that he had given
them a right to trust him and so made justifiable whatever participant responses are
distinctive of trust if he were to diverge from his usual perambulatory patterns.
But although the relation Kant’s invited neighbours would have to him would be like
the promisee/promiser relation in involving a distinctive second-personal standing, it
would not be same second-personal standing. By inviting his neighbours to trust him
to keep to his normal patterns, Kant would not necessarily thereby have given them
any right to his actually maintaining them, or, indeed, to expect him to do so in any
juridical or deontic sense that might justify resentment or moral blame. The only right
he would have given them would be to trust that he would do so.
We can well imagine Kant saying to his neighbours: ‘Look I’m not promising I will
continue my regular patterns, and so I’m not giving you any right to my doing so, but
I invite you to trust that I will.’ For this to be a genuine invitation to trust, we must
imagine that Kant is inviting his neighbours not simply to rely on his regularity, but to
have some combination of the second-personal attitudes towards him that Baier,
Jones, and Holton have identified as distinctive of trust. These might include, as Jones
argues, along with an optimistic attitude towards Kant’s continued regularity, some
expectation of Kant that he will be motivated by the knowledge that his neighbours are
counting on him to do so. In inviting his neighbours to trust him, Kant would be giving
them standing to feel disappointed in him or to take it personally were he to fail to keep
to his regular patterns or, at least, were he to fail to be sufficiently moved by the know-
ledge that his neighbours were counting on him. But given his explicit demurral,
accepting this invitation to trust would still not seem to give them a promisee’s rights.
It would not seem to warrant the distinctively deontic reactive attitudes of resentment
or moral blame or to justify a juridical claim for compensation.
It is consistent with this, of course, that there can be ways of inviting trust that create
deontic relations that are indistinguishable from that between promisor and promisee.
If, for example, Kant were to say, ‘Trust me, I will keep to my normal routine’, then this
would not be simply the issuing of an invitation to trust. It would amount to giving an
assurance and a second-personal transaction of that kind is virtually indistinguishable
from a promise (Darwall 2011).
One way of seeing the difference between promising and the issuing of an invitation
to trust is to notice that despite the fact that promisees frequently trust their promisors
to keep their promises, one can accept a promise without thus trusting. It is possible,
indeed, to accept a promise while relying even on the promisor’s breaking the promise.
This might be criticizable in at least some cases as showing bad faith, but it is nonethe-
less clearly possible. One may want to show that the promisor’s word is worthless and
bet the ranch on that’s being so. But while it is possible to accept someone’s promise
without trusting that the promise will be kept, it is not possible to accept an invitation
to trust without trusting. Nothing else could count as accepting such an invitation.14
14
I am indebted here to discussions with Stewart Cohen and other participants in a workshop at the
University of St Andrews in May 2015.
46 Stephen Darwall
Unlike the relation between promiser and promisee, that between truster and
trusted is not a(n impersonal) relation of right. It is a kind of personal relationship in
which truster and trusted make themselves vulnerable to one another personally rather
than juridically. Promise is, as Daniel Markovits has argued, an ‘arm’s length relation’
(Markovits 2011). It may be too strong to say, as Markovits does, that ‘the formal struc-
ture of promising is in an important way incompatible with the formal structure of
recognition among intimates’ (Markovits 2011: 295). As Seana Shiffrin has pointed
out, intimates sometimes need to avail themselves of the kinds of rights and obliga-
tions that promise involves (Shiffrin 2008; see also Darwall 2010). Even so, it seems
clear that the kind of mutual recognition that is most distinctive of personal relation-
ships like friendship and love is not intrinsically deontic or juridical. It is a mutual
openness to forms of personal relation, including trust, that we have no authority to
demand of one another, but that we nonetheless hope for and are understandably
personally hurt or let down by when it is not reciprocated.15
To see the point, consider our attitudes towards fellow motorists with whom we
have no personal relationship. There is here a kind of reliance from the participant’s
stance on one another’s goodwill and at least minimally responsible driving. We take
ourselves to be entitled to expect of other drivers that they will drive responsibly and to
be justified therefore in objecting and having second-personal holding-accountable
attitudes like resentment and moral blame when they do not. But when, however, one
is cut up by an overly aggressive driver who is a stranger, it would seem odd to have
participant responses that are distinctive of trust, to feel let down or disappointed in
the driver, or personally hurt or betrayed. These would seem to imply some form of
personal relation that would not exist in such a case.
4. Trust as a Second-Personal Attitude of the Heart

Considering the trust theory of promising has enabled us to appreciate how the truster/
trusted relation differs from the promiser/promisee relation, even when the former
results from an accepted invitation to trust. The central difference, for our purposes, is
that whereas the latter is a deontic relation of bipolar obligation and claim right, the
former is not. My thesis is that trust is a species of second-personal attitude through
which we lay ourselves open to others in a way that is distinctive of personal relation-
ship and attachment.16 It is, we might say, an ‘attitude of the heart’, using ‘heart’ in its
customary metaphorical sense to refer to that aspect of the human psyche through
which we are heartened or disheartened, inspired or deflated, encouraged or discouraged,
15
On this point, see Darwall (2016b). It is consistent with all this, of course, that personal relationships
frequently involve personal obligations that do warrant legitimate demands, even if the attitudes of love
and trust that structure these relations do not themselves presuppose mutual accountability for compliance
with such demands.
16
For an excellent discussion of the ways in which attachment involves emotional susceptibilities to
regret, see Wallace (2014).
filled with hope and joy or deflated with despair, emptiness, or sadness. This is most
obvious when we put something personal at stake, if only our own feelings, in placing
our trust in others. But I believe that a version of the same phenomenon occurs even in
epistemic cases when we trust what others tell us, especially when we take it to heart.
Perhaps the clearest case of a non-deontic second-personal attitude of the heart is
love of the kind that, as Strawson puts it, ‘two adults can sometimes be said to feel
reciprocally, for each other’ (Strawson 1968: 79).17 Strawson classes reciprocation-
seeking love as a reactive attitude because it is felt from a second-person standpoint
(see also Abramson and Leite 2011). Avowals of love formulated with the second-person
pronoun do not purport simply to transmit information that could in principle
be expressed or appreciated third personally. They are essentially second-personal
transactions—confessions, declarations, or professions of love. Love is most naturally
expressed in such a second-personal way that seeks uptake and reciprocation, thereby
constituting a loving relationship.
It should be obvious, however, that love is not a deontic or juridical attitude. Unlike
blame and resentment, which presuppose some authority to make claims and demands
of others and hold them accountable for compliance, love, even in friendship, is noth-
ing we can claim from someone or hold her accountable for. Love is like God’s grace,
something we can neither earn nor deserve but that can only be freely given as the
gift of an open heart. This does not mean that there are no distinctively personal
obligations in relationships of love and friendship. It just means that there can be no
obligation to have feelings or the attitude of love itself, although there may well be to do
what one can to nurture and cultivate them.
Blame and resentment seek reciprocation by attitudes and conduct that express
respect for the authority to claim they implicitly presuppose. But whereas respect can
arise by reciprocation through acceptance of the claim for respect, love cannot arise
through acceptance of a claim for love. Even if there could be a legitimate claim to love,
the most that could arise through accepting the claim’s legitimacy would be respect for
someone’s authority to make it and for them as having this authority. And this helps us
see why there cannot be a legitimate claim to love itself, even if there can be legitimate
claims to forms of conduct that people in a loving relation warrantedly expect of one
another. Love is nothing that could be given as a result of accepting a valid claim; love
cannot arise from respect. Love is an attitude of the heart that is not up to us in the way
that respect for someone as an equal moral person is.
Love is perhaps the quintessential attitude of the heart. Love lays the heart open to
another heart, hoping it will be openly received, seeking love in return, and making us
vulnerable to the other in ways that can fill our hearts with the joy of love’s requital or
bring heartache or heartbreak. Strawson mentions ‘hurt feelings’ as a kind of reactive
attitude (Strawson 1968: 75) since it is an attitude to which we are susceptible from the
participant stance when we relate to others personally in these ways.
17
In the next few paragraphs, I draw from Darwall (2016b).
48 Stephen Darwall
That trust is also, like love, a second-personal attitude of the heart can be seen by its
implication in love. Lovers and friends entrust their hearts and feelings to one another.
Though they have no legitimate claim to their love or friendly feeling’s return, they can
hope and even trust that it will be, and are naturally disposed to feel hurt and personal
disappointment when it is not. But trust reveals itself as an attitude of the heart even
when it is not an expression of love.
Trust is a form of confidence in someone and therefore a source of encouragement.
Someone can be heartened by our trust, both moved and encouraged by it. Similarly,
others can be disheartened by our failure to trust them. And if our trust is accepted but
then not fulfilled, or worse, if though accepted, it is trifled with, then we are likely to
feel hurt and disheartened as well. On the other hand, if it is fulfilled, then this can
buoy both our spirits and the one we trusted reciprocally also. (So we might as easily
call trust and love ‘attitudes of the spirit’.)
Of course, things may be more complicated. Trust can be welcomed in some ways
but burdensome in others: a source of confidence and encouragement, on the one
hand, and a yoke of expectation on the other. But personal relationships are compli-
cated in just these ways. And that trust can be so also is simply a reflection of the role
trust plays as an attitude of the heart in constituting personal relationships.
Earlier I introduced the idea of reciprocating attitudes and suggested that trust was
like deontic attitudes in calling for reciprocation. When we trust, I said, we invite the
person we are trusting to accept our trust and trust in it, to trust that we are indeed
trusting him. We so easily take this for granted that it does not occur to us. But some-
one may especially need our trust when he is insecure and suspicious, both of his own
abilities and character but also that others could be well disposed towards him. Indeed,
his suspicions may extend to others, doubting that others can be trusted to trust or, at
least, they can be trusted to trust him. Like love, then, trust aims at mutuality and
invites it second-personally.
Nor is this the end of it. When we trust someone, we implicitly invite him to trust
himself also. We regard him as trustworthy and bid for him to see himself this way too.
Something analogous is true of love. To love someone is to see him as lovable and to
invite him to see himself that way as well. Someone’s not loving himself is particularly
disturbing to those who love him, not just because of its effects on his well-being, but
because it can make it difficult, if not impossible, for him to accept their love.
The analogy carries through to deontic reactive attitudes also. When we hold some-
one accountable through an attitude like moral blame, we bid for her to hold herself
accountable as well. Moreover, if we judge that the person is not capable of holding
herself accountable, say because she lacks the capacity to feel the reciprocating attitude
of guilt, then we cannot justifiably, or perhaps even intelligibly, hold her accountable
through moral blame.18 And a version of this analogy carries through to trust. If we
know that someone does not trust herself, it will at least be more difficult for us to trust
18
This is what I call ‘Pufendorf ’s Point’ in Darwall (2006).
her. That we nonetheless can, and that she can be encouraged by this to move towards
self-trust, is yet a further indication of an attitude of the heart.19
References
Abramson, Kate and Adam Leite (2011). Love as a Reactive Emotion, Philosophical Quarterly
61: 673–99.
Baier, Annette (1986). Trust and Antitrust, Ethics 96: 231–60.
Bell, Macalester (2013). Hard Feelings: The Moral Psychology of Contempt. Oxford: Oxford
University Press.
D’Arms, Justin and Daniel Jacobson (2000). Sentiment and Value, Ethics 110: 722–48.
Darwall, Stephen (2006). The Second-Person Standpoint: Morality, Respect, and Accountability.
Cambridge, MA: Harvard University Press.
Darwall, Stephen (2010). Responsibility Within Relations, in Partiality and Impartiality:
Morality, Special Relationships, and the Wider World, ed. Brian Feltham and John Cottingham.
Oxford: Oxford University Press. Reprinted in Darwall 2013b.
Darwall, Stephen (2011). Demystifying Promises, in Promises and Agreements: Philosophical
Essays, ed. Hanoch Sheinman. Oxford: Oxford University Press. Also in Darwall 2013a.
Darwall, Stephen (2012). Bipolar Obligation, in Oxford Studies in Metaethics, v. vii, ed. Russ
Shafer-Landau. Oxford: Oxford University Press, pp. 333–67. Also in Darwall (2013a).
Darwall, Stephen (2013a). Morality, Authority, and Law: Essays in Second-Personal Ethics I.
Darwall, Stephen (2013b). Honor, History, and Relationship: Essays in Second-Personal Ethics II.
Darwall, Stephen (2016a). Empathy and Reciprocating Attitudes, in Forms of Fellow Feeling, ed.
Neil Roughley and Thomas Schramme. Cambridge: Cambridge University Press.
Darwall, Stephen (2016b). Love’s Second-Personal Character: Holding, Beholding, and
Upholding, in Love’s Reasons, ed. Esther Kroeker and Katiren Schaubroeck. New York, NY:
Routledge.
Darwall, Stephen (2016c). Løgstrup on Morals and ‘The Sovereign Expressions of Life’, in What
is Ethically Demanded? Essays on Knud Ejler Løgstrup’s ‘The Ethical Demand’, ed. Hans Fink
and Robert Stern. Notre Dame, IN: University of Notre Dame Press.
Dill, Brendan and Stephen Darwall (2014). Moral Psychology as Accountability, in Moral
Psychology and Human Agency, ed. Justin D’Arms and Daniel Jacobson. Oxford: Oxford
University Press.
Faulkner, Paul (2007). On Telling and Trusting, Mind 116: 875–902.
Faulkner, Paul (2014). The Moral Obligations of Trust, Philosophical Explorations 17: 332–45.
Friedrich, Daniel and Nicholas Southwood (2011). Promises and Trust, in Promises and
Agreements: Philosophical Essays, ed. Hanoch Sheinman. Oxford: Oxford University Press.
Hawley, Katherine (2014). Trust, Distrust, and Commitment, Noûs 48: 1–20.
19
I am grateful to Paul Faulkner, Thomas Simpson, Juan Piñeros Sanchez, participants in a workshop on
promising organized by Hallie Liberto at the Center for Human Values at Princeton University in February
2015, as well as philosophers at the University of St Andrews in May 2015, especially Katherine Hawley and
Stewart Cohen, and two anonymous reviewers for the Press.
50 Stephen Darwall
Hohfeld, Wesley Newcomb (1923). Fundamental Legal Conceptions, ed. Walter Wheeler Cook.
New Haven, CT: Yale University Press.
Holton, Richard (1994). Deciding to Trust, Coming to Believe, Australasian Journal of
Hume, David (1985). An Enquiry Concerning Human Understanding, in Enquiries Concerning
Human Understanding and Concerning the Principles of Morals, ed. L. A. Selby-Bigge, 3rd
edn., rev. P. H. Nidditch. Oxford: Clarendon Press.
Jones, Karen (1996). Trust as an Affective Attitude, Ethics 107: 4–25.
Løgstrup, Knud Ejler (1997). The Ethical Demand. Notre Dame, IN: University of Notre Dame
Press.
McKenna, Michael (2014). Responsibility and Conversation. Oxford: Oxford University Press.
Markovits, Daniel (2011). Promise as an Arm’s Length Relation, in Promises and Agreements:
Philosophical Essays, ed. Hanoch Sheinman. Oxford: Oxford University Press.
Mason, Michelle (2003). Contempt as a Moral Attitude, Ethics 113: 234–72.
Raz, Joseph (1972). Voluntary Obligations and Normative Powers, II, Proceedings of the
Aristotelian Society. Supp. Vol. 47: 79–102.
Shiffrin, Seana (2008). Promising, Intimate Relationships, and Conventionalism, Philosophical
Review 117: 481–524.
Strawson, P. F. (1968). Freedom and Resentment, in Studies in the Philosophy of Thought and
Action. London: Oxford University Press.
Wallace, R. Jay (2014). The View From Here: On Affirmation, Attachment, and the Limits of
Regret. Oxford: Oxford University Press.
Watson, Gary (1987). Responsibility and the Limits of Evil: Variations on a Strawsonian
Theme, in Responsibility, Character, and the Emotions: New Essays in Moral Psychology, ed.
F. D. Schoeman. Cambridge: Cambridge University Press.
4
On the Risks of Resting Assured
An Assurance Theory of Trust
Edward S. Hinchman
What do you risk when you trust? On some influential accounts, trust distinctively
risks not mere disappointment but betrayal.1 What exactly is it for trust to be betrayed?
If the trusted simply fails to do what you’ve trusted her to do, that looks like an occa-
sion for mere disappointment. Though there may be something in your relations with
the person that warrants your feeling betrayed, betrayal does not appear to lie in the
performative lapse as such. Where else may it lie? I’ll argue that the risk of betrayal lies
at a deeper level: in the risk that her action or inaction—whether disappointing or
not—will manifest a failure to engage your needs in the way that you’re trusting her to
engage them. Though trust includes reliance, and reliance aims not to be disappointed,
I’ll argue that the aim distinctive of trust is more complex and is compatible with dis-
appointment. Disappointed trust may thus fall short of betrayed trust, and betrayed
trust may not derive from disappointed trust. It is betrayed trust, not disappointed
trust, that violates the mutual understanding at the core of interpersonal trust. And it is
betrayed self-trust, not disappointed self-trust, that violates the self-understanding at
the core of intrapersonal trust. To understand this normative structure at the core of
both social and personal agency, we must understand how each form of trust distinc-
tively risks not mere disappointment but betrayal.
What is it to risk betrayal of your trust? Annette Baier argues that the risk is moral:
The assurance typically given (implicitly or explicitly) by the person who invites our trust,
unlike that typically given in that peculiar case of assurance, a promise or contract, is not assur-
ance of some very specific action or set of actions, but assurance simply that the trusting’s
welfare is, and will one day be seen to have been, in good hands. (Baier 1994: 137)
Baier thus posits a contrast between an invitation to trust and a promise. I reject this
contrast, but I more fundamentally reject her moral emphasis. Here I side with those
1
See, for example, Baier (1994: Chapters 6–9); Holton (1994); Jones (2004); Walker (2006: Chapter 3);
Hieronymi (2008); McGeer (2008); McMyler (2011: Chapter 4); and Hawley (2014).
52 Edward S. Hinchman
who have argued against moralizing trust.2 But these critics also reject Baier’s emphasis
on betrayed as opposed to disappointed trust. The critics link their rejection of what
they regard as moralism with a claim that the risk of betrayal adds nothing, as such, to
the risk of disappointment, thereby omitting the interpersonal element that Baier
marks with the concept of an assurance. Baier is on the right track in emphasizing
betrayal, I’ll argue, but the assurance at the core of an invitation to trust targets the
trusting’s rationality, not the trusting’s welfare or any other distinctively moral status.
I do not emphasize rationality to the exclusion of morality: I’m going to argue that a
promissory assurance gives rise to a promissory obligation insofar as it targets the
promisee’s rationality, and a promissory obligation is paradigmatically moral. I claim
merely that the rational obligation is more fundamental. Though it does not follow
from how trust risks betrayal that trust is a moral relation, it does follow from how
trust risks betrayal that trust is a rational relation.
Since it is easier to understand how the obligation is thus fundamentally rational in
the context of our self-relations, I’ll explain the rational obligation at the core of inter-
personal trust by developing an analogy with intrapersonal trust. Both species of trust,
interpersonal and intrapersonal, differ from mere reliance insofar as they underwrite
distinctive rational relations. The analogy marks how betrayal reflects the form of
understanding at the core of any trust relation. Interpersonal trust underwrites a form
of interpersonal reasoning that rests on a non-evidential mechanism whereby one per-
son can make a reason available to another—‘non-evidential’ because the reason is not
grounded in evidence of reliability. Intrapersonal trust underwrites the medium
wherein an individual person maintains enkratic rational coherence.
In each case, trust is betrayed, rather than merely disappointed, when one party—
whether a person or an aspect of one’s own self—violates the shared understanding
that generates norms governing the trust relation. Interpersonal trust is betrayed when
the trusted violates the shared understanding of how the trust matters to the trusting,
where how the trust matters reflects how the trust provides or fails to provide the
trusting with reasons. And intrapersonal trust is betrayed when the trusted—an aspect
of one’s own self—warrants akratic self-mistrust, thereby violating one’s own under-
standing of the requirements of enkratic coherence. In speaking of a rational relation,
I don’t claim that reasons and rational requirements amount to the same thing. I’ll
merely use each to cast explanatory light on the other, treating rationality as a broad
category that includes both (if you want, disjunctively). The understanding at the core
of both species of assurance, conceived as shared between parties to the trust relation,
informs that relation in this key respect: when the understanding is violated the trust
is betrayed.
The chapter has the following structure. In section 1, I’ll develop examples that
vindicate the possibility of betrayed though undisappointed trust. In section 2, I’ll show
what that possibility reveals about the rational relations invited by an interpersonal
2
See, for example, Hardin (2002: Chapter 3); Nickel (2007, Section 6); and Rose (2011, Chapter 9).
on the risks of resting assured 53
assurance. In section 3, I’ll develop examples that vindicate the possibility of unbe-
trayed though disappointed trust. In section 4, I’ll show what that reveals about the
rational relations invited by the intrapersonal assurance at the core of an intention. In
section 5, I’ll generalize the conclusion of section 1: the self-trust that informs intra-
personal assurances can be betrayed without being disappointed. In each kind of case,
whether interpersonal or intrapersonal, the key to understanding the nature of trust
lies in understanding how trust makes you vulnerable to the risk of betrayal.
1. Trust Undisappointed yet Betrayed

Though betrayal may bring its own disappointments, the concept of betrayed though
undisappointed trust is perfectly coherent. Let me illustrate with some simple cases—
cases that do not, I’ll stipulate, involve any explicit or implicit assurance. These cases will
mark a useful contrast with the more complex cases that we’ll consider in later sections.
To get us started, consider this case:
Lovely Leaf. Andrew and Bernice are out hiking, and Andrew finds a leaf that
strikes his fancy. Nursing it in his palm, Andrew asks Bernice to keep the leaf in her
backpack, since Andrew’s pack is full. ‘Why?’ she asks. ‘No particular reason,’ he
replies. ‘I merely like it.’ Bernice accepts the leaf and places it in her pack. Though
Bernice makes no explicit or implicit promise (let’s stipulate), Andrew nonetheless
trusts her to take good care of the leaf.
Will Bernice betray Andrew, or betray his trust, if she fails to take adequate care of the
leaf? The obvious answer is that it depends on how she fails. If she underestimates the
leaf ’s fragility and tucks it a bit too tightly against the tent poles, that would reflect on
her competence and thus on her worthiness of Andrew’s reliance, but it is difficult to
see how that would amount to any betrayal of Andrew. How might we get a betrayal?
An obvious betrayal emerges if we simply imagine Bernice’s incompetence willed:
Lonely Leaf. As in Lovely Leaf, except that Bernice places the leaf on the ground
next to her pack, instead of in her pack, confident that Andrew will not notice that
she has left it behind and will soon forget about it.
Bernice thereby betrays Andrew’s trust, since she doesn’t even try to do what he is
trusting her to do. But what if she not only tries but succeeds in doing it? Could Bernice
nonetheless betray Andrew’s trust? Consider this case:
Less Lovely Leaf. As in Lovely Leaf, except that when Bernice places the leaf in her
pack, she thinks to herself ‘What an idiot—I’ll shut him up by putting the leaf in here
with my stuff where it probably won’t come to any harm.’
One might think that this could not amount to betrayed trust as long as Andrew is
not aware of Bernice’s contemptuous attitude. I’m not sure that’s right: why can’t her
betrayal simply be unknown to him? Let’s nonetheless set that issue aside by consider-
ing a realistic case in which Andrew is aware of Bernice’s attitude:
Unlovely Leaf. As in Lovely Leaf, except that Andrew and Bernice are now hiking
back to their campsite in fading daylight. When he asks her to keep the leaf, Bernice’s
‘Why?’ reveals impatience. ‘And why,’ she adds, ‘are you occupying yourself with
such trivia when we’re trying not to get lost in the dark?’ ‘I like it,’ he replies. Rolling
her eyes in disgust, Bernice accepts the leaf and places it in her pack with sarcastic
gestures of play-acted ‘carefulness’.
Could Andrew think to himself ‘She’s contemptuous of my needs, but I nonetheless
trust her to keep the leaf safe from harm’? He could, if he regards the sarcasm as merely
meant for display. But if he regards her as genuinely contemptuous of his needs, or of
this particular stated need, then he cannot rationally or reasonably trust her with the
leaf—no matter how reliable he may regard her as a custodian of her pack, which he
now sees includes this precious cargo. Let me emphasize that this needn’t be a question
of how Bernice ‘really feels’ about Andrew: the question is how she is disposed to act,
not the true state of her feelings. If he regards her as disposed to act without appropri-
ate concern for his needs, then, while he can rationally rely on her to keep the leaf from
harm, he cannot rationally trust her with the leaf.
I do not claim that one couldn’t get away with describing Andrew as ‘trusting’
Bernice with the leaf. We sometimes say ‘trust’ when we merely mean ‘rely’—as when
we wonder whether to ‘trust’ this toaster with a slice of bread. I claim merely there is a
distinction worth drawing here, and that it makes good sense to draw it as the distinc-
tion between trust and mere reliance. If we challenge Andrew when he says that he
‘trusts’ Bernice with the leaf, it would make good sense for him to retreat to the claim
that he is relying on her. It would be irrational for him to trust her, over and above rely-
ing on her. Though the simple cases that we’re considering lack an assurance, I’ll argue
in section 2 that we need the concept of an assurance to understand the normative
point of this distinction.
Beyond mere reliance, trust appears to involve optimism about the trusted’s respon-
siveness to a subset of one’s needs—not optimism merely that the trusted will do what
one trusts her to do. But why should that be so?3 Is trust a moral relation, part of which
involves appropriately just recognition or acknowledgement of the trusting’s vulner-
ability to harm? It seems not. Andrew need not be worried about any moral dimension
of Bernice’s failure to recognize or acknowledge his needs. As I’ll explain in section 3, he
may be worried about a dimension of her performance: that she may fail to disappoint
his expectations of performance if such disappointment is what it takes, given an
unforeseen shift in his needs, for her not to betray his trust. In general, one may ration-
ally fail to trust where one doesn’t expect the trust to be disappointed because the trusted
3
Following Jones (2004), I don’t think this appearance actually reveals a necessary condition on trust.
But the appearance is plausible (for theories of trust that emphasize it, see Baier (1994) and Jones (1996)),
and a theory of trust should explain that plausibility.
may need to disappoint one’s trust in order not to betray it. We’re now considering
undisappointed yet betrayed trust, but in section 3 we’ll turn to unbetrayed yet disap-
pointed trust. This complexity will reveal how a trust relation can amount to a rational
relation that unfolds through time, thus paralleling a core dimension of intrapersonal
trust. Before we explore that complexity, we need to see how interpersonal trust rela-
tions more generally amount to rational relations. As I’ll now argue, they do so by serv-
ing as a medium whereby the trusted can make reasons available to the trusting.
2. Trust as the Medium of Interpersonal

Reason-Giving
On an assurance view of trust, as I conceive it, the distinction between trust and other
forms of reliance mirrors Paul Grice’s distinction between natural and non-natural
meaning. Grice drew that distinction by contrasting an evidentiary mechanism—‘Those
clouds mean rain’, ‘Those spots mean measles’—with a mechanism that works through
recognition of the speaker’s intentions. In non-natural meaning, Grice argued, a speaker
intends to give her addressee a reason to produce a certain response grounded not in
evidence of her reliability (though she may well expect or hope that there is such evi-
dence) or in any other evidentiary basis (though she may well believe that such evidence
is available) but specifically in the addressee’s recognition of her intention to give him
this reason. I’m not sure that Grice’s approach can yield a plausible account of meaning,
but I believe it provides the orienting insight for a powerful theory of the dimension of
human sociality that we mark with the word ‘trust’. What my approach inherits from
Grice’s is an emphasis on the distinctive way in which speech acts aim to give reasons: not
through an evidentiary mechanism but through a structure of mutual recognition and
understanding. What the assurance theory adds to this Gricean approach is the further
claim that such cases reveal how trust differs from mere reliance.
I focus not on speech acts in general but on propositional assurances: paradigmatic-
ally, testifying, advising, and promising. Observing, with Grice, that such assurances
aim to give a reason simply through the addressee’s recognition of that aim amounts to
observing that the speaker is inviting the addressee’s trust. Such an assurance is an
invitation to trust in this respect: S invites A to regard himself as having a reason to act
or believe grounded, in part, in how S undertakes an obligation in issuing the invita-
tion. In such a case, reliance is mere reliance unless it is thus invited by a propositional
assurance; thus invited, it can then serve as a basis of obligation, on the side of the
speaker, and practical or epistemic reasons, on the side of the addressee. The reasons in
question are grounded not in the trust relation itself but in the speaker’s status as rele-
vantly reliable. Adapting Grice’s terminology, we may say that while both ‘natural’ and
‘non-natural’ reason-giving depend on the speaker’s reliability, there is a crucial
difference in how they depend on reliability. In light of this difference, there is also a
further difference in what counts as relevant reliability. I’ll explain these differences in
turn, thereby explaining what it is for an assurance to ‘invite’ the addressee’s trust.
How could reliability ground a reason non-evidentially? First consider how

evidence of reliability could ground a reason, as in this new case:
Guide’s Assertion. Aaron knows that Bonnie is a trained wilderness guide, but he
also knows that she despises him and would never help him find the campsite where
necessary food and shelter await them. Fortunately, he doesn’t have to ask her: he
spies Bonnie ahead on the trail below him, and training his binoculars on her mouth
he lip-reads her as muttering under her breath that the campsite is to the north, via
the rightmost fork of the trail. Believing her reliable about the location of the camp-
site, he comes to believe what she thereby asserts. And he is correct in both beliefs:
she is reliable, and her assertion is true.
Here, reliance on another’s assertion does not amount to trust. How might we get an
instance of trust? Consider this emendation:
Guide’s Testimony. As in Guide’s Assertion, except that Bonnie now addresses
Aaron, looking him in the eye and telling him that the trail is to the north, and Aaron
trustingly believes her.
The same contrast emerges when we shift from testimony about the campsite’s location
to advice about how to reach it. First, the mere reliance case:
Guide’s Practical Assertion. As in Guide’s Assertion, except that Aaron lip-reads an
assertion not about the campsite’s location but about where Aaron ought to go: ‘That
lost-looking guy ought to keep heading north.’
And now the trust case:
Guide’s Advice. As in Guide’s Testimony, except that Bonnie makes the assertion in
Guide’s Practical Assertion, now addressed to Aaron.
Further modifications yield a parallel contrast for promising. First, a statement of
intention that does not amount to a promise:
Guide’s Assertion of Intention. As in Guide’s Assertion, except that Aaron lip-reads
an assertion of intention: ‘I’ve got to keep heading north, in order to get back to the
campsite before dark.’
And now the promise:
Guide’s Promise. As in Guide’s Testimony, except that instead of testifying to the
whereabouts of the campsite, Bonnie promises to take Aaron there.
In each of these three pairs of case, we have a contrast between the assurance and a
mere assertion. In each pair, Aaron can get reasons from the assurance and also from
the assertion. He can get an epistemic reason to believe that the campsite is to the north
in both Guide’s Assertion and Guide’s Testimony. He can get a practical reason to go
north in both Guide’s Practical Assertion and Guide’s Advice. And he can get planning
reasons—reasons to do or to plan to do things premised on Bonnie’s leading him to the

campsite—in both Guide’s Assertion of Intention and Guide’s Promise. That Aaron can
get these reasons does not distinguish the cases. What distinguishes them is how he
gets the reasons.
We can understand what’s distinctive of each species of assurance—testimony,
advice, promise—by contrasting the mechanism whereby it makes reasons available
with the evidential mechanism whereby reasons are made available in the mere
Assertion cases. One concise way to draw the contrast is to observe that each Assertion
case can be quasi-Gettierized: in each, it is possible for Bonnie to be relevantly reliable,
and for her assertion to be true, though Aaron’s reliance on her does not track this reli-
ability and therefore does not track the truth of the belief that he acquires by this
mechanism. Here is one way that could happen:
Quasi-Gettierized Assertion. As in Guide’s Assertion, with two new bits: (i) Aaron
believes Bonnie reliable because he believes that wilderness guides are trained to
‘sniff out’ campsites by the scent of their campfire, and (ii) as it happens Bonnie has a
sinus infection, though she nonetheless remains reliable. She cannot ‘sniff out’ any-
thing in her condition, but she does reliably judge the direction of the campsite by
visual acuity, drawing on a skill that she developed in her former career as a fire-
spotter for the forest service. Aaron has no idea that she was ever in the forest ser-
vice, or that she has such visual skills. It is thus mere luck that his belief that she is
reliable turns out to be true. This luck ensures that he fails to derive from her asser-
tion any actual reason—as opposed to the mere appearance of a reason—to believe
what she asserts.
Since Aaron relies on Bonnie’s assertion in order to form a belief, the case amounts to a
Gettier case of the classic sort: Aaron forms a justified true belief that fails to count as
knowledge because his justification fails to be related in the right way to what makes
the belief true. If Bonnie had been ‘sniffing out’ the campsite, as Aaron believes she is,
then his evidence for the reliability of her ability to ‘sniff out’ destinations would justify
his belief in a way that is appropriately related to what makes the belief true. But though
she is normally thus reliable—and therefore his belief is justified—what makes his
belief true runs through a different causal mechanism, which intuitively undermines
any claim he might make to knowledge.4
It is irrelevant to the case’s status as a classic Gettier case that Bonnie happens to be
reliable in a way that does not engage Aaron’s belief-forming mechanism. But it serves
my purposes to distinguish a category of quasi-Gettier cases with this structure: A does
not have a particular reason that he takes himself to have on the basis of B’s assertion
because, though he has good evidence that B is relevantly reliable, and B is relevantly
4
This appeal to reliability is not the appeal to reliability characteristic of reliabilism. The reliabilist
appeals to the reliability of Aaron’s—not Bonnie’s—belief-forming process. In Quasi-Gettierized Assertion,
Aaron’s belief-forming process is clearly not reliable.
reliable, A’s evidence does not track B’s reliability. In Quasi-Gettierized Assertion, this
reason is a reason to believe that the campsite is to the north, but we could produce
quasi-Gettierized versions of the two other Assertion cases as well, wherein Aaron
would (i) fail to have a reason to act in accordance with Bonnie’s practical assertion
that he ought to head north, and (ii) fail to have reasons to do or plan to do things
premised on the assumption that Bonnie is leading him to the campsite. As we
originally imagined the cases, we assumed that Aaron’s evidential relation to Bonnie’s
reliability gives him these reasons. But when we quasi-Gettierize the cases, we see that
he fails to have the reasons: she is reliable in a way that vindicates his reliance on her—
he is relying on her to guide him to the campsite, by reliably asserting the truth about
its location, or about what he ought to do, or about the state of her own intentions—but
not in a way that would give him the reason that he takes himself to have.
It is distinctive of the Assurance cases, by contrast, that they cannot be thus quasi-
Gettierized. When Bonnie turns to address Aaron—assuring him that the campsite is
north, or that he ought to head north, or that she will lead him there—we lose grip on
the idea that Bonnie could be reliable in a way that would vindicate Aaron’s reliance on
her, and yet not in a way that would provide the reason that he presumes she gives him.
Of course, he could fail to have that reason, but only because she is not reliable in a way
that would vindicate his reliance on her. That possibility is present in quasi-Gettierized
cases as well: Bonnie may simply prove an unreliable guide to the campsite. When we
shift from mere assertion to addressed assurance, we’re not talking about a breakdown
in her status as a reliable guide but in her status as a reliable interlocutor. The form of
address yields the crucial difference that Aaron is now relying on her not merely to lead
him to the campsite but to pull her weight in an interlocutory relation. Because of this
difference in the nature of Aaron’s reliance, the reason that he gets from Bonnie if she is
reliable must not be the same sort of reason as he got in the original Assertion cases. Of
course, in each case Bonnie’s assurance includes an assertion, so it may be that Aaron
gets both sorts of reason. But they would nonetheless remain different sorts of reason.
What other sort of reason could he get? I have characterized propositional assur-
ances as ‘invitations’ addressed to your interlocutor—‘invitations’ that he or she should
trust or rely on you in specific ways. It’s time to cash in that metaphor. When you give a
propositional assurance, you invite your addressee to rely on you as a source of reasons,
and you present yourself as worthy of precisely that species of reliance. How would you
be thus worthy? What grounds the reason, I’ll argue, is your undertaking an obligation
to do justice to your addressee’s needs in respects relevant to the understanding at the
core of the trust relation that you invite. Interpersonal trust can be betrayed, rather than
merely disappointed, because of the normative role played by this shared understand-
ing. When your addressee accepts your invitation to trust, the now-shared understand-
ing at the core of this trust relation comes with important normative consequences: if
you uphold your end of the trust relation by being relevantly trustworthy, your addressee
gets a reason (to believe through your testimony, to act through your advice, or to plan
through your promise); if your addressee upholds his end of the trust relation, by
trusting you in appropriate ways as determined by the understanding that you invite
him to share in trusting you (we’ll inquire into what such appropriateness comes to),
you count as undertaking an obligation to be thus trustworthy. One can betray a trust
relation because one can betray the shared understanding at its core—a normative
structure with attendant obligations and entitlements.
Let me schematize this core strand in my argument. To understand the possibility of
betrayal, I’ll argue, we must understand how the trust relation binds trusting and
trusted into something like a contract. My first thesis is that the basis of this contract-
like institution lies in the shared understanding of the point of the trust relation—that
is, of what the trust relation can do. My second thesis is that the principal thing that an
interpersonal trust relation can do is this: provide the trusting with a distinctive source
of reasons. In purporting to provide the trusting with such reasons, the trusted under-
takes an obligation to be trustworthy in ways that would ground the reasons—that is,
to give them their rational force. As we’ll see when we get to promising, the obligation
may also be moral—but it is more fundamentally rational. (The same applies to obliga-
tions of truth-telling in testimony and advice.) I’ll explain the structure of this norma-
tive relation more precisely in section 3. That will motivate shifting our attention, in
section 4, to the parallel structures that inform intrapersonal trust.
3. Trust Disappointed yet Unbetrayed

In section 1, we considered cases of undisappointed yet betrayed trust. Now that
we’ve considered how propositional assurances can provide reasons through the
medium of trust, we’re ready to consider cases wherein trust is disappointed yet
unbetrayed. I’ll begin by explaining why we cannot get such cases without assurance.
This will reveal why only an account that stresses assurances can explain what it is for
trust to be betrayed.
Consider again Lovely Leaf, in which Andrew trusts Bernice to care for a leaf that
strikes his fancy. We saw how Andrew’s trust could be undisappointed yet betrayed.
Could his trust be disappointed yet unbetrayed? No. If Bernice disappoints his trust by
failing to produce the result—a leaf kept safe—that Andrew is relying on her to
produce, she betrays his trust. To say that she betrays his trust is not to say that she
intended to betray it or that she betrays it with anything but the best intentions. She
may make her best effort not to betray his trust, yet betray it anyway. As we saw,
Andrew’s trust in Bernice is not mere reliance on her to keep the leaf safe. She may keep
the leaf safe yet betray his trust, if she does not manifest appropriate concern for his
needs. (Again, she need not feel concern; she need merely be disposed to act as if she is
concerned.) So the manifestation of concern constitutes a necessary condition on
unbetrayed trust. But it does not constitute a sufficient condition; at least, it does not in
simple cases like Lovely Leaf. In such simple cases, leaving the trust undisappointed is
also a necessary condition on the trust’s counting as unbetrayed.
What of testimonial or advisorial cases? Here too we get this structure: unbetrayed
trust is not merely undisappointed trust. Here too we need to add a condition of con-
cern: the testifier must manifest appropriate responsiveness to the trusting’s context-
sensitive epistemic needs, and the advisor must manifest appropriate responsiveness
to the advisee’s context-sensitive practical needs. But merely manifesting this concern
does not suffice for the trust to go unbetrayed. As in Lovely Leaf, the trusted must actu-
ally produce the result that vindicates the trust—in the testimonial case by asserting
the truth, and in the advisorial case by steering the advisee towards what the advisee
actually has reason to do. As in Lovely Leaf, two conditions are individually necessary
and jointly sufficient for unbetrayed trust: (i) the trusted must actually produce the
result that the trusting is relying on her to produce, and (ii) the trusted must do so in a
way that manifests appropriate responsiveness to a subset of the trusting’s needs: spe-
cifically, those needs recognition and acknowledgement of which inform the agree-
ment at the core of the trust relation.
Can there be a case in which the first condition drops out? I’ll now argue that prom-
issory trust has that structure. Unlike the other forms of trust that we’ve considered,
in promissory trust, the two conditions do not come apart: as we might call them, the
condition of execution does not come apart from the condition of concern. What
explains this difference, I’ll argue, is the distinctive way in which a promise is a prop-
ositional assurance: without this distinctive element, the trust relation cannot have
this normative structure. In section 4, I’ll argue that promises share this structure
with the intrapersonal assurance at the core of an intention. What gives promises and
intentions their special normative structure is their distinctive r elation to time. When
you form an intention, you’re aiming to do justice to your ongoing needs in respects
relevant to your understanding of the point of the intention. If your needs change in
certain ways as a result of changing circumstances, your rational obligation to do what
you intend to do may merely lapse, rather than being outweighed or overridden. You
may thus abandon the intention because your obligation to f ollow through has lapsed;
as we’ll see, it would be a mistake to say that the obligation lapses only because you
have abandoned the intention. And if your promisee’s needs change as a result of
changing circumstances—I don’t mean just any needs, but specifically those that gave
point to the promise—then your obligation to do what you promised lapses, rather
than being outweighed or overridden. These are planning needs and thus are partly
constituted by the promisee’s actual plans and dispositions to plan: they give point to
the promise insofar as they explain why the promisee is relying on the promise. In this
respect, promises and intentions share their normative structure. Though a promis-
sory obligation is also typically a moral obligation, the parallel with intention reveals
that it is more fundamentally a rational obligation. We can grasp the nature of these
obligations, I’ll argue, by understanding how the trust relations at the core of both
promises and intentions can be disappointed yet unbetrayed.
Let me introduce my approach to promising with some new cases. Though it would
be simpler to use Guide’s Promise as a template, all of the Guide cases in section 2 are
somewhat cartoonish, and the conclusion for which I’m now arguing emerges most
clearly in cases that reflect the human complexity of everyday life. If our lives did not
contain such human complexity, the concept of a promise might have likewise lacked
the complexity that gives it the element I’m emphasizing.
Two Promises. Ben is serving on a search committee with Andrea and Andres, and
Ben has promised each colleague to read the remaining dossiers today. Andrea and
Andres are depending on the promise; each is planning in concrete ways (for example,
drafting paperwork for the dean) that depend on Ben’s acting on this promise. But as
Ben begins to read the dossiers, he gets a call from the daycare: his child is sick, and
since his spouse has firmer commitments today, he’ll have to fetch the child and
nurse him at home—a full-time job. So he cannot act on his promise. Should he
apologize? Well, of course he should—to Andres. As it happens, Andrea is Ben’s
spouse, and it does not appear to make sense for Ben to apologize to his spouse in
these circumstances. Ben’s sick child is Andrea’s sick child; if Ben doesn’t nurse him,
Andrea will have to, thereby thwarting many of her planning needs, including the
needs that informed Ben’s promise.
How might we characterize the difference in Ben’s predicaments vis-à-vis his two
promisees? Ben’s predicament vis-à-vis Andres appears to be one in which violating a
promissory obligation is morally permissible in the light of broader considerations
(and Ben may even be, in that light, morally required to violate it). Ben’s promissory
obligation does not simply lapse here. But Ben’s predicament vis-à-vis Andrea appears
different in just that respect: his promissory obligation appears to be cancelled, not
overridden or outweighed, by his obligation to care for his child. The difference seems
to lie in how Andrea herself participates in that normative circumstance insofar as she
shares custody of the child. And that difference explains the difference in whom Ben
owes an apology to. An apology to Andrea would reveal confusion about how that
normative circumstance structures his promissory obligation. But there could be no
such confusion vis-à-vis Andres. (I’ll presently justify these intuitions about apology.)
Does the difference merely reflect Ben’s greater intimacy with his spouse? We can
easily imagine the cases reversed. Say Ben’s non-spousal colleague, with whom Ben is
not even friends, undergoes the shift in practical needs, while Ben’s spouse’s needs
remain constant in relevant respects. We could fill in the details as follows:
Role Reversal. Ben is looking after the child of his non-spousal colleague, Andres,
while Andres teaches a class. (If that seems to presuppose intimacy with Andres—or
a second, competing promise—we could say that Andres’s child is old enough to be
left unattended in Andres’s office, and Ben merely happens to be the only other
person on the hall.) As before, Ben has promised Andres to read the dossiers during
this time but must suddenly whisk the child off to urgent care instead. When Andres
gets Ben’s voicemail and calls him back, ought Ben to apologize for having failed to
read the dossiers? Of course not. Doing so would show that something has gone
wrong in Ben’s relations with this colleague. But now Ben’s spousal colleague,
Andrea, whom Ben has also promised to read the dossiers, phones to ask if Ben
has yet done so. Though Andrea understands Ben’s predicament, Ben should
nonetheless apologize.
The difference that emerged in Two Promises thus does not depend on the question
of intimacy.
Does the difference rest on some norm of professional or workplace ethic? We can
easily construct parallel cases outside the workplace. Imagine Ben has promised his
spouse and his neighbour to plant two trees this afternoon, one in each of the adjoining
yards, but a sick child prevents him from executing the promises. To whom does Ben
owe an apology? Without unusual background assumptions, we’ll say that Ben owes
his neighbour an apology but not his spouse. (Make it his neighbour’s child to address
the objection from spousal intimacy.) No professional or workplace norm explains
this difference.
At a methodological level, one might object that I have framed my argument in
terms of intuitions about when one person owes another an apology. What is the basis
of these intuitions? Their basis lies, I’ll argue, in the normative relation between
promisor and promisee, and specifically in the nature of the authority that the promisee
exercises in holding the promisor to the promise. I’ll first say what I believe that
authority amounts to. Then, in section 4, I’ll use the analogy between promising and
intending to explain the basis of the authority.
What grounds the intuition that Ben does not owe Andrea an apology when his cru-
cial care for their child prevents him from executing his collegial promise to read those
dossiers? The question asks us to consider the nature of Andrea’s promissory right to
demand performance from him. In a typical case with this structure, the promisee will
let this promisor off the promissory ‘hook’, if not in prospect then in retrospect; even if
through distraction or forgetfulness she doesn’t, it will be clear all along that if she had
attended to the circumstances she would have done so. But imagine that as it happens
Andrea doesn’t let Ben off the hook, and moreover that she wouldn’t even if fully aware
of the circumstances. Must it be implausible to think that she nonetheless ought to let
him off the hook? Must it be implausible to think that she ought to let him off the hook
by the terms of that very promissory agreement? There is no general reason why this
must be implausible. If professional anxiety gets the better of Andrea and she holds her
spouse to his collegial promise—overlooking that they share custody of this child and
that the failure to attend to the child’s needs would have a direct impact on her planning
agency—then it seems she is making a mistake. It is possible that she is not making a
mistake, but only if she really doesn’t care about the child enough to have adopted plans
and policies characteristic of a care-giving commitment. What grounds the intuition
that Ben doesn’t owe Andrea an apology for having failed to execute the promise is that,
assuming that she does care about their child in the ways typical of parenthood and that
her spouse’s failure to execute the promise reflects his success in tracking her parental
planning needs, she simply does not have any promissory right or authority to demand
performance from him. Ben doesn’t owe her an apology because she has no right to one.
What grounds these intuitions about apology, then, is our grasp of what is at stake in
the promissory agreements between Ben and Andrea and between Ben and Andres.
Compare that understanding with the understanding that informs the non-promissory
trust relation between Bernice and Andrew in Lovely Leaf. Part of what it is for Bernice
not to have promised in Lovely Leaf is for it not to be the case that her obligation has the
diachronic complexity that we see illustrated in the promissory trust relation between
Ben and Andrea in Two Promises. This is a substantial claim about what a promise adds
to a sub-promissory trust relation. There is, I claim, a distinctively diachronic dimen-
sion to the norm informing a promissory trust relation. Andrew trusts Bernice to do
justice to those among his needs that inform the point of his relying on her at the time
at which he gets her to acknowledge how he is relying on her. Andrea trusts Ben to do
justice to her ongoing needs in respects relevant to the point of his promise. This refer-
ence to the ‘point’ of the relation is in each case reference to the implicit or explicit
agreement between them. In a sub-promissory agreement, the point of the relation
lacks the diachronic complexity characteristic of the point of a promissory agreement.
As I’ll now explain, this difference derives from how these two types of trust relation
admit of betrayal.
4. Trust as the Medium of Intrapersonal

Rational Coherence
How could a promise serve as a source of planning reasons? I have characterized the
promissory assurance as an invitation to trust, but an invitation to trust does not by itself
give rise to the reasons that it represents itself as making available, nor does the trust
itself.5 What gives rise to a reason is the trusted’s worthiness of this trust. In the testimo-
nial and advisorial cases, this includes reliability in not disappointing the trust, in add-
ition to whatever further condition might codify how the trusted will not betray the trust.
But in the promissory case, as we’ve seen, it is a mistake to posit an anti-disappointment
condition alongside the anti-betrayal condition. One might worry that this disanalogy
undermines my claim that promissory trustworthiness can serve as a source of planning
reasons. If trustworthiness does not include reliability simply in doing what one has
promised to do, how could it serve as the basis of the promisee’s planning reason?
We can assuage the worry by articulating an analogy between the interpersonal
reason-givingness of a promise and the intrapersonal reason-givingness of an inten-
tion, which shares the just-noted feature of promising that distinguishes it from testi-
mony or advice as a source of reasons. Though no developmental claim is crucial to my
assurance view of trust, we can vividly pose the analogy by asking how we learn to treat
5
On whether trust can itself ground reasons, see Faulkner (2011). For objections to the idea, see
Hinchman (2012). For replies to those objections, see Faulkner (2012).
a promise as a source of planning reasons. It is not an implausible hypothesis that we

learn how promissory trustworthiness provides a source of reasons by reflecting on
how the promissory dynamic between S and A resembles an intrapersonal dynamic
that might unfold within A himself—between the earlier self AI, that is A insofar as he
forms and retains an intention, and the later self AP, that is A insofar as he follows
through by planning and acting on that intention.6 From this angle, S’s promise-based
obligation to A revealingly resembles AI’s intention-based obligation to AP. The paral-
lel most fundamentally lies in a parallel between the interpersonal agreement or
‘understanding’ at the core of a promise and the shared intrapersonal understanding—
shared as between earlier and later selves—at the core of an intention.
The fundamental parallel lies here: both promises and intentions generate obliga-
tions that are distinctively sensitive to requirements on responsible redeliberation—
that is, on redeliberation whether to do what you’ve committed yourself to do that
manifests an appropriately responsible attitude towards your obligation to do it. We
considered the interpersonal dimension of this issue in section 3, as we observed how
the promisee’s ongoing needs engage both parties’ expectations of the promisor’s per-
formance. In the intrapersonal dimension, the question of expectation appears to run
in parallel: you are rationally required to redeliberate if the future does not unfold as you
relevantly expected it would when you formed the intention—either by raising relevant
new considerations that you did not consider when you deliberated or by revealing rele-
vant problems with how you considered what you considered in that deliberation. Why
do we need these appeals to relevance? We need the appeals to r elevance because not
every falsification of your expectations affects the rationality of following through on an
intention—a point that obviously applies to the promissory case as well.
We can illustrate with the following case and its several variations. Imagine you
intend to have a picnic this afternoon at your local park, and consider the bearing of
several unexpected developments on the rationality of following through on your
intention. As you’re preparing to leave for the picnic, (i) dark storm clouds suddenly
loom on the horizon, (ii) you can’t find your favourite picnic blanket, (iii) you can’t get
your hair to look right, and (iv) you notice that your neighbour has acquired a fourth
poodle. Let’s say that development (iv) is the most surprising among these, since
surely—you’d naturally assumed—three poodles is quite enough for any single pet
owner. Do you therefore reconsider the intention to picnic? Obviously not, since the
fourth poodle has nothing to do with your plans.7 What then of development (iii)?
Well, that probably doesn’t matter either, though it’s possible your picnic plans require
that you be photogenic in ways undermined by the problem you’re having with your
hair. The missing picnic blanket is probably relevant. Though most picnickers would
treat that merely as a matter of means, some might regard the disappointment as rele-
6
A more controversial hypothesis would view A’s grasp on intention as developmentally derived from
his ability to enter into proto-promissory relations with caregivers.
7
Change the poodles to black cats, and some might superstitiously deem the development relevant to
their picnicking plans.
vant to the plan to picnic itself—in which camp are you? Nearly all picnic planners
would regard unexpected storm clouds as requiring a redeliberation whether to picnic.
But even here there may be exceptions—it all depends on how important the picnic is
to you and on how well you imagine you’ll be able to cope with unexpected weather.
Perhaps your picnic is a reunion of Navy SEALs who would welcome the challenge.
The case reveals how much we take for granted in intending. We distinguish relevant
from irrelevant expectations—that is, expectations whose falsification should lead us
to redeliberate from expectations whose falsification is irrelevant to redeliberation—
without formulating the distinction in terms of explicit conditions on the intention.
We nonetheless have an implicit understanding of what we’re up to in intending,
and that understanding—shared between intending and acting self—in turn deter-
mines how the distinction applies. As we’ve seen, there is a parallel role for shared
understanding—specifically for the understanding informing the promissory agree-
ment between promisor and promisee—in distinguishing circumstances in which the
promisor should do what she has promised to do from circumstances in which her
doing so would reveal that she has misunderstood the implicit point of the promise.
At the core of the parallel lies a distinctively temporal norm: in each case, the one
relied on—you qua intender or S qua promiser—bears responsibility to keep track of
and do justice to needs that arise partly from how the one relying has planned on the
basis of that reliance. The parallel focuses our attention on one key respect in which the
two acts look thus forward to future need. When A intends to φ at t, A typically gives
himself a reason to expect that he will φ at t, an expectation that in turn enables him to
plan what to do in the meantime. This reason is grounded in an aspect of A’s reliability
in intending to φ at t, where such reliability in intending is very different from reliabil-
ity in a mere disposition to φ at t. And when S promises A that she will φ at t, S’s reliabil-
ity in promising gives A a reason to expect that she will φ at t, similarly useful in
planning. The key to understanding what’s distinctive of an intention, as opposed to a
mere disposition, and of a promise, as opposed to a mere statement of intention, lies in
understanding how temporality informs these distinctive species of reliability.
To clarify this aspect of intrapersonal rationality, we must understand how it mani-
fests a dynamic of self-trust and self-mistrust. What if you are untrustworthy in form-
ing an intention? Then, by my account thus far, you do not get any planning reasons
from the intention—so it’s useless as an intention. But wait: that’s your intention! So
what should you do? Insofar as you do have this intention, you should follow through
on it. But insofar as the intention is not worthy of your trust, it appears that you should
not follow through on it. Something has to give. Here’s my solution: you have a narrow-
scope rational obligation to follow through on your intention but only a wide-scope
rational obligation to follow through on the practical judgement that informs it. Yes,
insofar as you intend, you ought to follow through on the intention. But insofar as you
are not trustworthy in making the practical judgement that informs the intention, you
ought not to retain the intention. This complexity in the structure of intrapersonal
rational coherence arises not from the rational nexus between an intention and its
execution but from the rational nexus between a practical judgement and the choice or
intention wherein you commit to that judgement.
The key question is this: how could you manage to ‘mistrust’ your own intention?
When you mistrust a promise, the promise does not thereby disappear: the promisor
continues to count as promising, but you mistrust that promise. When you ‘mistrust’
an intention, by contrast, it seems you thereby no longer have that intention. So the
question of your trustworthiness—the question whether to trust yourself—poses this
more perplexing question: how could you manage to mistrust yourself in this respect?
If we are to make sense of the analogy with promising, when you mistrust yourself at a
time, t, your mistrust must target something true of you at t—in the way that a prom-
isee’s mistrust targets the promisor’s promise. But if you cannot simply mistrust your
own intention—that is, yourself insofar as you have this intention—how can you mistrust
yourself? Though I lack space for a full treatment, I’ll sketch an answer in section 5.8 I’ll
thus explain what it is for intrapersonal trust to be betrayed without thereby being
disappointed.
5. Self-Trust Undisappointed yet Betrayed

In section 4 we considered how the self-trust informing an intention could be disap-
pointed yet unbetrayed. It remains to consider how such self-trust could be undisap-
pointed yet betrayed. As before, the answer lies in seeing how the intrapersonal relations
at the core of an intention run in normative parallel with the interpersonal relations at
the core of a promise. Each normative relation generates these two possibilities. Just as a
promisor may disappoint your trust without thereby betraying it, so a promisor may fail
to disappoint your trust—that is, may follow through on the promise—while thereby
betraying your trust by violating the promissory agreement that informs it. And just as
you may ‘disappoint’ the self-trust in your own practical judgement that forms the core
of an intention without thereby betraying that self-trust—because you abandoned the
intention in rational responsiveness to an unexpected change in your circumstances—
so you may fail to ‘disappoint’ that self-trust, by following through on the intention,
while thereby betraying it. As in the promissory case, you can betray your own self-trust
in following through on an intention by failing to remain true to the intrapersonal
agreement that informs the intention.
On the proposal I’ll now sketch, your intention has two elements: a practical judge-
ment, and your trusting commitment to that judgement. When, in a slight revision of
Donald Davidson’s (1970: 30) example, you intend to go to sleep without brushing
your teeth tonight, you judge, having considered matters to your satisfaction, that you
ought to skip brushing, and you commit yourself to that judgement in the diachronic-
ally action-guiding way of intention (for example, altering your sink-side prepar-
ations). Why distinguish these two elements? Imagine, as Davidson does, that you
8
The sketch summarizes the account developed in Hinchman (2009; 2010; and 2013).
cannot bring yourself to act on your intention: at some point before going to sleep, you
come to mistrust yourself insofar as you have the intention, and the self-mistrust pre-
vents you from acting on it. Do you thereby mistrust the intention itself? As we’ve
begun to see, that doesn’t quite make sense. Though you may have worries or reserva-
tions about an intention that you persist in holding, if you simply ‘mistrust’ your inten-
tion, in the way that you might simply mistrust another’s promise, you thereby cease to
hold the intention. An intention manifests trust in your judgement. While there may
be special contexts or respects in which you can trust yourself in some substantial way
and yet at the same time count as mistrusting that trust, trust in your judgement does
not appear to admit of that possibility. You cannot trust your practical judgement that
you ought to φ in such a way as to count as intending to φ yet at the same time also mis-
trust that trust in your judgement. One way to abandon an intention is to abandon the
judgement that informs it, but that is not the only way. You might abandon an inten-
tion by mistrusting the judgement that informs it. Our q uestion is therefore how this
works: how might you mistrust a judgement that you nonetheless retain?
What is a practical judgement? How does a practical judgement guide the formation
of an intention? Let’s work with the influential account of judgement pioneered
by T. M. Scanlon (1998: 25–30; 2007), which assimilates practical commitment to a
species of doxastic commitment, equating your all-things-considered practical

judgement that you ought to φ with a doxastic judgement that you have conclusive
reason to φ.9 The account identifies the specifically practical element in a practical
judgement with an element in the content of that judgement: the idea that you have a
conclusive practical reason. One might wonder what it is for you to judge that you have
a conclusive practical reason, but I’ll take that notion for granted. As we’ve now seen,
when you mistrust yourself insofar as you intend to φ, you specifically mistrust, not the
intention itself, but the practical judgement that informs it. Adopting the Scanlonian
view, we’ll say that you mistrust your judgement that you have conclusive reason to φ.
What is it to mistrust this judgement? For reasons of space, I must state my proposal
abstractly.10 When I speak of ‘mistrusting’ your own judgement, I don’t mean mistrust-
ing your faculty of judgement. If ‘mistrusting your judgement’ could only mean mis-
trusting your faculty of judgement, then when you mistrust your judgement you’d
be—deliberatively speaking—just stuck. You’d have to stop deliberating and merely
wait for the bout of self-mistrust to pass. But we don’t think you’re just stuck when you
mistrust your judgement; we think you can mistrust your judgement in this or that
respect and resume deliberating by trusting your judgement in other respects. We
individuate these ‘respects’ with propositions. While self-mistrust typically targets a
subject matter, we individuate subject matters with propositions, and a subject matter
can be so narrow that it coincides with a single proposition. To say that you mistrust
9
I give grounds for rejecting Scanlon’s view of practical judgement in Hinchman (2013: Section VI),
but there is no harm in assuming it for present purposes.
10
Again, for more details, see the works cited in note 8.
your judgement in the respect individuated by the proposition p is thus to say that you
mistrust your judgement on the question whether p. Does it follow that you cannot
count as judging that p? If you were redeliberating whether p, then you would not
count as judging that p. But from the fact that you mistrust your judgement on the
question whether p, it does not follow that you are redeliberating whether p. You might
be deliberating whether to redeliberate whether p. We can put the point like this: to
reconsider your judgement that p is not yet to reconsider whether p. To mistrustfully
reconsider your judgement that p is to wonder—not idly but with appropriate
engagement—whether to reconsider whether p.
Why might you raise the issue, wondering whether to reconsider whether p? One
occasion for raising the issue arises when circumstances have unexpectedly changed since
you formed the intention and you wonder whether those changes warrant reconsider-
ing it. Now say you nonetheless persist in your intention without reconsideration, in a
way that violates the intrapersonal point of the intention. You form an intention to
hike up a mountain trail as a way of enjoying the beautiful weather, but you do not
reconsider the intention as the sunshine gives way to a downpour, since you have
lapsed unthinkingly into goal-directed determination to reach the peak. By your own
lights you ought to have reconsidered your intention—perhaps reaffirming it, with a
new aim of reaching your goal, perhaps abandoning it in accordance with the original
point of the intention—but you do not reconsider it, rather like the promisor who does
what she promised to do despite an unexpected change in the promisee’s relevant
needs. You don’t decide to hike up that mountain in a downpour; you merely persist
in your intention despite this change in your circumstances. But the change, given
your understanding of the point of your hike, makes your persisting in the intention a
violation of the trust that informs it. Given that understanding, you ought to have
reconsidered whether to continue hiking up that mountain as soon as the weather
changed. You didn’t reconsider, and you still aren’t reconsidering; it seems that your
intention has somehow ‘got the better of you.’ (Are you afraid of looking like a ‘quitter’?
Are you so focused on your train of thought that you haven’t noticed that you’re
drenched and shivering?) In thus ‘getting the better of you’ it betrays your self-trust.
So what should you do? Even if you continue to judge, without any further reflection,
that you ought to hike that trail, at the very least you ought to consider whether to
reconsider, thereby withdrawing your trust in that judgement. Reasoning ‘upstream’ in
this way11—should you abandon your judgement because you mistrust it?—manifests
responsiveness to the possibility of betrayed self-trust. Such self-mistrust does not, of
course, reveal betrayed self-trust, since it is possible that you do care appropriately
about what is at stake for you in your deliberative context and that your self-mistrust is
therefore mistaken. But it is also possible that your self-mistrust is not mistaken: it is
possible that you really have betrayed the invited self-trust relation.
11
For this metaphor, see Kolodny (2005), though Kolodny argues against the possibility of such reason-
ing (534–9). I reply to Kolodny’s specific objections in Hinchman (2013).
The point runs in parallel with the possibility of betrayal in interpersonal trust. If you
mistrust a promisor, your mistrust does not itself reveal that the promisor has betrayed
the trust that she invites in promising. But your mistrust does manifest responsiveness
to the possibility that she has betrayed your trust. In each case, interpersonal and intra-
personal, the responsiveness at the core of trust is a rational responsiveness because it
targets the possibility that your trust has been betrayed. Responsiveness to betrayal and
responsiveness to reasons or rational requirements thus go hand in hand. Trust is cru-
cially unlike other forms of reliance, whether interpersonal or intrapersonal, because of
this broad but basic link between trust and rationality.
References
Baier, Annette (1994) Moral Prejudices (Cambridge, MA: Harvard University Press).
Davidson, Donald (1970) How is Weakness of Will Possible? reprinted in his Essays on Actions
and Events (Oxford: Oxford University Press, 1980).
Faulkner, Paul (2011) Knowledge on Trust (Oxford: Oxford University Press).
Faulkner, Paul (2012) Replies, Abstracta VI, 117–37.
Hardin, Russell (2002) Trust and Trustworthiness (New York: Russell Sage Foundation).
Hawley, Katherine (2014) Trust, Distrust, and Commitment, Noûs 48:1, 1–20.
Hieronymi, Pamela (2008) The Reasons of Trust, Australasian Journal of Philosophy 86:2,
213–36.
Hinchman, Edward (2009) Receptivity and the Will, Noûs 43:3, 395–427.
Hinchman, Edward (2010) Conspiracy, Commitment, and the Self, Ethics 120:3, 526–56.
Hinchman, Edward (2012) Can Trust Itself Ground a Reason to Believe the Trusted? Abstracta
VI, 47–83.
Hinchman, Edward (2013) Rational Requirements and ‘Rational’ Akrasia, Philosophical Studies
166:3, 529–52.
Holton, Richard (1994) Deciding to Trust, Coming to Believe, Australasian Journal of
Philosophy 72:1, 63–76.
Jones, Karen (1996) Trust as an Affective Attitude, Ethics 107:1, 4–25.
Jones, Karen (2004) Trust and Terror, in P. DesAutels and M. Walker (eds), Moral Psychology
(Lanham: Rowman & Littlefield), 3–18.
Kolodny, Niko (2005) Why Be Rational, Mind 114:455, 509–63.
McGeer, Victoria (2008) Trust, Hope, and Empowerment, Australasian Journal of Philosophy
86:2, 237–54.
McMyler, Benjamin (2011) Testimony, Trust, and Authority (Oxford: Oxford University Press).
Nickel, Philip (2007) Trust and Obligation-Ascription, Ethical Theory and Moral Practice 10:3,
309–19.
Rose, David (2011) The Moral Foundation of Economic Behavior (Oxford: Oxford University
Press).
Scanlon, T. M. (1998) What We Owe to Each Other (Cambridge, MA: Harvard University Press).
Scanlon, T. M. (2007) Structural Irrationality, in G. Brennan, R. Goodin, F. Jackson, and M. Smith
(eds), Common Minds (Oxford: Oxford University Press), 84–103.
Walker, Margaret Urban (2006) Moral Repair (Cambridge: Cambridge University Press).
5
Betraying Trust
Collin O’Neil
Introduction
A standard observation about trust is that it can be betrayed (Baier 1986). What
philosophers usually mean by this is that trust disposes us to feel betrayed by someone’s
failure to do as trusted. Accordingly, their efforts have been largely focused on explain-
ing this fact by describing the nature or psychology of trust. But there is a more
straightforward reading of this observation that calls for a different sort of explanation.
Consider friendship, which can also be betrayed. Certainly friendship makes us ready
to feel betrayed by someone’s failure to act as a good friend. But friendship does more
than that—it makes it possible for us to be betrayed by someone’s failure to act in certain
ways. This fact cannot be explained just by describing the attitudes characteristic of
friends. A moral explanation is needed: in particular, an explanation of how friendship
grounds obligations whose violation constitutes a betrayal, specifically of loyalty.
Trust, like friendship, does not merely ready us to feel betrayed, but can also be
betrayed. Although the wrong of a betrayal of trust may be under-theorized, it is a
familiar complaint. ‘But I trusted you’ is its everyday expression. To explain how trust
can make it possible to be betrayed by someone’s failure to do as trusted, a description
of the attitude of trust does not suffice. My aim in this chapter is to make sense of the
idea that trust can itself be betrayed by showing how trust can give rise to a distinctive
obligation whose violation we would be apt to describe thickly as a betrayal of trust.
In the first section I set out a number of formal conditions that such an obligation
must satisfy if its violation is to correspond to our concept of a betrayal of trust. In the
second section I ask whether trust could give rise to what I call a ‘first-order’ obligation
to justify trust, and argue that if there is to be a trust-based obligation that meets the
formal conditions, it must instead be conceived of as a ‘second-order’ obligation—that
is, an obligation to respect an already existing obligation.
In the third and fourth sections I evaluate two possible accounts of the grounds of a
second-order obligation to justify trust. The first involves a form of trust that can have
the status of a gift because it does not involve an expectation of trustworthiness,
and the second involves a form of trust that cannot count as a gift but can count as an
betraying trust 71
honour precisely because it does include such an expectation. Only trust that confers
an honour will ground a second-order obligation of the appropriate kind. The account
of betrayals of trust I ultimately propose is that betrayals of trust consist in the viola-
tion of a second-order obligation to vindicate the honour bestowed by an expectation
of trustworthiness.1
1. The Formal Properties of the Obligation

To characterize the kind of trust-based obligation whose violation would count as a
betrayal of trust, we must begin with an examination of our concept of a betrayal of
trust. We do have a clear sense of the form of this kind of wronging, even if we do not
already have a clear idea of its substance. After delineating its formal features, we can
then evaluate accounts of its substance on the basis of how well they fit its form.
First, a betrayal of trust violates a special obligation. Under the category of ‘special’
I gather together three features. One is that a betrayal of trust is not merely a wrong but
is a wronging of someone. This means that the relevant obligation must be a directed
obligation, in the sense of an obligation to someone. Another is that a betrayal of trust
is necessarily a wronging of someone in particular, namely the truster. So the obliga-
tion must be directed to the truster. Finally, only one person can commit a betrayal of
trust, namely, the trusted person. This means that the relevant obligation must belong
only to the trusted person, such that it is impossible for anyone except the person who
is trusted to violate it.
Second, a betrayal of trust violates an obligation to do specifically as trusted. If you
trust me to pick you up, it is possible for me to betray your trust by failing to pick you
up, even if I take steps to ensure that your interests are protected by arranging another
ride for you, or by giving you a timely warning. For this to be the case, the relevant
obligation must be to do specifically as trusted.2
Third, a betrayal of trust is only a failure to do as trusted. If you trust me to pick you
up, but don’t trust me not to steal from you, then only my failure to pick you up can
betray your trust. The obligation must be exclusively an obligation to do as trusted.
Fourth, a betrayal of trust violates a trust-based obligation. A betrayal of trust
cannot occur unless one is trusted: it is always a defence to a charge of having betrayed
someone’s trust that there was no trust to betray. So the relevant obligation must be
based or grounded in trust. Furthermore, the elements that distinguish the form of
1
The basic elements of the positive account appear in O’Neil (2012: 307–17), although the present ver-
sion modifies the earlier account in a number of ways. The main aim of that article was to explain ‘abuses’
of trust, a related wrong I do not discuss here.
2
Although the obligation must be to do specifically as trusted, this does not mean that what one is
trusted to do must also be specific. As Margaret Walker (2006: 80–1) has noted, the focus of trust can range
from the very specific, such as descriptions of particular actions, to the more general, such as descriptions
of tasks.
72 Collin O’Neil
trust that can be betrayed (whatever those turn out to be) from other forms of trust
must be essential to the generation of the obligation.
Fifth, it is also essential for a betrayal of trust to occur that the trusted person know
of the trust. The trusted person cannot betray trust that he or she is unaware of. So the
relevant obligation must not only be grounded in trust, but also in the trusted person’s
awareness of that trust. However, in requiring knowledge I am not assuming in
advance that the trust must also have been invited or accepted by the trusted person.
To sum up, the relevant obligation should have five formal properties, if its violation
is to count as a betrayal of trust. It should be:
(1) special: an obligation only of the trusted to the truster;
(2) specific: an obligation to do specifically as trusted;
(3) exclusive: only an obligation to do specifically as trusted;
(4) trust-based: an obligation grounded in trust;
(5) knowledge-based: an obligation grounded in knowledge of the trust.
Although these five properties follow fairly straightforwardly from our concept of a
betrayal of trust, there is a further question that I think our concept leaves open. The
question is whether a betrayal of trust can sometimes occur without any other kind of
wronging also occurring, or whether a betrayal of trust can occur only when and partly
because another kind of wronging is also being committed. Corresponding to these
two different conceptions of betrayals of trust are two different types of trust-based
obligations. I will call a type of obligation ‘independent’ if at least some of its instances
are first-order obligations. First-order obligations are not obligations to respect another
kind of obligation: their violation does not presuppose the violation of another kind of
obligation. I will call a type of obligation ‘dependent’ if all of its instances are second-
order obligations. Second-order obligations are obligations to respect another kind of
obligation: their violation presupposes the violation of another kind of obligation.3
The most familiar special obligations are independent in the sense that some,
though not all, of their instances are first-order. Some promissory obligations, like
those deriving from so-called ‘redundant promises’ (Shiffrin 2011) not to steal from
someone or not to lie to someone, cannot be violated without violating an obligation of
another type. These promissory obligations are second-order. But the usual point of
promises is to create a first-order obligation, namely an obligation to do something
one is not already obliged to do.
Obligations of friendship, like promises, have some second-order instances. There is
an obligation of friendship to avoid wronging one’s friend. Stealing from a friend is not
only a violation of his or her property rights, it is also disloyal. But of course friendship
also grounds a variety of first-order obligations, such as obligations to spend time with
one’s friend, to support one’s friend, and so on, that can be violated without violating
another kind of obligation.
3
Thanks to Thomas Simpson for helping me to clarify this distinction.
betraying trust 73
Although these familiar types of special obligations have both first-order and
s econd-order instances, the fact that they do have second-order instances suggests the
possibility of a type of special obligation that has only second-order instances, a
dependent type of obligation. So we should not rule out in advance a conception of
betrayals of trust according to which they are always at the same time a violation of
another kind of obligation. I will eventually recommend such a conception of betrayals
of trust, but only after examining and rejecting the view that trust can generate first-
order obligations that satisfy the formal criteria I set out earlier.
2. Varieties of Non-Moralized Trust and First-Order

Obligations
2.1 Mere reliance
Matthew Harding denies that there could be ‘an obligation to do what you trust me to
do, precisely because you trust me to do it’ (2011: 77), even when trust is invited.4
Friedrich and Southwood, by contrast, cite an ‘obligation not to betray the trust [one]
has invited’ (2011: 280). Both sides have in mind a first-order obligation, so it makes
sense to begin by examining this possibility. To do this I will first need to draw a
distinction between two broad categories of trust.5 One category is non-moralized, in
the sense that it need not represent the trusted person as under an obligation. Merely
relying on someone to do something is non-moral in this sense, but so is a richer form
of trust that I will label ‘goodwill trust’. Goodwill trust involves an expectation that the
trusted person will do as trusted out of (possibly morally optional) concern or affec-
tion for the truster.
Moralized trust is a category of trust that does, by contrast, essentially involve a rep-
resentation of the trusted person as under an obligation. The distinction is important
because, if trust is capable of grounding first-order obligations, only a non-moralized
form of trust could play this role. A form of trust that represents the trusted person as
under a prior obligation to do as trusted could generate at most a second-order obliga-
tion to respect that prior obligation.6
I will begin by asking whether mere reliance could generate an obligation that has
the formal features introduced earlier. Although trust is usually distinguished from
4
In what follows I’m indebted to Harding’s argument against such an obligation, although I will ultim-
ately defend its possibility at the second-order.
5
Although I favour a form of moralized trust as the form of trust that can be betrayed, I would not say
that to deserve the name ‘trust’ an attitude must be the kind that can be betrayed. Not only is this unlikely
to be an essential feature of trust, but as Thomas Simpson (2012) has argued, trust may not even have an
essence.
6
To believe that it could generate a first-order obligation would imply that, if my trust falsely represents
you as under a prior obligation to do as trusted, it could still create an obligation to do as trusted. But it is
implausible to think that my trust, which is false or inaccurate in this respect, could still ground an obliga-
tion to do as trusted.
74 Collin O’Neil
mere reliance, it will be instructive to see why mere reliance fails to deliver the right
kind of obligation. Next I will consider the richer variety of trust, goodwill trust, to see
whether the features that distinguish it from mere reliance enable it to deliver the obli-
gation that mere reliance cannot. Finally, I will examine whether adding an invitation
to trust will yield the needed obligation.
Trusting someone to do X usually involves some kind of vulnerability to the trusted
person’s failure to do X. One form of vulnerability occurs when we rely on someone to
do X. Let’s suppose that Sara becomes aware that Andy is relying on her to feed his lab
mice. He has registered for a conference that would take him away from the lab and if
the mice go unfed they will die and set back his dissertation by at least two years. I will
assume that, although Sara has become aware of his reliance, she has done nothing to
invite or encourage it. I will also assume that she is not, prior to his relying on her,
under any obligation to feed his mice: she is not the lab’s animal technician.
As Karen Jones (2012: 80) points out, even when Andy’s reliance is ‘poorly placed
or presumptuous’, it does not follow that Sara is morally free to turn her back on him.
Since the consequences for Andy of his reliance could be serious, Sara may have an
obligation to respond appropriately to his reliance. But does this obligation to respond
appropriately have the right formal features? If we assume for the moment that mere
reliance can be considered trust, then her obligation would be trust-based, in the
sense that it would not have existed but for Andy’s relying on her. But is it also special
and specific?
Keeping in mind that Sara has done nothing, intentionally or negligently, to encour-
age Andy’s reliance, any obligation she acquires as a result of his reliance would not be
special, but would derive from a general duty of mutual aid or helpfulness. A third-
party, Jonathan, who was aware of Andy’s reliance on Sara and in a similar position to
help, could have the same obligation to protect Andy’s interests, even though Andy is
relying on Sara not Jonathan. Yet only Sara could betray Andy’s trust.
Nor is the obligation specific in the way it must be to capture our concept of a
betrayal of trust. Even where it exists this obligation to respond usually admits of some
latitude: it is not necessarily an obligation to do specifically as relied upon (Harding
2011: 78). Perhaps it would be difficult for Sara to feed his mice, but she could arrange
for someone else to do it, or at least warn Andy that she is not planning to feed his mice
if there is still time for him to make other arrangements. Although Andy is relying on
her specifically to feed his mice, not to feed his mice or see to it his mice are fed or to
warn him, she acquires no obligation specifically to feed his mice, but at most an obli-
gation to protect his interests or to give him an opportunity to protect them himself.
Yet it is possible to betray trust by failing to do specifically as trusted, even when the
truster’s interests are protected.
2.2 Goodwill trust
Sara’s obligation to respond to Andy’s reliance is, although trust-based, neither special
nor specific. If we add the elements that distinguish trusting in a richer sense from
merely relying, will this yield an obligation that is special and specific? Goodwill trust,
betraying trust 75
which adds an expectation of responsiveness to one’s vulnerability from concern or

affection, is the candidate I will evaluate.
Goodwill Trust: A trusts B to do X if A relies on B to do X (or is otherwise vulnerable
to B’s failure to do X) partly because A expects B to be directly and favourably moved
to do X by B’s recognition that A is relying on B to do X (or that A is vulnerable to B’s
failure to do X).7
If Sara learns that not only is Andy relying on her to feed his mice, but that he is doing
so in part because he expects her to respond to his reliance from concern or affection,
does she acquire a special and specific obligation to feed his mice, whereas before she
had at most a general and indefinite obligation to help? Again, we are assuming that,
although she is aware of Andy’s expectations, Sara has neither intentionally nor negli-
gently encouraged them. And if this richer form of trust is to have the potential to
ground a first-order obligation, we should assume that Sara does not already have an
obligation to feed his mice, and also that Sara does not already have an obligation to
display the concern or affection that Andy is expecting.8
What difference could this richer form of trust make? Andy’s expectation of con-
cern or affection might indicate an additional vulnerability, beyond the risk to his dis-
sertation. There tends to be overlap between the people from whom we expect goodwill
and the people we like, so it may be the case that Andy cares about receiving her good-
will in itself. Perhaps her displaying concern or affection for him is important to him
because he would like to be friends with her, and this would assure him that a friend-
ship is possible. So let’s assume that the absence of this motivation on her part would be
bad for him; its presence good for him.
If there were an obligation to respond to this particular vulnerability, not just to
protect Andy’s dissertation, it would arguably be special. Only Sara could be in a
position to address Andy’s vulnerability to the absence of concern or affection from
Sara herself. While Jonathan’s feeding the mice would protect Andy’s dissertation, it
could not possibly provide Andy with a display of Sara’s concern or affection. Such an
obligation might even be specific. Whereas the risk to his dissertation from his reliance
could be mitigated without doing specifically as trusted, Sara’s delivering a timely
warning to Andy that she is unwilling to feed his mice may not display the degree of
concern or affection Andy desires.
However, it is doubtful that Andy’s vulnerability to the absence of her concern or
affection as such grounds any such obligation. Sara and Andy are not, we are assuming,
already in the kind of relationship that might ground an entitlement to concern or
affection. Given this, the fact that a particular demonstration of Sara’s goodwill is
important to Andy as such would not by itself seem to ground an obligation on her part
to respond to it, much less deliver it. While the vulnerability created by Andy’s reliance
may provide the focus for a general obligation of mutual aid or helpfulness, this other
7
This formulation is based on Jones (1996), and Friedrich and Southwood (2011).
8
Thanks to Paul Faulkner for suggesting the need for this assumption.
76 Collin O’Neil
vulnerability to the absence of her goodwill as such does not appear to provide a focus
for any obligation at all.
2.3 Trust that has been invited

We’ve seen that even rich trust, in the sense of goodwill trust, does not deliver the right
kind of trust-based obligation to do as trusted, at least when such trust has not been
encouraged by the trusted person. But what about trust that is invited? Friedrich and
Southwood (2011), among others,9 have argued that promissory obligations derive
from a more fundamental obligation not to betray trust that one has invited. They say,
In [inviting trust], and having the overture accepted (or not rejected) the promiser incurs an
obligation not to betray the trust she has invited. The distinctive wrong involved in breaking a
promise is precisely a matter of violating this obligation. (2011: 280)
Could adding an invitation secure the kind of obligation whose violation would count
as a betrayal of trust?
The form of trust they have in mind is the goodwill trust introduced earlier. So let’s
suppose that Sara invites Andy to trust her to feed his mice, in the sense that she con-
veys her recognition of the importance this has to him and invites him to rely, on that
basis, on her feeding his mice. In this case, Sara may now have a special obligation to
respond to his vulnerability, not merely a general obligation of mutual aid or helpful-
ness. Although Jonathan, who has not invited Andy’s trust, may still under some cir-
cumstances have an obligation to respond, it is not the same obligation as Sara’s. Sara’s
obligation would be stronger, in the sense that it would require her to go to more
trouble to respond to Andy’s vulnerability than Jonathan.
But does Sara’s invitation to trust yield a specific obligation to feed Andy’s mice? It
would seem to depend on the kind of invitation to trust she makes.10 Suppose that Sara
knows that Andy is hesitating to register for a conference because his mice need to be
fed. She wants to let him know that it would be reasonable for him to rely on her feed-
ing his mice, but she also wants to preserve the option of changing her mind if some-
thing comes up, although she thinks it highly unlikely anything will. So she says,
‘I know how important those mice are to you, and if you’d like, I’ll plan on coming in to
feed them. Just so we’re clear, however, I am reserving the right to change my mind. But
rest assured, I’m very unlikely to do so, so you should go ahead and register for the
conference.’
Sara is inviting Andy’s trust, in the sense that she is letting him know that it would be
reasonable for him to rely on her feeding his mice and even encouraging him to do so,
although she is clearly not making a promise, since she is explicitly reserving the right
to change her mind. Reserving her right in this way does not make it unreasonable for
him to rely on her, since she is highly unlikely to change her mind. In fact, if Sara chafes
9
See Thomas Pink (2009).
10
The following line of argument derives from Owens (2012: 218).
betraying trust 77
under the yoke of promises, by reserving her right she may make it even more likely
she will feed his mice than if she promised, and Andy may know this (Owens 2012:
217). Assuming that Andy does trust her in response to this invitation, I do think she
acquires a special obligation to protect him from loss. But this does not yield an obliga-
tion to do specifically as trusted. She has violated no obligation to Andy if she does
change her mind and gives him a timely warning, or arranges for someone else to feed
them, and so on. The obligation here is trust-based, and the invitation makes it special,
but it is not specific.
When the invitation to trust does not reserve a right in this way and makes what we
would call a promise, then it does yield a special obligation to do specifically as prom-
ised. But this is not a trust-based obligation (Harding 2011: 77). Although Andy may
trust Sara, his trusting her is not essential to the generation of this obligation. As even
Friedrich and Southwood concede, all that is required for the invitation to trust (that
makes a promise) to create a specific obligation to feed the mice is that Sara perform a
speech act of acceptance (2011: 279). The very same obligation could have arisen had
Andy not trusted Sara to feed his mice and even had Sara known that Andy did not
trust her to feed his mice. Since a betrayal of trust cannot occur without trust, any more
than a betrayal of loyalty could occur without a special relationship, the obligation
generated by the invitation to trust that makes a promise cannot be the obligation
whose violation counts as a betrayal of trust.
We are now in a position to draw a sceptical conclusion about the possibility of an
independent trust-based obligation whose violation counts as a betrayal of trust. If
trust could ever create a first-order trust-based obligation to do as trusted, it would be
trust that was invited. But we’ve just seen that if the invitation does not make a promise,
the resulting trust-based obligation may be special but is not specific. If the invitation
does make a promise, the resulting obligation may be special and specific, but is not
trust-based. In neither case do we have an obligation with the right formal properties
to explain betrayals of trust.
We should therefore abandon the hypothesis that a betrayal of trust consists in the
violation of a first-order obligation. But we should not give up on the idea that a
betrayal of trust is a distinctive wronging. Although Andy may not have needed to
trust Sara to be wronged by her failure to keep her promise, his trust could still be
essential to an additional complaint: not only has she broken her promise, she has also
betrayed his trust.
Here is the way forward. Andy trusts Sara to feed his mice, and it must be possible
for any failure to do as trusted, even a timely warning, to betray his trust. This is the
specificity requirement. Since Sara has promised to feed his mice, any failure to feed
his mice, including a timely warning, will break the promise. But as we’ve seen, this
promissory obligation is not trust-based. What is needed to explain why a timely
warning could also betray Andy’s trust would be a trust-based obligation with the
same content as the (non-trust-based) promissory obligation. One way to secure same-
ness of content would be to conceive of the trust-based obligation as a second-order
78 Collin O’Neil
obligation to respect the promissory obligation. Any violation of the promissory obli-
gation, such as a timely warning, would also violate this trust-based second-order
obligation to respect the promise, thereby betraying Andy’s trust. But what form of
trust could ground such a second-order obligation, and why would it do so?
3. Moralized Trust and Second-Order Obligations

3.1 The feeling of resentment and the feeling of betrayal
Earlier I examined non-moralized forms of trust, since only forms of trust that do not
essentially involve a representation of the trusted person as under an obligation could
possibly ground a first-order obligation. Since we’re now pursuing the question of
whether and how trust might ground a second-order obligation, we can discuss moral-
ized forms of trust. These accounts are often motivated by, as Katherine Hawley puts it,
the idea that ‘there is a distinction between trust in a rich sense—trust which can be
betrayed—and mere reliance’ (2014). I will eventually describe a moralized form of
trust that I believe can ground a second-order obligation whose violation constitutes a
betrayal of trust. But before I do, I want to point out the gap that exists between explain-
ing how trust can dispose us to feel resentment or betrayal and explaining how trust
can be betrayed.
On the most common understanding of the claim that trust can be betrayed, it refers
to a certain emotional vulnerability. When we trust someone to do X, we are ready to
feel a certain reactive attitude, should we discover that they failed to do X. When we
merely rely on someone’s doing X, by contrast, we are at most ready to feel disap-
pointed. Usually the relevant reactive attitude is taken to be a feeling of resentment. To
feel resentment in particular towards someone is to feel that they have wronged you.11
Although most writers have taken the feeling of resentment to be the marker of trust
(or have treated the feeling of resentment and the feeling of betrayal interchangeably),
recently others such as Cogley (2012) and Frost-Arnold (2008) have drawn a distinc-
tion between the feeling of resentment and the feeling of betrayal, taking the feeling of
betrayal to be the marker of genuine trust, or at least of an important kind of trust.
They suggest that the feeling of betrayal is about a more restricted class of wrongings
than resentment. Whereas resentment can take the violation of any directed obligation
as its object, the feeling of betrayal is only about the violation of a subset of directed
obligations: in particular, obligations the violation of which we would be inclined to
describe more thickly as betrayals.12
11
A number of authors provide what I am calling ‘moralized’ accounts of trust, or of important kinds of
trust. Some accounts are explicitly moralized in this way (Cogley 2012; Frost-Arnold 2008; McLeod 2002;
Nickel 2007; O’Neil 2012). Normative expectation accounts (Faulkner 2011; Holton 1994; Jones 2004;
McGeer 2008; Walker 2006) are more challenging to classify.
12
Possibly Hawley (2014) shares this view, since she restricts the moral content of trust to commit-
ments. But she does not see commitments as a type of obligation, and also does not require that the com-
mitment be directed to the trusted person.
betraying trust 79
Cogley confines the subset to obligations deriving from friendships and friendly
acquaintanceships.13 Frost-Arnold is more inclusive, and recognizes not only such
obligations of loyalty, but also promissory obligations and fiduciary obligations (2008:
73–4). I shall adopt the more inclusive view and understand betrayals as the violation
of any obligations deriving from certain kinds of transactions or relationships. Thus
there will be betrayals of loyalty, betrayals of fidelity, fiduciary betrayals,14 and more.
The wronging I am seeking to explain—a betrayal specifically of trust—belongs in this
category. Although a betrayal of trust may co-occur with any other betrayal, it is not
the same betrayal as any of these others.15
The readiness to feel resentment and the readiness to feel betrayed can each be satis-
factorily explained by moralizing trust. To explain why a form of trust would dispose
us to feel resentment, a feeling that is about a wronging, we can say that trust represents
the trusted person as under an obligation that is directed to the truster. To explain why
a form of trust would dispose us to feel betrayal, a feeling that is about a betrayal, trust
should represent the trusted person as under a directed obligation deriving from some
transaction or relationship with the truster.
Ascribing moral content to an attitude of trust may explain why, when we trust in
this way, we are prone to feel betrayed by a failure to do as trusted. However, ascribing
moral content to trust will not, as such, explain how the attitude of trust could alter, as
opposed to merely reflect, the moral landscape. This is the sort of explanation needed
to understand how trust can be betrayed.
While it may be true that ascribing moral content to trust will not in itself explain
how trust can make a moral difference, there is another aspect of moralized trust that
clearly can alter the moral landscape. Although this aspect of trust can have consider-
able moral significance, I don’t believe it is capable of making the kind of moral differ-
ence needed to explain betrayals of trust. Here is why.
Trust can make a moral difference qua vulnerability. Trusting someone to do X is
associated with being vulnerable to the trusted person’s failure to do X, in the sense
that a failure to do X would harm or be bad for the truster. Earlier I explained why
being vulnerable to someone’s failure to do X would not create an obligation to do X
specifically, but at most an obligation to take steps to protect that vulnerability. But
vulnerability can make another kind of moral difference when there is already an
obligation to do X, such as to keep a promise or not to steal. When A trusts B not to
13
Specifically, he believes betrayals (of loyalty) are possible when individuals enjoy a relationship that is
deep enough to make it the case that they owe each other goodwill, in the sense of concern or affection
(Cogley 2012: 40–1).
14
Fiduciary betrayals are often called betrayals of trust, but this is misleading since a non-trusting or
even distrusting patient or client could still be the victim of a fiduciary betrayal, though not a betrayal of
trust.
15
A betrayal of trust requires trust, and one can violate promissory obligations, obligations of friend-
ship, and fiduciary obligations that one is not trusted to keep. Betrayals of trust can also occur without
promises, friendships, and fiduciary relationships. I may trust a stranger in the airport not to run off with
my luggage while I go to the bathroom, and if the stranger knows of my trust, the theft could betray it.
80 Collin O’Neil
wrong her, A will usually, in virtue of trusting B to respect that obligation, incur add-
itional vulnerabilities beyond the ‘vulnerability’ to the bare wronging. If A trusts B not
to steal her laptop, she may for example rely on B not stealing her laptop by not backing
up her work in the cloud. If B does steal her laptop, then she can complain, not only of
the theft, but also of the additional costs of reliance. These costs make the theft worse.
This is an important way in which trust can make a moral difference. It can aggravate a
prior wronging by making its consequences worse for the truster, thus making the
wronging more serious.16
There is no need to identify any distinctive trust-based obligation to explain this
kind of moral difference. But it would be a mistake to think that trust’s association with
vulnerability to harm exhausts its moral significance. The complaint ‘I trusted you’
makes sense even when the truster did not suffer harm as a consequence of trusting,
and even when the truster was never even at risk of suffering such harm. Even when
there is harm because of trust, trust seems to do more than merely increase the degree
of the wronging. When we trust, at least in a certain kind of way, there is the possibility
of a distinct and additional kind of violation. There is a reason we have a concept of a
betrayal of trust. We would not be content with an account of the moral relevance of
friendship that said that wronging someone in the context of friendship is a more
serious wrong than a similar wronging of a stranger, whether because wrongs to a
friend tend to be more devastating, or even because wrongs to a friend are worse in
themselves. We think there is also a distinct and additional kind of violation when one
mistreats a friend, which is why we label such wrongs betrayals of loyalty. The same is
true of betrayals of trust.
3.2 Trust as a gift

I’ve explained the gap between moralizing trust and giving an account of betrayals of
trust. In the remaining sections I will try to close it. To develop my account I will need
to appeal to a distinction within the category of moralized trust. The distinction I have
in mind is in terms of the presence or absence of certain expectations accompanying
the representation of the obligation.
Some have suggested that only trust that involves an expectation of trustworthiness
is genuine (D’Cruz 2015; Hieronymi 2008; Keren 2014). Others have drawn attention to
the phenomenon of ‘therapeutic trust’, a type of trust that represents the trusted person
as under an obligation, but without an expectation of trustworthiness. Therapeutic
trust, as the name suggests, is trust that ‘is undertaken with the aim of bringing about
trustworthiness’ (Jones 2004: 5). For example, parents may decide to trust their irre-
sponsible teenage children with the house for the weekend, ‘hoping by such trust to
elicit in the fullness of time, more responsible and responsive trustworthy behavior’
(McGeer 2008: 241).
16
See Jones (2004: 7–16), and Walker (2006: 88–98) for accounts of the variety of ways in which trusting
can make one vulnerable to harmful consequences.
betraying trust 81
Therapeutic trust, on all accounts, is something we can decide to do, at least when
we already believe someone has an obligation to us. Since we cannot decide to expect
trustworthiness, therapeutic trust cannot involve such an expectation. It is sometimes
suggested that we could bootstrap an expectation of trustworthiness from a decision
to manifest therapeutic trust (McGeer 2008). Even so, this expectation could not be
part of what it is to therapeutically trust, since that expectation would disqualify it
from being something we can decide to do.
There are various ideas about the mechanism by which therapeutic trust might
work. For example, McGeer believes that hope is a critical component, and that by
giving trusted people a hopeful vision of themselves as trustworthy, we inspire them to
realize this vision (2008). However, my focus will be on the two characteristics of
therapeutic trust that may equip it to ground a certain trust-based obligation. First, as
noted, therapeutic trust is something we can voluntarily decide to do. And second, it
can be a benefit to be therapeutically trusted. As McGeer observes,
[B]y way of such hopeful scaffolding, we also give trusted others something substantial in
return—namely, a motivationally energizing vision of what they can do or be . . . Hence, trusted
others are very often gratified by, even grateful for, the trust we invest in them. (2008: 249)
The fact that therapeutic trust is a voluntary act undertaken to confer a benefit on the
trusted person gives the benefit a certain status, that of a gift or favour. This status sug-
gests a possible way of accounting for a trust-based obligation. We should be grateful
for gifts and favours, especially when the benefactor incurs some risk in providing the
benefit the benefactor incurs, as therapeutic trusters do by trusting the untrustworthy.
And arguably gratitude for trust that has the status of a gift or favour can generate a
trust-based obligation to do as one is trusted to do.
Therapeutic trust among other things represents the trusted person as under an
obligation to do X. Insofar as therapeutic trust is a gift or favour, it may be ungrateful
for the trusted person to violate the obligation to do X. As A. D. M. Walker has argued,
although gratitude leaves us latitude when it comes to reciprocating gifts or favours, it
also makes a definite set of demands with respect to our future treatment of the bene-
factor: all ways of wronging the benefactor are ungrateful (Walker 1988: 203–4).
Suppose that Sara promises Andy that she will feed his mice. Andy accepts the
promise and thereupon represents her as under an obligation to feed his mice. He does
not regard her as trustworthy, but he decides to therapeutically trust her to keep this
obligation, as a favour to her. In light of Andy’s favour, Sara may now be under an add-
itional obligation of gratitude not to violate the promissory obligation, an obligation
that derives from the general obligation not to wrong one’s benefactor in conjunction
with the promissory obligation. The trust (gift) is inessential to the generation of the
promissory obligation, but it is essential to the generation of the second-order obliga-
tion of gratitude not to wrong the truster (benefactor) by breaking the promise.17
17
Of course, the second-order obligation of gratitude would not exist unless there really was a promis-
sory obligation. It is not ungrateful to do something that a benefactor thinks would wrong him; it is only
ungrateful to actually wrong him.
82 Collin O’Neil
Is this a satisfying account of betrayals of trust? One obvious problem is that the
account would be too revisionary. Surely therapeutic trust is not the only form of trust-
ing relation that can be betrayed—some philosophers are even hesitant to classify it as
a genuine form of trust. There is also a problem with the form of the obligation it cre-
ates. I’ll review the formal properties an obligation must possess to explain betrayals of
trust. The obligation must be special. Although the prior obligation represented by the
therapeutic trust need not be special, the obligation of gratitude not to wrong the
truster is special. Only the trusted person (beneficiary) could be under this obligation,
and this obligation could only be to the truster (benefactor). The obligation is also
clearly trust-based and knowledge-based, in the sense that it would not obtain but for
the distinguishing elements of therapeutic trust—the elements that make it a gift or
favour—and the knowledge that one has received such a favour.
Now let’s think about specificity and exclusivity—the conditions that there be an
obligation to do specifically as trusted, and only an obligation to do as trusted. On this
account, the obligation of gratitude is specifically an obligation to do as trusted.
Returning to the example of promising, let’s assume that Andy’s trust represents Sara
as having a promissory obligation to feed his mice. If she does indeed have this obliga-
tion, then it would be ungrateful for her to fail to act as trusted, even if she provided a
timely warning. The reason is that, in virtue of her receipt of his therapeutic trust, she
acquires a second-order obligation of gratitude not to wrong Andy, and a failure to
keep her promise wrongs Andy.
But the account does not satisfy exclusivity. As Walker suggests, any way of wronging a
benefactor is ungrateful. Andy may only therapeutically trust Sara to keep her promise,
and breaking her promise would therefore violate a second-order obligation of gratitude
not to wrong him. He may not, let’s suppose, therapeutically trust her not to mistreat him
in another way, for example, not to borrow his car without permission. Nevertheless, his
status as a benefactor makes it ungrateful for her to wrong him in any way, not merely the
way in which he therapeutically trusts her not to wrong him. Obviously borrowing his
car without permission cannot betray Andy’s trust, since by hypothesis he does not
therapeutically trust her not to do this. Yet on this account of the trust-based obligation,
this wrong would still betray his trust. So whatever the merits of this account as a possible
source of trust-based obligations, it has not identified the source of the kind of trust-
based obligation whose violation would count as a betrayal of trust.
3.3 Trust as an honour

The absence of an expectation of trustworthiness is what enables therapeutic trust to
function as a gift or favour. Forms of trust that involve this expectation cannot be
willed. But without this expectation, therapeutic trust lacks another kind of signifi-
cance for the trusted person that is often associated with trusting.18 Horsburgh draws
18
As Hieronymi (2008: 223) notes, ‘Being trusted generally carries for us a kind of significance: whether
or not someone trusts is taken to reveal something of the other person’s opinion of us.’
betraying trust 83
the contrast between therapeutic trust and trust that contains an expectation of trust-
worthiness as follows:
Such trust is placed in response to what is believed to be a moral need, and so does not presup-
pose any favorable estimate of those in whom it is reposed . . . [A]lthough trust of this kind can
be disappointed it is not seriously threatened with disillusionment, for it does not rest upon
sanguine assumptions as to the nature of the person towards whom it is directed.
(Horsburgh 1960: 349)
Therapeutic trust, insofar as it represents the trusted person as under an obligation,

may make us ready to feel resentment or feel betrayal, depending on the type of obliga-
tion represented. But it will not make us ready to feel disillusioned by a failure to do as
trusted. What makes us ready to feel disillusioned by a failure is a favourable estimate
of the trusted person, something missing from therapeutic trust.
This favourable estimate, the expectation of trustworthiness, is also a benefit,19 but
its significance for us does not, unlike the benefit conferred by therapeutic trust,
depend on its being voluntarily conferred. Indeed, it wouldn’t have this significance for
us at all if it were possible to choose to expect trustworthiness from someone. The way
trust that involves an expectation of trustworthiness benefits the trusted person is by
honouring its recipient. If we could choose to believe that someone was trustworthy,
there would be no honour in being believed trustworthy. The different kinds of benefits
conferred by trust that does a favour and by trust that confers an honour can both be
seen as falling under different parts of gratitude, understood as the general virtue of
properly valuing benefits. What counts as properly valuing a benefit varies with the
nature of the benefit. As we’ll see, it is the fact that gratitude for an honour makes a
more circumscribed demand than gratitude for gifts or favours that makes it suitable
as a ground for the obligation whose violation is a betrayal of trust.
What exactly must we expect from someone for our expectation to honour them?
I’ve been referring to an expectation of trustworthiness, but now I’ll explain what we
must mean by ‘trustworthiness’ for our expectation to honour the recipient. When
A represents B as under an obligation to do X, normally there would be no honour if
A did not expect B to respect this obligation.20 So for A’s trust to honour B it must
normally at least involve an expectation that B will respect this obligation—that B will
not wrong A. I shall understand obligations as deliberative constraints. To respect an
obligation to do X we must avoid attaching weight to reasons (or at least a range of
reasons) that recommend against doing X in our practical deliberations. So if Sara
promises Andy that she will feed his mice, she could wrong him by failing to feed his
mice. But even if she decides to feed his mice she could still wrong him by giving weight
19
Pettit (1995) classifies it as an ‘attitude-dependent good’.
20
I say ‘normally’ because respecting an obligation is sometimes unjustified. In such contexts there may
be an honour in expecting the trusted person not to respect the obligation.
84 Collin O’Neil
in her deliberations to reasons that recommend against feeding the mice, such as the
inconvenience.21
To honour someone we represent as under an obligation to us, then, is normally to
expect that they will respect the obligation, where this requires that it play the right
role in their deliberations.22 But expecting the trusted person to actually respect this
obligation is not enough—one can expect someone to respect an obligation without
honouring them. One way is if one expects them to respect the obligation in a context
in which respecting it would be unjustified, such as expecting someone to respect a
seriously immoral promise. There is also another way.
Notice that we could think that someone actually will not wrong us while thinking
that they would wrong us if the situation were different in a range of morally unimportant
ways. As D’Cruz has argued, ‘if you think the babysitter will fail to take good care of the
child if her friends come visit, then you do not really trust her even if you think it is
unlikely that her friends will stop by’ (2015: 475). Expecting someone to succeed in
respecting an obligation does the trusted person no honour if one sees the success as
lucky. If you think that someone will respect a certain obligation, but only because in
the actual situation there are no temptations to do otherwise or she is unaware of those
temptations, then you see their success as too fragile to do them credit. The same is true
if you think they will succeed but only because they happen to have had a good night’s
sleep or happen to be in a good mood. So I shall follow D’Cruz in thinking that we must
see their success as stemming from a reliable disposition for it to do the trusted person
an honour.23
Here, then, is the moralized form of trust that I believe normally honours the trusted
person:
A regards B as under a certain obligation to A to do X (or not do X) and:
(1) A expects B to respect this obligation, because
(2) A believes B has a (minimally) reliable disposition to respect this obligation.
Note that the disposition must only be reliable enough for the trust to honour the
trusted person. It need not be perfectly reliable or even very reliable. Even if I think
that you would fail to respect the obligation in the face of fairly strong temptations or
distractions, my expectation that you will succeed can still honour you. The more
reliable the disposition I believe you have, the greater the honour. But I’m only concerned
here with the point at which an expectation of success would no longer confer an
honour at all—the point at which I cease to see your success as coming from you and
instead from luck.
21
The concept of obligation and the idea of respecting an obligation in this sense come from Owens
(2012: 85–95).
22
It may also be possible to respect it without fulfilling it (Owens 2012: 90), but I will ignore this
complication.
23
D’Cruz (2015: 473) sees this as a general requirement on trust, whereas I am treating it only as a con-
dition on a (moralized) form of trust that honours its recipient.
betraying trust 85
It is also important to notice that the disposition is construed narrowly: it is a

isposition to respect a particular obligation to the truster. A may trust B not to steal
d
his laptop, without trusting B to keep a secret. The fact that A does not trust B to keep a
secret does not undercut the honour in trusting B not to steal. The honour in A’s trust is
limited in scope: it is only about a reliable disposition in connection with a particular
obligation.
If there is a distinctive emotional vulnerability associated with this form of trust, it
would be a feeling of resentment or betrayal combined with a sense of disillusionment,
as Horsburgh suggests. But characterizing the feeling is not my primary concern.
What is morally significant about this form of trust is that it confers an honour, albeit
limited in scope, on the trusted person. An honour, like a gift or favour, is a benefit. But
these are very different kinds of benefits, and the demands of gratitude, construed as
the general virtue of properly valuing benefits, vary accordingly.
The demand that gifts and favours make is indefinite when it comes to reciproca-
tion, but definite and unlimited when it comes to avoiding wronging the benefactor.
Someone who properly values a gift or favour will be disposed to reciprocate in some
way or other, or at least acknowledge it. But wronging the benefactor is incompatible
with properly valuing the gift or favour. And this demand extends to any way of wrong-
ing the benefactor: the demand is unlimited in this sense.
Honours, by contrast, do not demand reciprocation. Properly valuing an honour,
however, does demand something definite. It demands that you vindicate that honour,
that is that you live up to the good opinion it expresses. If you are honoured with a job
or a responsibility, in the sense that the choice to offer you the job or responsibility
expresses a good opinion of you, not a desire to do you a favour, then you fail to prop-
erly value that honour if you are not disposed to make yourself deserving of it, what-
ever else you do. Properly valuing the honour also seems to make no further demands.
Were I to slack off at work, this would normally display a failure to value the good
opinion of my employer. But if I know the good opinion is limited to my talents, not my
work ethic, then I could not in slacking off be accused of failing to appreciate their
good opinion.
4. Betrayals of Trust: The Account

I’ve identified a form of trust that confers an honour. Is this also the form of trust that
can be betrayed? I’ll begin by evaluating the substance of such an account before turn-
ing to the form of the obligation it yields. As we saw earlier, if therapeutic trust were the
only trust that could be betrayed, then this would yield a far too revisionary account of
betrayals of trust. Therapeutic trust is an unusual form of trust. But trust that involves
an expectation of trustworthiness is not uncommon. Indeed, as I mentioned, there are
a number of writers who argue that trust that involves an expectation of trustworthi-
ness is the only genuine kind of trust.
86 Collin O’Neil
The idea that, in betraying trust, we are failing to live up to a demand placed by an
honour is also attractive. When we add the complaint, ‘I trusted you’, to a complaint
about a wronging, part of what we mean may sometimes be that the wronging was
more damaging because of some vulnerability connected to our trust. But the com-
plaint ‘I trusted you’ can also be understood as a reaction to a special kind of insult. We
have a sense, when our trust is betrayed, that our good opinion has been treated dis-
missively or repudiated. As I mentioned earlier, properly valuing our good opinion is
not a matter of caring about our good opinion by regarding it as desirable and wanting
to keep it. If the trusted person had taken great pains to conceal from us the fact they
had failed to do as trusted, or after failing to do as trusted made efforts to regain our
good opinion, this would show that they valued our good opinion in the sense of
wanting to keep it or reacquire it. But this would not assuage the special kind of insult
delivered by failing to live up to trust. To properly value our good opinion requires
vindicating it. And it can be properly valued in this way even if the trusted person does
not care about keeping it. Someone might strive to live up to our good opinion, with-
out bothering to let us know or even carelessly allowing us to form the false impression
that our good opinion was unjustified.
So I think that the substance of this account—that a betrayal of trust consists in fail-
ing to live up to the honour bestowed by trust—is plausible. But a satisfying account of
betrayals of trust must also yield second-order obligations with the right form. Does it?
The obligation to properly value this form of trust by vindicating the honour it expresses
is special. Only the trusted person can have this obligation, since only the trusted per-
son is the recipient of this honour. And the trusted person is under this obligation only
to the truster, since it is the truster who honours the trusted person. The obligation is
also trust-based, in the sense that the obligation would not exist but for the trust and in
particular the elements of trust that distinguish it from related attitudes. The expect-
ation of trustworthiness is essential. Without it, there is no honour to live up to.
It is also knowledge-based. There is no insult in failing to live up to an honour one
did not realize had been bestowed. Now, given that this form of trust requires not only
an expectation that the trusted person will not wrong us, but also that they still would
not wrong us were the situation different in a range of ways, it will not always be o bvious
whether we are trusted. The fact that the truster does not actually monitor or constrain
the trusted person is not decisive. The truster could fail to monitor or constrain because
they see us as reliably disposed against wronging us, or they could fail to monitor or
constrain because they do not think the situation calls for it, even while thinking that if
the situation were different in morally unimportant ways, monitoring or constraining
would have a point. In the case of intimates, I think we often know whether and how
we are trusted, since we have had many interactions. But in one-off interactions, such
as between strangers, it may be the case that we cannot know we are trusted unless, in
the actual situation, there are obvious opportunities and temptations to default and the
truster expects us to succeed in respecting the obligation despite these temptations
and distractions.
betraying trust 87
The obligation will also be an obligation to act specifically as trusted. The form of
trust represents the trusted person as under a first-order obligation to do X. The
honour in trust is precisely an expectation of trustworthiness with respect to this
obligation. So living up to this honour is a matter of respecting this obligation. Suppose
that Sara promises Andy that she will feed his mice and also knows that Andy trusts
her to keep the promise. Any failure to keep this promise, even with a timely warning,
would violate this second-order obligation since this would wrong him, and in trust-
ing Sara he is doing her the honour of expecting her not to wrong him.
The account of betrayals in terms of ingratitude for the gift or favour delivered by
therapeutic trust was also able to yield a specific trust-based obligation to do as trusted.
But it failed the exclusivity condition because it yielded additional obligations as well
as an obligation to do as trusted. An account of betrayals of trust in terms of ingrati-
tude for the honour delivered by trust succeeds where an account in terms of ingrati-
tude for a gift or favour delivered by trust fails. The honour bestowed by expecting
trustworthiness from someone in connection to a particular obligation is an honour
that is limited in scope. And the demand is just to live up to this particular honour. Sara
would be ungrateful for the honour bestowed by Andy’s expectation of trustworthi-
ness from her in connection to her promissory obligation if she did not properly value
this honour by living up to it. But in stealing from him or in wronging him in any other
way, she would not thereby show that she failed to properly value this limited honour.
Conclusion
I have indicated why I think the substance of this account of betrayals of trust is plaus-
ible. The obligation generated by this kind of trust also satisfies the necessary formal
conditions. In originally setting out these conditions I left two questions open. One
question was whether trust could sometimes create first-order obligations or was only
capable of generating second-order obligations, which I’ve already addressed. The
other was whether trust must be invited or accepted before it can create an obligation,
which I will address now.
Notice that the trust-based obligation I’ve described is special, even without an act
of invitation or acceptance from the trusted person. An obligation is special if it is
impossible for anyone else to have it, and since the trusted person is the only one
honoured by the trust, the trusted person is the only one who can have the obligation.
One concern is that if a new obligation could be unilaterally legislated by the truster,
this would be inconsistent with the trusted person’s status as a moral equal (van der
Vossen 2015: 65–7). But I do not think this worry applies in the case of this trust-based
obligation. First, since the form of trust that creates this obligation incorporates an
expectation of trustworthiness, the truster cannot decide to trust. Thus, even if
the trusted person cannot decline this obligation, this does not make the trusted
person’s obligations subject to the truster’s will. Second, the trust-based obligation is a
88 Collin O’Neil
second-order obligation, which means that it inherits its content from an already
existing obligation. The truster may be creating a new trust-based obligation, but since
it overlaps with the prior obligation, there is no new imposition. Because of this, the
trusted person does not require the protection that conditioning the trust-based
obligation on his or her invitation or acceptance would confer. Therefore, so long as the
trusted person is genuinely under the prior obligation represented by the trust, trust is
an honour that cannot be declined.24
References
Baier, Annette. 1986. Trust and Antitrust. Ethics 96(2): pp. 231–60.
Cogley, Zac. 2012. Trust and the Trickster Problem. Analytic Philosophy 53(1): pp. 30–47.
D’Cruz, Jason. 2015. Trust, Trustworthiness, and the Moral Consequence of Consistency.
Journal of the American Philosophical Association 1(3): pp. 467–84.
Friedrich, Daniel, and Nicholas Southwood. 2011. Promises and Trust. In Promises and
Agreement: Philosophical Essays, ed. Hanoch Sheinman. Oxford: Oxford University Press.
Frost-Arnold, Karen L. 2008. The Epistemological Importance of Trust in Science. PhD
dissertation, University of Pittsburgh.
Harding, Matthew. 2011. Responding to Trust. Ratio Juris 24(1): pp. 75–87.
Hawley, Katherine. 2014. Trust, Distrust and Commitment. Noûs 48(1): pp. 1–20.
Hieronymi, Pamela. 2008. The Reasons of Trust. Australasian Journal of Philosophy 86(2):
pp. 213–36.
Holton, Richard. 1994. Deciding to Trust, Coming to Believe. Australasian Journal of Philosophy
72(1): pp. 63–76.
Horsburgh, H. J. N. 1960. The Ethics of Trust. Philosophical Quarterly 10(41): pp. 343–54.
Jones, Karen. 1996. Trust as an Affective Attitude. Ethics 107(1): pp. 4–25.
Jones, Karen. 2004. Trust and Terror. In Moral Psychology: Feminist Ethics and Social Theory,
ed. Peggy DesAutels and Margaret Urban Walker, pp. 3–18. Oxford: Rowman & Littlefield.
Jones, Karen. 2012. Trustworthiness. Ethics 123(1): pp. 61–85.
Keren, Arnon. 2014. Trust and Belief: A Preemptive Reasons Account. Synthese 191(12):
pp. 2593–615.
McGeer, Victoria. 2008. Trust, Hope and Empowerment. Australasian Journal of Philosophy
86(2): pp. 237–54.
McLeod, Carolyn. 2002. Self-Trust and Reproductive Autonomy. Cambridge, MA: MIT Press.
Nickel, Philip J. 2007. Trust and Obligation-Ascription. Ethical Theory and Moral Practice
10(3): pp. 309–19.
O’Neil, Collin. 2012. Lying, Trust, and Gratitude. Philosophy & Public Affairs 40(4):
pp. 301–33.
Owens, David. 2012. Shaping the Normative Landscape. Oxford: Oxford University Press.
24
Thank you to Paul Faulkner, David Owens, and Thomas Simpson for valuable written comments on
earlier drafts; to Marcello DiBello for helpful discussion; and to the audience at the Mount Sinai Working
Papers in Ethics and Moral Psychology Series, organized by Nada Gligorov, where I presented an earlier
version of this chapter.
betraying trust 89
Pettit, Philip. 1995. The Cunning of Trust. Philosophy & Public Affairs 24(3): pp. 202–25.
Pink, Thomas. 2009. Promising and Obligation. Philosophical Perspectives 23(1): pp. 389–420.
Shiffrin, Seana V. 2011. Immoral, Conflicting and Redundant Promises. In Reasons and
Recognition: Essays on the Philosophy of T.M. Scanlon, ed. R. Jay Wallace, Rahul Kumar, and
Samuel Freeman. Oxford: Oxford University Press.
Simpson, Thomas W. 2012. What Is Trust? Pacific Philosophical Quarterly 93(4): pp. 550–69.
Van der Vossen, Bas. 2015. Imposing Duties and Original Appropriation. The Journal of
Political Philosophy 23(1): pp. 64–85.
Walker, A. D. M. 1988. Political Obligation and the Argument from Gratitude. Philosophy &
Public Affairs 17(3): pp. 191–211.
Walker, Margaret U. 2006. Moral Repair: Reconstructing Moral Relations after Wrongdoing.
Cambridge: Cambridge University Press.
6
‘But I Was Counting On You!’
Karen Jones
The goal of this chapter is to understand the normative status of the complaint, ‘But
I was counting on you!’, where this complaint is not reducible to or replaceable by other
nearby complaints such as ‘But you promised!’ or ‘How could you be so callous!’
Do reasonable (in some sense yet to be explained) countings on give rise to legitim-
ate normative expectations and if so, in virtue of what under-girding norm or norms,
of what normative kind?
There are many reasons to want a better understanding of the normative force of this
complaint, including the variety of contexts in which we are potentially vulnerable
to it, the way its preemptive reminder, ‘I’m counting on you’ is used to scaffold agency,
and the fact that vulnerability to it marks cases where we have moved beyond mere
convergence of agency to form a genuine ‘we’. But the chief reason to be interested in it,
I will argue, is that it is the signature complaint of those who think that their trust has
been betrayed. Thus, by coming to understand whether, and if so when and why, the
complaint has normative force, we can explore the shape, source, and force of central
norms governing trust and trustworthiness.1 Many moralize those norms and analyse
trust as ascribing, or even grounding, an obligation to be trustworthy (Lagerspetz 1998;
Nickel 2007) and view trustworthiness as requiring moral integrity or commitment
to morally defensible values (MacLeod 2002; Potter 2002). Against such moralizing
moves, I argue that the norms of trust and trustworthiness are not themselves
moral, have the potential to sit uneasily with moral norms, and require external
moral regulation.
The chapter proceeds as follows: Section 1 unpacks the question and explains why it
matters. Section 2 examines and rejects Scanlon’s attempt to ground the normative
force of the complaint in a moral principle enjoining us to exercise care in the expectations
that we lead others to form about what we will do. I argue that there is no such moral
1
Will tracing the norms that govern use of the complaint ‘But I was counting on you!’ give a complete
mapping of the norms governing trust and trustworthiness? Almost certainly not, but I don’t address this
question here, which is why my thesis is only that by tracing the force of this distinctive complaint we can
come to an understanding of central norms governing trust and trustworthiness.
‘but i was counting on you!’ 91
principle. Section 3 proposes an alternative ground for the normative force of the com-
plaint ‘But I was counting on you.’ Its force, when it has it, derives from the pressing
interest that we have as finite, reflective, and social creatures in being able directly to
recruit the agency of another. It is this interest that gives the concepts of trust and trust-
worthiness their distinctive conceptual role. Norms of trustworthiness are not redu-
cible to norms of morality and have the potential to sit uneasily with the demands of
morality. By exploiting a structural analogy with friendship, I argue that, nevertheless,
they do have genuine normative force. In the fourth and final section, I begin the task
of mapping the contours of that subset of norms governing t rustworthiness and rea-
sonable trust that reveal themselves in what we take to be acceptable rejoinders to the
complaint, ‘But I was counting on you.’
1. The Significance of the Complaint

To count on something or someone is to embed in your plans and goals an expectation
that, if false, means you risk being left worse off than you otherwise would have been.
Counting on is related to the notion of predictive expectation, or expectation that
(Hollis 1998), but having a predictive expectation is neither necessary nor sufficient for
counting on something or someone. The toppling of predictive expectations can be met
with mere surprise, even pleasant surprise. Not so when the things that we count on
coming to pass fail to occur. That failure comes, by definition, at a cost and so is met with
disappointment, frustration, let down, or—in cases where what we are counting on is
the agency of another and our relationship is one of trust—with feelings of betrayal.
I can count on something coming to pass even when I lack the predictive expectation
that it will. I might simply have no choice but to count on something or someone despite
grave doubt. For instance, I might plan my escape from the fire on the assumption that
the rope, old and frayed though it seems, will nonetheless bear my weight, because the
possibility of death from a fall is better than the certainty of death from the approaching
fire. It seems, however, that so long as I am engaged in planning and not mere fantasy,
I cannot count on something which I am quite certain will not come to pass.2 Just as
countings on can be forced, they can also be unwitting. Sometimes we realize what we
were counting on only when it doesn’t happen and our plans are wrecked.
The distinct complaint that I am interested in understanding is one that is made only
in respect of a subset of failed countings on, those involving fellow agents. Interestingly,
however, the complaint or its preemptive reminder, ‘I’m counting on you!’ can be
directed towards agents, such as small children, who do not yet have the capacities
that fit them for being held accountable, but who have the potential to develop such
2
Compare Holton (1994: 65) on reliance in the face of doubt. Is counting on, then, simply reliance? On
some ways of understanding reliance, perhaps. However, the notion of reliance is closely tied to action—it
seems I can choose to rely, whatever I believe, simply by, in our example, grabbing the rope and jumping.
The notion I’m trying to capture is tied to deliberation and planning in a way that it seems reliance is not.
92 Karen Jones
capacities. Developmentally, the preemptive reminder is a significant part of the scaf-

folding that goes into making and sustaining our capacity for trustworthiness and for
responsible agency in general. Consider the commonplace example of a small child
being told, ‘You are a big girl now and Mummy is counting on you.’ By addressing the
child in this way, the mother is presenting to them an image of themselves as compe-
tent and worthy of the trust that is being placed in them. That image—of something
not yet quite actual, but affirmed as a possibility for the child—motivates them to live
up to the expectations being placed on them (McGeer 2008). Nor is the preemptive
reminder made only to the young. Reminding people that we are counting on them
can galvanize agents whose wills we fear are weakening and bootstrap responsiveness
to the fact of our dependency among those whose past track record would suggest that
counting on them is unduly risky.
The role that the complaint, ‘But I was counting on you!’ and its preemptive reminder
plays in creating and sustaining the agency of others is reason enough to be interested
in understanding when it has normative force and from whence that force derives, but
it is not the only reason. The sheer ubiquity of the potential for this complaint or its
reminder provides further grounds. It can be made in contexts involving any of the
following: shared activity, collective intention, conventions regarding how ‘we’ do
things around here, and even abrupt changes to regularities in our behaviour that are
known to be known. The potential for making such a complaint is the hallmark of any
case where we assume we have a genuine ‘we’ rather than mere convergence of agency.
Understanding whether, and if so, when and why, it has normative force is thus central
to understanding truly collective endeavour in its various forms. It is also a hallmark of
relationships that are assumed to be non-coercive, as is shown by reflecting on the
oddness of a master using it towards a non-compliant slave.
I have a further, theory-driven, interest in understanding the complaint’s distinctive
force. On my preferred ‘counting on’ theory of trust and trustworthiness, the signature
reason to which the trustworthy are responsive is the fact that others are counting on
them; trust in turn comprises, among other things, an expectation of such responsive-
ness (Jones 2010; 2012; section 3 below). On a counting on theory of trust and trust-
worthiness, the question of the normative force of demands to be trustworthy and the
question of who may appropriately make them towards whom amounts to the ques-
tion of when and why complaints against those who let us down have normative force.
The relationship between prima facie legitimate complaints of non-responsiveness to
dependency and trust will become clearer in the next section when I examine Scanlon’s
proposal, which fails to distinguish, as it must, between trust and mere reliance.
2. Scanlon’s Explanation of the Normative Force

of ‘But I Was Counting On You!’
In the literature, the nearest thing to an explicit discussion of the normative force of
the complaint, ‘But I was counting on you’, is to be found in Scanlon’s proposal for the
principles that should govern the creation of expectations in other people concerning
what one will do in contexts where they will suffer a loss if they rely on them. These are
cases where the other is counting on us, and so, if Scanlon is right, the principles will
identify cases in which our complaint has normative force. On Scanlon’s account, the
force of the complaint is moral, since the complaint, when legitimate, is grounded in
those principles governing the creation of such expectations that are part of what we
owe to one another. Scanlon identifies the following principles:
Principle D: One must exercise due care not to lead others to form reasonable but
false expectations about what one will do when there is reason to believe that they
would suffer significant loss as a result of relying on those expectations.
Principle L: If one has intentionally or negligently (that is to say in violation of
Principle D) led someone to expect that one is going to follow a certain course of
action X, and one has reason to believe that that person will suffer significant loss as
a result of this expectation if one does not follow X, then one must take reasonable
steps to prevent that loss. (Scanlon 2003: 239–40)
Recall that, according to Scanlon, we test whether an action is right or wrong by find-
ing out whether it would be: ‘permitted by principles that could not reasonably be
rejected by people who were moved to find principles for the regulation of behavior
that others, similarly motivated, could not reasonably reject’ (Scanlon 1998: 4). The
principles must be, in Pettit’s phrase, ‘contractually irresistible’ (Pettit 2000: 149). To
work out whether a principle meets the test of contractual irresistibility, we first iden-
tify the reasons for having such a principle and then ask, could the principle nonethe-
less be reasonably rejected?
Scanlon’s argument in favour of Principle D is remarkably brief: ‘Its validity consists
just in the fact that one can reasonably refuse to grant others license to ignore the costs
of the expectations that they lead one to form’ (Scanlon 2003: 239). No possible
grounds on the basis of which we might allow the other to ignore the effects of the
expectations that they create are investigated. But such grounds are easily found;
indeed, we frequently absolve one another from all accountability for failing to fulfil
an important class of expectations that we have led others to form. Consider share
trading and other zero-sum games. Share trading occurs in the context of expectations
about how others will behave, as do decisions about when to use the tennis courts if
you want to find them uncrowded, and so on. These expectations can be epistemically
well grounded in solid inferences about what others will do on the basis of past behav-
iour or awareness of their reasons for acting. Not only can the other form a justifiable
belief about what we are likely to do, we recognize that they are likely to form such
expectations and embed them in their plans in ways that matter to their success.
Nevertheless, we do not take ourselves to be obliged to warn them off should we intend
to act other than as expected. In examples of this kind, others are counting on us and
we recognize that fact. However, the context is one in which we are each going about
94 Karen Jones
our business. Though others are counting on us, there is no further expectation that we
will respond to the fact of their counting on us. We are relying on the other’s behaviour,
but not trusting them to behave in that way.
If Principle D fails the test of contractual irresistibility, then Principle L, which pre-
supposes D, fails also. The share trading example suggests that Principle D is drawn too
broadly because it enjoins reliability rather than trustworthiness, and reliability is not
something we can demand of one another. A simple solution suggests itself: revise D,
narrowing it down, so that it becomes a principle enjoining trustworthiness.
Principle RT: One must exercise due care not to lead others to form reasonable but
false expectations about one’s competence and willingness to be responsive to the
fact of their dependency when there is reason to believe that they would suffer sig-
nificant loss as a result of relying on those expectations.
I call this principle, Principle RT, because it is a principle enjoining what I have
elsewhere identified as rich trustworthiness:
B is richly trustworthy with respect to A just in case: (i) B is willing and able reliably to signal to
A those domains in which B is competent and will take the fact that A is counting on her, were
A to do so, to be a compelling reason for acting as counted on and (ii) There is a non-trivial
number of relatively central domains in which B will be responsive to the fact of A’s depend-
ency in the manner specified in (i). (Jones 2012: 74)
The notion of rich trustworthiness requires further explication. As finite social crea-
tures we want there to be other agents who will willingly meet us in our dependency;
that is, agents who will be responsive to the fact that we are counting on them in a
particular domain. However, this is not all that we want: our knowledge and time is
limited and it can be hard to identify those who have the competence and willingness
to respond to our dependency in a particular domain. Better yet if people are not only
responsive to our dependency in some domains but if they also help us by identifying
themselves, by signalling that they are up for, or not up for, a particular kind of depend-
ency from people like us. Those who take this further step in meeting us in our
dependency I call the ‘richly trustworthy’.
Is RT contractually irresistible? I have, in effect, just made the case in favour of it
being so. As finite social beings, we have an interest in being able to recruit the
assistance of others to expand the effectiveness of our agency. Principle RT would let us
do this safely for it tells us that we each must take due care in monitoring the way others
form expectations about those areas in which we will be competent and willing to
respond to their dependency. When they are at risk of forming an expectation that we
are either unable or unwilling to meet, we ought to signal out.
Are there nevertheless grounds on which the principle could be reasonably rejected?
It seems to me that there are. First, the principle, as moral principle, is universalized;
that is, it demands due care with respect to any arbitrary other. Rich trustworthiness,
in contrast, is defined as a relation between two individuals, or groups of individuals.
It is not clear that the interest that we have in there being others who will help us locate
where to place our dependency is an interest that extends universally. We need there to
be at least some such people, but from that it doesn’t follow that we have a reason to
demand this from everyone.
Second, and more decisively, the principle is extremely burdensome and burden-
someness is a recognized basis on which one might reasonably reject a principle
(Scanlon 1998: 205). Scanlon and those influenced by him (e.g. Alonso 2009) don’t
appreciate this because they haven’t thought about how we form expectations about
what others will do and thus they haven’t thought about what it would take to monitor
those expectations and head off ones that, though reasonable, we will not or cannot
meet. The principle needs to be tested for reasonableness in the light of a realistic
account of expectation formation. The story of how we generate or overrule expect-
ations in others is the story of signalling and it is very complex.3 Signalling operates
through mechanisms from formal (certification) to informal (clothes and comport-
ment). It takes place against a vast social background, including norms and shared
understandings of what can be expected of whom. This means that we are always
already signalling and what we are signalling exceeds our control and, often enough,
our awareness. The capacities required for good signalling are numerous, including
self-directed abilities such as the ability to recognize the zones of one’s competence and
to monitor new expectations for compatibility with current ones, as well as other-
directed abilities such as the ability to recognize the expectations different kinds of
people are liable to form of us, and what they will count as a signal, whether out or in.
Because of the social embeddedness of signalling, there is reason to believe that it is
impossible to be a good signaller with respect to all potential recipients and the burden
even of trying is large. For this reason, Principle RT can be reasonably rejected on the
basis of burdensomeness.
It might be thought that if the problem lies in the way signalling exceeds the control
of both sender and recipient then the solution lies in a more restrictive principle that
governs only explicit intentional signalling. Consider:
Principle E: One must not intentionally create in others false expectations about
what one will do, especially in contexts in which they are liable to count on you
doing it.
Alternatively, in case we are worried that this principle would rule out contexts in
which bluff and double bluff are not morally objectionable (share trading, perhaps?),
we could modify it to:
Principle EIT: One must not intentionally create in others false expectations that
one will be responsive to their dependency in a domain, especially in contexts in
which they are liable to count on one being responsive.
3
For a fuller discussion of signalling, see Jones (2012: 74–6). For an exploration of failures in uptake of
signalling, see Jones (2013a).
96 Karen Jones
I label this principle, Principle EIT, because it amounts to saying that you should not
explicitly invite another’s trust when you are not going to respond to it with
trustworthiness.
There is a good argument that Principle EIT passes Scanlon’s test: unlike Principle
RT, it is not unduly burdensome and fulfilling it is within the agent’s control. The case
in favour of it is as strong as the case in favour of principles ruling out deceitful actions,
for it identifies a subclass of such action. However, it is a principle that is clearly too
weak to ground the normative force of the charge, ‘But I was counting on you’ or its
preemptive reminder, in anything like the range of cases in which we are inclined to
make it. In only a small portion of cases is what we complain about some kind of deceit,
as Principle EIT suggests. Moreover, where such deceit is in play, the complaint can be
reduced to or replaced by other more familiar moral complaints, such as ‘How could
you lie to me!’
The thought that, if grounded in Principle EIT, the charge is reducible to other moral
complaints might start to make one sceptical that it ever has any independent norma-
tive force, that such force as it has is always derivative of other morally salient features
of the situation. Further reflection about who directs that charge towards whom with
what effect strengthens the case for scepticism. The charge can, and often enough does,
function ideologically to maintain existing patterns in behaviour, patterns which, in
societies characterized by relations of dominance and subordination—and I take
all actual human societies to be so characterized—maintain unjust social relations.
Consider the following ordinary encounter, occurring in a context characterized by
unequal divisions of domestic labour: ‘But you always get the milk!’ The tacit com-
plaint here is that failure to get the milk without due warning amounts to letting the
other down and is therefore, at least to a degree, morally problematic. There is a famil-
iar dynamic in which any regularity in behaviour, which is known and known to be
known, comes to be invested with normative expectations and not merely predictive
ones. Because you know that I know that this is what you usually do, you should know
that I am likely to embed in my plans the assumption that you will do it in ways that
have the potential to make me incur a loss if you do not. Knowing all that, if you fail to
do as I am counting on, you ignore my dependency and so are fit target for the com-
plaint, ‘But I was counting on you!’ The threat of the complaint generates inertia in
favour of things continuing as they are now. It makes ‘trickster’ behaviour—calculated,
often covert, attempts to subvert unjust social relations—subject to moral censure
insofar as such behaviour disrupts the status quo and this is so, even if all things con-
sidered disrupting those relations is the right thing to do (Frost-Arnold 2014). No
other complaint could be plausibly made in this context and making this one looks
suspect. Perhaps, then, the complaint is suspect whenever it purports to have inde-
pendent standing.
Showing that complaints of failure to live up to others’ expectations can be used
ideologically does not, of course, show that they don’t have genuine normative force
when used correctly. Nor do I take myself to have established that there is no moral
option for explaining that force when, and if, the complaint has it. I’ve only shown that
the nearest thing to a defence that has actually been offered does not work. Perhaps
other defences could work better.4 Rather than anticipating what those other defences
would look like, and responding to them seriatim, I propose to sketch an alternative
account of the force that the complaint has that is not grounded in morality, but rather
in a distinct set of norms for trust and trustworthiness that answer to the pressing
interest that we have in there being people whose agency we can directly recruit to
enhance the effectiveness of our own. Thus any account that moralizes the force of
these norms mistakes the interest we have in them, or so I shall argue.
3. An Alternative Account of the Normative

Force of ‘But I Was Counting On You!’
Earlier, I noted that preemptive reminder of the possibility of complaint contained in
the exhortation, ‘I’m counting on you!’ is a routine part of the way in which we scaffold
the agency of others, whether that is developmental scaffolding, or scaffolding that
bolsters the capacity and resolve of adult agents.5 We pay attention to instilling and
maintaining sensitivity to the fact that others are counting on us. It is one of our core
‘people-making’ practices. Why is that? What value or interest do these practices serve
and is that value or interest best served by training up a limited sensitivity that comes
with a moral filter or by training up a sensitivity that is not constrained in this way, and
thus a sensitivity that may require external moral regulation, rather than having such
regulation built into it? If, as I shall argue, that value is not moral and so is not best
served by training up a filtered sensitivity, then our norms for trustworthiness will
exist in potential tension with moral norms, much as the norms of friendship do, on
some accounts of friendship.
Let’s turn briefly to the case of norms of friendship, which offers a structural analogy
for how to think about norms of trust and trustworthiness, their relation to moral
norms, and the source of their normative force. Suppose we wanted to know what
norms should govern relationships of friendship and how those norms stand with
respect to other norms, such as moral, prudential, or epistemic norms.6 The way to
answer this question is first to inquire what friendship is. Knowing that, we then ask
4
The virtue ethicist might try to ground it in the claim that trustworthiness is a virtue. When we dem-
onstrate untrustworthiness, we are subject to censure, just as we are when we behave in a cowardly fashion.
Elsewhere I have argued that trustworthiness is not a virtue, so I take this option to be closed (Jones 2012).
A more promising line is to ground the complaint in norms that are defended on either rule or indirect act
consequentialist grounds. Though I cannot defend this claim here, I doubt such grounding will work
because the norms do not, even by and large, track the greater good.
5
That there’s a parallel here should not come as a surprise, if, as McGeer (2008) argues, maturity is not
something that once achieved belongs to us alone. Creating and sustaining agency are continuous
processes.
6
For a defence of the claim that norms of friendship can conflict with epistemic norms, see Keller
(2004).
98 Karen Jones
what traits, skills, and norms enable us to be good qua friend. Our inquiry could not
end there, however, for the identified norms would only have genuine force if there was
something good about friendship—some worthwhile point or purpose to the practice
of friendship, or something valuable in it, considered in itself. Given that friendship is
universally taken to be a constituent of a flourishing life, that friendship has value is a
platitude. An answer to the ‘what is’ question that analysed friendship as a relationship
without value of any kind could be rejected on that ground alone. But not all roles and
relationships which generate norms are like friendship in this respect. Though one can
be good qua master or qua slave and there are norms regarding dominance and sub-
missiveness which enable excellence in those roles, the roles are themselves corrupt, as
are the norms that go along with them. Since the norms for master and slave support
relationships that lack all value, the norms have no genuine normative force.7
The nature and value of friendship is disputed, with views ranging from the highly
moralized, which sees friendship as about scaffolding virtue, to the resolutely non-
moralized, which sees friendship as about the kind of intimacy that is made possible by
openness to having one’s interests and agency directed by the other (Cocking and
Kennett 2000). Intermediate positions are possible, too. On views of the former kind,
there is no tension between friendship and morality, because friendship itself is of
moral value, so potential conflicts between, for example, what we owe our friends and
what we owe strangers are conflicts between competing moral values, resolvable (if
they are—I do not mean to take a stance on the possibility of genuine moral dilemmas)
within morality. Similarly, acts from friendship are moral acts, and one cannot be a
good friend without also being a good person.
On a view of the latter kind, in contrast, there is the possibility of tension between
friendship and morality, because the value of the kind of intimacy that friendship
makes possible is not itself a moral value—such relationships can even lead you into
moral danger. Further, again in contrast with moralized conceptions of friendship,
acts from friendship might or might not be moral acts, and one can be a good friend to
someone without also being a good person: ‘I might be a perfectly good friend. I might
just not be a perfectly moral one’ (Cocking and Kennett 2000: 287). However, so long
as there is indeed a value to having relationships of this kind, then the norms that make
one good qua friend will have genuine normative force, just not a force based in moral
values, as it was on the earlier account.
To accept that there can be norms of friendship which are not reducible to moral
norms and yet have genuine normative force is not to take a stand on what should be
done when friendship and morality conflict. For all that’s been argued, in cases of con-
flict, the reasons of friendship must give way to the reasons of morality, since morality
is overriding. Nor have I defended this non-moralized conception of friendship. My
point is to understand the structure of an argument that shows the possibility of norms
7
This strategy for identifying and evaluating the force norms is modelled on Haslanger’s (1993) explor-
ation of norms of femininity and masculinity.
with genuine normative force, whose force is not moral. Whatever you think about the
norms of friendship, I will argue that you should think that the norms governing rela-
tionships of trust are norms like this. Moreover, this is no accident because relation-
ships of trust and trustworthiness enable the kind of direct recruitment and shaping of
the agency of another that this account of friendship claims is characteristic of friend-
ship. From this, however, nothing follows about what, all things considered, an agent
should do when norms of trustworthiness conflict with norms of morality.
This brief inquiry into the nature and status of norms of friendship provides a tem-
plate for answering the question whether norms governing trust and trustworthiness
have genuine normative force and, if they do, what the source of that force is. Our
answer must have three stages: (i) an answer to the question of what it is to trust and to
be trustworthy; (ii) an inquiry into those skills, norms, and so on that make you a fit
participant for trust relations, so understood; and (iii) an inquiry into the value of rela-
tionships of this kind. In this section I’ll focus on (i) and (iii) since, as with friendship,
it is a constraint on an adequate account of trust and trustworthiness that it be able to
explain the value we take trust relations to have, so these questions become intertwined
(Jones 2013b). I return to the question of the skills and norms that make one a fit
participant in trust relations in section 4.
My preferred counting on theory offers the following definitions of the paired
concepts of trust and trustworthiness:
Three-place trustworthiness—B is trustworthy with respect to A in domain of inter-
action D, if and only if she is competent with respect to that domain, and she would
take the fact that A is counting on her, were A to do so in this domain, to be a compel-
ling reason for acting as counted on. (Jones 2012: 70–1)
Trust—A trusts B in domain of interaction D if and only if, A has an attitude of
optimism that B’s competence and responsiveness to the fact that she is being
counted on will extend to cover that domain.
As already discussed in section 2, those willing to take the further step and signal those
domains in which they are three-place trustworthy I call ‘richly trustworthy’.
The account is defended not by fit with intuitions about particular cases, since
intuitions vary and are unstable, but rather by a conceptual role argument that begins
by looking at the normative point of having such concepts. What conceptual work do
we want them to do and why do we bother to make people who are not just reliable, but
who are trustworthy and capable of placing their trust well? As finite social creatures
other agents are a particularly salient source of risk to us, but they also provide a rem-
edy for our finitude, for together we can do what we cannot do alone, whether because
the activity itself is necessarily a shared one (waltzing), or because it requires divisions
of time, labour, and skill (most activities in our daily lives). Our reflectiveness provides
a remedy, by enabling us to recruit our sociality to solve the problem of our finitude.
Because we have a theory of mind, we can make decisions that take into account the
mental life of others. There is thus available to us a distinctive way of responding to the
100 Karen Jones
fact of other agents’ dependency through recognizing that very dependency. Nor does
the power of our reflectiveness end there. We each know that the other is able to take
into account the ways in which we depend on them and we can come to count on the
other responding to our counting on them. They, in turn, can recognize that we are
doing this and can respond to this new way in which they are being counted on and
may do so even when they would not have responded to first-level dependency
(Jones 2012: 64). This dual structure of dependency—counting on the other, in a
domain, and counting on them to respond to the fact that we counting on them—is the
heart of trust.
We can get clearer about the value of trusting relationships in which trust is met
with trustworthiness by asking ourselves what would be missing in a world without
trust or trustworthiness as I understand them (Jones 2010: 78–9). Let’s not cheat in our
thought experiment by imagining a world in which people never help each other and
in which they lie, cheat, steal, and murder—a world in which life is nasty, brutish, and
short. Let’s instead imagine a world in which the majority of people follow what we
take to be the standard set of norms, moral and otherwise, setting aside only norms, if
any such there be, that pertain to trust and trustworthiness. Let’s make it even harder
by supposing that the behaviour of others is fully predictable: they never misjudge
what they have most reason to do, they are never akratic, and so they always act as they
have most reason to act. Further, each knows what the other will take to be a reason,
and each knows the norms and principles that the other subscribes to. These last
assumptions amount to mutual transparency and full rationality.
This is a world in which, as far as the agency of others goes, I need never risk being
let down. I can predict what the other will intend to do, and so, provided the world
cooperates enough for action to match intention, I need never fear building into my
plans a false assumption about what others will do. My plans are safe from wreckage by
other agents. This world would no doubt be much safer than the actual world, yet still
there is something missing in it, and something that we have reason to value. Because
no one recognizes the fact of dependency to be a reason, agents would lack the capacity
to directly enlist the agency of another in the service of their ends. Moreover, they
could not have the confidence that, sometimes, another would meet them in their
dependency in a domain when norm-governed and other reasons had run out for
them. Being able to count on people meeting us in our dependency makes possible the
distinctive vulnerability of intimacy.
In this world, you could work with the agency of others, or around it, but it is not
available for you to recruit in the distinctive way that responsiveness to the fact of
dependency makes possible. In deciding whether to count on them to do Z, you have
to decide whether, independent from the fact of your counting on them, they have
sufficient antecedent reasons (e.g. moral or self-interested reasons, including those
deriving from threat advantage) to do Z. If they do not, then you cannot count on them
in that way. When we add trustworthiness to this world we introduce sensitivity to the
reason, ‘she’s counting on me’ and with it, the possibility of giving the one who is being
counted on a direct reason to act as they are being counted on to act that did not exist
prior to that very counting.8
The possible world we have been exploring is distant from the actual world. The
importance of being able directly to recruit the agency of others is even clearer once we
relax some of the wildly implausible idealizing assumptions of our thought experi-
ment. We are far from ideally rational. Trust, when it elicits trust-responsiveness, can
help potentially conflicted and weak-willed agents to stick to norms that they profess
to endorse by giving them an additional reason for compliance—having professed to
follow them, someone else is now counting on them to do so, so they had better follow
through. In the trust literature, there’s dispute between those who find trust, and with it
dependency on trustworthiness, ubiquitous (Baier 1986; 1991), and those who think
only a small subset of relations are trust relations (Hardin 2002). Nor can this disagree-
ment be understood as the effect of adopting a more or less restrictive account of trust
and trustworthiness, for Baier, who finds trust and dependence on trustworthiness
ubiquitous, adopts a goodwill-based analysis of trust, according to which trust tacitly
imputes goodwill to the one-trusted. The account is highly restrictive and thus seems
to imply that trust relations should comprise only a selective sub-set of social relations.
The observation that trustworthiness plays backstop to assist agents who might
otherwise be wavering in their commitments to norms suggests a way of reconciling
apparently conflicting judgements about how common trust relations are. In a sense
they are indeed ubiquitous: trustworthiness can function as a backstop in multiple
contexts, giving an additional reason to act as one is being counted on to act, a reason
that might not be needed to tip the balance of favour of so-acting but that stands in the
wings, ready to be called on when required. In another sense they are not: trustworthi-
ness often remains merely on standby.
Not only are actual agents often conflicted and weak-willed, we are far from mutu-
ally transparent. We often do not know what the other values or what they will assume
they have reason to do independently of any reasons that we may give them by count-
ing on them. However, if they are responsive to our dependency, then we can at least
know that we have the power to give them a reason to act as we are counting on them to
do, by the very fact of counting on them and communicating that we are doing so.
Proper use of this power is regulated by the norms governing trust.
I have argued that the purpose of the concepts of trust and trustworthiness is broadly
normative: we focus on the distinctive kind of active dependency characteristic of
trust and the active responsiveness to it characteristic of trustworthiness so as to pro-
mote them as ways of reducing the risk of extending our agency through dependency
on others. Trust when met with trustworthiness allows us to enhance the effectiveness
of our agency. As finite and social agents we have a pressing interest in being able to do
this. However, the interest that we have in being able to extend our agency in this way is
8
Compare the way in which requests can generate reasons which do not exist prior to the request
(Enoch 2011).
102 Karen Jones
not a distinctly moral interest. It is about making agency effective by drawing on the
agency of others rather than about making it good. Extended agential power is an
ends-independent value. Whatever our ends, we want to be able to recruit the agency
of others: even counter-ethical projects require recruits.
If the value of the kind of relationship that trust and trustworthiness enables is not a
moral value, and the norms of trust and trustworthiness are tailored to make you fit for
such relations, then it follows that the norms of trust and trustworthiness have the
potential to be in tension with norms of morality. Norms of trustworthiness can, in a
context, enjoin you to follow through with concealing a wrong act, such as hiding a
body, if you are being counted on to get your friend out of trouble and if your relation-
ship with that friend is such as to support an expectation of assistance of that kind
(Cocking and Kennett 2000). Given this is so, there need be nothing morally objec-
tionable in failing to be trustworthy in at least some instances. Though not accusable of
a moral failing, you might nonetheless be vulnerable to the complaint ‘But I was count-
ing on you!’ That complaint points not to the existence of a pro tanto moral reason (he’s
counting on me) to help hide the body that is, in the context, outweighed, but rather to the
existence of a different class of reasons, reasons derived from norms of trustworthiness.
Even the amoralist is able sincerely to make the complaint, if the force of that com-
plaint is not moral. And this is indeed what we see among those textbook amoralists,
the Mafioso who, bound by in-group norms of trustworthiness, are able to extend their
agency in ways that imperil the well-being of others. You can be perfectly trustworthy
with respect to some people without being a good person; indeed being trustworthy
with respect to some people requires that you not be a good person. The virtuous will
be trustworthy to the virtuous, but may or may not be trustworthy (whether richly or
three-place) to the non-virtuous.
4. (Some) Norms of Trust and Trustworthiness

Following the template derived from thinking about friendship, the norms that should
govern trust and trustworthiness are those norms that fit us for participation in trust
relationships; that is, they make us good qua truster and qua one-trusted. This formu-
lation requires unpacking, for, though it is clear enough what it means to be good qua
friend, it is not at all clear what it means to be good qua truster and one-trusted. The
point can be made somewhat more intuitive by saying that the norms for trust and
trustworthiness are those that enable us to enter into and maintain good trust relation-
ships. But even this way of putting it is apt to be misleading, for one might be tempted
to think that good trust relationships are those that enable the pursuit of morally good
or at least morally neutral ends. This way of unpacking the criteria of success for trust
relations would be, in effect, to subject successful trust relationships to a moral filter
and would yield a moralized conception of the norms for trust and trustworthiness.
Instead, we should think of successful trust relations (i.e. those that are good qua trust
relations) as those that realize the distinct value of such relations. I have argued that
distinct value is the value of enabling the extension of our agency by the non-coercive
recruiting of the agency of others to remedy our finitude. Successful trust relations are
thus those that match up would-be dependents with those willing and able to be
counted on in the way required. We find the norms for trust and trustworthiness
by asking, ‘what norms enable this matching?’ Since that matching is temporally
extended, we should expect there to be norms governing the various stages of trust
relations including entering, maintaining, and exiting them. However, given trust rela-
tions vary considerably, from those between friends to those between strangers, and
even within these broad categories, differ in the particulars, we should not expect these
norms to be rules, let alone lexically ordered rules (Baier 1986; 1991), but rather expect
them to be rough general precepts, guidelines, and vaguer images or stereotypes of the
skills and capacities of the wise truster and of the trustworthy person. This is so for
friendship, too, where we find not friendship rules—what could those be?—but looser
precepts or guidelines such as ‘friends give each other the benefit of the doubt’.
Whereas friendship is governed by one set of norms, trust relations are governed by
two, one for the truster and one for the trusted. This is because friendship is a sym-
metrical relation but trust is not. I can think I am someone’s friend when they are no
friend of mine, but I cannot be someone’s friend without their also being mine. This is
so even for friendships among unequals. Trust relations can be fully symmetric, as
when two individuals trust each other and respond with trustworthiness in the same
domain of interaction—thieves performing the same function in a break-in, for example.
But they can be reciprocal without being symmetrical, as when each stands to the other
as truster and trusted, but regarding different domains (this is probably the typical
case); and they can also be non-reciprocal when the dependency is in one direction
only. Because of this difference there is only one set of norms to follow to be good qua
friend, but two sets to follow to be good qua participant in trust relations, one for each
side of that relationship.
Since our task is to understand when the complaint ‘But I was counting on you!’ has
normative force, I will focus on that subset of norms of trust and trustworthiness that
regulate legitimate use of this complaint. These norms centre on the appropriate com-
munication of who is being, or who may be, counted on by whom, for what and they
regulate both entering into (or renegotiating, which can be understood as renewing)
and maintaining trust relationships. Even here the terrain is large and messy, so what
I have to offer is foray, rather than complete mapping.
Begin from the trust side: we can start to map the contours of the relevant norms by
looking at what would count as a reasonable rebuttal of the complaint, ‘But I was
counting on you!’ There are two strategies for defence against this complaint. The first
amounts to recognizing its force, but defending what you did in the light of the broader
constellation of your reasons. When taking this strategy, residue always remains and
so apology is in order. The second strategy seeks to rebut the charge by claiming that no
reasonable norm of trustworthiness was broken, so no apology required. This defen-
sive strategy is one of countercharge: the fault lies with the truster rather than the
104 Karen Jones
trusted. Successful counter-charges thus reveal the norms that make one fit for enter-
ing trust relations in the role of truster.
There are a number of countercharges which, depending on the facts of a particular
case, have the potential to be successful: denial; ignorance; presumption; and due
warning. Let’s look at each in turn. One response to the charge that you have let some-
one down is straight out denial. If trust demands performance of a specific action on
the part of the one-trusted, denial would be difficult, if not impossible. If, however,
trust assigns discretionary powers to the one-trusted, then there can be legitimate dis-
pute as to whether or not that trust has been fulfilled. What might, at first look, appear
a failure in responsiveness to the fact of someone’s dependency might, on closer
inspection, be revealed as a creative way of meeting it. It need not always be easy to tell:
discretionary powers can be exploited to mask subtle breaches of trust (Baier 1986:
238). Nevertheless, discretionary powers are worth the risk, for they enable us to trust
when we lack the expertise to specify what would count as a successful meeting of that
trust, as often we do. Discretionary powers enable the one-trusted better to meet us in
our limitation and so we should expect there to be norms for trusters enjoining them
to give the one-trusted the benefit of the doubt in interpreting their behaviour and not
to lightly make the charge of having been betrayed (Baier 1991). Trusters who are
quick to blame will be poor at sustaining trust relationships.
The second rebuttal is ignorance: ‘How could I possibly have known that you were
counting on me?’ and it points to the existence of norms enjoining trusters to take
responsibility for communicating those dependencies to which they expect others to
be responsive. When there are standing shared assumptions about who can count on
whom for what, such communication need not be explicit. But when there are no such
standing assumptions or their existence is unclear, explicit communication can be
required. As with the defence of denial and, as we will see, with the other defences as
well, sometimes there can be dispute about whether this norm has been violated and
hence whether the complaint is rebutted; however, this is what we should expect given
that there are at least two perspectives on any trust relationship, and given the skill and
judgement that it takes to recognize standing assumptions and the forms of signalling
that can overturn them.
The remaining two counter-charges, presumption and due warning, amount to
accusations that the truster has failed in their uptake of a properly executed signal, a
signal that reveals that the one-trusted is, after all, richly trustworthy: I signalled, you
failed to hear and went ahead anyway; responsibility for being let down thus lies with
you. They are given voice by the rejoinders, ‘Whatever made you think I’d be up for
that?’ and ‘But I told you not to count on me!’ Presumptuous trust is trust extended in
the absence of any signal that it will be met with trustworthiness, or extended on the
basis of imagined signals. Communicating presumptuous trust threatens to use the
other’s trust-responsiveness to manipulate them into a dependency relationship that
they would not otherwise accept. The norms of trust and trustworthiness distribute
the burden of responsibility for matching those willing and able to respond to a
dependency and those seeking such responsiveness between the two parties. They
do not require the one-trusted to respond to manipulative or warned off trust with
trustworthiness. It might be thought that the due warning defence only sometimes
serves to get its maker off the hook, even when such warning has been indisputably
given; after all, we are not always allowed to signal out. Some people, we might think,
are entitled to count on us for certain things because of our responsibilities towards
them. This is true, but not in a way that undercuts the method of finding the contours
of norms of trust through an investigation of the ways in which signalling out rebuts
the specific complaint, ‘But I was counting on you!’ When signalling out is precluded
because of other features of the situation, the fault lies not in failure to follow the norms
of trustworthiness, but in failure to follow some other additional norm that applies in
the context. The point of focusing on the specific complaint is to isolate the norms
for trust and trustworthiness from the wide range of other norms with which they
invariably interact so that we can see their shape more clearly.
I have argued that many, though not all, of the norms that make us good qua trusters,
and those that regulate the proper use of the complaint ‘But I was counting on you!’ are
about making us good receivers of the signalling efforts of the richly trustworthy.
Moving now to the trustworthiness side, we should expect to find that many of the
norms that make us good qua one-trusted are about making us skillful senders of such
signals; that is, they are norms for how to be richly trustworthy. Failure to meet these
norms will invite the charge ‘But I was counting on you.’ This is indeed what we find.
We are already familiar with the broad contours of some of them from our investiga-
tion of Scanlon’s proposal to treat such norms as moral norms. The first norm pre-
cludes explicitly inviting trust and then not fulfilling it. Considered as a moral norm,
I argued that it counts as an instance of norms prohibiting deceit, rather than as an
independent moral norm enjoining trustworthiness. But with a non-moralized con-
ception of its source now available, it can also be viewed as a self-standing norm that
helps make us richly trustworthy and so fit to participate in successful trust relations.
This means that sincere complaint when the norm is violated is available to the amoral-
ist and can be legitimately addressed to a fellow amoralist, even by their own lights.
A norm enjoining follow-through given explicit invitation to trust allows those with
poor track records to enter into trust relationships. It does this by offering to take from
the truster responsibility for possible future failure of the relationship. A would-be
truster can be justified in trusting on the basis of an explicit invitation even when,
given a track-record, they would not be justified in forming the predictive expectation
that the one-trusted will behave as they are being counted on to do. Given that trust
relationships are often mutual, even if seldom symmetric, its no surprise that we
should want there to be a way to redeem poor past performance so as to be able to par-
ticipate in future trust relations. This norm provides such a mechanism.
A further norm of signal-sending enjoins us to be attuned to our location in the web
of standing signals and to take measures to signal against those standing signals, where
appropriate whether because we lack either the competence or the will to be responsive
106 Karen Jones
to the class of dependencies they would typically invite. This is no easy task and it
might be thought that it is as vulnerable to the burdensome objection as the modified
Scanlonian principle, Principle RT (see section 2), since it is a version of it, albeit stated
in the weaker form of guideline rather than strict principle.9 However, the burden-
someness objection is at home in Scanlon’s project of trying to defend moral norms as
contractually irresistible. Contractually irresistible norms generate moral obligations.
On the alternative view, they are norms that enable us to realize the distinct good of
trust relations, and not a matter of moral obligation; thus critique for having failed to
meet them comes with a different flavour and with a lesser stringency. For this reason,
I think the burdensomeness objection does not carry over to my view.
A further norm is about signalling, but indirectly: however good your signalling
skills you will signal falsely if you are deluded about your capacity. Accordingly, we are
enjoined to cultivate the self-reflective capacities required to know the borders of our
competence and of our capacity to meet multiple expectations of responsiveness to
dependency. Lacking these skills we risk getting into trust relationships that we will
not be able to fulfill and hence making ourselves vulnerable to the charge, ‘But I was
counting on you!’
I have argued that those norms governing trust and trustworthiness that regulate
the proper use of the complaint ‘But I was counting on you!’ can be divided into norms
that fit us to enter into (or renegotiate) trust relations and to maintain them. There are
norms governing both truster and the trustworthy. To recapitulate, on the trust side,
we find at least the following norms governing entry into, or renegotiation of, trust
relationships: (i) pay attention to the signalling of others regarding the kinds of
dependencies that they might be up for from people like you; (ii) be aware of different
understandings of the background network of social assumptions that form part of the
standing fabric in which signalling takes place so that you don’t misinterpret signals;
(iii) take responsibility for communicating those dependencies, responsiveness to
which cannot be assumed given shared understandings of who can be counted on by
whom for what. Together i–iii amount to the injunction to meet partway any who
would be richly trustworthy towards you. They fit us for trust relations by enabling
match between would-be truster and those who would respond with trustworthiness.
The flip-side norms for trustworthiness are: (i) be aware of your location in standing
social systems of signalling and, as required, counter-signal; (ii) be aware of different
understandings of the background network of social assumptions that form part of the
standing fabric in which signalling takes place and of what will be counted as counter-
signalling; (iii) do not explicitly invite trust and then fail to follow through; (iv) develop
reflective awareness of capacity so as to avoid overly optimistic signalling.
Norms governing maintaining trust relationships that are revealed by considering
rejoinders to the charge ‘But I was counting on you!’ but that do not relate to signalling
are, from the trustworthiness side, Do not abuse discretionary powers to mask subtle
9
Thanks to Suzy Killmister for pressing this objection.
untrustworthiness, and, from the trust side, Do not be quick to make the complaint
‘But I was counting on you!’
5. Conclusion
I have argued that there are distinct norms governing trust and trustworthiness, norms
that are in part revealed by investigating when the charge of ‘But I was counting on
you!’ has normative force and when the charge can be successfully evaded. I have
claimed, however, that although this normative force is genuine, it is not moral but
rather derives from the pressing interest that we have in there being people who will be
responsive to the fact of our dependency and who will meet us in the task of working
out who we can depend on for what. The argument rested on an analogy between trust
relationships and friendship, on a non-moralized conception of what friendship
entails. Just as on a non-moralized conception of friendship, what can be demanded of
friends is not subject to a moral filter and morally bad people can be good friends, so
too what can be demanded of those we trust—at least where that trust is placed in
accordance with the norms for establishing trust relationships—is not subject to moral
filter and morally bad people can be genuinely trustworthy.
I have not broached the question of what to do when norms of trust and
trustworthiness collide with norms of morality, as they can on this account. If you hold
that morality is overriding, then the answer to this question will be clear; however, if
you do not—and I do not—then the answer will be far from clear. Being trustworthy
might require you to do the morally wrong thing and there might be no broader
perspective from which to adjudicate the conflicting normative perspectives to arrive
at a judgement as to what, all things considered, is the thing to do. My non-moralized
account of the normative force of trustworthiness norms will have practical import
only to those who hold this latter view. Nevertheless, whatever your view about the
status of morality, it is important to refrain from the various moralizing moves that
inform much of the current literature because moralizing moves mistake the ends-
independent value of trust relations and so oversimplify the complexity of the values
our norms are designed to serve.
References
Alonso, F. 2009. Shared Intention, Reliance, and Interpersonal Obligations. Ethics 119:
444–75.
Baier, A. 1986. Trust and Anti-trust. Ethics 96: 231–60.
Baier, A. 1991. Trust and Its Vulnerabilities and Sustaining Trust. Tanner Lectures on Human
Values 13. Salt Lake City: University of Utah Press.
Cocking, D. and J. Kennett. 2000. Friendship and Moral Danger. Journal of Philosophy 97:
278–96.
Enoch, D. 2011. Giving Practical Reasons. Philosophers’ Imprint 11: 1–22.
108 Karen Jones
Frost-Arnold, K. 2014. Imposters, Tricksters, and Trustworthiness as an Epistemic Virtue.

Hypatia 29: 790–807.
Hardin, R. 2002. Trust and Trustworthiness. New York: Russell Sage.
Haslanger, S. 1993. On Being Objective and Being Objectified. In A Mind of One’s Own:
Feminist Essays on Reason and Objectivity, ed. Louise Antony and Charlotte Witt. Boulder,
CO: Westview Press.
Hollis, M. 1998. Trust within Reason. Cambridge: Cambridge University Press.
Holton, R. 1994. Deciding to Trust, Coming to Believe. Australasian Journal of Philosophy 72:
63–76.
Jones, K. 2010. Counting on One Another. In Trust, Sociality, Selfhood, ed. Arne Gron and
Claudia Welz. Tubingen: Mohr Siebeck.
Jones, K. 2012. Trustworthiness. Ethics 123: 61–85.
Jones, K. 2013a. Distrusting the Trustworthy. Reading Onora O’Neill, ed. David Archard et al.
London: Routledge.
Jones, K. 2013b. Trust. In The International Encyclopedia of Ethics, ed. Hugh LaFollette. Oxford:
Blackwell Publishing.
Keller, S. 2004. Friendship and Belief. Philosophical Papers 33: 329–51.
Lagerspetz, O. 1998. Trust: The Tacit Demand. Kluwer: Dordrecht.
McGeer, V. 2008. Trust, Hope, and Empowerment. Australasian Journal of Philosophy 86:
237–54.
McLeod, C. 2002. Self-Trust and Reproductive Autonomy. Cambridge, MA: MIT Press.
Nickel, P. 2007. Trust and Obligation-Ascription. Ethical Theory and Moral Practice 10:
309–19.
Pettit, P. 2000. Two Construals of Scanlon’s Contractualism. The Journal of Philosophy 97:
148–64.
Potter, N. 2002. How Can I be Trusted? A Virtue Theory of Trustworthiness. Lanham, MD:
Rowman and Littlefield.
Scanlon, T. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press.
Scanlon, T. 2003. Promises and Contracts. Reprinted in The Difficulty of Tolerance. Cambridge:
Cambridge University Press.
7
The Problem of Trust
Paul Faulkner
Trust, I propose, should be distinguished from mere reliance, even if it essentially

involves reliance.1 Here are some examples of trust. Leaving one’s closed diary on a
desk where one knows one’s partner will see it. Not asking for a second quote when a
mechanic says one’s car needs lots of work. Shaking on a deal. Following a stranger’s
directions. Purchasing an item that will be delivered later. And these are also examples
of reliance. The difference is that although reliance can be willing—one can be confi-
dent that someone or something will prove reliable—it can also be forced. By contrast,
trust, Bernard Williams observes, ‘involves the willingness of one party to rely on
another to act in certain ways’ (2002: 88). This willingness is essential because to trust
someone is to take an optimistic view of that person and their motivations. It is ‘to
lower one’s guard, to refrain from taking precautions against an interaction partner,
even when the other, because of opportunism or incompetence, could act in a way that
might seem to justify precautions’ (Elster 2007: 344). Such precautions might include
keeping one’s diary in a locked desk, and consolidating the handshake with a legally
enforceable contract. To trust is to forgo precaution and garner reassurance just from
the thought that the other is trustworthy.
The attitude of trust is then part identified by this thought. We can ‘trust’, or so we
say, our car to start or our alarm to go off. But cars and alarms aren’t trustworthy and
we don’t, properly speaking, trust them. Rather, this is just reliance: we rely on them
and do so believing that they are, or will prove, reliable. Reliability is a feature of trust-
worthiness but does not equate to it. Similarly, and famously, Kant’s neighbours could
rely on his punctuality in the same way that they could rely on a clock; they could ‘trust’
Kant to walk by the window at eight in the morning. But trusting Kant to do this,
Russell Hardin argues, would require his ‘having their interests at heart in deciding to
take his walk’ (2002: 5). It would require thinking of Kant’s walk as a trustworthy
action, or one done, at least in part, in response to the dependence exhibited by his
neighbour’s ‘trust’. But of course Kant did not take his walk for this reason and his
1
See Baier (1986).
110 Paul Faulkner
neighbours did not think he did. There might have been reliance and reliability but
there was no trust or trustworthiness.
Trustworthiness is more than reliability in that it is evaluative as well as descriptive.
The trustworthy act, like the opportunist one, is done in response to someone’s
depending on one in certain ways, but unlike the opportunist act, it is the appropriate
thing to do—where ‘appropriate’ here amounts to a quasi-moral evaluation. Thus, we
think, you shouldn’t read another’s private diary or give an inflated quote, and you
should honour agreements and help strangers who ask for directions. The thought that
the person one interacts with is trustworthy, or will prove so, then provides rational
reassurance: there is no need for precaution, if this thought is true.
So trust is unproblematic if one knows that the trusted is trustworthy. But, arguably,
it becomes problematic as our epistemic grounds for judging that the trusted is trust-
worthy lessen. However, the resulting problem of trust is fundamentally practical
rather than epistemic: it is a problem of the rationality of reliance broadly—a problem
of the rationality of cooperation. Interactions that potentially raise the issue are those
that have a worst outcome: one delivers and the other defaults. One leaves one’s diary
on the desk and it gets read, one doesn’t get a second quote and is stung, and so on. So
precautions can seem justifiable. And what one seeks to do in taking precautions is to
remove the problematic reliance. But if the reliance cannot be removed, what would
make it rational are any grounds that provide reassurance that the other is, or would
prove, reliable or cooperative. Of course, knowledge that the person is trustworthy
would do this, but reliance could be rationally sustained on slighter grounds. In respect
of this, Williams (1988: 118) identifies four general motivations people can have to
cooperate or rely on one another. Fear of sanctions is one; the others are: particular
self-interest, a positive evaluation of cooperation, and a positive evaluation of friendly
relations. Sanctions can be informal or formal. Informal sanctions consist in some
form of social exclusion, such as loss of reputation or simple ostracism, and the further
losses this causes. Formal sanctions can be anything from a fine to a prison sentence;
professionals who fail in their role, for instance, can lose their legal right to practise.2
We are sensitive to all these motivating reasons for cooperation. Moreover, we also
engage with one another in ways that provide some grounds for empirical judgement
of motivation. Certain features of an individual or their behaviour provide what Elster
(2007: 347) respectively calls ‘signs’ or ‘signals’ that allow a judgement of reliability. The
problem of trust is then this: even if the net is cast widely, so that all that is needed to
rationalize reliance is some grounds for judging reliability, we can still lack grounds for
this judgement.
1. The Trust Game

This problem can be outlined by reference to an experiment called the Trust Game.3 In
this game there are two players: a trusting party or ‘investor’, X and a trusted party or
2
See Blais (1987). 3
See Glaeser et al. (2000).
the problem of trust 111
‘trustee’, Y. In the simplest version X has a certain endowment, say £100, and options of
keeping this or transferring part or all of it to Y and keeping whatever remains. What
gets transferred gets multiplied, say by a factor of four, and Y then has the option of
keeping the resulting sum or making a back-transfer of part or all of it and keeping the
remainder. In this case, the best X could do, would be to give Y everything and hope
Y splits his £400 windfall. If Y does share, both do well and the game has resulted in a
cooperative outcome. But how would it be in Y’s interest to return anything?
One way in which it might be is if the game were iterated. In such a situation,
defection would be repaid by a lack of future cooperation, and one would thereby
lose the benefits of this cooperation. So X’s interest in receiving a back-transfer is
also in Y’s interest since it allows for another exchange. And this, Russell Hardin
proposes, is all that trust amounts to: it is merely a matter of encapsulated interest.4
The problem with this view is that X’s interests can be encapsulated in Y’s interests
only insofar as the exchange is ongoing. If the relationship is a one-off, then there is
no value to be got from acting as if it were ongoing.5 But then considering such cases,
why think that Y will cooperate? And here we return to the motivations that people
can have to be cooperative, or to rely on one another, and our knowledge of these.
However, suppose now that X’s position is one of ignorance; that is, suppose—as is
meant to be the case in the experimental set-up—that X knows nothing about Y and
so is ignorant of what does or might move Y. Once this assumption of ignorance is
added it becomes hard to see how anything other than defection from the start could
be the rational thing to do. Admittedly, we are never entirely ignorant—even to pre-
sume that Y prefers more money to less is a presumption about preference. However,
once the rationality of reliance is grounded on the judgement that the other party
will prove reliable, then it seems as though reliance becomes irrational when the
grounds for such a judgement are not available. In the face of ignorance, it is surely
better to keep what monies one has.
This problem of trust is then a sceptical problem insofar as the facts that generate it
commonly hold true; that is, insofar as (i) we need to rely on another but recognize that
doing so could have a worst outcome—we rely and the other proves unreliable—and
(ii) we know that this interaction is one-off (or one of a determinate number), and (iii)
we are entirely ignorant of the other’s individual motivations, but recognize a general
motivation to be unreliable. Given these facts, reliance would seem to be irrational,
and if it is, then a fortiori so too is trust. Yet presently, and thankfully, trust is part of the
fabric of our society: thus a key experimental result is that people, with some interest-
ing qualifications, do by and large cooperate in the Trust Game even though it is played
as a one-off game under conditions of ignorance, and hold trustees to the expectation
4
‘I trust you because I think it is in your interest to attend to my interests in the relevant matter’
(Hardin 2002: 4). On the view of trust being presupposed here, this is insufficient because the judgement
of trustworthiness that part constitutes it is non-evaluative.
5
Equally, if it is not a one-off but has a determinable end, once a final interaction is established the
rationale for cooperation unravels by backwards induction.
112 Paul Faulkner
that they ought to cooperate in return.6 Thus, as with other sceptical problems, the
issue is how to reconcile the philosophical result that reliance seems frequently
irrational with the everyday fact that trust is pervasive. The difficult question is to
account for how such reliance is rational under these conditions.
2. Gauthier’s Internal Solution

In practice, it is simple to render reliance unproblematic: any potentially problematic
situation, one described by (i), can be altered such that facts (ii) and (iii) no longer hold
of it; that is, devices can be introduced which through adding future interactions or
some other means, such as sanctions, give the trustee known reason to be cooperative.
But, David Gauthier observes, to solve the problem of trust in this way is simply to
‘miss the point’: what is needed is an account of how the mere ‘mutual advantageous-
ness’ of cooperation, the fact that both investor and trustee benefit from it, renders it
rational. Thus Gauthier distinguishes what might be called internal and external solu-
tions. An external solution establishes the rationality of reliance, in a situation where
(i) is true by falsifying (ii) or (iii). An internal solution shows how reliance is rational
even given the truth of (i), (ii), and (iii). And he seeks to provide an internal solution.
The solution, Gauthier suggests, is to abandon the conception of rationality pre-
supposed in generating the problem of trust. What is presupposed is a maximising
conception of rationality: individuals pursue what is most in their interest. So where
(i), (ii), and (iii) hold, and the investor is ignorant of the trustee’s particular interests,
he can only think that the trustee would keep all the monies if he had the chance.
However, rather than see the rational outcome as an equilibrium point—the product
of maximizing choices—the rational outcome, Gauthier (2013: 195–6) proposes, is
the optimal one, or that outcome that could not be better for one subject without
being worse for others. The rational thing to do is to adopt a strategy that optimizes
rather than maximizes, where to do this is to be what Gauthier (1986: 167) calls a
‘constrained maximizer’ as opposed to a ‘straight-forward maximizer’. A constrained
maximizer pays attention to the interests of others, not merely their choices, or at
least does so conditional on others reasoning likewise. And it is rational, Gauthier
argues, to be a constrained maximizer because one does best this way. The reason for
this is simple: we have a basic disposition to cooperate only on terms that are fair
and, as such, constrained maximizers have more opportunity for cooperation.
‘Straightforward maximizers are disposed to take advantage of their fellows should
the opportunity arise; knowing this, their fellows would prevent such opportunity
arising’ (Gauthier 1986: 173). So it is rational for the investor to rely on the trustee to
return half the windfall and for the trustee to make this fair return because were
either to have the disposition to act otherwise they would have less opportunity for
6
See Glaeser et al. (2000).
cooperation in general. Any gain to be made in this case would be offset by the
greater loss that this exclusion would entail.
Allow that Gauthier has established the rationality of being a constrained maxi-
mizer. A constrained maximizer will cooperate and so rely on another only if the other
is equally judged to be a constrained maximizer. As such the straightforward maxi-
mizer ‘must seek to appear trustworthy, an upholder of his agreements’ (Gauthier 1986:
173). And constrained maximizers must cultivate ‘their ability to distinguish sincere
cooperators from insincere ones’ (Gauthier 1986: 181). A difficulty, then, for this solu-
tion to the problem of trust is that it must presume this constrained maximizer ability
outstrips the ability of straightforward maximizers to beguile. And this presumption
just seems to be the denial of (iii): the rationality of reliance thereby hinges on a belief
about the trustee’s motivational dispositions, namely that the trustee is a constrained
maximizer, not a straightforward one, and recognized as such. Thus in a situation of
ignorance, were (iii) true, the rational thing to do would surely be to presume that the
other is a straightforward maximizer and avoid relying on them.
Not necessarily argues Gauthier. The inability to detect straightforward maximizers
would not matter if there were not so many of them. In demonstration of this Gauthier
offers a sophisticated account of how the rationality of being a constrained maximizer
would be preserved if the proportion of constrained maximizers in the population
were significantly large.7 However, the issue then becomes what reason we have for
thinking that our population is like that. And the problem is that if our reason for
thinking this is empirical, Gauthier’s argument for the rationality of cooperation
would seem to rest, at base, on the denial of (iii): our situation is not really one of
ignorance if we know enough to empirically justify the assumption that the other is a
constrained maximizer. However, arguably we do not have any a priori reason for
making this assumption.
To argue this, consider the case of Blue Jays and Viceroys.8 Although they are good
for Blue Jays to eat, Viceroys look like Monarchs that are not. This mimicry is an evolu-
tionary lie. The Monarch signals truly that it is a poor meal; the Viceroy mimics this
signal to communicate something false. The position of an investor is then analogous
to the position of a Blue Jay faced with a brightly coloured butterfly. Just as the Blue Jay
gains from eating the butterfly if it is a Viceroy, and loses if it is a Monarch, so the investor
gains from making the transfer if the trustee is a constrained maximizer and loses if the
trustee is a straightforward maximizer. And the ability of investors to discriminate
constrained maximizers from straightforward maximizers, like the ability of Blue Jays
to discriminate Viceroys from Monarchs, is limited. Suppose then a best case starting
state: this limited discriminator capacity is unproblematic because there is no decep-
tion. Only Monarchs are colourful; only constrained maximizers purport to be so;
7
‘The more constrained maximisers there are, the greater the risks a constrained maximiser may ration-
ally accept of failed cooperation and exploitation’ (Gauthier 1986: 176).
8
See Sober (1994).
114 Paul Faulkner
Blue Jays eat all drab butterflies; and the worst-case scenario is never realized for the
investor. From this starting point, deception will evolve: colourful Viceroys and
straightforward maximizers who learn ‘to appear trustworthy’, will flourish. But as
levels of deception rise, levels of reliance go down; but they should decrease only so far,
because if reliance vanished, there would be no gain to be had from deception. So at
what point does the population reach a stable equilibrium? ‘Interestingly enough’, Eliot
Sober (1994: 80) observes, ‘it turns out that the stability of the system depends on the
type of dynamics one assumes.’ What this means is that there is no a priori reason for
thinking that constrained maximizers predominate. They might, just as most colourful
butterflies might be Monarchs, but if this is so, this ‘stability depends on some highly
contingent properties of the evolutionary process’ (Sober 1994: 80). Which is to say, it
depends on some highly contingent properties of our society. So the rationality of
reliance has to be ultimately grounded on an empirical belief about our social situation.
This might be a solution—and in some ways it is close to the one I sketch below—but so
far it is not the internal one hoped for.
3. Pettit on Our Trust Responsiveness

Philip Pettit presents a different internal solution to the problem of trust. Take any
potential engagement where (i) is true, such as the Trust Game; the standard reason
that an investor might rely on a trustee to share their windfall, Pettit argues, would be
the judgement that the trustee is trustworthy (where by this Pettit means little more
than ‘is reliable’). And the standard reason one has for this judgement is the further
belief that the trustee is motivated by loyalty, virtue, or prudence, where these motiv
ations need not be exclusive. So what grounds are then available for this judgement if
the investor is ignorant of the trustee’s particular motivations? That is, if facts (ii) and
(iii) hold. To answer this question, Pettit (2002: 353) suggests that one needs to recog-
nize that people are motivated by both ‘action dependent goods’, such things as com-
modities and money, and ‘attitude dependent goods’ or such things as ‘being loved,
being liked, being acknowledged, being respected, being admired, and so on’.
Moreover, the desire for these goods is universal in that it can be assumed ‘that each of
us desires the good opinion of others’ (Pettit 2002: 354). However, if this is the case,
there are grounds for thinking that the trustee will prove reliable, and so grounds for
rationally relying on the trustee, even given the truth of (i), (ii), and (iii).
Pettit lays out these grounds with the following argument:
(1) There are situations where an act of trust will signal to a trustee, and to witnesses, that the
trustor believes in or presumes on the trustworthiness of the trustee . . . and so thinks well of
him to that extent.
(2) The trustee is likely to have a desire, intrinsic or instrumental, for the good opinion of the
trustor and of witnesses to the act of trust.
(3) The desire for that good opinion will tend to give the trustee reason to act in the way in
which the trustor relies on him to act.
Conclusion. And so the trustor, recognizing these facts may have a reason to trust someone,
even when he actually has no reason to believe in the other’s pre-existing trustworthiness.
(Pettit 2002: 357)
As a statement of a reason we can have for relying on others, this conclusion is well
observed. But as a solution to the problem of trust, the argument faces a couple of dif-
ficulties. First, to resolve the problem of trust, the desire for the good opinion of others
needs to be universal in the sense that it can be taken to be a component in every inter-
action. Thus, premise (2) is too weak as stated with the qualifier ‘likely to have a desire’
and the conclusion correspondingly so with the qualifier ‘may have a reason’. To
strengthen premise (2), it would have to be supposed that the desire for the good opin-
ion of others were intrinsic rather than instrumental, since it would not be plausible to
claim either that the good opinion of others was instrumentally valuable in every case,
or that it was instrumentally valuable in every case where the trustor didn’t have an
intrinsic care for the good opinion of the trustee. Suppose then that this is true: in
every case the trustee will have an intrinsic desire for the good opinion of the trustor
and of witnesses to the act of trust. Even with premise (2) thus strengthened, it is still
insufficient for the trustor to have a reason to trust—that is, rely—in the moot cases.
The issue now is that even if the trustee always has a desire for the good opinion of
others, and so a reason to be ‘trustworthy’, this reason need not suffice for being so.
This is because in most cases the trustee will have more interests than this at play in the
interaction. So what needs also to be true is that the trustor has grounds for thinking
that this desire for good opinion of others is either the only operative desire or the
dominant one. And while the trustor could well have grounds for thinking this, the
trustor would not have such grounds when the interaction takes places under condi-
tions of ignorance; that is, when (iii) is true. Put another way, if (iii) is true, the trustor
lacks the unique reason for relying on the trustee that Pettit hypothesizes. So the
hypothesis of this reason offers no internal solution to the problem of trust. Again it
depends on a particular empirical belief whose possession implies that one of the facts
generating the problem does not hold.
4. Our Valuing Trust and Trustworthiness

In the remainder of this chapter, I hope to elaborate on a solution to the problem of trust
suggested by Bernard Williams. In this section, I outline the solution.
As observed, Williams (2002: 88) takes trust ‘in its most basic sense’ to be no more
than ‘the willingness of one party to rely on another to act in certain ways’. This willingness
implies some expectation of motive but ‘it does not imply that those motives have
to be of some specific kind’ (Williams 2002: 88). In particular, it does not imply that
the motives of the trusted are those of the trustworthy person: the judgement that the
trusted was sensitive to the possibility of punitive sanctions, for instance, might
suffice for the prediction that the trusted would prove reliable, and so rationalize reliance.
116 Paul Faulkner
However, Williams goes on to argue, such judgements of motive do not resolve what
I’ve called the problem of trust. Williams’s interest in this problem lies with a particular
instantiation of it: what motives we have for telling others the truth, or being sincere.
Any functioning society will involve cooperative engagements which demand infor-
mation be communicated between individuals. So sincerity is clearly desirable from
the social point of view, but it is equally not always in an individual’s best interest. For
instance, Williams imagines the case of the hunter who has just made a kill, which he
would prefer to keep for himself and his family. Sincerity is not in this hunter’s interest.
The problem this case illustrates is that
The value that attaches to any given person’s having this disposition [sincerity] seems, so far as
we have gone, largely a value for other people. It may obviously be useful for an individual to
have the benefits of other people’s correct information, and not useful to him that they should
benefit of his. So this is a classic example of the ‘free-rider’ situation. (Williams 2002: 58)
Whilst it is always in an audience’s interest to be informed, sincerity need not best serve
a speaker’s interest and as audiences we know that this is the case. So, what I have called
the problem of trust threatens. The source of this problem, Williams argues, is that
sincerity—trustworthiness in speech—has only been given instrumental value: its
value is given by that good which follows from information being pooled, such as the
good brought about by cooperative endeavour. And when it is only valued in this way
there will always be the possibility of a fissure between interacting parties’ interests
such that trust exposes the trusting, or those who have a trustworthy disposition, to
‘free-riders’. An adequate solution must not allow the realization of this possibility
to be systemic. What this shows, Williams (2002: 59) claims, is ‘that no society can get
by . . . with a purely instrumental conception of the values of truth’ of which sincerity,
which is trustworthiness in speech, is one. What any society thereby requires is that
such trustworthines be intrinsically valued, so that it is thought to be ‘a good thing
(many other things being equal) to act as a trustworthy person acts, just because that is
the kind of action it is’ (Williams 2002: 90).
Three points need to be made about this idea of intrinsic value. First, to value trust-
ing and trustworthy acts intrinsically is to take an evaluative stance: it is a good thing to
act as the trusting and trustworthy person acts, which is to say that one ought to act this
way. So if solving the problem of trust requires thinking of trust in this way, trust must
be taken to involve more than mere reliance—as claimed in section one. Second, it
follows, and Williams stresses this implication, that there can be no solution to the
problem of trust that starts, as Gauthier’s and Pettit’s do, from an individual’s interests.
This is because no argument from such a starting point, namely one ‘which sets out
from a game-theoretical formulation of the problems of trust could possibly show that
trustworthiness had an intrinsic value’ (Williams 2002: 90). The most that could be
concluded is that it can be in one’s interest to act as if trust and trustworthiness had
intrinsic value; but this is consistent with pretending they had this value, rather than
valuing them, and so would not eliminate the issue of free-riders or solve the problem.
However, third, talk of intrinsic value raises a dilemma. On the one hand, intrinsic
value seems rather mysterious, amounting to the claim that ‘there is nothing else to be
said about its [trust’s] valuableness—it is good because it is good, and that is all there is
to be said about it’ (Williams 2002: 90). On the other, if some account of its value
is offered ‘such as securing cooperative activity which is in everyone’s interest’
(Williams 2002: 90) then it seems as though one is giving a reductive account: its value
is instrumental after all.
What is needed is a non-reductive explanation of what it is for something to have
intrinsic value. And a sufficient condition for this, Williams proposes, is that
first, it is necessary (or nearly necessary) for basic human purposes and needs that human
beings should treat it as an intrinsic good; and, second they can coherently treat it as an intrin-
sic good. (Williams 2002: 92)
A genealogy, he then claims, satisfies both of these conditions. Sincerity is ‘necessary

(or nearly necessary) for basic human purposes’ in that were individuals not to have
the disposition to be sincere or trustworthy in speech, no cooperative endeavour
would be possible and there would be no such thing as society, at least as we know it.
Except this claim is too strong as stated: cooperation in communication could be
established on the same grounds as cooperation more generally, through knowledge of
the motivations we know people can have to cooperate. But what could be claimed is
that it is only through individuals having the disposition to be trustworthy that
cooperation could be sustainable once conditions (i), (ii,) and (iii) are in place. So
trustworthiness’s having intrinsic value is necessary for what I called an internal solu-
tion to the problem of trust. And a genealogy then shows how sincerity ‘can coherently
be treated as an intrinsic good’ through showing how our having the disposition to be
trustworthy ‘makes sense to [us] from the inside, so to speak’ (Williams 2002: 92). And
this is what I hope to do in the remainder of this chapter: to offer a conceptual analysis
of trust and trustworthiness—an account of our understanding of these notions—that
makes sense of our being motivated to act in trusting and trustworthy ways just
because we value acting in these ways, where this motivation can be undimmed by an
interaction occurring under conditions (i), (ii), and (iii).
However, before continuing with this analysis, it is worth observing that the kind of
generalized disposition to trust and be trustworthy, which follows our intrinsically
valuing trust and trustworthiness, is quite different to the dispositions to ‘trust’ and be
‘trustworthy’ as characterized by Gauthier and Pettit. The rational disposition to have,
or strategy to follow, according to Gauthier is to be a constrained maximizer when and
only when one judges one’s interaction partner is so and to be a straightforward maxi-
mizer otherwise. To be a straightforward maximizer would entail being untrustworthy;
it would demonstrate a failure to give the appropriate deliberative weight to a trusting
party’s interests. Thus, Gauthier’s proposal suggests that one can be motivated to be
trustworthy in some situations and motivated to be untrustworthy in others, which
implies that the disposition to be trustworthy can be switched on and off in response
118 Paul Faulkner
to the situation one finds oneself in.9 However, it is not plausible to suppose that a gen-
eralized disposition to trust or be trustworthy can be conditional in this way. Take the
Trust Game. Were an investor to have a generalized disposition to trust, the investor
would be disposed to make a transfer irrespective of any judgement of the trustee’s
likely response. To simply keep the money would seem the ‘wrong’ thing to do. Of
course, this is not to say that there is no knowledge that would stop the investor from
making a transfer, but it is to suggest that the disposition to do so cannot be condi-
tional in the way Gauthier describes. Rather, in having this disposition the investor
would see himself as having a reason to make a transfer, a reason that in some circum-
stances might be unfortunately outweighed, but a reason irrespective of these circum-
stances, not a reason that only exists conditional upon them.
Sticking with the Trust Game, the rationalization of trust proposed by Pettit (in the
strengthened form suggested) proceeds from the premise that we have a basic desire
for the good opinion of others. While this may be true, both of the trustee and in gen-
eral, the trustworthy response to the investor’s transfer does not start deliberation
from the fact that a return transfer will receive plaudits. Rather, the trustworthy person
starts deliberation from the fact that the investor trusted or made a transfer and is
consequently in a position of vulnerability. And were the investor to have a generalized
disposition to trust, this disposition would, in part, be that of presuming the trustee
would start deliberation from this fact. To think otherwise and, in particular, to sup-
pose that the trustee deliberates from the premise that a return transfer would garner
praise would be to think badly of the trustee; it would be to think that their motivations
are not ‘right’, or not those of the trustworthy person.10
5. Trust as a Two-Place Relation

The challenge, as Williams puts it, is to give an ‘explanation without reduction’; to say
why trust and trustworthiness are valued without making this value instrumental. This
challenge, I now want to suggest, cannot be met all the while trust is thought of as a
three-place predicate whose instances are X trusting Y to φ. While trust can be three-
place, we trust one another to act in various ways—not to read our diary, to give a fair
quote and so on. This form of trust, its contractual form one might call it, cannot be
fundamental; or at least it cannot be so if trust is something that is intrinsically valued.
For the value, for X, in trusting Y to φ is fundamentally the (action dependent) goods
that come from Y φ-ing. Its value is fundamentally instrumental (the value attached to
keeping one’s indiscretions hidden, or paying a fair price, for instance). Any value
intrinsic to trust must adhere simply to X’s having the trusting attitude; that is, in X’s
9
The same thing is implied by a strategy of tit-for-tat which would require one’s trustworthiness to be
conditional on the previous play of one’s interaction partner.
10
Thus, there is something oddly self-defeating about the desire for good opinion. Compare Elster
(2007: 351).
trusting Y, or just in X’s trusting. That is to say, insofar as trust is intrinsically valued, it’s
fundamental form must be, as one might say, attitudinal; it must be two-place, or even
one-place but not three. It is then X’s having a generalized disposition to trust in this
way—to simply trust, or to trust Y—that explains why it is that X trusts Y to φ.
Focusing, then, on attitudinal trust, what is this generalized disposition to trust? It
is, I propose, the disposition to presume that one’s interaction partner, the trustee in
the Trust Game for instance, is trustworthy. This then raises the question of what it is to
be trustworthy. Here, Williams observes,
one thing it needs to be is the disposition of an agent to be reliable, not in the sense that you can
rely on him to help you (that is a different disposition, helpfulness), but in the sense that he will
help you if he has told you that he will help you or, perhaps, if he has led you to believe that he
will. (Williams 2002: 92)
The answer I think is much closer to the one Williams rejects. His focus is wrongly on
trust in its three-place or contractual form, and he understands the trustworthy
disposition by reference to this form. But if it is the attitudinal form that the disposition
of trustworthiness needs to be understood by reference to, then trustworthiness is
much closer to helpfulness in that it is essentially a benevolent attitude, or an attitude
of goodwill, that is manifest by an appropriate response to another’s dependence.
Appropriate in the sense of right: the trustworthy person can be relied on to do ‘the
right thing’. It is this benevolent attitude and associated goodwill that then allows
for contract.
That these attitudinal forms of trust and trustworthiness are fundamental can then
be supported, I think, by five independent bits of evidence.11 The first piece of evidence
comes from everyday language. First, both the two-place predicate ‘X trusts Y’ and the
three-place ‘X trusts Y to φ’ have unique and irreducible meanings. It is true that some-
times we use ‘X trusts Y’ as shorthand for ‘X trusts Y in some particular way’; for
instance, asked why she left her diary visible on the desk, X might reply that she trusts
Y, and by this mean that she trusts Y not to read it. However, this is not the most
straightforward use of ‘X trusts Y’, which is that of a description of X’s attitude towards
Y as a trusting or trustful one.12 And by implication that Y, the object of X’s attitude of
trust, is someone who can be trusted. By contrast, ‘X trusts Y to φ’ is a metaphysically
hybrid notion (like knowledge, as opposed to belief, is ordinarily understood) in that it
describes an action—X’s relying on Y to φ—and says of that action, that it is done with
a certain attitude, which is best described as trustful. That is, it reports the fact of X’s
reliance and X’s attitude to relying; it is not a direct description of X’s attitude and does
not carry the implication that Y is someone to be trusted. Now while ‘X trusts Y’ might
imply a disposition to rely on Y in various ways, and one to rely on Y to φ, it cannot be
reduced to such a disposition and formalized as ‘∀φ, X trusts Y to φ’. For ‘X trusts Y’
11
For further argument see Domenicucci and Holton (2016).
12
See Becker (1996: 44–5).
120 Paul Faulkner
might be true, while ‘∀φ, X trusts Y to φ’ will almost certainly be false: there is always a
limit to what we will trust others to do. Moreover, this does not seem to be merely a
quantification issue, since there is no restricted range of φ, R, for which ‘∀φ∈R, X trusts
Y to φ’ stands as an adequate formalism of ‘X trusts Y’. For while it might be true that a
complete lack of willingness to rely on Y would falsify the claim that ‘X trusts Y’, there
is no particular way in which X must rely on Y for this claim to be true. However, while
‘X trusts Y’ and ‘X trusts Y to φ’ are unique statements, there is some implication from
the former to the latter but not vice versa. If X does trust Y, then there must be some φ
for which X trusts Y to φ. But that X trusts Y to φ does not, in any way, imply that X
trusts Y more generally, even if this would often also be true. Thus, of the two predi-
cates, the two-place one is arguably more fundamental.
Similar things may then be said when comparing the one-place predicate ‘X trusts’
with the two-place predicate ‘X trusts Y’. The former equally seems to have a place in
everyday language: ‘we do’, Uslaner (2002: 22) observes, ‘speak of “trusting people”
generally’. And this form does not seem reducible to ‘∀Y, X trusts Y’ for similar reasons.
It will not be that X trusts everyone, and there is no determinate range of people that X’s
trust must range over. Rather, ‘X trusts’ seems to make a different claim: that X has faith
in people, in some ‘generalised other’, as Uslaner (2002: 24) says, not faith in any spe-
cific person or description. But again, while ‘X trusts’ and ‘X trusts Y’ seem to be differ-
ent and unique statements, there is some implication from the former to the latter but
not vice versa. If X trusts, there must be some Y that X trusts, but that X trusts Y does
not in any way imply that X trusts more generally. So of the two predicates, the one-
place one is arguably more fundamental. Thus, the heart of our notion of trust seems to
be simply an attitude of trust, which may, but need not, take specific persons as its
object, and which can support, but need not, the act of relying on persons.
The second and third bits of evidence come from considering trust in conjunction
with distrust. ‘To understand trust’, Katherine Hawley (2014: 1) says, ‘we must also
understand distrust, yet distrust is usually treated as a mere afterthought, or mistakenly
equated with an absence of trust.’ The mere absence of trust might report nothing
about one’s attitudes but rather stem from the fact that there is no cause for reliance.
The car mechanic I trust when I don’t seek a second quote, I don’t trust to deliver my
mail. However, my lack of trust here is its mere absence: I don’t trust the mechanic in
this regard not because I don’t think him up to the job but because that is not his job. So
I don’t rely on him in this respect. Distrust, however, is not the mere absence of trust: it
is an attitude in its own right, and one might expect, as Hawley (2014: 4) proposes,
there to be analytic connections between the attitudes of trust and distrust; such as, for
instance, that if distrust is an appropriate attitude to take, then trust is not. However,
that there are such analytic connections is hard to maintain if the fundamental notion
of trust is taken to be three-place or ‘X trusting Y to φ’. Given that trust in this sense is
metaphysically hybrid, any failure of trust can always be down to the failure of the
action component. (My not trusting my mechanic to deliver my mail because I don’t
rely on him to do this.) But then trust could be inappropriate because of some
i nappropriateness in this action component; it would, for instance, be wrong to trust

my mechanic to deliver my mail. However, this wrongness does not imply it is right to
distrust my mechanic. So to keep the parallel between trust and distrust, the focus
needs to be on the attitudinal conception of trust: trusting Y and distrusting Y.
Moreover, this is implied by the fact that there is no three-place distrust predicate:
we do not say ‘X distrusts Y to φ’—I don’t distrust my mechanic to deliver my mail!
The absence of this predicate form then suggests that it is the two place ‘X trusts Y’ that
is fundamental.
Trust and distrust, it is often said, are contraries but not contradictories.13 And this is
true all the while trust is conceived contractually, or as three-place. In this case, a lack
of trust need not imply distrust because there might be a lack of trust because there is a
lack of reliance; there is no contract, as it were, or commitment as Hawley (2014: 10)
would say. However, a lack of trust can imply distrust. Where trust is the background
attitude—where it is two-place or one-place—if trust is lost what remains is not merely
its lack but distrust. Suppose X trusts Y. This trust is manifest in X’s disposition to rely
on Y in various ways. And were X to report that he trusts Y, what X would thereby
describe is an attitude towards Y that is a basic attitude that one can take towards a
person, which involves making positive presumptions about their goodwill towards
oneself. Remove these positive presumptions, so that it can no longer be taken for
granted that Y will act in certain ways and will not act in others and what is left is dis-
trust. For example, you might not seek a second quote simply because you trust your
mechanic, and if so, you just presume the quote is honest; you might leave your diary
lying on the desk simply because you trust your partner, and if so, so you just presume
they won’t read it; or suppose you trust your partner, and if so you will just presume
they are not cheating on you; and so on. Remove trust in these cases, so you no longer
presume the quote honest, the diary safe, or your partner faithful and these situations
are now ones of distrust.
Relatedly, we tend not to trust people not to do things. For instance, you don’t trust
your partner not to have an affair, not because they can’t be trusted in this but because
such trust is peculiarly self-defeating. To trust them not to have an affair would be to
draw their attention to the fact that you do not presume they will not, which amounts to
not trusting them in this respect. Equally, your partner would not reassure you were
they to say ‘don’t worry I won’t be unfaithful’. This should be unspoken, part of what is
presumed by mutual trust. The same goes for one-place trust; in having a non-directed
attitude of trust we presume things about how people in general will behave towards us.
For instance, we presume they won’t be unpromptedly aggressive. This presumption,
Williams observes, can be sustained by reasoning ‘in desperate circumstances’, but in
‘better times’ we just take it for granted. And it needs to be taken for granted because
‘[o]ne is not likely to be reassured by someone who says, “I promise not to murder you” ’
(Williams 2002: 89). Thus a proper account of the relation of trust to distrust, and the
13
See Jones (1996: 15).
122 Paul Faulkner
recognition that these can be contradictory, requires a purely attitudinal conception of

trust. And it is then hard to see how it is not this attitude that is, as Williams (2002: 88)
says, the basic form of trust ‘on which all social interaction depends’.
The fourth piece of evidence concerns the relationship between trust and trust-
worthiness. The attitude of trust is, in part, identified by its relation to the thought that
the trusted is trustworthy. Reassurance comes from this thought. However, this con-
nection between trust and trustworthiness is broken if trust is conceived contractually,
or as three-place. Under this conception, say in a case where X trusts Y to φ, the thought
that Y is trustworthy is, at least, that Y will reliably φ. However, it might be that this is
not the trustworthy thing to do and, indeed, can be quite the opposite. This might be
illustrated by a case where the trusting party is in error. Imagine a hot parched land and
X arriving thirsty at Y’s homestead. He asks Y for water from the well that stands in
front of Y’s house, and Y responds by telling him that he can’t have that water and then
goes inside. In fact Y has gone to fetch X some clean water from the tank at the back of
the house, the water in the well standing in front of the house having been poisoned by
livestock that fell into it and died at the start of spring. Not knowing this, X will judge Y
untrustworthy, and if trustworthiness is identified by reference to trust Y would be so.
But of course, Y’s response is the right and trustworthy one. This point is made and
developed by Knud Ejler Løgstrup in his discussion of trust:
The other person’s interpretation of the implication of the trust offered [that is, the trusting
party Y’s interpretation] . . . is one thing, and the demand which is implicit in that trust . . . which
I must interpret is quite another thing. (Løgstrup 1997: 21)
Responding to trust cannot be ‘merely a matter of fulfilling the other person’s expect
ations and granting his or her wishes’ (Løgstrup 1997: 21). This is because, in the trust
situation, such as that of the poisoned well, ‘what we are speaking of is a demand for
love, not for indulgence’ (Løgstrup 1997: 21). Thus the demand on the trusted—what
Løgstrup calls the radical ethical demand and might be called the demand that X be
trustworthy—is generated by the fact of the trusting party’s dependence.14 It is not gen-
erated by X’s attitudes—that is, by his trust. But this is to say that trustworthiness can-
not be defined with respect to trust if trust is conceived contractually, or as three-place.
The analytical connection between trust and trustworthiness is preserved if trust is
taken to be merely an attitude, or as two-place. For suppose, in the poisoned well case,
that X simply trusts Y. In trusting Y, X will think that Y is trustworthy. And in thinking
this, X will not place any specific expectation on Y, but will rather just expect it of Y that
Y does the right or trustworthy thing.
Connected to this point is Katherine Hawley’s observation that trust can be
unwanted. In this regard she gives the example of trusting her colleagues to buy her
champagne, in a situation where, for whatever reason, she is to be honoured. Now
it might be that her colleagues plan to buy her champagne but, Hawley (2014: 7)
14
Although Løgstrup doesn’t ever talk of trustworthiness; see Faulkner (2016).
observes, ‘[s]till, they do not invite or welcome my trust in this respect; instead, they
want to give me a treat, not merely to act as trustworthiness requires, and certainly not
to risk betraying me if they forget to buy the champagne’. This observation is good, but
her trust is unwanted, in part, I suggest, because it implies the falsehood that the col-
leagues would be untrustworthy if they did not supply it. This is false precisely because
being trustworthy is not a matter of wish fulfilment. It is a matter of doing the appro-
priate thing, which might still be to buy champagne, and Hawley’s colleagues want to
be trusted to do this. So to trust them is to trust in the two-place sense and such trust
would not be unwanted. What is objectionable is the implicit contract, not the back-
ground attitude.
The fifth and final piece of evidence for the priority of two-place over three-place
trust comes from a consideration of infant trust. Any account of trust, Annette Baier
(1986: 244) proposes, should accommodate infant trust. And this generates the con-
straint ‘that it not make essential to trusting the use of concepts or abilities which a
child cannot be reasonably believed to possess’. Suppose now that X trusts Y to φ. In
trusting Y to φ, X will take an optimistic view of Y and her motivations; and in so taking
this view X will, at the very least, presume that Y will φ, and φ because X manifestly
depends on her doing so. To do what X trusts for this reason would be to be trustworthy.
So in trusting Y to φ, X presumes that Y is trustworthy (or at the very least X presumes
this: often, if not ordinarily, this presumption will be an item of knowledge or firm
belief). Thus X’s trusting Y to φ involves a complex of reasoning. It involves imagining
the trust situation from Y’s perspective, imagining Y’s recognition of X’s dependence,
and imagining Y seeing this as a reason to do what X depends on Y doing. Now it is
arguable that this kind of second personal reasoning is both prosaic and f undamental
to moral thought.15 However, it is not the kind of reasoning that an infant could engage
in. By contrast, suppose that X trusts Y; for instance, an infant X trusts his mother Y. In
trusting his mother, X will have the thought that she is trustworthy and, at the very
least, presume this thought to be true. But this thought need not be articulated in these
terms and its possession involves no second-personal awareness; it is merely the
thought that his mother will do the right thing, had in a context of dependence. And
even this thought need not be articulated: it amounts to no more than confidence or
faith in his mother’s actions. This does seem to be the kind of thought that an infant
could have. Suppose then that Baier’s constraint on accounts of trust is plausible. What
this implies is either that an account of trust must satisfy her constraint, or acknowledge
that two-place trust is a more basic form than three-place.
6. The Trust-Based Solution

Where does this leave the problem of trust? This problem, I argued in section two, is
confronted when three facts hold true of a situation where one might rely on someone:
15
See Darwall (2006) and Faulkner (2014).
124 Paul Faulkner
(i) reliance has a worst-case outcome—one relies and the other proves unreliable;
(ii) both parties know the interaction to be one-off, or of a determinate number; and
(iii) one is entirely ignorant of the other party’s motivations, but know there is some
general reason to be unreliable. An internal solution to this problem is then one that
explains how it is that reliance is rational in such situations with a worst-case outcome
even when (ii) and (iii) hold. An internal solution is what is philosophically wanted
because we do rely in these circumstances, and any other solution would fail to make
sense of the rationality of doing so. Both Gauthier and Pettit advance internal solu-
tions, which for the reasons discussed, I think fail. And this chapter advances an
internal solution, which might be called the trust-based solution. According to this
solution, the problem of trust is insinuated by the background assumption that it is the
subject’s beliefs and desires—their preferences—that explain action; and that rational
action is a matter of maximizing preference satisfaction. This insinuates the problem
because it excludes the reason that can, and often does, motivate reliance, which is
simply the attitude of trust. Thus, with respect to the Trust Game, and the question of
why rely on the trustee when (ii) and (iii) hold, the answer can be just that the investor
is trusting. And this answer can be not merely a description of the investor’s psycho-
logical disposition; it can also capture the investor’s reasons.
This raises the question of how it is that a trusting attitude can rationalize reliance.
And this question is, in fact, two: how does a trusting attitude rationalize reliance?
And, why does a trusting attitude rationalize reliance? With respect to how, the answer
is via what is involved in thinking well of others, which is what having a trusting atti-
tude amounts to. Suppose that X is trusting (one-place) or specifically trusts Y (two-
place). If this is true, then in a case where X has a choice of relying on Y, X will presume
that Y will do the right or trustworthy thing. In the Trust Game, this presumption
would be that the trustee will return a fair share of the money gained. Making this pre-
sumption is constitutive of having a trusting attitude: in taking this attitude one thinks
well of others, and this presumption articulates this thought in a context of potential
reliance. Thus the presumption is not based on any evidence of Y’s reliability and, as
such, it would be available even if X knew nothing about Y’s particular motivations. It
would be available that is, in a situation where (i), (ii), and (iii) hold true. However, the
question is then whether this presumption is available to the investor in the Trust
Game when the investor knows that there is some general reason for the trustee to
be unreliable. The answer, I think, is that it is a feature of trust that it has a certain
resilience to doubt. Were X to possess knowledge of Y’s particular motivations to the
effect that Y’s preference was always for profit, trust would become hard (though I con-
jecture, it might still be possible to give Y the benefit of the doubt). However, in the
absence of any such particular grounds for worry, the nature of trust as an optimistic
attitude involves giving no credit to a mere general worry.16 It follows that the pre-
sumption that Y will do the right or trustworthy thing would be available, and that, as a
16
In this respect, one might say that trust is the default; see Stern (2016).
consequence, it would be rational for X to make the transfer. If X’s attitude is trusting—
if X trusts (one-place) or trusts Y (two-place)—then it is rational for X to rely on Y in
certain respects. In particular, if X’s attitude is trusting, it is rational for X to rely on Y to
make a back transfer, even where (ii) and (iii) hold true.
Why, then, does a trusting attitude rationalize reliance? The short answer: because
trusting and being trustworthy are actions that we intrinsically value; and in valuing
behaviours that are so described we think that, other things being equal, we have a
reason to act in these ways, and acting in these ways is the right thing to do. Moreover,
valuing these behaviours cannot be translated into a statement about our preferences,
even if it might imply such; value is not, as Gauthier (1986: 49) says, merely ‘a measure
of individual preference’. The solution to the problem of trust is not this: change
the preference ordering by adding a desire to be trusting and trustworthy. The prefer-
ence ordering is rather kept as described: the problematic cases of reliance are ones
where (i) holds true; they are cases with a worst outcome, which is that one relies and
the other proves unreliable. These are cases where some reassurance would seem to
be needed. Thus the solution is not to change these cases by adding such a desire to be
trusting that what was the worst outcome ceases to be so. Rather, in intrinsically valu-
ing trusting and trustworthy behaviours, we recognize reasons for acting in these ways
that are not reducible to our preferences for so acting. Thus, in intrinsically valuing
these behaviours, we think that rational deliberation in a potential reliance situation
should start from the presumption that the other party will behave trustworthily.17
And, when trusted, we think that rational deliberation should start from the fact of the
trusting party’s need. We then explain and justify our own and others’ behaviour by
reference to these prescriptions. Thus the claim about intrinsic value can be under-
stood as a claim, as Williams says, about how these actions and their accompanying
attitudes make sense to us from the inside.
Moreover, deliberating as the trusting and trustworthy person cannot be captured
in terms of a preference for following a norm—the norm, for instance, of trusting, or
presuming that the other party is trustworthy. This is because deliberating in these
ways involves seeing the potential reliance situation in a certain light, where this percep-
tual metaphor highlights the immediacy of the normative judgement. Take the Trust
Game. The right, or trustworthy, way of viewing an investor’s transfer is as being a reason
to make a back-transfer, and this is how a trustworthy trustee would see it. Then in
presuming that the trustee is trustworthy, the trusting investor would likewise see the
situation generated by their transfer as one wherein the trustee has this reason. That is,
the perception of this situation will contain within it a judgement about what ought to
be done and, correlatively, about how action is to be explained.18 There is no inference
from the thought that some norm applies in this case conjoined with the desire to
17
In this respect, as Løgstrup (1997: 18) observes, ‘we do not normally advance arguments and justifica-
tions for trust as we do for distrust’. Again, see Stern (2016).
18
See McDowell (1978: 100–1).
126 Paul Faulkner
f ollow this norm; and to suppose that there is would be to miss the immediacy of what
is essentially a non-inferential judgement.
In conclusion, let me make three observations about this trust-based solution. First,
the claim is not that a preference-based explanation of action generally fails; on the
contrary, this form of explanation is good in very many cases. Moreover, it can be
the best way of predicting behaviour in a potential reliance situation. The trustee in
the Trust Game might be Gauthier’s straightforward maximizer. Now trust tends to
exclude doubt.19 So if one’s attitude is trusting, then, other things being equal, one will
put aside the worry that another is not similarly motivated—if this worry occurs at all.
However, other things might not be equal: one might have concrete grounds for doubt.
And if this were so, the only way to render reliance rational would be to support it with
a belief about outcome. At this point, a calculation of preference would be needed,
since an attitude of trust could not rationalize reliance given sufficient evidence that it
is not mutual. However, what makes this normative claim true, ordinarily, is that given
such evidence, the attitude of trust would no longer be available.20 It would be replaced
by an attitude of distrust. What our intrinsically valuing trust and trustworthiness
then explains is why distrust is not so pervasive.
Second, the philosophical justification of our valuing trust and trustworthiness,
which is Williams’s genealogical argument, is not what Korsgaard (1986: 22) calls an
‘ultimate justification’; that is, it does not purport to show that ‘all rational persons
could be brought to see that they have a reason to act in the way required’, which, in this
case, is to act as the trusting or trustworthy person does. If Gauthier’s or Pettit’s justifi-
cation of reliance—‘trusting’ as they say—were successful, then such an ultimate justi-
fication would be available. But a significant burden of this chapter has been to support
Williams’s contention that the problem of trust is endemic to this conception of rea-
son. What the solution argued for here involves is an expansion of the scope of what
counts as a reason for acting. Korsgaard, of course, argues for such an expansion in the
hope of ultimately justifying moral principles. It would take more time to argue this
point but the problem this strategy confronts, at least in application to the present case,
is that it must regard strategic reasoning as a failure of rationality. But there seems no
rational deficiency in reasoning in a tit-for-tat manner.
Third, the ‘morality’ of trust is thus a veneer in the sense that we need not have
achieved this way of looking at things, and this way of looking at things could easily
break down. In Williams’s terms, we are lucky to live in ‘better times’ (Williams 2002:
89).21 However, this is not to make intrinsic value purely a matter of our valuing: our
having this set of values is further valuable in the sense that it would be a bad thing if
we saw the world otherwise. These are better times. In this respect, the shift in how the
19
This point is well made by Möllering (2009: 140): ‘Vulnerability is a precondition for trust . . . [but]
trust captures the highly optimistic expectation that vulnerability is not a problem’.
20
‘Ordinarily’ because there may be cases—a parent’s relation to their child, for instance—where one is
bound to trust almost come what may.
21
Equally, it is a contingent matter if truth-telling is predominant; see Sober (1994).
trust situation is conceived that is involved in having a trusting attitude is not merely a
matter of having a certain upbringing—it is a matter of having a good upbringing. It is
just that this notion of goodness cannot be worked into correctness since there could be
other ways of securing the motivations needed to render trust non-problematic.22, 23
References
Baier, A. 1986. Trust and Antitrust. Ethics 96:231–60.
Becker, Lawrence. 1996. Trust as Noncognitive Security about Motives. Ethics 107:43–61.
Blais, M. 1987. Epistemic Tit for Tat. Journal of Philosophy 82 (7):335–49.
Darwall, Stephen. 2006. The Second-Person Standpoint: Morality, Respect and Accountability.
Domenicucci, Jacopo and Richard Holton. 2016. Trust as a Two-Place Relation. In The
Philosophy of Trust, ed. P. Faulkner and T. Simpson. Oxford: Oxford University Press.
Elster, Jon. 2007. Explaining Social Behaviour: More Nuts and Bolts for the Social Sciences.
Faulkner, Paul. 2014. A Virtue Theory of Testimony. Proceedings of the Aristotelian Society
114 (2):189–212.
Faulkner, Paul. 2016. Trust and the Radical Ethical Demand. In What is Ethically Demanded?
Essays on Knud Ejler Løgstrup: The Ethical Demand, ed. H. Fink and R. Stern. Notre Dame,
IN: University of Notre Dame Press.
Gauthier, David. 1986. Morals By Agreement. Oxford: Clarendon Press.
Gauthier, David. 2013. Achieving Pareto-Optimality: Invisible Hands, Social Contracts, and
Rational Deliberation. Rationality, Markets and Morals 4:191–204.
Glaeser, E., D. Laibson, J. Scheinkman, and C. Soutter. 2000. Measuring Trust. Quarterly
Journal of Economics 113:811–45.
Hardin, R. 2002. Trust and Trustworthiness. New York: Russell Sage Foundation.
Hawley, Katherine. 2014. Trust, Distrust and Commitment. Noûs 48 (1):1–20.
Jones, K. 1996. Trust as an Affective Attitude. Ethics 107 (1):4–25.
Korsgaard, Christine. 1986. Skepticism about Practical Reason. The Journal of Philosophy
83 (1):5–25.
Løgstrup, Knud Eljer. 1997. The Ethical Demand. Notre Dame, IN: University of Notre Dame
Press.
McDowell, John. 1978. Are Moral Requirements Hypothetical Imperatives?. In Mind, Value
and Reality, ed. J. McDowell. Cambridge, MA: Harvard University Press. Original edition,
1978.
Möllering, Guido. 2009. Leaps and Lapses of Faith: Exploring the Relationship Between Trust
and Deception. In Deception: From Ancient Empires to Internet Dating, ed. B. Harrington.
Stanford, CA: Stanford University Press.
22
See Velleman (2010) for relevant ethnographic data.
23
This chapter has had a particularly long production time and has gone through many incarnations.
Thanks are owed to audiences in Arizona, Bristol, Edinburgh, Griefswald, Manchester, Sheffield, Stirling,
Warwick, and York. Particular thanks are owed to Don Fallis, Katherine Hawley, Richard Holton, Arnon
Keren, Guy Longworth, Neil Manson, Matt Nudds, Tom Simpson, and Bob Stern.
128 Paul Faulkner
Pettit, Philip. 2002. The Cunning of Trust. In Rules, Reasons, and Norms, ed. P. Pettit. Oxford:
Clarendon Press.
Sober, Elliott. 1994. The Primacy of Truth-Telling and the Evolution of Lying. In From a
Biological Point of View: Essays in Evolutionary Philosophy, ed. E. Sober. Cambridge:
Stern, Robert. 2016. Løgstrup on the Priority of Trust. In The Philosophy of Trust, ed. P. Faulkner
and T. Simpson. Oxford: Oxford University Press.
Uslaner, Eric M. 2002. The Moral Foundations of Trust. Cambridge: Cambridge University
Press.
Velleman, David. 2010. Regarding Doing, Being Ordinary. Unpublished paper.
Williams, B. 1988. Formal Structures and Social Reality. In Trust: Making and Breaking
Cooperative Relations, ed. D. Gambetta. Oxford: Blackwell.
Williams, B. 2002. Truth and Truthfulness. Princeton, NJ: Princeton University Press.
8
Trust and Collective Agency
Bernd Lahno
1. Introduction
Trust is an attitude that enables us to expose ourselves willingly to a certain kind of
risk: the risk that another person may do something harmful to us. In almost all
cooperative endeavours individuals are required to make themselves vulnerable to the
action of their cooperative partners in some way or other. This is one major reason why
trust is so important and valuable to us.
In collective agency the individual mode of cooperation is transcended. People take
on the goals and values of a group and submit their decisions to a scheme of actions
that is collectively held to be best in view of these goals. But the risk remains. If I act as
part of a collective agent, I run the risk that others will not. The consequences may
well be against the interests of the group as well as against my individual interest that
would have been my guide had I not identified with the group. Moreover, my positive
expectations to be part of a collective enterprise would be frustrated and this may
cause severe hurt. By not committing to the group I can prevent or at least lower some
of the harm to be expected from a failed attempt to form a collective agent. But the
price would, of course, be to forgo the collective enterprise from the outset. Again,
trust would seem to be the ground on which people are enabled to come together and
form a collective agent in spite of the individual risk incurred.
In this chapter I will argue that trust is in fact inherent in collective agency. The
we-perspective of an individual identifying with a collective includes a certain percep-
tion of other group members which may easily be identified as a particular form of
trust. My argument will be based on the analysis of simple cases featuring a two-person
group engaged in some collective activity such as dancing or paddling a canoe. But
I am confident that the argument can be extended to the general case.
Although it is generally acknowledged that collective agency involves some vulner-
ability on the side of a complying member of a group to deviant behaviour of his fellow
members, one hardly finds any reference to or even analyses of trust in the classic
130 Bernd Lahno
literature on collective intentionality.1 I am aware of two exceptions. The first excep-

tion is joint work by Raimo and Maj Tuomela (2005), based on Maj Tuomela’s concept
of ‘rational genuine trust’.2 It is argued on conceptual grounds that rational cooper-
ation in the we-mode entails a weak form of trust, while on contingent grounds it fos-
ters the cultivation of a relationship between the members of the collective in which
‘genuine trust’ can thrive (see Tuomela and Tuomela 2005: 79f). The second exception
is a recent paper by Bernhard Schmid (2013).3 Schmid argues that the perception
of other group members in collective intentionality can neither be satisfactorily
accounted for in pure terms of cognitive representations as in belief or reliance nor in
terms of normative expectations only. The intentional structure of a joint action,
according to Schmid, actually includes as an integral component an attitude that com-
bines both cognitive and normative elements, namely trust.
My account here has much in common with these two approaches. However, there
are also essential differences. Like Schmid (but unlike the Tuomelas) I will argue that
(genuine) trust is in fact a constituent rather than a precondition for collective agency.
Like the Tuomelas (but unlike Schmid) I will presuppose a specific concept of genuine
trust. But unlike both I am not only concerned with the conceptual or contingent con-
nection between collective agency and trust. A major concern of my analysis here is
also to explore the specific form of trust that we find in collective agency. What are the
specific demands on trust in collective agency and how does trust adapt to these
demands? My aim is to contribute—by asking these kind of questions—to a better
understanding of collective agency and its inherent characteristic form of trust. This
will hopefully also improve our understanding of other forms of genuine trust in indi-
vidual interaction characterized by similar problems.
My argument will proceed as follows. After briefly sketching the concept of trust
as an emotional attitude that I will use throughout this chapter (1) I will introduce a
simple game to represent the fundamental problem of coordination in collective
agency (2). My discussion will start from the theory of team reasoning (TR) as
developed by Robert Sugden (1993; 2003) and Michael Bacharach (1999; 2006).
I will discuss in particular Sugden’s claim that some form of ‘assurance’ is needed to
1
Wilfried Sellars (1980), one of the first to notice that there are such things as we-intentions, makes no
reference to trust at all. Tuomela and Miller (1988) argue that we-intentions cannot be reduced to
I-intentions because of a potential conflict between the two. Such a conflict between we- and I-intentions
implies vulnerability, but trust as a possible cure is not mentioned. John Searle (1990) emphasizes:
‘Collective intentionality presupposes a background sense of the other as a candidate for cooperative
agency, i.e. it presupposes a sense of others as more than mere conscious agents, but as actual or potential
members of a cooperative activity.’ But he does not refer to trust as a source of such a perception of the
other. According to Bratman (1992) a shared cooperative activity is characterized by mutual responsive-
ness, commitment to the joint activity, and commitment to mutual support, but, again, no reference to
trust is made. Similarly, Margaret Gilbert (1997) thinks that individuals forming a plural subject are mutu-
ally connected by a conditional commitment. This commitment, she argues, is a source of reliability. But
trust remains unnoticed.
2
See Tuomela (2006) for a comprehensive introduction to this account of trust.
3
There is also a short indication of this argument in Schmid and Schweikard (2013).
trust and collective agency 131
engage in collective action, which is obviously related to trust (3). In contrast to

Sugden, I will argue that genuine trust is a constituent element of collective agency
rather than a precondition of forming a collective agent (4). The argument proceeds
by investigating what the attitude between members of a collective must be like to
perform a complex collective project. I argue that this attitude must be one of trust in
integrity. It serves to overcome potential conflict of interest. Trust in integrity is
often encapsulated in routines and based on a normative framework that is pro-
duced by the group’s social structure and the institutional background (5). Finally,
I point to a more problematic aspect of trust in some cases of collective agency: this
trust is categorical in the sense that it is not based on the individual agent’s values
and goals but, quite the contrary, the values and goals of the individual are consti-
tuted by his trust in the collective (6).
2. Genuine Trust
As mentioned above, trusting behaviour is characterized by the risk that another per-
son may act in undesired ways. A rational individual willingly incurs such a risk only if
she has reason to believe that others will respond cooperatively. Such a person is said to
have trusting expectations. Trust, in turn, is an attitude that disposes us to engage in
trusting behaviour. From this it is only a small step to identifying trust and trusting
expectations. Moreover, many philosophers and social scientists in fact take trust to be
essentially a cognitive expectation that a partner will behave in desired ways.4 Indeed,
the word ‘trust’ is now and again used in common intercourse to refer to a mere belief
that others will act cooperatively. But the concept of trust as used in everyday language
is ambivalent. Most people would respond somewhat reluctantly when confronted
with the case of a marriage trickster ‘trusting’ his victim to be generous to him.
Although they will realize that the trickster may ground his action on well-founded
beliefs, they may hesitate to call it trust. At least, most people will argue that this is not
‘real’ or ‘genuine’ trust.
In what follows I will base my argument on a concept of ‘genuine trust’ that is spe-
cific enough to mark the difference between trust in a more narrow sense and pure
reliance. On the other hand it should be wide enough to encompass the most interest-
ing cases of what people would usually acknowledge as ‘real’ or ‘genuine trust’. I will not
argue for this specific account of genuine trust in this chapter.5 However, I am confident
that those readers who do not agree with the concept of genuine trust introduced here
will at least agree that my definition captures an interesting attitude which plays an
essential role in social interaction and collective agency and is, therefore, worth
investigating.
4
See, e.g., Coleman (1990), Gambetta (1988), Dasgupta (1988), Hardin (1991).
5
But see Lahno (2001a; 2001b; 2002).
132 Bernd Lahno
Here is a short characterization of the concept of (genuine) interpersonal trust that

I will use in the following analysis:
Trust as an emotional attitude towards a person includes a participant attitude and a feeling of
connectedness to him or her by shared aims, values, or norms. This attitude allows the trusting
person to incur risks concerning the actions of the trusted person, as they are perceived as being
guided by the normative foundation of trust, which is felt to be shared.
The two essential features that define this account of trust as an emotional attitude are:
(T1) The other is perceived as a person that can be held responsible for his acts.
Genuine interpersonal trust necessarily involves a personal stance or, more specifically,
a participant stance as defined by Holton (1994) following Strawson (1974). A trusting
person is disposed to react to a misuse of her trust in a particular and emotional way.
This is due to the fact that the other is seen as a responsive person consciously engaged
in interaction with the trustor. As the author of her acts, she is held responsible and,
thus, the expectations of the trusting person are normative in character. In contrast,
perceiving the other from some distance like a mechanism governed by natural behav-
ioural laws, an objective attitude in Strawson’s sense that would allow for pure factual
expectations only, is incompatible with genuine interpersonal trust.
(T2) The other person is perceived as someone connected to oneself by shared
interests and/or normative convictions.
This is the normative foundation of trusting expectations. She who trusts another makes
herself vulnerable because she perceives her partner as being connected to herself by
shared aims or values. Thus, in the eye of the trustor, a situation of trust calls for the real-
ization of such aims or for observing shared norms as part of a cooperative enterprise.
Both of these key elements in interpersonal trust, a participant stance and ‘connected-
ness’ in interests and/or normative convictions, are emotional in character. They essen-
tially characterize the way the partner and the relevant part of the world are perceived in
trustful interaction. Certainly, this way of seeing things will as a rule imbue the trustor
with typical trusting expectations. So there is a causal relationship between trust and
belief. Yet, by inducing certain patterns in the way the world is represented in thought
and certain contents of thought are associated with each other trust primarily deter-
mines how a trusting person thinks. Thus, it cannot be understood as the immediate
result of rational consideration, and trust should not be confused with the expectations
a trusting person usually has, which are but a result of his trusting attitude.
The two conditions exclude pure reliance based on cognitive belief only as a genuine
form of trust. At the same time they are weak enough to include interactions that are
not based on deep personal relationships as in business or between voters and politi-
cians in potential arenas of trust.6
6
Another virtue of this concept of trust is that it may be modified and extended in plausible ways to
account for peculiar or extreme forms of trust as trust in god or trust in institutions. See Lahno (2002) for
a discussion of different forms of trust in different social contexts.
3. The Coordination Puzzle

Suppose two friends, Adam and Berta, are canoeing down a white-water river in a two-
person boat. Adam and Berta are experienced paddlers but the waters are rough.
Moreover, although the two checked the map before they started the trip, the river
changes its course and shape regularly during high flow after the thawing period in
spring. So they do not and cannot know all the peculiarities of the course and, there-
fore, make every effort to look out for unpleasant surprises.
Suddenly, after passing a bend they find themselves directly heading towards a big
rock in the middle of the stream. If they do not react instantly the canoe will bump into
the rock and capsize. The scenery is wild and noisy. There is no chance—and, in any
case, no time—to communicate. The two have to take action. But given the wild cur-
rent neither of the two can navigate the boat alone. If they want to pass the rock on the
right, both have to start immediately paddling jointly to the right; otherwise, they
would have to paddle jointly to the left. Assume that it is easy to recognize for both
that—given the current at the exit of the bend—it is much easier to take the route to the
right. While passing on the right side is obviously easily manageable, the left route is
unclear and might be highly demanding.
B
r l
r 2 | 2 0 | 0
A
l 0 | 0 1 | 1
Figure 8.1 Hi-Lo game
Figure 8.1 represents the situation as it might be perceived by each of the individuals
as a simple strategic form game. If both paddle right (r) they will get their best result (rep-
resented by a utility of 2 units for both); if both paddle left (l) they will still gain one unit of
utility while if they fail to coordinate they will end up with the worst result possible repre-
sented by a utility of 0 for each. A game of this form is usually called a Hi-Lo game.
It is perfectly clear what A and B should—and actually will—do in such a situation.
Taking the right bypass is obviously the best they can do and so each will choose r; that
is, each will do his/her part to realize this optimal solution. However, following the
theory of individually rational decision-making, the situation seems to lose much of its
unproblematic obviousness.
The game has two equilibria (r, r) and (l, l) with (r, r) dominating (l, l) in the Pareto
sense. So r seems to be the unique rational choice in this situation. But notice that the
Pareto-optimality of (r, r) is a property of a common scheme of action, which no indi-
vidual can realize alone by his or her choice. Let us take a closer look at the situation
through the eyes of an individually rational actor A. Although taking the right bypass,
(r, r), is obviously the best choice for the two, there is no clear best option for A. The
134 Bernd Lahno
simple reason is that A’s optimal choice depends on what B does. A has reason to
choose r only if he has reason to believe that B will also choose r (with sufficient prob-
ability). But B is in the same position. Knowing that B is rational like himself, A there-
fore needs a reason to believe that B has reason to believe that A will choose r. But
again, B is in the same position . . . Obviously such reasoning can go on endlessly with-
out ever reaching solid ground. There is no independent reason for any expectation of
any order that either A or B may have available. Instrumental rationality just does not
suffice to determine a unique rational expectation on the given information and, thus,
does not suffice to determine what to do.
Note that the problem does not arise due to a lack of information. In fact, both actors
have all the information available and relevant for choice at their disposal. There sim-
ply is no information that could solve the problem except the information that one of
the actors coordinates in some peculiar way. So, there is no information to induce
rational coordination unless there is an independent reason for one of the players to
choose a particular option.
One is tempted to argue that there is such a reason to choose r in our example,
namely that coordination on r is best for both. But again, this is a reason for the two-
person team only as a unit, as neither of the two can realize this result by his or her
decision alone. It is a reason for Adam only if it is a reason for Berta, and this is all one
can say. We are trapped again in an endless chain of reasoning that is unsuited for the
production of a non-hypothetical, substantial reason to choose one or the other
alternative.
That simple coordination cannot be explained by rational individual decision-making
as conceptualized in traditional decision theory has been noticed many times by
different scholars, most notably by Robert Sugden (1991; 1993) and Margaret Gilbert
(1989; 1990).7 Obviously, the problem is a problem in the theory of rational decision-
making rather than a problem for actual decision-making. Even hard-nosed econo-
mists would not expect Adam and Berta to have any difficulties in deciding what to do.
They will, of course, do what is obviously the right thing to do and conjointly steer
right to pass the rock.
4. Team Reasoning and the Need for Assurance

We are not interested in critically assessing the theory of rational choice here. Our
focus is rather on how forming a collective agent transforms and extends the behav-
ioural potential of human actors. Analysing the problem of coordination from a
rational choice point of view is, nevertheless, of great interest to us because it may
reveal the particular features that make the difference. The core of the problem as
detailed above is that individually rational actors make their decisions in isolation,
7
See Lahno (2007) for a more detailed account of the problem. Sugden (1991) gives an overview on the
‘classic’ literature. He traces the discovery and first analysis of the problem to Hodgson (1967).
taking the decisions of others as independent prerequisites of choice. The fact that
these prerequisites are themselves the result of choice is taken account of by strategic
reasoning. But strategic reasoning of the individual kind cannot overcome the funda-
mental divide between your choice and my choice. In coordination problems individ-
ual rationality in the sense of traditional decision theory tells me what to do only on
condition of what I expect you to do. However, my knowledge that you are like me does
not suffice in determining what you will do and, thus, does not suffice in determining
what I should do. If we form a team the perspective is fundamentally altered. The guid-
ing question is not ‘What should I do?’, but rather ‘What should we do?’ The answer to
this question is often as simple as in our coordination problem: ‘We should pass the
rock on the right side!’ So each of us chooses to do his part in the common scheme thus
defined and as each of us knows that we are alike, we are assured that the collective
scheme is actually realized by our individual acts. The we-perspective determines indi-
vidual choice by defining the right collective action and, concurrently, gives us confi-
dence that we will actually bring it about although each of us controls only part of what
we do together.
This is the basic idea from the theory of collective intentionality as developed in dif-
ferent forms by various scholars during the last few decades.8 We will base our discus-
sion on a particular variant here: the theory of team reasoning (TR) as developed by
Robert Sugden (1993; 2003) and Michael Bacharach (1999; 2006).9 TR presents itself
as a good starting point because of its particularly modest metaphysical and ontological
presumptions and its affinity with classic decision theory. The core idea is that
collectives (‘teams’) should be added to the classic theory of rational choice as possible
agents. A team is a group of individual agents, which I will call ‘teamers’ here. A team
action is a profile of actions by the teamers. It is realized if each teamer chooses her
action according to the scheme as given by the profile. So collective actions are wholly
composed of individual actions (what else?) and collective decisions are grounded in
individual decisions based on individual reasoning. The characteristic feature of
decision-making according to TR is that the individual decisions which define the
collective act are based on a peculiar way of (individual) reasoning: a teamer takes on
the we- or team-perspective to determine what to do. She asks ‘What should we do’ and
does whatever the answer to this question identifies as her part.
This decision-making procedure presupposes that there is a common scheme of
evaluating the consequences of interaction from a team-perspective. Each team is
assumed to be characterized by its team preferences on the possible outcomes of (indi-
vidual) action. Given such an evaluation scheme, the question ‘What should we do?’
8
See Schmid and Schweikart (2013) for an overview.
9
A good recent overview including some hints at differences between the accounts of Sugden and
Bacharach is found in Sugden and Gold (2007). More recently, Sugden revised his theory of team reasoning
fundamentally in giving up the idea that the team reasoning decision-making role can be characterized as
the maximization of expected team utility (Sugden 2015). But he still upholds the idea discussed below that
assurance is fundamentally needed to engage in team reasoning.
136 Bernd Lahno
can be answered in the standard instrumental way by identifying the combination of

actions that yields the best result according to team preferences. TR thus extends clas-
sic decision theory by two essential elements:
(1) Teams characterized by a team utility function are added as potential agents;
(2) A specific decision-making rule determines choices of teamers: choose your
part in the optimal scheme of team action relative to team preferences.
Team reasoning is understood as one mode of reasoning among others. The idea is
that it may be activated in situations that suggest conceiving choice as an issue of a
collective rather than of individuals. But even if a situation may be characterized as
such from the outside, TR does not necessarily demand that individuals actually
employ team reasoning. Team reasoning is conditional on a certain perception of the
situation from inside, that is, by the individuals that are to form the collective. It pre-
supposes that individuals conceive of themselves as members of a team and perceive
this team as a potential unit of (collective) agency characterized by its team goals (cf.
Sugden and Gold 2007: 125). The essential element of such group identification then is
that individuals take on the team goal as their own.
Whether or not an individual identifies with a team is taken to be a matter of fram-
ing, rather than a matter of rationality. Both Sugden and Bacharach agree that there are
empirical regularities of the form that group identification in individuals is promoted
by certain situations but prevented by others.10 But these issues are primarily the object
of empirical research in psychology or behavioural science, not an integral part of TR.
Sugden explicitly takes TR—just like standard decision theory based on individual
instrumental rationality—as a theory without empirical content (Sugden 2000: 203).11
He is exceptionally cautious when it comes to the question under what conditions TR
will actually be used by empirical individuals. In contrast to Bacharach he thinks that
group identification by all potential teamers does not suffice in identifying team
reasoning as the preferred or even unique rational mode of reasoning. Sugden argues
that it may still be rational to refrain from team reasoning, even though group identifi-
cation is common knowledge. To see this, consider a team playing Hi-Lo. Assume that
it is common knowledge that both individuals A and B identify with the team. Assume
further that both believe that the partner will choose Lo. Then—in view of an optimal
result for the team, but in contrast to the demand of the team reasoning decision-making
rule—both will rationally choose Lo (and their beliefs will be mutually confirmed).
10
Hindriks (2012) may serve to exemplify this idea. Based on literature from scholars of the so-called
social identity approach in social psychology, he argues that social categorization is a core factor in group
identification.
11
In contrast, Bacharach identifies at least one condition of group identification within his account of
TR, which he refers to as ‘strong interdependence’. It is beyond the scope of this chapter to give a detailed
account of strong interdependence. In a nutshell, a situation is one of strong interdependence if there is a
feasible outcome that Pareto-dominates every (other) possible solution consistent with the principles of
individual rational decision-making as specified by RC; see Bacharach (2006: 84). Note that Hi-Lo, as well
as many dilemma games such as the much discussed Prisoners’ Dilemma, display this peculiar property.
A person identifying with the group may thus, in the very light of group goals,
still rationally deviate from team reasoning, if (she believes that) all others do.
Consequently, to engage in team reasoning an individual has to be sufficiently assured
that others not only identify with the group but also use this mode of reasoning as well.
According to Sugden, therefore, any complete empirical explanation based on TR has
to refer to some form of ‘assurance’. Individuals must have good reasons to believe that
others are teamers; they must have good reasons to believe that the others also have
good reasons to believe that group members are teamers, and so on.12
Sugden’s argument is especially interesting for us here because it touches upon the
role of trust in forming a collective subject—albeit a specific cognitive form of trust
that Sugden calls ‘assurance’. It seems to me that Sugden is right in his analysis but
wrong in his conclusion. The analysis is correct: if group identification consists essen-
tially in individuals taking on the group goal as their own, then common knowledge of
group identification does not suffice in solving the problem of coordination. Each
individual still faces the same old problem: if I do my part in the optimal scheme of
action this will contribute to achieving the team goal only if others act alike. So my
preference for the team goals give me a reason to apply team reasoning only if I have
reason to expect that others apply the decision-making rule of team reasoning as well.
But again, others are in the same situation, and thus we are in the now well-known trap
of endless reasoning without solid ground.
Sugden’s suggestion for solving the problem is that some additional evidence should
be provided such that the individual agent is ‘assured’ that others (also) reason accord-
ing to the team reasoning decision-making rule. He also identifies a main source of
such assurance: common experience of a shared practice (Sugden and Gold 2007:
135). If people regularly observe that others act according to the beneficial scheme and
if this is a public experience—everybody observes the regularity, everybody is aware
that everybody else does, etc.—then each individual is sufficiently assured that suffi-
ciently many others not only identify with the group but also act on the group’s goal in
the way TR prescribes.
I do not want to deny here that the observation of a shared practice in the present
and/or past may well justify expectation of conforming behaviour in the future.
However, it seems questionable to me whether or not this can provide an adequate
foundation for team reasoning. I doubt that ‘assurance’ in Sugden’s sense can provide
the reasons that we need to engage in collective reasoning and agency. Note that shared
practice does not seem to play a significant role in our example. Is there any shared
practice at all to guide Adam and Berta while paddling down the river? And if there is
one, is it actually needed to ground their decision? Obviously, a shared practice of
passing a rock in the current on the right-hand side cannot be the correct basis for the
collective action as described, although it would, of course, produce the same outcome
12
This is a rough sketch of the condition only. For a precise formulation see Sugden and Gold (2007:
32ff), and Cubitt and Sugden (2003) for an elucidation of the concept of ‘reason to believe’.
138 Bernd Lahno
in the given case. A more suitable shared practice seems to be ‘pass a rock on the side
that seems to offer the more pleasant and easier paddling passage’. But I cannot see why
the experience of such a practice should be necessary to coordinate. A team may well
be confronted with the sudden and unforeseen occurrence of a rock in the middle of
their paddling line for the very first time and still react as a team in an optimal manner.
The most suitable practice that would seem to justify reasoning as a team according to the
decision rule in general seems to be a practice of reasoning as a team according to
the decision rule. But this would presuppose a practice of what is yet to be explained.
One might want to respond that simply observing a corresponding behavioural
regularity without any assumption about its motivational foundation suffices to
justify engaging in team reasoning. But such a justification would be inadequate
for two reasons.
First, ‘assurance’ by the bare observation of a behavioural regularity is based on
purely inductive reasoning. I observed that others regularly behave in such and such a
way in the past, so I infer that they will also act in just these ways in the future if the
circumstances are alike. In this way, inductive reasoning bypasses the pitfalls of
strategic reasoning and, thus, solves the coordination problem. But the price of this
argument is that a specific stance towards the others must be presupposed. Others are
treated as a piece of nature governed by law-like regularities. They are not perceived as
agents guided by reason just like the individual himself. But this precludes perceiving
them as genuine participants in interaction and, in particular, as candidates for joint
action. Thus, reasoning from a bare behavioural regularity presupposes an attitude
towards the others that seems incoherent with perceiving the others as teamers in a
joint enterprise.
Second, if ‘assurance’ by experiencing a behavioural regularity produces a compel-
ling reason to reason and act as a teamer, then it also directly justifies doing one’s part
in the collective scheme from a standard rational choice perspective. If I have reason to
expect that others act according to the collective scheme, then, given the structure of a
coordination problem, it is individually rational for me to conform. No reference to a
team reasoning decision rule is necessary here. No reference to the team or ‘we’ is
needed at all. The team reasoning decision rule loses its significant role in solving the
problem of coordination.
There is a general lesson to be learned here. Once the belief that others will conform
is justified, no decision-making rule other than the classic rule of individually maxi-
mizing expected utility is needed to coordinate. So, if this is the proposed route to
solving coordination problems, why should we bother with team reasoning as a mode
of reasoning at all? A straightforward account would suggest a minimal extension of
traditional rational choice theory instead. Such a minimal extension would comprise
the introduction of collective utility functions guiding individual behaviour and the
acceptance of ‘assurance’ as a valid source of evidence.
One problem of such a theoretical approach is that it does not properly cohere with
the way a situation as the one in our example is, in fact, perceived by the participating
individuals. It is very unlikely that Berta, if being asked to justify her decision to paddle
to the right, would answer something like ‘(I believed that) Adam would paddle to the
right and, given this, my best choice to realize my personal (or, alternatively, our
shared) goal was to choose paddling right as well.’ Instead we would expect her to say
something like ‘We wanted to get down the river fast and safely, and passing the rock
on the right side was the best way to do this. So I did what was my part in realizing this.’
The main reasons we would expect Berta to specify are the collective intention (pad-
dling down the river fast and safely), a belief about which collective scheme of action
would be optimal in realizing the intention (going to the right), and how her contribu-
tion is defined by the collective scheme. These reasons just specify how we conceptual-
ize making a decision as part of collective agency. Expecting for instance that Adam is
going to paddle to the right is not among these reasons. This is not to say that Berta
would not actually believe that Adam will do just this. She will, of course, endorse such
a belief, if asked for it.
As a matter of fact, if Berta finds out that Adam will not do his part in what is known
to both as the optimal scheme given their shared goals, she will refrain from complying
with the team reasoning decision rule herself. But this is because she will instantly give
up her perspective as a participant in the collective enterprise. A belief that others will
not comply with the optimal collective scheme of action is simply inconsistent with the
perception of these others and oneself as forming a collective unit engaged in a joint
endeavour. Believing that others comply with the joint project is a necessary condition
of joint agency in a strict logical sense. If an individual has no such belief, then there is
no ‘we’ and no joint action for her that she could be part of. I can perceive of myself as a
teamer only if I actually identify a group of individuals as a team, that is, if I perceive
other group members as being teamers too. But this does not mean that believing that
others act as teamers is my reason to conform with the optimal scheme. If I act as a
teamer, I do not need such a reason.
5. Trust as a Constitutive Element of Collective Agency

Sugden’s argument is misguided. Team reasoning—comprising taking on the team
goals and values as one’s own and applying the team reasoning decision rule—is, as
Sugden and Bacharach forcefully argue, a particular mode of reasoning that is acti-
vated and formed by a particular frame of the situation. Framing a situation in a pecu-
liar way is part of adopting a peculiar attitude or stance towards this situation. However,
we do not adopt a specific attitude because of argument. Our attitudes are not the
immediate object of choice. What we know about the world plays of course an important
role in the development of our attitudes towards the world. But we do not reason from
the information we have in choosing our attitudes towards the world. To some extent,
the interrelation between belief and attitude is even conversely directed. Our percep-
tion of the world and thus the information that we extract from our observations and
the way we process this information is crucially determined by our attitudes. Being a
140 Bernd Lahno
teamer includes a specific way of perceiving other team members and, therefore, a spe-
cific way of interpreting their behaviour. If we look at others as a teamer we do not look
for reasons to believe that they act as teamers, we presuppose that they are and inter-
pret their actions in this light.
It seems inadequate to ask ‘What are the decisive reasons to adopt team reasoning as
one’s effective mode of practical reasoning?’ Reasons may play an important role in
grounding beliefs that support the characteristic attitude underlying team reasoning.
But adopting an attitude is not the conclusion of a reasoning process. And this trans-
lates to team reasoning as an essential element in realizing the relevant attitude here.
Team reasoning is a source of reasons rather than its product. This does not mean that
team reasoning cannot and need not be explained. But the explanation that we should
be looking for is not of the explanation-by-reasons kind. Instead of asking ‘What are
our reasons to engage in team reasoning?’ we should be asking questions like: ‘What
are the empirical conditions which foster team reasoning as a dominant mode of
reasoning among the members of a social group?’ Or, in short, ‘What causes team
reasoning?’13 As I said before, it is not my aim here to explain how individuals may
form a collective unit of agency that can be assigned intentions and actions. This is an
issue for social psychology and the behavioural sciences. I am interested in the more
philosophical (and probably decision theoretic) task of understanding what the essen-
tial elements of collective agency are and how collective agency may shape our life.
From the discussion of our simple coordination problem in collective action we
may gain some very basic insights into the character of collective agency:
(1) The members of a group of individuals who form a collective unit of agency are
characterized by a certain attitude towards the group and the other members of the
group.
(2) The group is perceived as an independent entity with a more or less specific
domain of action and a collective preference order on the respective domain of
collective action or the possible consequences of action.
(3) An individual member of the group will take on the preference order of the
group as her own. This will motivate her to do her part in what she identifies as the
optimal collective scheme of action relative to the collective preference order.
(4) An individual member of the group will perceive others as being members like
herself. They will also take on the collective preference order as their own and be
motivated to do their part in realizing what is perceived to be the optimal collective
action relative to the shared collective preference order.
This implies that collective agency is fundamentally related to genuine trust.
According to (4) group members perceive each other as being alike and guided by the
13
There is empirical evidence that neither Bacharach’s strong interdependence nor ‘assurance’ in the
sense of Sugden (nor a combination of the two) are sufficient to bring about team reasoning as a source of
coordination; see Lahno and Lahno (2014).
same shared goals and group values (defined according to (2) and (3)). Therefore the
characteristic attitude between group members (according to (1)) includes connected-
ness in the sense of (T2) as a substantial constituent:
(5) Group members perceive other group members as being connected to them by
sharing the perspective and the goals and values of the group.
This perspective entails that individuals are perceived as participants in a common
enterprise rather than as isolated agents. Consequently, they will treat each other as
responsible agents as specified in (T1):
(6) The attitude of an individual participant in collective agency to other individual
participants is characterized by a participant stance.
Moreover the values and goals that are perceived as connecting the participants in
collective agency form a solid ground for corresponding (factual and normative)
expectations which may motivate mutual reliance on conformity. Thus, genuine trust
in the sense specified above is a necessary element and essential constituent of
collective agency.
Our discussion of the role of assurance for team reasoning as the preferred mode of
reasoning reveals that trust is a constitutive element of collective agency. The charac-
teristic trust that we find in collective agency is not independent of the fact that there is
collective agency. So it may not serve as its proper cause. The relationship between
collective agency and trust is logical rather than causal. Trusting others is a necessary
condition of collective agency in the strict logical sense. It is not a (causal) precondi-
tion in the sense that—as Sugden seems to believe—a collective agent can be formed
only if the characteristic trust is pre-existent between the potential participants. Trust
is inherent in collective agency, so forming a collective agent means establishing a
form of trust between trusting fellow participants. Whatever explains collective agency
also explains that there is trust between the individuals of the collective.
Some readers may be concerned about these far-reaching conclusions from a dis-
cussion of such a simple stylized situation as defined by the Hi-Lo game. It may, for
example, be objected that one should speak of ‘genuine trust’ only if an individual is
actually motivated to engage in trusting behaviour. But it seems inadequate to identify
a coordinative act in a simple coordination game as Hi-Lo as trusting behaviour. By
deciding to contribute to a certain coordination one incurs a risk of being harmed if
the other fails to decide on the same coordination. But one cannot evade this risk in a
coordination game. The risk of a coordination failure is present whatever an actor
decides. So she is vulnerable whatever her actions are; she does not make herself vul-
nerable. But even if it seems inappropriate to ascribe trusting behaviour to her it may
still be adequate to ascribe trust to her in the light of the fact that her attitude complies
with the defining characteristics of trust; it would in fact motivate her to engage in
trusting behaviour should the occasion arise. Moreover, real cases of collective agency
will, as a rule, be much more complex than the simple situation represented by our toy
142 Bernd Lahno
game Hi-Lo. They will typically involve the necessity for agents to make themselves
vulnerable to the actions of other participants. In the next section I will investigate
some of the more salient elements in more realistic occurrences of collective agency in
order to explore the demands on trust that they constitute.
6. Trust in Integrity
Coordination is hardly ever a singular isolated task as in Hi-Lo. It is usually an ongoing
challenge in a collective project that extends over time. After Adam and Berta managed
to pass the rock on its right side, a rapid may suddenly appear or some wood may block
the waterway. The canoeing trip will form a row of coordinative tasks that will, more or
less, be connected to each other. In most cases there will be various ways to solve the
problem and each collective solution to the problem (pass right; pass left; slow down,
turn around and float backwards, etc.) may be realized by different combinations of
individual acts. Many collective decisions will have to be made in a short time, and in
contrast to Hi-Lo a unique optimal route to coordination may not be salient or actually
exist in all cases. Teamers may develop routines and use external coordination devices
to cope with these problems. An experienced couple of paddlers will be able to ‘read’ the
movements of the waters and their boat. They will also be attentive to the reactions of
their partners and able to interpret them correctly. Routines may guide them in reacting
to the information given in these ways. As these routines are initiated by mutually
accessible events, they produce a synchronized concert of action.
A prominent external coordination device is the music that guides a dance. Music
provides the publicly available beat to synchronize the trained routines of the individ-
ual dancers in a dance formation. But, of course, it cannot do the entire coordinative
job. A dance couple will have to choose suitable routines during the evolving dance.
Sudden discontinuities will occur, caused by other dancing couples intersecting
with the evolving figure of the dancers, by the dance floor’s unevennesses, or simply
by occasional small stumbles of one partner. The couple will need to coordinate
spontaneously in their reactions to these kinds of occurrences. Sometimes the
collective may be able to reduce the coordinative effort by making more or less
detailed plans for the collective project. A professional dancing couple may agree on a
perfectly worked out and fine-grained choreography for their dance to minimize the
number of choices to be made in performing the dance. But no choreography can be
detailed enough to account for the entire fine-tuning necessary to produce a consum-
mate performance. Nor can any choreography provide for all the unexpected inter-
ferences that may predictably occur.
Habitual routines and external coordination mechanisms will often not suffice in
completion of a complex coordinative task. In the course of a collective action, situ-
ations will arise that require instant coordination without any indication of a salient
pattern on which to coordinate. There is a vast and unmanageable manifold in such

situations, only a part of which can be accounted for by routines. A dancing couple will
cope with these problems by exchanging small signals by faint bodily movements or
slight measures of guidance. These mechanisms are generalizable. To cope with sud-
den demands of instant coordination the partners in a collective project need simple
communicative means to establish a common understanding of the challenge and the
suitable way to master it. Moreover, if there are several equally convenient ways of
coordinating and no predetermined way to choose among them, it will be disadvanta-
geous and probably impossible in the time available to find a mutual agreement by
communication. The standard way to solve such a problem is to endow one of the
members of the group with the authority to guide the others.
Both instruments of coordination, hierarchy as well as communication, are them-
selves based on convention. Mastering complex coordinative tasks in this way thus
presupposes the more general and fundamental coordination necessary to establish a
suitable social structure of the collective agent. Individuals need to coordinate if they
act as part of a collective agent. By doing so they constitute the collective agent. But as
a precondition there is often a general agreement on roles and rules defined for
the members of the collective agent. This is itself a matter of convention and thus a
coordinative task.
Another problem of collective agency that is not reflected in our introductory
example and the discussion so far is the possibility of conflict among the members of
the collective. We assumed that the two paddlers enter the collective activity with
exactly coinciding goals and interests. It seemed most plausible, then, that individual
and collective preference orders just coincide and can, therefore, be represented by the
same utility function. As soon as some conflict between individual and collective
interests is perceived to be possible the situation changes fundamentally. Each partici-
pant in the collective activity must face, then, the possibility that other participants
may not make their decisions in view of those values and goals that the individual per-
ceives as the guiding standards for the group.
One paddler may be interested in a calm journey along the river enjoying the scen-
ery while the other seeks the thrill of white water. But, of course, the best route to cir-
cumvent a suddenly occurring barrier in the waterway may depend heavily on whether
the point of the project is thrill and adventure or tranquility and calm recreation.
Forming a collective agent presupposes that some tacit or explicit agreement is found
on the collective goals and values to guide the collective project. But, whatever the
agreement may be, in most cases some uncertainty will remain. When forming a col-
lective agent individuals will not give up their individuality in total. Participants may
well deviate from the collective scheme when individual interests are heavily threat-
ened. Moreover, even if each individual feels categorically committed to the collective
project, the possibility still remains that there is some unnoticed disagreement about
what the terms of the agreement that founds the project actually say.
144 Bernd Lahno
What is the upshot of all this for the demands of collective agency on participating
individuals and the role of trust?
(1) Collective agency is often based on collectively shared routines. Trust is encap-
sulated in these routines.
That I conceive of myself as a member of the collective means not only that I take on
the collective goals as my own, it also means that I am guided and motivated by the
routines of the group. The formation of such collective routines presupposes some
background atmosphere of trust and the routines then become part of the common
frame that forms the group’s outlook onto the world. That I perceive others as being
guided in the same way establishes a necessary background belief of my participating
in the shared routine. But my action is immediately guided by the collective routine,
not by some deliberation about the best response to what I believe others will do. Trust
is thus encapsulated in the group routines and trusting behaviour consists in auto-
matic action (di Nucci 2011).
(2) Collective projects over time require some institutional background. This insti-
tutional background provides a normative frame of trust.
Identifying with the group includes familiarity and a commitment to the relevant
social roles and institutional rules of the group. Perceiving others as members of the
group is trusting that they share these commitments. It constitutes a common under-
standing of how things are done in the group and indicates agreement on the distribu-
tion of rights and responsibilities among the members of the group.
(3) Collective agency is threatened by potential conflict of interest. Trust that such
conflict will not affect the common project is embedded in perceiving each other as
members of the group.
A typical collective project will involve some uncertainty about individual and collect-
ive values or goals and, thus, potential conflict. The problem might be alleviated by
communication and hierarchy to some extent. But the individual members of the
group remain vulnerable in principle. Whoever enters a collective project can never
totally exclude the possibility that others may not promote the collective good in what
she thinks is the required way. She could avoid this risk of being harmed by not enter-
ing the collective project in the first place. But if she really identifies with the group and
if she perceives others as doing the same, she will not worry about this risk.
Identifying with the group comprises more than just taking on the group’s goals as
one’s own. It means sharing a commitment to a whole web of norms and values. These
values and norms not only define the ends of the group, but also provide means to find
agreement on proximate goals and channel the way in which the collective project is to
be realized through individual contributions. Perceiving others as identifying with the
group, therefore, means perceiving them as sharing the defining normative framework
of the group. It is trust in integrity.
7. Being One with Another

That a collective project evolves over time and comprises a series of collective actions
each of which may presuppose the solution to some coordinative or cooperative prob-
lem on the individual level may have even more dramatic consequences for the
demands on mutual understanding and trust. Not only may new options and chal-
lenges emerge which are not taken care of by known routines, but from time to time
challenges may also arise that were not anticipated at all and for that, as a consequence,
no collective preference order is clearly defined.
Think of the two paddlers and assume that they were planning and expecting to go
out for a quiet tour along a calm creek. One of them is slightly more experienced
than the other but neither considers the fact that the small creek might turn into
rather wild waters during the thawing period in spring; and it is spring! So after pass-
ing a bend they suddenly realize what they did not expect: this is going to turn into
a—probably—rather dangerous adventure. The first paddler might feel excited and
look forward to the thrill while the second grows stiff in horror. The group is totally
unprepared. Each individual may have her personal feelings and inclinations. But
each is insecure as to how the partner feels and, what is more important, no collective
evaluation of the situation is instantly available. The collective agent is in danger of
forfeiting its existence because it misses an essential element of agency: a clearly
defined preference order. But the individuals cannot simply resolve the collective
entity. The situation forces them to stick together. If each acts independently catas-
trophe seems unavoidable.
Obviously, in such a situation the scope of the collective’s coordinative background
must be broadened. The signalling conventions by which the partners communicate
must not only be used to signal individual intentions, to affirm mutually collective
consent and to find an agreement on the collective scheme of action; they must first of
all serve to find an agreement on the collective goals before the scheme may be deter-
mined. The routines that normally guide individual action in forming the collective
act may now serve for giving decisive clues as to what goals could in principle be
accepted and efficiently pursued by the collective. Hierarchies may be used not only to
determine choice but also to select collective goals. Additionally, the guiding values
and norms, although some of them are themselves possibly the object of gradual
change, will serve as an elastic normative frame in which agreement on goals may
be formed.
There is hardly any constraint on the evolution of the collective agent in a collective
practice. The process of forming collective goals is not restricted to proximate goals. It
may also extend to ultimate goals, which change over time. Imagine the paddlers are
involved in many new experiences while paddling unexpectedly down the white
waters. They have a lot of practice and although they would never have dared to start a
white-water tour, this is just what they now want. They watch out for the torrent to miss
no wave to dive in and exploit its force for an exciting ride. A collective agent is not just
146 Bernd Lahno
a temporal unit created by individuals to carry out a singular collective act or project.
It may well be an enduring entity with evolving properties, faculties, goals, and values.
A collective agent may have a biography much like an individual. To constitute such a
collective agent its members must be equipped with appropriate communication and
coordination devices together with suitable capabilities to adjust to each other.
Forming a collective agent in such cases is a complex process over time in which not
only the new collective entity arises. The individuals themselves also undergo a funda-
mental change. Taking on the goals of the collective does not just mean acting as if
those goals were guiding individual choice; adjusting individual goals to each other is
not like finding a lowest common denominator, a compromise as a pragmatic platform
of cooperation. If individuals unite to form a collective agent they concurrently form
each other. Forming a collective agent then means being formed and probably turning
into a new person.
This may, in fact, have dramatic consequences for the form of trust involved in col-
lective agency. In an ordinary case of trust I rely on another person because I perceive
her as a person who will be guided by the values and norms that I endorse. I will not be
harmed because you are going to do the right thing. But when I am part of an evolving
collective agent my values and norms may sometimes be formed by how the collective
behaves. You may not be able to harm me then because, whatever you do, I will accept
it as something well done. If what is right is to a great extent determined by what you
do, my trust may turn into bare subordination. We know this radical form of uncondi-
tional, ‘categorical’ trust from the ‘basic trust’14 of infants or from trust in god. So we
know that it may have a beneficial function and, in fact, contribute to our flourishing as
an individual. But it is obvious that it may well be destructive and most dangerous. In
the occasional tendency of collective agency towards categorical trust we face the
glorious appeal as well as the horrifying grimace of Leviathan.
8. Conclusion
Trust is a constitutive element of collective agency. It is embedded in collectively
shared routines and part of the way individuals perceive each other as members of the
group. The values and goals of the group in conjunction with the system of roles and
rules that constitute the social structure of the group provide a normative foundation
of interpersonal trust. So, the trust we find is genuine trust in the sense specified above.
It is a complex attitude towards other participants in the collective project that cannot
be reduced to a pure cognitive belief about the probability of conforming behaviour.
Choice in collective action is rarely mediated by individual deliberation. It is the
immediate result of what the participants in the collective project perceive as the ‘right’
and ‘their’ way of doing things. And the trust embedded in their mutually coordinated
behaviour has the same kind of immediacy. A teamer trusts her fellow teamers but she
14
Erikson (1950); see Lahno (2003) for an analysis of trust in god and the concept of categorical trust.
may never have to think about trust. As we saw, this form of trust may have the most
remarkable and, in fact, dramatic consequences in defining the person that we are. But
on the surface it often appears as inconspicuous, ordinary, and even invisible. Maybe
this is one of the reasons why it has remained also largely unnoticed in the extensive
debate on collective agency.
References
Bacharach, Michael (1999) Interactive Team Reasoning: A Contribution to the Theory of
Cooperation. Research in Economics 53: 117–47.
Bacharach, Michael (2006) Beyond Individual Choice: Teams and Frames in Game Theory. Ed.
Natalie Gold and Robert Sugden. Princeton, NJ and Oxford: Princeton University Press.
Bratman, Michael E. (1992) Shared Cooperative Activity. The Philosophical Review 102(2):
327–41.
Coleman, James S. (1990) Foundations of Social Theory. Cambridge, MA: The Belknap Press of
Harvard University Press.
Cubitt, Robin and Robert Sugden (2003) Common Knowledge, Salience and Convention.
Economics and Philosophy 19: 175–2010.
Dasgupta, Partha (1988) Trust as a Commodity. In D. Gambetta (ed.). Trust: Making and
Breaking Cooperative Relations. Oxford and New York: Basil Blackwell.
di Nucci, Ezio (2011) Automatic Actions: Challenging Causalism. Rationality Markets and
Morals (RMM) 2: 179–200. Available from: http://www.rmm-journal.de/downloads/Article_
Di_Nucci.pdf [accessed 12 December 2015].
Erikson, E. H. (1950) Childhood and Society. New York: Norton.
Gambetta, Diego (1988) Can We Trust Trust. In D. Gambetta (ed.). Trust: Making and Breaking
Cooperative Relations. Oxford and New York: Basil Blackwell.
Gilbert, Margaret (1997) What Is It for Us to Intend? In R. Tuomela and G. Hoömstrom-
Hintikka (eds.). Contemporary Action Theory. Vol. 2: Social Action. Dordrecht: Kluwer.
Gilbert, Margaret (1989) Rationality and Salience. Philosophical Studies 57: 61–77.
Gilbert, Margaret (1990) Rationality, Coordination, and Convention. Synthese 84: 1–21.
Hardin, Russell (1991) Trusting Persons, Trusting Institutions. In R. Zeckhauser (ed.). Strategy
and Choice. Cambridge, MA: MIT Press.
Hindriks, Frank (2012) Team Reasoning and Group Identification. Rationality and Society
24(2): 198–220.
Hodgson, D. H. (1967) Consequences of Utilitarianism. Oxford: Clarendon Press.
Holton, Richard (1994) Deciding to Trust, Coming to Believe. Australasian Journal of
Lahno, Bernd (2001a) On the Emotional Character of Trust. Ethical Theory and Moral Practice
4: 171–89. Reprinted in: Elias L. Khalil (ed.) (2003) Trust: Critical Studies in Economic
Institutions. Vol. 3. Cheltenham and Northampton, MA: Edward Elgar.
Lahno, Bernd (2001b) Institutional Trust: A Less Demanding Form of Trust? Revista
Latinoamericana de Estudios Avanzados (RELEA) 15: 19–58.
Lahno, Bernd (2002) Der Begriff des Vertrauens. Paderborn: Mentis.
Lahno, Bernd (2003) Gottvertrauen. Analyse & Kritik 25: 1–16.
148 Bernd Lahno
Lahno, Bernd (2007) Rational Choice and Rule Following Behavior. Rationality and Society 19:
425–50.
Lahno, Amrei and Bernd Lahno (2014) Team Reasoning as a Guide to Coordination. MPRA
Paper No. 55670. Available from: http://mpra.ub.uni-muenchen.de/55670/ [accessed 11
December 2015].
Schmid, Hans Bernhard (2013) Trying to Act Together. In: Michael Schmitz, Beatrice Kobow,
and Hans Bernhard Schmid (eds.). The Background of Social Reality. Dordrecht: Springer.
Schmid, Hans Bernhard and David P. Schweikard (2013) Collective Intentionality. In: Edward
E. Zalta et al. (eds.). Stanford Encyclopedia of Philosophy. Summer 2013 Edition. Stanford,
CA: Stanford University Press. Available from: http://plato.stanford.edu/archives/sum2013/
entries/collective-intentionality/ [accessed 21 November 2014].
Searle, John (1990) Collective Intentions and Action. In: P. Cohen, J. Morgan, and M. E. Pollack
(eds.). Intentions in Communication. Cambridge, MA: MIT Press.
Sellars, Wilfried (1980) On Reasoning about Values. American Philosophical Quarterly 17:
81–101.
Strawson, P. F. (1974) Freedom and Resentment. London: Methuen.
Sugden, Robert (1991) Rational Choice: A Survey of Contributions from Economics and
Philosophy. The Economic Journal 101: 751–85.
Sugden, Robert (1993) Thinking as a Team: Towards an Explanation of Nonselfish Behavior.
Social Philosophy & Policy 19(1): 69–89.
Sugden, Robert (2000) Team Preferences. Economics and Philosophy 16(2): 175–204.
Sugden, Robert (2003) The Logic of Team Reasoning. Philosophical Explorations 6(3): 165–81.
Sugden, Robert (2015) Team Reasoning and Intentional Cooperation for Mutual Benefit.
Journal of Social Ontology 1(1): 143–66.
Sugden, Robert and Natalie Gold (2007) Collective Intentions and Team Agency. Journal of
Philosophy 104(3): 109–37.
Tuomela, Maj (2006) Rational Social Normative Trust as Rational Genuine Trust. In: L. V. Siegal
(ed.). Philosophy and Ethics: New Research. New York: Nova Science Publishers.
Tuomela, Raimo and Kaarlo Miller (1988) We-Intentions. Philosophical Studies 53(3): 367–89.
Tuomela, Raimo and Maj Tuomela (2005) Cooperation and Trust in Group Context. Mind &
Society 4: 49–84.
9
Trust as a Two-Place Relation
Jacopo Domenicucci and Richard Holton
How should we think of trust? It has become normal, at least in Anglophone analytic
philosophy, to think of it as fundamentally a three-place relation with an infinitival
component: A trusts B to do C.1 Here we aim to question that idea. We don’t deny that
this three-place relation exists, expressed in a perfectly natural English idiom, with a
useful role to play. Rather, we explore the idea that, in giving an account of trust, this
three-place relation provides us with the wrong place to start, and that we should start
instead with the two-place relation, A trusts B, and work from there.2 Likewise, we
suggest, for three-place relations that put something else in the third place: A trusts B
with C; A trusts B in the role of C. These too we deny are fundamental.
What do we mean by the ‘wrong place to start’? A few parallels—more or less apt—
might help fix ideas. Some think that we should understand knowledge as built out of
belief and truth and something else. Others object that this is the wrong place to start,
that with those ingredients we will never arrive at an account of knowledge. Some
think that we should understand enduring persons by starting with time-slices and
joining them together with relations of psychological connectedness or the like. Others
object that this is the wrong place to start; with such ingredients we will never arrive at
an account of persons.
We suggest that the same is true of trust. If we start with a three-place relation, and
try to understand the two-place in terms of it, we will not succeed. There are other
concepts where no one would deny such a claim. No one—or at least, hardly anyone—
thinks that we should understand what it is for Antony to love Cleopatra in terms of
1
There are many instances. In the earliest discussion we know of, Horsburgh (1960) distinguishes two-
and three-place trust, but goes on to prefer the former. For an early endorsement of the primacy of three-place
trust, see Hardin (1992). For one that has been influential on at least one of the current authors, see Holton
(1994). In his defence, though, Holton does in that piece talk about the importance of ‘trusting relationships’.
2
Baier, although explicitly working with a three-place account, rather than a two-place, worried that
this might be ‘forced and wrong’: ‘For there are some people whom one would not trust with anything, and
that is not because one has considered each good one might entrust to that one and rejected that possibility.
We want then to say that unless we first trust them we will not trust them with anything. I think that there
is some truth in this, which my account has not captured’ (Baier 1986: 258–9). For other considerations in
favour of a two-place account, see Faulkner (2016: §6).
150 Jacopo Domenicucci and Richard Holton
the three-place relation ‘Antony loves Cleopatra for her__’, or in terms of any other
three-place relation. Likewise, hardly anyone thinks that we should understand the
two-place relation of friendship in terms of some underlying three-place relation (here
we don’t even have any natural English expressions for the three-place). To this extent
at least, we suggest that trust might be like love and friendship.
Rather than providing support, these last two comparisons might be taken to show
what is wrong with the two-place account of trust. For unlike love and friendship, trust
can be partial: we can trust people in some ways and not others. In time we will need to
accommodate this. But not yet. We start, instead, with some positive reasons for
embracing the two-place account.
1. Arguments for the Two-Place Account

To warm up, we present three linguistic considerations.
(1) Other languages. In English, as we have accepted, the three-place construction
with an infinitival third component is completely natural. But in core Romance
languages—Latin, Italian, French—it is not readily available. In French ‘J’ai confiance
en toi pour X’ is colloquial, but not in the dictionaries; in Italian ‘Ho fiducia in te per X’
is simply unacceptable, as is the equivalent in Latin ‘Fidem habeo alicui ut X’. It is of
course possible that the fundamental trust relation cannot be expressed directly in
these languages, and so has to be expressed in roundabout terms (typically in terms of
counting on someone to do something, although that suggests something closer to
mere reliance). It is also possible that the English notion of trust cannot be translated
into these languages. But we suggest that a more plausible claim is that the Anglophone
accounts of trust have been unduly influenced by a construction that is available in
English, but that even there should not be seen as central.
(2) Distrust. Katherine Hawley (2014) has usefully pointed out that most accounts
of trust have made no mention of distrust. We follow her in thinking that distrust is a
contrary, not a contradictory, of trust: to distrust someone is not simply to fail to trust
them. Moreover, it is not merely a dispositional state. If we distrust someone, it is not
just that we would react in some particular way if certain circumstances arose; rather,
we are, in some way that needs elucidation, thinking badly of them. As the Oxford
English Dictionary (OED) says of the related notion of mistrust, it is ‘to suspect [their]
actions, intentions, or motives’. Strikingly though, even in English there is no three-
place syntactic construction of distrust. We do not say that we distrust someone to
do something. We simply distrust, or mistrust, a person.3 But if distrust is in some
important way a contrary of trust, and so inherits the basic form of trust, that suggests
that trust itself is primarily an attitude to a person.
3
There is an obsolete usage in which ‘mistrust’ does take a three-place construction; but here it means
something very different, something along the lines of ‘to suspect’: ‘He said he was taking his Way to
Boston, but is mistrusted to be going to Long or Rhode Island’ (OED).
trust as a two-place relation 151
(3) No sense of incompleteness. Many relations are genuinely, fundamentally, three-

place: consider ‘to give’ and ‘to tell’. Even if, as in the latter case, the expression
syntactically accepts a two-place construction—‘Jules told Jim’—this leaves a question
in the air: ‘What did Jules tell him?’ Or consider reliance, a state that has been widely
taken to have strong affinities to trust. If you simply say that you rely on some individ-
ual, then the immediate question arises: rely on them in what way, for what? Trust is
different. If Jules tells us that he trusts Jim, no parallel question is pressing. In fact, if we
do ask it—‘You trust him to do what?’—we raise a question mark about the trust itself;
by suggesting that it is partial, we suggest that it is not the full thing. (Use of the present
continuous—‘I am trusting him’—is rather different in most dialects of English, in that
it does suggest incompleteness. We return to this below.)
So, we have three broadly linguistic considerations that point towards a two-place
account of trust. But even taken together, they are little more than suggestive. To go
further we suggest that we think about the purpose of trust; about the role that it plays
in our lives.4
It has often been contended that trust involves a form of vulnerability on the part of the
person trusting; and often this is developed as a kind of ignorance. In some accounts this is
understood as ignorance of whether the trust is well placed: we trust insofar as we cannot
be sure that the person trusted will not let us down. That, however, is implausible: on such
an account, as our knowledge of a person grows, so our trust in them must diminish. We
suggest that the idea of vulnerability should be developed in a rather different direction. In
our view, trust centrally involves a preparedness to grant a certain power or control. In
trusting, we grant discretion, whether to act, or to judge, or even just to feel. Granting such
discretion is not all there is to trust; indeed it may not even be necessary. But it is central.
If this is right, then there is a good reason why trust cannot in general be understood
as a three-place relation, with the third place taken by an action. For very often one
grants discretion exactly because one doesn’t know what action should be taken: either
one doesn’t know how things will turn out, or even if one does, one lacks the expertise
to know how to respond. That’s the point about complete trust: the kind of trust we
might have in a parent, or a partner, or a child who is old enough. It is not that we envis-
age a particular action that we trust them to perform. We trust them simpliciter. Could
we be said to trust them to act in our best interests? Not necessarily, since there may be
cases in which they rightly judge that our interests aren’t paramount. Could we at least
be said to trust them to do whatever is best? Even that is too specific. Perhaps they will
simply do what is good enough (we don’t have to think that they are angels) and that
will be good enough for our trust to have been respected. Perhaps there are things that
we would not trust them to do, even though we do trust them. Most likely, especially if
the relationship is a healthy one, we haven’t even thought these things too much. The
fundamental fact is that we trust them. Exactly how that would translate into particu-
lar actions is something on which we need have no view.
4
For a presentation of that general approach, see Simpson (2012).
(The analogy is not perfect, but there is again a parallel here with love. Insecure
l overs sometimes play the ‘Would you love me if . . . ?’ game: Would you love me if I lost
my beauty? Would you love me if I lost my money? Would you love me if I became a
fascist? Notoriously, the revelatory returns are poor. Since love is not built out of affec-
tion for particular features, little is shown by asking whether it would survive the loss
of those features. Trust, we are suggesting, is much the same.)
It might still be insisted that the two-place relation is really some form of generaliza-
tion of the three-place: ‘I trust you’ is shorthand for ‘I trust you to F, for some class of F’.
Perhaps the speaker has not fully worked out the range of the quantification, but if
pressed they could say something.
We doubt that this is right. As evidence against it, consider again how the logic of
trust differs from that of a true three-place relation: again we take reliance as our foil.
Suppose that Jules asks Jim if he might borrow his car. Jules hesitates for a moment (it is
a rather nice car), and then replies:
(1) Ok, I trust you. You can borrow it.
That is a perfectly normal thing to say. In contrast, if Jules had said:
(2) Ok, I rely on you. You can borrow it,
we would conclude that he probably wasn’t a native speaker of English. Why the
difference? It’s not that we couldn’t put in three-place constructions here. Both of
the following are acceptable:
(3) Ok, I trust you to take good care of it. You can borrow it.
(4) Ok, I rely on you to take good care of it. You can borrow it.
Since we can’t start with (4) and convert it to (2) by simply eliding the third place and
implicitly generalizing over it, why think that that is what is happening in the move
from (3) to (1)? Instead (1) seems to work in a radically different way. Talk of trust is
used to explain or justify lending the car. It’s as though Jules had said ‘I trust you, there-
fore you can borrow it’.
One further observation lends force to this interpretation. The three-place uses with
‘rely’, and equally those with ‘count on’, are more natural if their aspect is changed to the
present continuous:
(5) Ok, I’m relying on you to take good care of it. You can borrow it.
(6) Ok, I’m counting on you to take good care of it. You can borrow it.
The implication is that the state of reliance is something that will accompany the bor-
rowing of the car. Unlike the state of trust that is referred to in (1), it is not something
that is antecedent to the borrowing, and thereby able to justify it. We can use a parallel
present continuous construction with ‘trust’:
(7) Ok, I’m trusting you to take good care of it. You can borrow it.
This does seem to be an essentially three-place construction. The two-place ‘I’m trust-
ing you’, unlike ‘I trust you’, does cry out for completion: ‘Trusting me to do what?’ is
the natural response. How then should we think of the three-place trust relation exhib-
ited in sentences like (3) and (7)?
Three-place trust surely bears some connection to the corresponding three-place
reliance: if you trust someone to do something you rely on them to do it. So a first
thought might be that you trust A to F iff you trust A, and rely on them to F. That does
ground three-place trust in two-place, but it doesn’t seem right. Trust looks to be par-
tial, in that one can trust an individual, but not trust them in all areas; we will discuss
how this is possible in a two-place account shortly. If that is right, then one might rely
on someone in the areas in which one does not trust them; but then three-place trust
cannot be simply trust plus reliance. More plausible is the idea that three-place
trust arises when the reliance is embedded within the trust: you rely on A as part of
the trust that you show them. Again, the two-place relation is fundamental, and the
three-place is constructed out of it. Seen in this way, it is not surprising that some
languages (such as English) have a distinct locution for the particular complex
phenomenon of three-place trust, while others (such as Italian) do not.
2. Trust, Reliance, and Reactive Attitudes

Let us think more broadly about the relation between trust and reliance. This has been
the subject of much philosophical scrutiny. Many have attempted to build trust con-
junctively out of reliance: if one trusts, then one relies, and some other condition
obtains. But if we are right that reliance is at heart a three-place relation, whilst trust is
at heart two-place, they cannot fit together in quite this simple way. The point comes
out when we think of the role of the reactive attitudes: the role of attitudes such as grati-
tude and betrayal in an account of trust. We agree with those who have argued that
such attitudes do mark a significant distinction between trust and mere reliance.5
Whilst we might rely on a machine to perform some task, we do not feel anything
approaching gratitude when it does, nor do we feel betrayed when it does not.6 A con-
junctive account might try to account for this by saying that to trust someone is to rely
on them to do something, and then to invest this reliance with the reactive attitude—a
readiness to feel grateful or betrayed depending on how things go. That, however, is to
read trust as implicitly three-place.
In contrast, on the two-place account that we are exploring, when one trusts, one
primarily takes a reactive attitude towards the person. As a result there will typically be
instances of three-place reliance and of three-place trust, which will indeed result in a
5
For the notion of reactive attitudes, see Strawson (1962); for its application to trust, Holton (1994).
6
This is also coherent with the neuroscientific evidence from Kosfeld et al. (2005) that oxytocin, the
neuropeptide that seems correlated with trust, is released when a subject is interacting with another human
being, not with a machine.
sense of gratefulness or betrayal depending on how things transpire. But these do not
exhaust the characteristics of trust, nor are they necessary for it. There may be instances
of trust even when there is no opportunity for reliance: we do not automatically stop
trusting someone when they lose the power to act in the ways in which we relied on
them. Taking the reactive attitude that is characteristic of trust can involve a host of fur-
ther attitudes: affective responses even in the absence of reliance; a readiness to form
certain beliefs and to desist from forming others; a readiness to testify to others; an
abandoning of certain lines of enquiry; and so on. We doubt that any of these is strictly
necessary for trust. If enough of the others are in place we may have a case of trust, even
if it is not as full as it could be. Trust is, to use an old-fashioned term, a cluster concept. 7
As as illustration, consider the place of belief. How central to trust is belief in
performance? Pamela Hieronymi has argued that it is essential, at least to ‘full-fledged’
trust.8 Jules and Jim agree to meet at a restaurant. Although Jules duly arrives a
little early at the appointed place, he lacks the belief—quite unreasonably we may
suppose—that Jim will turn up. Jim appears right on time. Suppose he were to discover
Jules’ lack of belief. Then he could, says Hieronymi, ‘rightly complain that [Jules’] lack
of confidence betrays a lack of trust’. There is clearly something lacking in Jules’ atti-
tude; his trust could have been greater. But is it true, as Hieronymi concludes, that he
didn’t really trust? Change the example a little. Suppose that Jules’ scepticism, although
still quite unfounded, turns out in fact to be corroborated: Jim never shows up.
Remonstrating later, Jules complains ‘I trusted you’. Could Jim rightly respond ‘No you
didn’t; you didn’t believe that I would come’? We think not.
In the original example Jim has a legitimate complaint that he is not believed, a com-
plaint that underpins the reasonable charge that he is not trusted as fully as he might
have been. Once wronged, it’s understandable that his complaint verges on the hyper-
bolic. But we should not be dazzled by the legitimacy of that way of voicing the com-
plaint into thinking that belief is really a necessary condition for trust; change the
aggrieved party and the intuitions change. Hieronymi draws a sharp distinction
between trusting someone to do something, which requires belief, and entrusting
them, which doesn’t. We doubt that there is such a clear binary division, but suggest
instead a continuum with many factors.9
We say much the same about distrust. Here again we have a reactive attitude, and
one with multiple parts. Again there is a sense of betrayal when any (perhaps
unavoidable) reliance is disappointed, although here it is coloured with grim expect-
ation rather than with surprise. There may be gratitude when reliance is upheld, per-
haps with a sense of guilt at having misjudged the person, or of cynicism if the initial
7
See Jones (2004) for some similar ideas. We agree with Jones that the account in Holton (1994) was too
focused on betrayal.
8
Hieronymi (2008).
9
Again the neuroscientific evidence, such as it is, suggests that belief is not terribly central. A higher
level of oxytocin, which seems correlated with a higher readiness to trust, appears not to increase the sub-
ject’s beliefs that the risks will be rewarded.
judgement is maintained. More commonly there will be a set of actions and attitudes
connected with avoiding any possible reliance, persuading others to do likewise, and
perhaps plotting revenge. And again, whilst there may be a belief that the person dis-
trusted will act, or fail to act, in certain ways, this is not essential. These consider-
ations reinforce the syntactic evidence that we mentioned earlier: trust is primarily an
attitude to a person.
3. The Extent of Trust

We have sketched a positive case for the idea that trust is centrally a two-place relation.
We now address the obvious objection.
A two-place account need not imply that trust is all or nothing: if we are right that it
is a cluster concept, trust will come in degrees. It might or might not involve certain
reactive attitudes; it might or might not involve belief, and so on, and the degree of
trust will change as these factors change. But the two-place account does seem to imply
that, to whatever degree one trusts someone, the attitude will be uniform with respect
to whatever they do. Yet that goes against the commonplace observation that our trust
can be different in different spheres: that someone might trust their partner, except
where alcohol is concerned; or that they might trust their plumber to fix the hot water
system, whilst not trusting them with their bank details, or their car, or their children.
Not every idiomatic use of the term ‘trust’ refers to a relationship of trust: no one
thinks that an account should capture, at least in a literal way, the dismissive ‘Trust
him!’ that follows a not unexpected lapse. But these two examples do seem to be genu-
ine cases of trust; they involve conceding discretion within the context of the kinds of
reactive attitudes that we have already discussed. So how is the two-place approach to
account for them?
The first thing to say is that the two cases we have given are rather different. The first
involves a general attitude of two-place trust, with a qualification: the trust gives out
when alcohol is involved. Such qualified trust is straightforwardly handled on the two-
place approach. The two-place attitude is still fundamental. One arrives at the qualified
attitude by starting with it and then knocking something off. Other cases follow a simi-
lar pattern, motivated by a welter of different considerations about competence or
motivation: ‘I don’t trust her when it comes to dogs’; ‘I don’t trust him once his relatives
are involved’ and so on.
The case of the plumber cannot be thought of in this way though. It is not as if one’s
attitude to one’s plumber is like one’s attitude to one’s partner, except qualified in vari-
ous ways (unless, of course, the plumber is one of the family). We don’t start with lots of
trust and then reduce; quite the reverse: we start with rather little, and then, perhaps, if
things go well, we add more.
Nevertheless, we suggest that the two-place model is more revelatory, even for the
case of the plumber. We still trust them, first and foremost, as a person; the variation
comes from the fact that, given a typical relationship with a plumber, what is required
to trust them as a person is radically less than is required to trust one’s partner.
To see the advantage of this way of looking at things, note that it isn’t typically true
that we just trust the plumber to fix the hot water system. We trust them in a host of
other ways too, more or less closely related to their professional role. If, when they are
in the attic, they notice that some tiles have come adrift, we trust them to tell us. If the
cat falls into the header tank while they have the top off, we trust them to fish it out,
or at least to sound the alarm.
This might suggest that our attitude to the plumber is still properly thought of as
three-place though; it’s just that the third place, rather than being taken by an infinitival
clause, should be taken by a description of their role. We trust the plumber in their role
as plumber. But we don’t think that that is quite right either. While their role as plumber
may provide the core to our trusting relationship to them, it will not exhaust it. There
will be other features that stem simply from them being an adult human being whom
we let into our house; and others that stem from the particular relationship we form
once we get to know them (as source, and a recipient, of opinion or advice on other
topics; a source, and a recipient, of help in other areas). Could we spell all this out? The
best we can say is: we trust them in the appropriate way given our relationship to them.
In short then, the relationship of trust that we have to different people is indeed rela-
tive: to reiterate, the trust one bears to one’s partner is different to that one bears to
one’s plumber since what is appropriate for trust in those two relationships is different.
Trust is thus relative to the relationship that we have to them. But once we see it like
that, we have lost anything useful that can go in the third place of a three-place relation.
You trust X in the way appropriate for X. What goes in the third place is just what goes
in the second. Of course, this isn’t static. As we get to know them better, our relation-
ship to the plumber may change: more trust may become appropriate, or, if things have
gone badly, less (although losing trust tends to be catastrophic rather than gradual).
Again, an analogy with friendship may be helpful. Friendships come in degrees: some
are closer than others. But we don’t conclude from that that friendship is really a three-
place relation; it is a two-place relation of variable strength and depth.10
4. Some Implications
We conclude with a brief discussion of some implications of our proposal.
Thinking about the appropriate form of trust for a relationship helps explain why
trust can be unwelcome. We might, for instance, not welcome the level of disclosure—
of personal secrets say—with which we are trusted. For a parallel, consider the case of
gifts. Some gifts are unwelcome. Often this is simply because one doesn’t want the
thing: it is ugly, tasteless, useless, or whatever. Sometimes, though, it is because one
10
There are broader issues in the background here about whether relativity should generally be under-
stood in terms of adding another argument place. For discussion, see Spencer (forthcoming).
doesn’t want to be given it by this person. Had it arrived as the result of a mistake by a
mail order company it would have been very welcome indeed. But coming from this
person it signals, and tries to create, an unwelcome degree of intimacy.
We suggest that much the same happens with unwelcome trust. If a relationship of
trust comes in degrees, then the degree can be changed. If both parties are in harmony,
the change can be smooth. But if one wants more than the other—perhaps the other
thinks they are going too fast, or perhaps they simply do not want such a relationship
with this person—then the extended trust will be unwelcome.
At the lower extreme, how thin can two-place trust be? Can we have it with those
we have never previously met? That is an empirical question, but there is good data
on it. Since the 1950s sociologists have been asking people the question: ‘Generally
speaking, do you believe that most people can be trusted, or can’t you be too careful
in dealing with people?’ This is clearly two-place trust; ‘generalized trust’ as Eric
Uslaner has termed it, arguing that it is essentially moral.11 The findings here have
been robust. Levels of generalized trust differ across generations, and within these,
very slowly, across time; they are not very responsive to experience; they correlate
with optimism and are higher in societies with greater equality. The findings are too
consistent to be dismissed. Note though that the question asks people about their pre-
paredness to trust. ‘Can most people be trusted?’ is a complex question, bridging both
the descriptive—‘Are you prepared to trust most people?’—and the normative—‘Is
such an approach justified?’ (‘Can I trust him?’ is normally a way of asking about the
other’s trustworthiness rather than about one’s own trusting capacities.) The talk here
is not really about generalized trust, but about a justified preparedness to extend it in
a general way to new cases. Nevertheless it is clearly two-place trust that is at issue:
what is extended is a moralized attitude to a person.
A different issue concerns the possibility of trust towards institutions: towards gov-
ernments, nations, companies, banks, and the like. If trust is first and foremost a two-
place relation to a person, what are we to make of this? Certainly we find plenty of talk
about trusting institutions. ‘We now have the trust in Greece’s government which was
lost over the past months,’ says Angela Merkel. Perhaps here she is just talking about
reliance (‘The EU can rely on Greece to repay its loans’), but we suspect she means
more than that. The banks, in the wake of the 2008 financial crisis, have asked how they
can rebuild the trust that they have lost. They don’t simply want their customers to rely
on them; given the realities of the financial world, the customers have little alternative.
What the banks really want is that they be allowed to do what they want to do without
regulation; that is, they want the kind of discretion that we give to people when we
genuinely trust them. What they want is something very much like two-place trust.
Can we make sense of such an attitude to a bank?
11
Uslaner (2002). In contrast, Uslaner takes the three-place relation ‘X trusts Y to F’ to be a mark of
strategic trust: trust undertaken on the basis of a calculation of its utility. This may sometimes be how it
is used, but we don’t see why the three-place construction can’t also be used to pick out a particular appli-
cation of generalized moral trust.
The question of trusting institutions breaks into two parts. The first is empirical. Do
people in fact have the same sorts of trusting attitudes towards institutions that they have
towards people? The second is normative: is it appropriate or good for them to do so?
There is some research on the first of these: some evidence that trust in institutions
(most centrally the government) does not correlate very strongly with ‘generalized’
two-place trust towards people, and that it tends to be more instrumental. The lack of
correlation does not by itself show that it is not two-place trust; it could just be that the
people who trust individuals are different to those who trust institutions. In contrast
the finding that institutional trust is more instrumental does suggest that it is of a
rather different nature: offered only in return for benefit, removed if that benefit is not
forthcoming, perhaps less ‘moralized’ in Uslaner’s terms (less likely to involve reactive
attitudes and their ilk in ours). The role is something that may vary across different
societies and different periods. We trust the grocers not to sell us adulterated bread,
but do we trust them not to charge more than their competitors? Would we be indig-
nant if we discovered that they did, or would we think it our own fault for failing to
check? It seems plausible that in a market society we are more likely to take the latter
view than we would in a traditional society. To that extent we have moved away from a
‘moral economy’.12
What of the normative question? This in turn appears to divide into two. First there
is the question of whether it is normatively appropriate to treat institutions in such a
way. It strikes us that there is no obvious inappropriateness involved in extending two-
place trust to an institution. Institutions can be person-like in many ways: more so
than most machines. They can plausibly have plans, act fairly or unfairly, show callous-
ness or compassion. We can interact with them as though they were persons without
absurd anthropomorphization. If there is something odd about trusting them it more
plausibly comes from consideration about the need for reciprocity. Trust seems to lie
somewhere between love and friendship in this regard. Friendship can be unequal in
many ways, but it cannot be totally one-sided: to discover that no friendly attitudes are
returned, is to discover that what one thought was a friendship in fact was not. Love, in
contrast, can persist spectacularly in full knowledge that it is unrequited. What about
trust? There is certainly something good about reciprocal trust, and something
unstable about many cases in which it is one-sided. If institutions cannot trust us—if
they lack the ability to have the relevant reactive attitudes—is there something wrong
with trusting them? Perhaps so, although we should be cautious in moving too quickly
here; certainly many streams of Christianity have thought that our relationship to God
involves something like one-way trust.
Lacking a definitive answer to this first question, what of the second? Is there prag-
matic advantage in trusting institutions, rather than simply making calculations as to
whether to rely on them? We see two clear goods that come from trusting people: an
12
For that term, and much stimulating though controversial discussion, see Thompson (1981).
instrumental one, that trust engenders further trust and further cooperation; and an
intrinsic one, that trust between people is a good in itself. It is far from obvious that
either of these apply to institutions. For instance, it seems to us that, rather than afford-
ing trust to banks, we should be monitoring more closely exactly what they are doing.13
But here too we should be cautious about generalizing from institutions which in
many people’s eyes have become villains. Take instead an institution to which people’s
attitudes were very different: the British National Health Service for instance. At its
foundation the NHS was regarded with widespread idealism, a wonderful institution
that took medical care out of the realm of profit and offered it to everyone. The NHS
certainly inspired loyalty, from staff and patients. Did they trust it? If it is right to say
that of any institution it is right to say it here: they certainly gave discretion over health-
care to the NHS, relied on it, and invested it with a wide range of reactive attitudes.
Things do not look quite so rosy now. Newspapers can lament ‘the breakdown of
trust that is going on between our health service and us’.14 Some will argue that there is
no reason for lament: there should never have been trust in the first place, just a careful
reckoning of the benefits to be gained from state-provided medicine. But as many
commentators have pointed out, state provision may not be robust once many people
opt out. If, as a result of a more critical attitude, a number of citizens decide to pay for
private medicine rather than trusting the NHS to provide it, the political will to pay for
the NHS may decline sharply.
We can focus the point with reference to Richard Titmuss’ famous discussion of
blood donation (Titmuss 1970). Titmuss argued that the voluntary donation of blood in
the UK resulted in more and better quality blood being available to the NHS than was
available in the USA under a system in which some donors were voluntary and others
were paid. His argument was that UK donors were primarily motivated by altruism;
once a market system was in place, such altruistic motivation would be undermined,
leaving only those who were desperate for cash prepared to sell their blood.
The details of Titmuss’ account have come in for some significant criticism; it
certainly isn’t a watertight piece of social science.15 But there are some core ideas there
that have a great deal of plausibility and are worthy of more research. The first is
Titmuss’s explicit contention that altruistic donation will be undermined by a market.
But behind that is the thought that the practice of voluntary blood donation Titmuss
described was made against the backdrop of an institution that was well trusted.16 It may
be that such institutional trust is necessary for the practice to get going: that individual
13
For thoughts along these lines see Warren (1999). 14 Daily Mirror, 2 September 2014.
15
For criticism of the methodology, see Rapport and Maggs (2002); for some evidence that the motives
of donors may be more mixed than Titmuss thought, see Ferguson et al. (2008).
16
Titmuss (1970) was well aware of this: ‘We cannot understand the National Blood Transfusion
Service without also understanding the National Health Service, its origins, development and values’
(p. 60). The NHS, he thought, ‘has allowed and encouraged sentiments of altruism, reciprocity and social
duty to express themselves; to be made explicit in identifiable patterns of behaviour by all social groups
and classes’ (p. 292).
altruism like this cannot effectively flourish without it. If it were so—we think it is a
genuinely open question—then we would have good reason for welcoming institutional
trust. Some bodies may need it if they are to carry out their purpose.17
References
Baier, Annette. 1986. Trust and Antitrust. Ethics 96: pp. 231–60.
Faulkner, Paul. 2016. The Problem of Trust, in New Perspectives on Trust. Oxford: Oxford
University Press.
Ferguson, E., K. Farrell, and C. Lawrence. 2008. Blood Donation is an Act of Benevolence
rather than Altruism. Health Psychology 27: pp. 327–36.
Hardin, Russell. 1992. The Street-Level Epistemology of Trust. Analyse & Kritik 14: pp. 152–76.
Hawley, Katherine. 2014. Trust, Distrust and Commitment. Noûs 48: pp. 1–20.
Hieronymi, Pamela. 2008. The Reasons of Trust. Australasian Journal of Philosophy 86:
pp. 213–36.
72: pp. 63–76.
Horsburgh, H. J. N. 1960. The Ethics of Trust. The Philosophical Quarterly 10: pp. 343–54.
Jones, Karen. 2004. Trust and Terror, in Moral Psychology, ed. Peggy DesAutels and Margaret
Urban Walker, pp. 3–18. Lanham, MD: Rowman and Littlefield.
Kosfeld, Michael, Markus Heinrichs, Paul J. Zak, Urs Fischbacher, and Ernst Fehr. 2005.
Oxytocin Increases Trust in Humans. Nature 435 (7042): pp. 673–6.
Rapport, F. and C. Maggs. 2002. Titmuss and the Gift Relationship: Altruism Revisited. Journal
of Advanced Nursing 40: pp. 495–503.
Simpson, Thomas. 2012. What Is Trust? Pacific Philosophical Quarterly 93: pp. 550–69.
Spencer, Jack. Forthcoming. Relativity and Degrees of Relationally. Philosophy and
Phenomenological Research.
Strawson, Peter F. 1962. Freedom and Resentment. Proceedings of the British Academy 48:
pp. 1–25.
Titmuss, Richard. 1970. The Gift Relationship: From Human Blood to Social Policy. London:
Allen and Unwin.
Thompson, E. P. 1981. Customs in Common. London: Merlin Press.
Uslaner, Eric. 2002. The Moral Foundations of Trust. Cambridge: Cambridge University Press.
Warren, Mark. 1999. Democracy and Trust. Cambridge: Cambridge University Press.
17
Thanks to audiences in Cambridge, Manchester, Paris, and Turin and in particular to Boudewijn de
Bruin, Paul Egré, Maurizio Ferraris, Zoe Fritz, Laurent Jaffro, Rae Langton, Marco Meyer, Alex Oliver, the
other members of the Cambridge Trusting Banks project, and the editors of the present volume. We are
very grateful for financial support from the PSL-Cambridge scheme.
10
Deciding to Trust
Benjamin McMyler
1. Introduction
In one very general respect, it should be uncontroversial that we can decide to trust.
We can decide to trust a mechanic to fix the car, decide to trust a neighbour to look
after the kids, or decide to trust a friend to show up on time. We often find ourselves in
situations in which we must deliberate about whether particular others are worthy of
our trust concerning particular matters, and when the upshot of this deliberation is
positive, it is natural to describe ourselves as having decided to trust.
In this general respect, interpersonal trust does not differ from belief. Just as we can
decide to trust someone as a result of deliberation concerning whether the person is
worthy of trust, so we can decide to believe a proposition as a result of deliberation
concerning whether the proposition is worthy of belief. We often find ourselves in situ-
ations in which we must deliberate about whether a particular proposition is true, and
when the upshot of this deliberation is positive, it is natural (or at least not unnatural)
to describe ourselves as having decided to believe the proposition (or as having decided
that the proposition is true).1
When philosophers make a point of the fact that we can decide to trust, however,
they seem to have something different in mind. The way in which we can decide to
trust is typically taken to mark an important difference between trust and belief. Here
are just a few representative examples of such a sentiment:
Suppose you run a small shop. And suppose you discover that the person you have recently
employed has been convicted of petty theft. Should you trust him with the till? It appears that
you can really decide whether or not to do so. And again it appears that you can do so without
believing that he is trustworthy. Perhaps you think trust is the best way to draw him back into
the moral community; perhaps you simply think it is the way you ought to treat one of your
employees. Of course your belief about the likelihood that he will steal will be one factor in
1
See, for example, Hampshire (1965: 97–8): ‘[T]here are countless thoughts that occur to me, and
that pass through, or that linger, in my mind, and of these only a small minority constitute beliefs. The
beliefs are those thoughts that I endorse as true. I do not merely find them occurring or lingering: I decide
in their favour.’
162 Benjamin McMyler
your decision whether to trust. It might be that if you really believe he will steal, you will not be
able to trust him. But you can trust him without believing that he will not. (Holton 1994: 63)
In affectively trusting S to φ A need not believe that S will φ. Moreover, this is crucial to our
being able to decide to trust. A can decide to trust S to φ just because trust need not involve the
belief that S will φ. (Faulkner 2014: 1979, original emphasis)
In some circumstances trust seems to be a matter of decision: we can sometimes directly con-
trol whether we trust, even though we lack that kind of direct control over our beliefs. Trusting
someone may eventually lead to belief, but it is possible for the trust to come before the belief.
So trusting someone to do something need not involve belief that she is trustworthy, nor belief
that she will do what she is trusted to do, nor even belief that it is likely she will do it.
(Hawley 2014: 2030)
There are two related but distinguishable claims in these passages. First, there is a claim
that we can trust a person S to φ without having certain associated beliefs, for example,
the belief that S will φ, or that S is likely to φ, or that S is trustworthy. Let’s call this non-
cognitivism about trust:
Non-cognitivism about trust: Trusting S to φ does not require believing that S will
φ (or believing that it is likely that S will φ, or believing that S is trustworthy).
Second, there is a claim that non-cognitivism about trust is connected to the fact that we
can decide to trust. The fact that we can trust S to φ without having such associated
beliefs makes room for deciding to trust. As Faulkner puts it, ‘A can decide to trust S to φ
just because trust need not involve the belief that S will φ.’ So the possibility of deciding
to trust requires that we can trust without belief. Holton, Faulkner, and Hawley thus all
seem to hold that trust can be a matter of decision in a way in which belief cannot.
In holding that we can decide to trust in a way that we cannot decide to believe, the
above authors are not denying that we can decide to believe in the sense that I articu-
lated above. They are not denying that we can form beliefs as the result of rational
deliberation. This is not the sense of decision at issue. In claiming that we can decide to
trust in a way that we cannot decide to believe, they appear to be appealing to the gen-
erally accepted idea that we cannot believe at will, voluntarily, in the way that we can
act. The sense of decision at issue appears to be the sense of decision relevant to debates
about doxastic voluntarism. The most natural way to interpret the claim that we can
decide to trust in a way that we cannot decide to believe is thus as stating that we can
trust at will, voluntarily, in the way that we can act. Let’s call this position voluntarism
about trust:
Voluntarism about trust: We can trust S to φ directly at will, in the way that we
can act.
In the above passages, Holton, Faulkner, and Hawley do not explicitly endorse what I am
here calling voluntarism about trust.2 Instead of explaining the precise sense in which we
2
For a more explicit recent endorsement of voluntarism about trust, see Frost-Arnold (2014).
deciding to trust 163
can decide to trust, the above passages contain an endorsement of non-cognitivism

along with a claim that this makes room for deciding to trust. Still, the most natural way
of interpreting how it is that non-cognitivism makes room for deciding to trust is by
construing the relevant sense of decision along the lines proposed by voluntarism. What
non-cognitivism supposedly makes room for is voluntary trusting.
The positions that I am calling non-cognitivism about trust and voluntarism about
trust are often run together in philosophical discussions of deciding to trust, such that
evidence for one is taken to be evidence for the other. I think that this is a mistake. In
this chapter I argue that even if one accepts non-cognitivism about trust, one should
reject voluntarism. There is good reason to think that we cannot trust directly at will, in
the way that we can act, and this is so regardless of whether trust requires belief.
First, I articulate a general constraint on something’s being directly subject to the
will. If something is directly subject to the will, then it can be done for any reason that
an agent takes to show it worthwhile. Things that are directly subject to the will do not
admit of ‘reasons of the wrong kind’, considerations that an agent takes to show doing
something worthwhile but for which an agent cannot in principle do that thing
(Hieronymi 2005). Second, I argue that trusting someone to do something fails to
meet this constraint. There are considerations that an agent can take to show trusting
S to φ worthwhile but for which the agent cannot in principle trust S to φ. Trusting
admits of reasons of the wrong kind. Trusting someone to do something thus is not
directly subject to the will. There may still be important differences between deciding
to trust and deciding to believe, and I discuss one such possible difference later in the
chapter. Importantly, however, such differences are not a matter of voluntariness. Trust
and belief are equally non-voluntary.
The case that I present against voluntarism is not entirely novel. Pamela Hieronymi
makes a similar case in ‘The Reasons of Trust’ (2008a), and what I say here is heavily
indebted to her work. Unlike my argument here, however, Hieronymi’s discussion is
bound up with a rejection of non-cognitivism about trust, with a defence of the idea
that trust, at least in the ‘purist’ sense that she attempts to articulate, requires belief.
I here formulate the argument against voluntarism in a way that is neutral with respect
to the question of whether trust requires belief. Regardless of whether one thinks that
interpersonal trust requires certain associated beliefs, one should reject the idea that
we can trust at will.
One clarification is in order before moving on. Since most philosophical discussions
of deciding to trust focus on cases of interpersonal trust, cases of trusting another per-
son to do something, this will be my focus here.3 As such I will have little to say about
3
As Hieronymi (2008a) notes, Holton’s (1994) discussion often seems to run together trusting another
person to do something and entrusting goods to others. Part of the plausibility of Holton’s case for volun-
tary trusting might thus arise from intuitions about entrusting goods. After all, we typically entrust goods
to others by means of performing particular actions, and these actions are clearly under our voluntary
control. However, I suspect that what makes these actions acts of entrusting is that they are motivated by a
trusting attitude, and I will here argue that such a trusting attitude cannot be directly willed. Note also that
other forms of trust such as self-trust, communal or institutional trust, trust in the
testimony of others, or entrusting goods to others. Nevertheless, I suspect that the
considerations that tell against the possibility of voluntary interpersonal trusting will
also tell against the possibility of other forms of voluntarily trusting.
2. Voluntarism about Trust

My argument against voluntarism about trust rests on a particular reasons-based
conception of voluntariness. This clearly is not the only way in which voluntariness
might be construed. It might even be an unfortunate use of the English term ‘volun-
tary’. However, I think that this conception of the voluntary best captures the sense in
which belief is said to be non-voluntary.4 If the voluntariness of trust is to be contrasted
with the non-voluntariness of belief, then presumably it is this sense of the voluntary
that is at issue:
Voluntariness: An activity is voluntary only if it can be done for any reason that
an agent takes to show it worthwhile.5
Actions are voluntary in this respect. An agent can raise her arm or cross the street or
run for president for any reason that she takes to show doing so worthwhile. This is not
to claim that all voluntary actions are actually done for reasons. An agent might per-
form an action for no particular reason. Still, to characterize the agent as performing
an action is to characterize her as doing something that could be done for any reason
that she takes to show acting worthwhile. If there were some reasons that she took to
show acting worthwhile, then she could, in principle, act for these reasons. These need
not be good reasons. The fact that one would like to see what the Oval Office looks like
might be a fine reason to take a tour of the White House, but it is a terrible reason to run
for president. Even so, if an agent takes this consideration to show running for presi-
dent worthwhile, however ‘irrational’ this might be, an agent can run for president for
this reason. There are no ‘reasons of the wrong kind’ for acting, no considerations
I assume in what follows that there is a distinctive general attitude of interpersonal trust the nature of
which we can attempt to elucidate. For an intriguing rejection of this assumption, see Owens (Chapter 13
in this volume).
4
For a defence of this claim, see Hieronymi (2006) and (2008b).
5
Hieronymi offers a more complicated account of voluntariness designed to explain how believing
indirectly for practical reasons, as the result of a project of self-management, differs from performing some
kind of extended voluntary action (like rearranging the room) by means of performing other more basic
actions (like pushing and pulling the furniture):
An activity is voluntary just in case you decide to do it for reasons you take to settle the
question of whether to do it, therein intend to do it, and, providing all goes well, do it by
executing that intention. That is, an activity is voluntary just in case one engages in the
activity by forming and executing an intention to do so, where one forms an intention to
engage in the activity in settling the question of whether to engage in it, where that question
could be settled by any set of considerations that one takes both to count in favor of the
activity and to be sufficiently strong (or, indeed, for no particular reason). (2008b: 366)
that an agent can herself take to show φ-ing worthwhile but for which she cannot
in principle φ.6
Beliefs are not voluntary in this respect. An agent cannot in principle believe that
p for any reason that she takes to show believing that p worthwhile. There are ‘reasons
of the wrong kind’ for believing, considerations that an agent takes to show believing
that p worthwhile but for which she cannot, in principle, believe that p.
In order to motivate the idea that we cannot believe directly at will, Jonathan Bennett
(1990: 88–90) points out that we typically cannot believe directly for the reason of what
he calls ‘inducements’, even though we might take such inducements to be considerations
that show believing that p worthwhile.7 If I offer you fifty dollars to believe that there
is a pink elephant in the room, or threaten to harm you if you do not believe that
there is a pink elephant in the room, you might very well take my offer or threat to be a
consideration that shows believing that there is a pink elephant in the room worth-
while. In this respect, you might take my inducement to count in favour of believing.
But can you believe that there is a pink elephant in the room directly for the reason
of my inducement?
We might imagine cases in which you take my inducement to be a consideration
that shows the content of the belief to be true. Perhaps you think that I wouldn’t make
such an offer or threat unless there actually was a pink elephant in the room, or per-
haps you irrationally think that anything that anyone expresses a desire for you to
believe is in fact true. You would then take my inducement to be something like
evidence or testimony of the presence of a pink elephant.
But what if you do not take my inducement to show the content of the belief to be
true? In such a case, it seems that you cannot believe that there is a pink elephant in
6
This claim has been disputed. Schroeder (2007; 2012) argues that actions do admit of reasons of the
wrong kind, but see Hieronymi’s (2013) response, and Schroeder’s (2013) rejoinder. Raz (2011: 50) also
seems to hold that there can be reasons of the wrong kind for acting, claiming that there can be what he
calls ‘non-standard reasons’ for action, though I find his argument for this unconvincing. As far as I under-
stand him, he seems to have something like the following in mind: imagine that the fact that φ-ing would
be courageous is a reason for φ-ing, but furthermore that actually φ-ing for the reason that it would be
courageous renders one’s act not courageous. The act is courageous only if it is performed for some other
set of reasons. Here we cannot act courageously for the reason that it is courageous, even though the fact
that it is courageous is a reason for so acting. So the fact that so acting would be courageous can only be a
reason to get ourselves (perhaps via habituation) to perform the action for other reasons, just as pragmatic
reasons for belief can only be reasons for getting ourselves to believe for other reasons (or for no reason).
But this seems to me a mistake. To describe an act as courageous is already to describe it in terms of the
reasons for which it is performed. What such cases reveal, then, is that acting for a reason admits of reasons
of the wrong kind (or non-standard reasons), not that actions themselves do. So there can be reasons of the
wrong kind (or non-standard reasons) for acting courageously or humbly or in some other way virtuously.
The fact that it would make me look good might be a reason to act virtuously, but to act for the reason that
it will make me look good will not then be to act virtuously. But this doesn’t show that there can be reasons
of the wrong kind (or non-standard reasons) for action. It just shows that there can be reasons of the wrong
kind for ‘ways of being’ that express our rational take on the world—courageous, humble, trusting, etc.
7
See also Alston (1988: 263). I say ‘typically’ in order to leave room for the kind of case I describe in the
next paragraph in which one takes the inducement to show the content of the belief true. I say ‘directly’ to
rule out cases in which one takes the inducement to be a reason to act in ways designed to bring about
the belief.
the room directly for the reason of my inducement. Even though you genuinely take
my inducement to show believing that there is a pink elephant in the room worthwhile,
if you do not take my inducement to show it to be true that there is a pink elephant in
the room, you cannot believe for this reason. You cannot believe that p for a reason that
you yourself do not take to show p true. Of course, you might take my inducement to
be a reason to attempt to bring yourself to believe that there is a pink elephant in the
room via some kind of project of self-management. You might take my inducement to
be a reason to seek out evidence of the presence of a pink elephant or to act in a way
designed to somehow cause yourself to believe this. You might take this to be reason to
hire a hypnotist or to take a pill designed to induce the belief. But even if such a project
proved successful, this would not amount to believing directly for the reason of my
inducement. Compare the case of intentional actions. If I offer you fifty dollars to raise
your right hand, and if you take my offer to show raising your right hand worthwhile,
then you can easily raise your right hand for the reason of my offer. You do not need to
perform some kind of act of self-management in order to bring yourself to raise your
hand. You can act directly for the reason of my inducement.8
One might think that the case of believing that there is a pink elephant in the room
differs from the case of raising your right hand in that, while you have good reason for
thinking that there is not a pink elephant in the room, you have no reason not to raise
your hand. But consider a different case. Imagine that I offer you fifty dollars to believe
that I grew up in Wisconsin or threaten to harm you if you do not believe this. Let’s
assume that you take my inducement to be a consideration that shows the belief worth-
while, and let’s also assume that my having grown up in Wisconsin is consistent with
everything else that you believe. You might even have other reasons, my accent, for
example, for thinking that I grew up somewhere in the Midwest, making this proposition
highly plausible. Still, if you do not take my inducement to show this belief true—if you
do not take it to be testimony or evidence that I grew up in Wisconsin—it seems that
you cannot believe that I grew up in Wisconsin for this reason. Again, you cannot in
principle believe that p for a reason that you yourself do not take to show p true.
Inducements like offers and threats are just a subset of a much larger class of consid-
erations that are reasons of the wrong kind for belief. There are all sorts of prudential
considerations that might show having a particular belief worthwhile but for which we
cannot directly believe. I might take the fact that it would make my parents happy if
I shared their political beliefs to be a consideration that shows these beliefs worth-
while. But if I do not take this fact to show these beliefs to be true, I cannot simply
believe for this reason.
Bennett presents these considerations concerning inducements in order to motivate
the idea that we cannot believe at will. Importantly, these considerations do not explain
8
I take this distinction between believing directly and indirectly in response to practical reasons to be
sufficiently intuitive for the purposes of this chapter. For further defence of this distinction, see Hieronymi
(2008b: 364–7).
why we cannot believe at will, and Bennett himself admits to failing to find such an
explanation. Nevertheless, these considerations concerning inducements do give us
good reason to think that we cannot believe directly at will.9
I think that similar considerations give us good reason to think that we cannot trust
directly at will. Voluntarism about trust, as I have defined it, is the view that we can
trust someone to do something directly at will, in the way that we can act. We can act
for any reason that we take to show acting worthwhile. There are no reasons of the
wrong kind for acting. I will argue that just as we cannot typically believe directly for
the reason of inducements like threats and offers, we cannot typically trust directly for
the reason of such inducements. There are reasons of the wrong kind for trusting,
meaning that we cannot trust in the way that we can act.
3. Against Voluntarism about Trust

Belief is not the only attitude that admits of reasons of the wrong kind. Hieronymi
(2006, 2008b, 2009) argues that the general class of judgement-sensitive or, as she
calls them, commitment-constituted attitudes all admit of reasons of the wrong kind.
Consider intentions. There can be prudential considerations that an agent takes to
show intending to φ worthwhile but for which the agent cannot directly intend. If
I offer you fifty dollars to intend to drink some poison in the future, without requiring
that you actually follow through, you might take this to be a consideration that shows
intending to drink worthwhile. But if you do not take this consideration to show
drinking the poison the thing to do, you cannot intend to drink the poison for this
reason.10 My offer is here a reason of the wrong kind for intending, meaning that we
cannot intend directly at will, in the way that we can act.11
This does not assume that intentions are a kind of or even require belief. The reason
for thinking that one cannot intend directly at will is not that intention requires belief.
Rather, it is that intentions seem to admit of reasons of the wrong kind, while things
done at will do not.
I think that similar considerations give us good reason to think that we cannot trust
at will. There can be prudential considerations that an agent takes to show trusting
someone to φ worthwhile but for which the agent cannot simply trust. Consider the
following variation on Holton’s shopkeeper example. Imagine that you are a store
manager and that I offer you a large sum of money to trust a particular employee not
9
Like Williams (1973), Bennett holds that believing at will is conceptually impossible, but Bennett
rejects Williams’ arguments for this conclusion and confesses to failing to find a convincing alternative
argument. Hieronymi (2006) presents an alternative argument for the conceptual impossibility of believing
at will designed to meet Bennett’s objections.
10
Of course, if you believe the ill effects of the poison to be outweighed by your need for the money, you
might take the inducement to be reason to drink the poison, in which case you could intend for the reason
of the inducement. But this would not be a case of intending to φ for a consideration that you to take to
show intending to φ worthwhile without showing φ-ing to be done.
11
This example is a variation on Kavka’s toxin puzzle (1983).
to steal from the till. Let’s stipulate that you genuinely take my offer to show trusting
the employee worthwhile. You don’t take the possible costs of trusting the employee to
outweigh the benefits of my offer, for example. Let’s also stipulate that you don’t take
my offer to be a consideration that shows the employee herself to be worthy of trust.
You don’t interpret the offer as evidence of the employee’s trustworthiness, for
example. Can you trust your employee for the reason of my offer?
It seems to me that you cannot. If you do not take my offer to be a consideration that
shows something about the employee, about her worthiness of trust concerning this
particular matter, then you cannot trust the employee for this reason, even if you take
this reason to show the attitude of trusting worthwhile. You cannot trust S to φ for a
reason that you yourself do not take to show S worthy of trust. Of course, you might
take my offer to be a reason to act as if you trust or to try to bring yourself to trust by
some indirect means, by seeking out evidence of the employee’s trustworthiness or by
otherwise trying to convince yourself of the employee’s worthiness of trust. But even if
successful, such a project of self-management would not amount to trusting directly
for the reason of the offer. It would be trusting for some other reason, or possibly for no
reason. My offer thus appears to be a reason of the wrong kind for trusting, meaning
that we cannot trust directly at will, in the way that we can act.
As in the case of intentions, this does not assume that trust requires belief. The reason
for thinking that one cannot trust directly at will is that trust admits of reasons of the
wrong kind, while things done at will do not. If trust did require belief, then insofar as
we cannot believe directly at will, this would help to explain why we cannot trust directly
at will, but many non-cognitivist accounts of the nature of trust are n evertheless con-
sistent with the idea that trust admits of reasons of the wrong kind. Take, for example,
the affective attitude account of trust defended by Karen Jones. According to Jones:
Trust is an attitude of optimism that the goodwill and competence of another will extend to
cover the domain of our interaction with her, together with the expectation that the one trusted
will be directly and favorably moved by the thought that we are counting on her. The attitude
of optimism is cashed out not primarily in terms of beliefs about the other’s trustworthiness but
rather—in accordance with certain contemporary accounts of the emotions—in terms of a
distinctive, and affectively loaded, way of seeing the one trusted. (1996: 4)
The attitude of optimism that Jones here refers to, an attitude that she describes as a
distinctive and affectively loaded ‘way of seeing’ the one trusted, doesn’t appear to be
something that can be adopted at will. If it could be adopted at will, then it could be
adopted for the reason of inducements like threats and offers. But while you might take
a threat or offer to show adopting such an affectively loaded attitude worthwhile, if you
do not take it to show the other worthy of trust, then you cannot ‘see the other’ in this
affectively loaded way simply for the reason of the inducement. Such ‘seeing’ is no
more voluntary than believing.
The same can be said of many emotions. Just as we cannot believe for any reason that
we take to show believing worthwhile, we cannot feel anger, fear, love, hope, or pride
for any reason that we take to show these attitudes worthwhile. Imagine that I offer you
a large sum of money to feel proud about an achievement of yours or angry about
another’s action. If you do not take my offer to show your achievement worthy of pride
or the other’s action worthy of anger, then you cannot adopt the attitude for this rea-
son. We cannot feel a certain way in the way that we can act, for any reason that we take
to show feeling this way worthwhile. This is not to say that we cannot control our emo-
tions or that we cannot decide to feel a certain way about the world or about other
people. It is only to say that the relevant sense of control or decision is not the sense that
applies to intentional action. As Jones puts it:
Affective attitudes look toward features of the world that would make them justified and can
no more be adopted in the face of a known and acknowledged absence of such grounds than a
belief can be adopted in the face of a known and acknowledged lack of evidence. Because trust
involves an affective attitude, it is not something that one can adopt at will: while one can trust
wisely or foolishly, trust cannot be demanded in the absence of grounds for supposing that the
person in question has goodwill and competence and will be likely to take into account the fact
that one is counting on them. This is not to say that there can never be an element of decision
in adopting beliefs or attitudes. We can, for example, decide that the evidence we now have is
enough to support the belief, but we can’t just decide to believe regardless of the evidence.
While trust cannot be willed, it can be cultivated. We cultivate trust by a selective focus of
attention toward the grounds for trust and away from the grounds for distrust. (1996: 16)
Even if trust is a non-cognitive, affective attitude, this makes it no more subject to the
will than belief, and this is why, as a matter of both personal and public policy, trust is
something that must be cultivated rather than incentivized. While we can give others
all sorts of reasons to trust, not just any reasons are reasons for which they can directly
trust. If we simply give others incentives, if we simply make it in their interest to trust,
all they can do in response is to take these incentives to be a reason to act in ways
designed to cultivate in themselves a trusting attitude. They cannot trust directly for
these reasons. If we want to give them reasons for which they can directly trust, we
need to provide them with what Jones calls ‘grounds for trust’, with considerations that
show people to be worthy of trust. These are the only considerations on the basis of
which others can, in the relevant sense, decide to trust.
At this point, it is worth noting a peculiarity of my argument thus far. I have posited
a general constraint on what it is for an activity to be voluntary, and then I have
appealed to the particular case of inducements to argue that the attitude of trust, like
many other attitudes, fails to meet this constraint. This might seem strange. After all,
those committed to voluntarism about trust do not claim that we can always trust at
will. They accept that there are many occasions on which we cannot, particularly occa-
sions on which we positively believe or have strong evidence for believing that the
other is untrustworthy. They claim only that we can trust voluntarily on certain occa-
sions, so one might think that even if we cannot trust at will in cases of inducements,
this does not show that we cannot trust at will in these other cases. It does not show that
we cannot trust at will in Holton’s original case of the shopkeeper, for example, and it
does not show that we cannot trust at will in other cases of therapeutic, coping, or
corrective trust.12
In denying that we can trust at will, I am committed to denying that the kinds of
cases that voluntarists present are genuine cases of voluntary trusting. I think that such
cases should be read in one of two general ways.13 Either they are cases in which an
agent has reasons that she genuinely takes to show the person trusted worthy of trust,
or they are cases in which the agent only has reasons that she takes to show the attitude
of trust worth having. If the former, then the agent can decide to trust for these reasons,
but so doing is not to trust at will any more than believing that p for reasons that one
takes to show p true is believing at will. If the latter, then the agent cannot trust directly
for these reasons. The agent can act as if she trusts, or act in ways designed to cultivate
in herself a trusting attitude, but she cannot trust directly for these reasons.
My proposal, then, is that all cases of purportedly voluntary trusting can be plausibly
interpreted in one of these two ways. I cannot go through every such case here, so let’s
just consider Holton’s case of the shopkeeper.14 The most interesting part of Holton’s
discussion of the shopkeeper, I think, is his claim that the shopkeeper might decide
to trust the employee for what appear to be moral reasons, as a way of drawing the
employee back into the moral community or as a reflection of how one ought to treat
one’s employees, without taking these reasons to show it to be true that the employee
will not steal from the till and hence without believing that the employee will not steal.
The question, then, is whether these moral reasons are considerations that the shop-
keeper takes to show the employee worthy of trust. If they are, then the shopkeeper can
trust directly for these reasons, though not at will. If they are not, then the shopkeeper
cannot trust directly for these reasons.
I think that we might be able to imagine a case in which the shopkeeper thinks that
she morally ought to trust her employees, takes this to show her employee to be worthy
of trust, but doesn’t take this to show it to be true that the employee will not steal from
the till. In such a case, the shopkeeper could decide to trust directly for this moral con-
sideration even though she doesn’t take this moral consideration to be an epistemic
reason for believing that the employee will not steal. But even though the reasons for
trust here would not be epistemic reasons for an associated belief, this would not be a
case of trusting at will. This would be a case of deciding to trust for a reason that one
takes to show the person trusted worthy of trust. This is the sense of deciding to trust
that I outlined at the beginning of this chapter, a sense of decision that applies equally
to belief. So if this is what Holton has in mind, then what he says about the shopkeeper
is actually consistent with my argument against voluntarism. The shopkeeper is not
12
For examples of purportedly voluntary trusting that fall into these three categories, see Frost-Arnold
(2014).
13
Hieronymi (2008a) proposes the same general strategy for explaining cases of purportedly voluntary
trusting.
14
For similar treatments of other cases of purportedly voluntary trusting, see Hieronymi (2008a), Keren
(2014), and Marušić (2015).
trusting voluntarily even though she is trusting for moral reasons that she does not
take to support certain associated beliefs.
Things would be very different if the shopkeeper took the relevant moral considerations
to show the attitude of trust worth having without showing the employee worthy of trust.
If the shopkeeper takes this moral consideration to show something about the attitude
of trust rather than about the person trusted, then it will not support the d istinctive
way of seeing the one trusted that Jones refers to. One cannot see the one trusted in this
particular way for considerations that one does not take to show the person trusted to
be this way. If such moral considerations are only taken to show having the attitude
worthwhile, then they are like inducements. They do not show the person trusted
worthy of trust, and so one cannot trust for these reasons. If this is the kind of scenario
that Holton has in mind, then I think that the shopkeeper cannot decide to trust directly
for these moral reasons. She can take these moral considerations to be reasons to act in
ways designed to cultivate in herself trust in her employee, but she cannot trust directly
for these reasons.
Along similar lines, Hieronymi (2008a) argues that considerations of the value or
importance of trust are reasons of the wrong kind for trusting, reasons that we might
take to show the attitude of trust worth cultivating but for which we cannot directly
trust. I gather that what she has in mind here, in referring to considerations of the value
or importance of trust, are considerations that an agent takes to show the attitude of
trust valuable, important, or worthwhile without thereby showing a particular person
worthy of trust. So construed, considerations of the value or importance of trust are
like inducements. But while considerations of the value or importance of trust might
be construed in this way, I think that what we have seen about Holton’s case of the
shopkeeper shows that there is another way in which they might be construed. If one
thinks that one ought to trust one’s employees with respect to certain matters, or that
one ought to trust one’s spouse or children or friends in certain ways, and if one takes
this consideration to show something about the relevant persons towards which the
attitude is directed rather than something about the value or importance of housing
the attitude, then it seems to me that one can trust directly for this reason. Such
considerations of value or worth might be reasons of the right kind for trust even
though they are not reasons of the right kind for certain associated beliefs.
So I think that Holton and others might be correct that deciding to trust differs
from deciding to believe at least in the sense that certain considerations that are
reasons for trusting might not be epistemic reasons for having certain associated
beliefs. But this does not show that we can trust voluntarily. There are good reasons
for thinking that we cannot trust voluntarily, at least in the sense that I have outlined,
even if some of the reasons for which we can directly trust are not epistemic reasons
for certain associated beliefs.15
15
I have argued that the attitude of interpersonal trust is not action-like. It is sometimes claimed that
trust can be both an action and an attitude. Faulkner (2011: 23), for example, claims that ‘the act of trusting’
4. Trust and Agency

I have argued that just as the fact that we cannot believe directly for the reason of
inducements gives us reason to think that beliefs cannot be directly willed, so the fact
that we cannot trust directly for the reason of inducements gives us reason to think
that trust cannot be directly willed. However, the fact that these attitudes cannot be
adopted directly for the reason of inducements does not explain why these attitudes
cannot be voluntary. I want to end by briefly sketching the form that I think such an
explanation should take.
As I have already noted, Hieronymi holds that there is a large class of judgement-
sensitive or commitment-constituted attitudes that admit of reasons of the wrong
kind and thus that cannot be directly willed. She characterizes these attitudes as atti-
tudes that, as she puts it, embody an agent’s answer to a question (2009: 139). Believing
that p, for example, is an attitude that embodies an agent’s answer to the question
whether p. An agent believes that p if and only if she has settled positively the ques-
tion whether p. Intending to φ is an attitude that embodies an agent’s answer to the
question whether to φ. An agent intends to φ if and only if she has settled positively
the question whether to φ.
Insofar as such attitudes embody answers to questions, a reason can be taken to bear
on these attitudes in two distinct ways. Either it can be taken to bear on the question
the answering of which the attitude embodies, or it can be taken to bear on a different
question concerning the obtaining of the attitude itself. For example, a reason can be
taken to bear on believing that p by bearing either on the question the settling of which
is an act of ‘putting oneself in a position of depending on something happening or someone doing some-
thing’ while ‘the attitude of trusting’ is ‘an attitude towards this dependence’. Many accounts that construe
trust as a form of reliance, such as Holton’s, also seem to imply that trust can be an action. Clearly, if trust
is an action, then it can be voluntary. However, I find it odd to construe trust as an action. While we might
describe an act of Φ-ing as a trusting act or an instance of reliance as an instance of trusting reliance, it
seems to me that to do so is to characterize the act in terms of the attitude that motivates it or is expressed
by it. To say that an act is a trusting act is to say that it is an act that is motivated by or expresses the attitude
of trust. To say that an instance of reliance is an instance of trusting reliance is to say that it is reliance that
is motivated by or expresses the attitude of trust. Whereas actions (including acts of reliance) are events,
trust, it seems to me, is a state, a psychological attitude. I have argued that this psychological attitude is not
voluntary. Frost-Arnold (2014) construes trust as an attitude that involves either belief or what Bratman
(1999) calls ‘acceptance’, an attitude of accepting something in a context for the purposes of practical rea-
soning. Bratman holds that since acceptance differs from and does not require belief, acceptance can be
voluntary. As I have argued, however, just because an attitude does not require belief does not mean that it
is voluntary. Intention, for instance, is not voluntary. So while I cannot argue for this in detail here, if
acceptance is an attitude, I doubt that it is voluntary. Notably, while Cohen (1989) also develops a similar
conception of acceptance, Cohen construes acceptance as a mental act rather than a state or attitude. While
mental acts can clearly be voluntary, for the reasons given above, I do not think that trust can be plausibly
construed as such a mental act. Interpersonal trust is not like the mental acts of bracketing certain consid-
erations for the purposes of practical reasoning or assuming or imagining something for the sake of a
reductio, all of which can be done for any reason that an agent takes to show doing them worthwhile. As
I suggest below, trusting someone to do something requires taking a stand on something outside of
the attitude of trust itself, taking a stand on something about the world. Bracketing or supposing things
for the sake of reasoning does not require taking a stand on anything outside the activity of bracketing
or supposing.
the attitude of belief embodies, the question whether p, or by being taken to bear on a
question concerning the obtaining of the belief itself, for example, the question
whether believing that p is desirable or to be brought about. Importantly, reasons can
be taken to bear on a question concerning the obtaining of an attitude without being
taken to bear on the question the answering of which the attitude embodies. This is so
when an agent takes a consideration to bear on the question whether believing that p is
desirable or to be brought about without taking it to bear on the question whether
p. The fact that such attitudes embody answers to questions thus explains why they
admit of reasons of the wrong kind and, by extension, why they cannot be voluntary.
Intentional actions are not attitudes that embody answers to questions, and as such
they do not admit of reasons of the wrong kind. This is true of so-called mental actions
as well as of bodily actions. Some mental operations appear to be directly subject to the
will. One can imagine a starry sky, for example, in the way that one can act, for any
reason that one takes to show it worthwhile. Consider again the case of inducements. If
I offer you fifty dollars to imagine a starry sky, or if I threaten to harm you if you do not,
and if you take my inducement to show imagining a starry sky worthwhile, you can
easily do so for the reason of my inducement. There do not appear to be cases in which
you might genuinely take my inducement to show imagining worthwhile but in which
you cannot in principle imagine for this reason.
The agency that we exercise over mental actions like imagining thus differs from the
agency that we exercise over attitudes that embody answers to questions. Mental
actions are under our voluntary control while attitudes that embody answers to ques-
tions are not. This is not to say that we are passive with respect to such attitudes in the
way that we are passive with respect to states like pains and sensations. The only way
that we can exercise agency over our pains and sensations is by acting in ways designed
to affect them. We can act in ways designed to affect attitudes that embody answers to
questions as well—we can act in ways designed to cultivate in ourselves beliefs, inten-
tions, and other attitudes—but our agency with respect to these attitudes is not limited
to such acts of self-management. We can also exercise agency over these attitudes by
settling or answering the particular questions that the attitudes embody. We can exer-
cise agency over our beliefs by settling questions of the form whether p, and we can
exercise agency over our intentions by settling questions of the form whether to φ.16 In
this respect, we can decide to believe that p and decide to intend to φ in a way that
we cannot decide to feel pain or decide to feel cold, even though such decisions are
not voluntary.
I think we should apply this same general schema to interpersonal trust. If trust is a
kind of commitment-constituted attitude that embodies an agent’s answer to a ques-
tion (or set of questions), then this will explain why trust cannot be voluntary even
16
The agency exercised in settling questions is, arguably, more fundamental than the agency exercised
in intentional action. If φ-ing intentionally requires intending to φ, and if intending to φ requires having
settled positively the question whether to φ, then φ-ing intentionally requires having settled positively the
question whether to φ.
though we can decide to trust in a way that we cannot decide to feel pain. If the attitude
of trust embodies an agent’s answer to a question (or set of questions), then we can
exercise agency over our trusting by settling the question (or set of questions) the
answering of which the attitude of trust embodies. However, it will also be the case that
we cannot settle the question (or set of questions) that the attitude of trust embodies
for reasons that we ourselves do not take to bear on this question (or set of questions),
even if we take these reasons to bear on a different question concerning the obtaining
of the attitude, such as the question whether trusting S to φ is desirable or to be culti-
vated. It is the commitment-constituted character of interpersonal trust that makes
room for reasons of the wrong kind for trusting and, by extension, explains why trust
cannot be voluntary.
Applying this general characterization of commitment-constituted attitudes to
interpersonal trust leaves a lot to be filled in. Most significantly, it leaves open the pre-
cise form of the question (or set of questions) the answering of which the attitude of
interpersonal trust embodies. Hieronymi (2008a) argues that trusting S to φ requires
settling positively the question whether S will φ. The question whether S will φ is a
question of the form whether p, and so Hieronymi construes trust as a species of belief,
as belief held for particular reasons. This is not the only possibility here, however, and
this way of filling in the commitment-constituted character of interpersonal trust is
not required for appreciating why it is that trust is non-voluntary. Trust might instead
be construed as a kind of intention, and it might be construed as a kind of sui generis
affective attitude along the lines that Jones proposes. In fact, the general explanation
offered here of why we cannot trust at will seems to be what Jones herself is gesturing
at when she claims that affective attitudes ‘look toward features of the world that
would make them justified and can no more be adopted in the face of a known and
acknowledged absence of such grounds than a belief can be adopted in the face of a
known and acknowledged lack of evidence’ (1996: 16).
For the purposes of appreciating the non-voluntary character of interpersonal trust,
I think that we do not need to worry about how precisely to cash out the questions the
answering of which the attitude of trust embodies. Here is a very general way of putting
the point that is consistent, I think, with what both Jones and Hieronymi are claiming.
Attitudes like belief, intention, hope, fear, and trust all involve a particular way of rep-
resenting the world or a particular kind of ‘take’ on the world. This take on the world is
subject to reason in the sense that something like the Anscombean why-question has
application with respect to it.17 One can rightly be asked why one takes the world this
way—why one believes that p, intends to φ, fears X, or trusts S to φ—where this ques-
tion looks not merely for an explanation of how it came about that one houses this
attitude but for one’s reasons for taking the world this way, for the considerations that
17
Anscombe (1957) discusses the relevance of this why-question only with respect to intentional
actions. For discussion of the broader relevance of this Anscombean point, see Moran (2001: Chapter 4)
and Hieronymi (2006) and (2009).
one takes to show p true, to show φ-ing to be done, to show X to be feared, or to show S
worthy of trust. The fact that these attitudes involve such a reasons-responsive take on
the world is then what explains the fact that they cannot be directly willed. We cannot
adopt these attitudes for reasons that we ourselves do not take to show the world to be
the way that the attitude takes it to be. If we do not take an inducement to show p true,
to show φ-ing to be done, to show X to be feared, or to show S worthy of trust, then we
cannot believe, fear, or trust for this reason, even if we take the inducement to show the
attitude itself worth having. We cannot adopt these attitudes in the way that we can act,
for any reason we take to show them worthwhile. So however we cash out the precise
sense in which trust embodies an agent’s answer to a question (or set of questions), and
whether or not we think that trust requires belief, the fact that interpersonal trust is an
attitude of this general type is what ensures that trust cannot be voluntary.
The fact that trust is non-voluntary does not mean that we are passive with respect
to our trusting; that we can only manage our trust in the way that we can manage our
pains and sensations. We can directly exercise agency over our trusting by settling the
question (or set of questions) that the attitude of trust embodies, by actively taking the
world to be a certain way. We can, in this respect, decide to trust others. Insofar as
deciding to trust others is deciding to take the world to be a certain way, we cannot
decide to trust others for reasons that we ourselves do not take to show the world to be
this way, but this is not a limitation on our powers. It is simply a reflection of the fact
that interpersonal trust is an attitude towards the world. It is a distinctive attitude, to be
sure, an attitude towards a soul, and this makes it liable to abuse and betrayal in a way
that many other attitudes are not (Baier 1986: 235). It might also make it the case that
certain moral considerations can be reasons of the right kind for trusting others even
though they are not reasons of the right kind for certain associated beliefs. Nevertheless,
I hope to have shown that even if we can decide to trust for such reasons, deciding to
trust is an exercise of agency that cannot be voluntary.18
References
Alston, W. (1988), The Deontological Conception of Epistemic Justification, Philosophical
Perspectives 2: 257–99.
Anscombe, E. (1957), Intention. Cambridge, MA: Harvard University Press.
Baier, A. (1986), Trust and Anti-Trust, Ethics 96: 231–60.
Bennett, J. (1990), Why Is Belief Involuntary? Analysis 50: 87–107.
Bratman, M. (1999), Practical Reasoning and Acceptance in a Context, in Faces of Intention.
Cambridge: Cambridge University Press, 15–34.
Cohen, J. (1989), Belief and Acceptance, Mind 98: 367–89.
Faulkner, P. (2011), Knowledge on Trust. Oxford: Oxford University Press.
Faulkner, P. (2014), The Practical Rationality of Trust, Synthese 191: 1975–89.
18
This chapter has benefitted immensely from comments from the editors of this volume as well as from
participants at a conference for this volume organized by the editors at the University of Oxford.
Frost-Arnold, K. (2014), The Cognitive Attitude of Rational Trust, Synthese 191: 1957–74.
Hawley, K. (2014), Partiality and Prejudice in Trusting, Synthese 191: 2029–45.
Hampshire, S. (1965), Freedom of the Individual. New York: Harper & Row.
Hieronymi, P. (2005), The Wrong Kind of Reason, Journal of Philosophy 102: 437–57.
Hieronymi, P. (2006), Controlling Attitudes, Pacific Philosophical Quarterly 87: 45–74.
Hieronymi, P. (2008a), The Reasons of Trust, Australasian Journal of Philosophy 86: 213–36.
Hieronymi, P. (2008b), Responsibility for Believing, Synthese 161: 357–73.
Hieronymi, P. (2009), Two Kinds of Agency, in Mental Actions, ed. L. O’Brien and M. Soteriou.
Oxford: Oxford University Press, 138–62.
Hieronymi, P. (2013), The Use of Reasons in Thought (And the Use of Earmarks in Arguments),
Ethics 124: 114–27.
Holton, R. (1994), Deciding to Trust, Coming to Believe, Australasian Journal of Philosophy 72:
63–76.
Jones, K. (1996), Trust as an Affective Attitude, Ethics 107: 4–25.
Kavka, G. (1983), The Toxin Puzzle, Analysis 43: 33–6.
Keren, A. (2014), Trust and Belief: A Preemptive Reasons Account, Synthese 191: 2593–615.
Marušić, B. (2015), Evidence and Agency: Norms of Belief for Promising and Resolving. Oxford:
Oxford University Press.
Moran, R. (2001), Authority and Estrangement. Princeton, NJ: Princeton University Press.
Owens, D. (2017), Trusting a Promise and Other Things, in The Philosophy of Trust, ed.
P. Faulkner and T. Simpson. Oxford: Oxford University Press, 214–29.
Raz, J. (2011), From Normativity to Responsibility. Oxford: Oxford University Press.
Schroeder, M. (2007), Slaves of the Passions. Oxford: Oxford University Press.
Schroeder, M. (2012), The Ubiquity of State-Given Reasons, Ethics 122: 457–88.
Schroeder, M. (2013), State-Given Reasons: Prevalent If Not Ubiquitous, Ethics 124: 128–40.
Williams, B. (1973), Deciding to Believe, in Problems of the Self. Cambridge: Cambridge
University Press.
11
Trust and Evidence
Thomas Simpson
What is the rational relation between trust and evidence? The dominant view among
philosophers is that, in some sense, the importance of trust is that it goes beyond the
evidence. If you are following the evidence, then you are not trusting. ‘One does not
actually trust someone to do something if one only believes they will do it when one
has evidence that they will’ (Faulkner 2007: 876). According to Victoria McGeer, sub-
stantive trust ‘go[es] beyond what the evidence supports’ (2008: 240). This chapter
argues the contrary. Not only is it compatible with trust that one is following the evi-
dence. Further, and stronger, there are times when one’s trust is appropriate only if
one is following the evidence. I call this latter claim the scope-restricted evidentialist
constraint. It applies frequently and in important contexts.
Two tasks are undertaken in this chapter. First, I clarify the theses expressed by the
rough opening remarks above. Second, I argue for the evidentialist constraint. The
basic argument for the constraint is given by the following case.
Antarctic Resupply. Lief is preparing to trek to the South Pole alone, pulling his
food and equipment with him on a sledge. He is aiming to break the record for the
fastest unsupported journey. The weight of his sledge would jeopardize the attempt
if it had provisions for the return journey as well. So Katherine—an old Antarctic
hand, who runs an adventure support company—promises that she will be con-
tactable via satellite phone. When called, she will drop by parachute a package of
supplies at the Pole, and they arrange a contingency plan if communications should
fail. If Katherine does not follow the plan, it is all but certain that Lief will die. Lief is
very keen to survive the expedition. He sets out south.
I propose that Lief trusts Katherine. The claim does not rely on Lief ’s setting off. So
long as he would be prepared to do so, he trusts her. Moreover, in trusting her, he
should come to a judgement about whether Katherine will follow the plan. In coming
to this judgement, he should take account of all and only those epistemic reasons that
bear on the likelihood of her fulfilment of his trust; that is, his total evidence as to
whether she is trustworthy. The case has a structural feature that is generally a pplicable,
178 Thomas Simpson
regarding the conditions under which evidence-following trust is appropriate: namely

when a concern for the outcome provides second-order reasons both to consider all
the epistemic reasons that bear on the likelihood of fulfilment of the trust, and not to
consider non-epistemic reasons.
By making the case for the evidentialist constraint, I aim to show that there are
philosophically robust reasons for taking trust sometimes to be subject to the con-
straints of epistemic rationality. This is an important result in its own terms. It is also
important because of the role that trust too often plays in applied ethics. Because of its
near-ubiquity in social interaction and its sweet moral redolence, trust is often invoked
in applied contexts with rhetorical flourish but without statement of when it is appro-
priate. Lacking that, its invocation risks obscuring more than it reveals. Many of the
contexts which applied ethicists are interested in, and in which trust is appropriate, are
ones in which the key constraints that trust ought to be subject to are epistemic ones.
So in identifying some of the conditions under which such trust is appropriate, a fur-
ther result of the enquiry is clarification of the demands that putative trustors are—and
are not—subject to.
The structure of the chapter is as follows. I first clarify the evidentialist constraint,
surveying briefly the position that prominent accounts of trust take on it. I then seek to
deflate some of the seeming tension between different accounts, arguing that there is
no single kind of mental attitude that trust must take. The positive case for the eviden-
tialist constraint follows, concluded by a defence against objection.
1. Identifying the Claims

Take a stock example of trust. Jack asks his friend Jill if she will lend him her first edi-
tion volumes of Sterne’s Tristram Shandy, promising to return it to her within a week.
Should she trust him?
The example has the following contours. Implicit is that trust is a tripartite relation,
relating a trustor, to someone trusted, regarding some action that the trusted will per-
form.1 Schematically, A trusts B to φ. For simplicity, take ‘A’ and ‘B’ to refer only to
people. Regarding the relatum φ, it is not actions only that B may be trusted to perform;
one can trust people to have certain attitudes or dispositions, for instance (Blackburn
2010). A could also presently trust B to have performed some action in the past.
Identifying the possible objects and tenses of trust is tangential to my task, however, so
use ‘φ’ to refer to anything that B could be appropriately trusted over.
Implicit also is that trust is a rationally responsive mental attitude; ‘should’ assumes
such a normative orientation. What is it for a mental state to be rationally responsive?
It is for it to be evaluable according to how it responds to reasons. When a person—qua
agent, or knower—apprehends the force of a reason, she should modify her mental
states accordingly, to reflect its force. (The implication of agency may be misleading;
1
The observation is a commonplace; e.g. Baier (1994: 101); Good (1988: 33).
trust and evidence 179
such modification may well not be under a person’s conscious control.) Typical
instances of rationally responsive mental attitudes exhibit some degree of responsive-
ness. Reasons may come from a variety of domains: epistemic, moral, instrumental,
aesthetic, and so forth. Examples of rationally responsive mental states include: beliefs,
desires, moral judgements, aesthetic judgements, intentions, and decisions. Trust is
rational when it is correctly responsive to reasons.
What is the relevance of evidence? Evidence provides epistemic reasons for believ-
ing a proposition, which propositional mental states are rationally responsive to.
Assume a probability-raising conception of evidence: e is evidence for p iff it confirms
p, and evidence against iff it disconfirms p. That is, p is more or less likely given the
evidence than not. Evidence is presently available iff, at t1, one is aware of it. It is
possibly available iff, at t1, it is a practical possibility that one may come to be aware of it
at t2. One’s total evidence for p is constituted by all one’s available evidence for whether
p or ~p; as such, total evidence is relative to a person. Total evidence may be construed
synchronically, according to one’s total evidence that is presently available, or dia-
chronically, according to one’s total evidence that is possibly available. Suppose, as
is plausible, that anything that constitutes an epistemic reason for belief—that is,
which is truth-conducive—satisfies the definition of evidence stipulated above. One’s
epistemic reasons for belief at t1 are then given only by one’s evidence, construed syn-
chronically. The evidentialist about belief, who is committed to the view that belief that
p should be determined by one’s total evidence only, thus claims that belief should be
responsive to all and only one’s epistemic reasons.
Apply the foregoing to trust. In wondering whether she should trust B over φ, one of
A’s concerns is often enough whether he will φ. Any concern for evidence is thus a
concern for her total evidence as to whether B will φ. B’s being likely to φ, we may sup-
pose, is necessary for his trustworthiness. There is, then, a basic division between two
kinds of views on the relation between trust and evidence, according to whether one
endorses the following:
Evidentialist constraint. Trust is rational only if, on one’s total evidence, it is likely
that B will φ.
Applied to our stock example, Jill should trust Jack only if, on her total evidence, it is
likely that Jack will return the first edition to him (in a week, and as well as fulfilling any
other commitments that he undertook, such as not damaging the book, and so on).
Her total evidence is construed synchronically. For it to be ‘likely’ that B will φ on one’s
total evidence, that evidence must be above the threshold of indeterminacy, when it is
likely neither that B will φ nor not φ. There are, I think, three basic considerations in
favour of the evidentialist constraint.
First, it makes sense of the intuitive relation between trust and trustworthiness. At
least prima facie, it is puzzling why you would want to trust someone who is not trust-
worthy. The evidentialist constraint gives an especially clear account of a necessary con-
dition for rational trust. Second, because it is natural to identify trust with belief given
180 Thomas Simpson
the evidentialist constraint, trust explains cooperation in a way consistent with the
simple Humean belief–desire model of action (see Hardin 2002: 7). Trust simply is the
belief that another is likely to cooperate; conjoined with the desire for its results,
cooperation ensues. Third, and in more broad-brush terms, it brings a concept that is
central to our social lives, and is seemingly in tension with the idea that rational agents
pursue their self-interest, within the orbit of utility-maximizing decision theory. No
queer moral notions need be invoked. Accordingly, the two most explicit exponents of
the view are economists, the former by discipline and the latter by inspiration. For
Partha Dasgupta, trust involves ‘correct expectations about the actions of other people
that have a bearing on one’s own choice of action when that action must be chosen
before one can monitor the actions of those others’ (Dasgupta 1988: 51). For Russell
Hardin, trust is ‘a cognitive notion, in the family of such notions as knowledge, belief,
and the kind of judgment that might be called assessment. All of these are cognitive in
that they are grounded in some sense of what is true. . . . The declarations ‘I believe you
are trustworthy’ and ‘I trust you’ are equivalent’ (Hardin 2002: 7, 10). Having identi-
fied trust as a type of belief, both writers undertake the challenge of picking out what
marks out trust from non-trust beliefs. The strategy is to restrict trust to specific
propositional content.
Both Hardin and Dasgupta identify trust as a specific category of mental attitude,
namely cognitive; the evidentialist constraint then follows from evidentialism about
belief. But both claims are inessential to the endorsement of the constraint. For one, a
conjunctive account of trust is possible, on which trust may comprise a belief plus
some other mental attitude(s). Trust need not even comprise a belief. The evidential-
ist constraint would still hold so long as trust consists in a rationally responsive type
of mental attitude, and that attitude requires some kind of belief for it to be rationally
sustained. Some emotions may satisfy this condition. Rational fear plausibly requires
a belief about danger, but does not itself consist in the belief. For another, one could be
anti-evidentialist about belief and still endorse the constraint, thinking belief some-
times to be permissibly responsive to non-epistemic reasons, but also holding that
those non-epistemic reasons do not apply in the case of trust. Regardless, my concern
is with the evidentialist constraint, not the identification of the kind of mental atti-
tude that trust consists in.
The evidentialist constraint, taken simpliciter, is surely false. Richard Holton’s nuanced
and influential treatment of trust shows why. He takes a game from drama classes as an
example, where someone is invited to fall backwards with her eyes closed. She presumes
that the others will catch her, but does so in the absence of any formed belief that they
will. It feels as if she is at that moment deciding whether or not to fall. In deciding whether
to fall, she decides whether to trust. Nor is this to confuse acting from trust with the men-
tal attitude of trust itself. Holton proposes that this phenomenology of decision should
be taken at face value and concludes that some instances of trust are voluntary. Because
belief is not voluntary, so trust does not involve belief. Instead, he proposes that trust is
constituted by a disposition to rely, along with adoption of the ‘participant stance’: a
readiness to feel betrayal should trust be disappointed, or gratitude should it be upheld

(Holton 1994: 63, 67; the participant stance draws on Strawson 1974). Holton’s account is
an instance of the second, broad view of the relation of trust with evidence, which denies
the evidentialist constraint. Call it non-evidentialism about trust.
According to non-evidentialism, Jill’s trust may properly be responsive to reasons
other than those that bear on whether Jack will return the first edition, and which may
suffice for rational trust in the absence of evidence for Jack’s performance. Such reasons
are not hard to find. As Holton notes, trust may be used as a form of moral training. It
may also serve to signal the view that the trusted is regarded as a member of the moral
community, with the approbation this implies—witness the silver given by Bishop
Myriel to the escaped convict Jean Valjean at the outset of Les Misérables, on the
promise that he would use it to become an honest man (Hugo 1862: II, 12). This is an
instance of supererogatory trust. There are arguably instances of obligatory trust too.
H. J. N. Horsburgh thought there is a general prima facie duty to trust because the dis-
couraging effects of distrust are so severe (Horsburgh 1960). Similarly, D. O. Thomas
thought that one ought to trust friends and intimates simply because they are
friends and intimates (Thomas 1979; for qualification, see Hawley 2014b). These
considerations comprise, I think, the principal motivation for non-evidentialism
about trust: they make explicit its responsiveness to moral reasons. A result of this
responsiveness to moral reasons is the often striking insensitivity of trust to counter-
evidence. Had Othello trusted Desdemona, he would have ignored all Iago’s reports.
Karen Jones takes it as a datum for an account of trust that it ‘can give rise to beliefs that
are abnormally resistant to evidence’ (Jones 1996: 15; also Baker 1987: 1).
Non-evidentialism comprises a buzzing variety of views about what kind of attitude
trust consists in. These variously describe trust as an affective state which shapes how
one interprets others’ action (Jones 1996; Becker 1996; Lähno 2001); or as an ascrip-
tion of obligation to the trusted (Nickel 2007); or as a basic acceptance of trustworthi-
ness (E. Simpson 2013); or as an expectation that the trusted will prove trustworthy
and be so because the trustor’s reliance counts as a reason for them (Faulkner 2011); or
as a necessary presupposition of moral life (Bernstein 2011); or as a belief about the
trusted’s commitment, plus reliance on them to fulfil it (Hawley 2014a). Accordingly,
in the example, Jill may trust Jack as each of these conditions is fulfilled. On each of
these accounts, the reasons of trust are non-epistemic. Although, as noted, a non-
cognitive mental attitude may still require a belief for it to be rationally held, in practice
each of these accounts is given in order to accommodate the possibility of rational
trust even when the evidence does not render it likely that B will φ. While not comply-
ing with the evidentialist constraint, these accounts need not be wholly unconstrained
by evidence. Some build in a defeasibility condition, such that possessing evidence on
which it is likely that B will not φ renders trust irrational. ‘This resistance to evidence is
not limitless’ (Jones 1996: 16; also Holton 1994; Faulkner 2011). Accounts with a
defeasibility condition allow trust that is counter to the evidential constraint just when
the evidence of the likelihood of B’s φ-ing is indeterminate.
182 Thomas Simpson
It is unsurprising that those who observe the responsiveness of trust to moral and
other non-epistemic reasons should construe trust as a non-cognitive mental state.
What is surprising is a related but distinct family of views that are now current, on
which trust is taken to be both cognitive and responsive to moral reasons. So distin-
guish non-cognitive non-evidentialism from cognitive non-evidentialism. What is the
motivation for this latter view? Exponents are, I think, impressed by the need for belief
that trust often seems to involve. It is not just that the trustor acts as if she believes that
the trusted will be trustworthy; rather, she actually believes that he is. It is that belief
which constitutes the trust. They are also impressed by the moral dimensions of trust,
such that interpersonal relationships ground the resulting belief, and this not merely
through the epistemically uncontroversial process of giving privileged access to evi-
dence. Cognitive non-evidentialism thus seeks to endorse both claims simultan-
eously. How is the view possible? In practice, it delimits a subset of evidence as that
which provides epistemic reasons for trust, as opposed to one’s total evidence. The
view is not that some evidence is ‘screened off ’ for epistemic reasons, as higher-order
evidence might do to lower-order. Rather, while screening-off occurs, it happens for
non-epistemic reasons.
One reason for doing so is the partiality that special relationships may justify. On
Thomas Hurka’s account, trust ‘involves believing beyond one’s evidence in the virtues
of particular people, such as one’s friends and family’ (2001: 108; for defence of doing so,
see Stroud 2006). In the example, it is because Jack is Jill’s friend that she is entitled to
believe that he will be trustworthy in returning the book, thus trusting him, even though
the total evidence for the likelihood of his returning it may be less than compelling.
Another reason is general, arising in virtue of the demands of respect for others. For
Pamela Hieronymi, it is respect for others’ agency, expressed by one’s taking their
practical reasons for action to be the basis for their action (thus excluding akrasia from
consideration for example). It is this which gives rise to a trusting belief, and specifically
not calculations of likelihood on which I ‘treat you as an object’ (2008: 226). For Ben
McMyler it is respect for the addressive, second-personal nature of interpersonal
relations. Trust ‘always involving believing that the person will φ . . . in a way that is
epistemically supported in virtue of the interpersonal relationship existing between
truster and trusted. . . . The reason that justifies the belief is irreducibly s econd-personal’
(2011: 115, 138; the second-personal terminology derives from Darwall 2006; both
Hieronymi and McMyler draw on Moran 2006). On this view, it is because Jack
addresses himself to Jill, promising to return that book and thereby giving himself
practical reason to do so, that she should trust him. In accepting his commitment as a
reason to believe that he will be trustworthy, she expresses her respect.2
Regardless of the variation, non-evidentialism is naturally expressed by locutions
such as ‘going beyond the evidence’, or ‘going against the evidence’. Each account
2
Arnon Keren’s (2014) ‘preemptive reasons’ account of trust should likely be classified as a form of
cognitive non-evidentialism. See Section 7.
ermits Jill to trust Jack in circumstances where her total evidence makes it equivocal
p
or even unlikely that he will return the book. Some even require, if Jill trusts Jack, that
she be in such a situation. This may be because evidence is just irrelevant to the ques-
tion of the rationality of that kind of mental state that constitutes trust, because it is a
kind of mental state rationally unresponsive to epistemic reasons. Or it may be because
the specified subset of evidence renders it sufficiently likely that he will return the
book, in contrast to her total evidence, on which it may be indeterminate whether he
will return it, or probable that he will not. In either case, if you are following the evi-
dence, you are not trusting.3
I endorse the following variant of the evidentialist constraint.
Scope-restricted evidentialist constraint. In some circumstances, trust is rational
only if, on one’s total evidence, it is likely that B will φ.
For brevity, I elide ‘scope-restricted’ when possible. I now turn to argue for the scope-
restricted view.
2. Plural Forms of Trust

How deep is the disagreement between the evidentialist constraint and non-evidential-
ist accounts of trust? In my view, much less deep than it seems. The reasons given in
favour of each count against the other only if a supplementary, suppressed premise is
granted. That premise is: ‘trust’ is univocal. As such, ‘trust’ always has the same referent.
I deny this. Listening to the way the word is used provides a compressed argument.
Sometimes ‘trust’ is naturally understood as referring to a sort of affective attitude (‘I
will trust my wife; I will not be jealous’); at other times to a conative one (‘Come what
may, I will trust you to the end’); and at yet others to cognitive ones (‘I know you are
honourable; I trust you’). Surely correctly, Annette Baier comments that ‘Trust, if it is
any of these [affective, cognitive and conative], is all three’ (1994: 132). Indeed, some-
times it is not a mental state but action itself which is described as trust (‘The patrol
followed the scout, trusting him to spot an ambush before it was too late’). It may be
that there is an explanation of why it is felicitous to use ‘trust’ in each of these instances
on which the term has an invariant meaning, perhaps expressible by necessary and
sufficient conditions. It is all but certain, however, that any such conditions do not
require the referent of ‘trust’ to be always the same kind of mental state. Further, I
doubt that there are any such invariant conditions of use. The vulnerability of existing
analyses of trust to counter-example is inductive evidence of this.
3
The cognitive non-evidentialist might object that they are following the evidence—just not all of it.
I take it, however, that the instruction ‘follow the evidence!’ pragmatically implicates that one should follow
one’s total evidence. Think of scenarios where it might be used, such as a judge instructing a jury, or as a
rebuke from one juror to another during their deliberations: ‘You’re not following the evidence!’ So
although cognitive non-evidentialism formally satisfies the instruction, it does not satisfy its implicature.
184 Thomas Simpson
The equivocity that ‘trust’ exhibits is not the same as that of ‘bank’. It is mere linguis-
tic accident that ‘bank’ is ambiguous between a financial institution and the side of a
river. It is no accident that ‘trust’ functions as an umbrella term that may refer to a
variety of mental attitudes that, while non-identical, nonetheless share a range of simi-
lar features. It is a separate question why these phenomena are referred to by the same
term. In outline, the explanation is partly analogical, due to resemblance, and partly
genealogical, due to convergent contexts of use. I argue the above points and give the
explanation at length elsewhere (Simpson 2012).
Granting this, a preliminary conclusion can be drawn regarding my opening ques-
tion. There is no general rational relation between trust and evidence. Because there are
different forms of trust, so there are different rational relations. Evidence has a different
bearing on affective attitudes, as it does on conative ones, as it does on cognitive ones, as
it does on action, because the reasons that each of these are rationally responsive to dif-
fer. There is no single set of ‘justification conditions of trust’, as Karen Jones seeks (1996:
4, 20–5). If ‘trust’ is an umbrella term, then there is no tension in fully endorsing as
instances of trust the scenarios identified by Holton, Horsburgh, and Thomas, in which
moral reasons result in trust even when trustworthiness is uncertain, whilst also recog-
nizing that there are other occasions where trust is rational only given epistemic reasons
possessed by the trustor for believing the trusted to be trustworthy. Scope-restricted
versions of the evidentialist constraint and non-evidentialist accounts—where each is
qualified to apply in separate domains—are compossible. The considerations that count
in favour of non-evidentialism do so at the expense of evidentialism only if scope-
restriction is not possible. But such qualification is wholly and independently plausible,
given the variety of kinds of attitudes that can constitute trust.
The effect of such a qualification is to lower the argumentative bar. To establish the
scope-restricted evidentialist constraint, it is not necessary to show that there are no
instances of non-evidence-following trust. Instead, it is sufficient to establish that, in
some circumstances where one trusts, one should do so only if the evidence renders B’s
fulfilment of the trust likely. Note that this entails the weaker claim, that following the
evidence is compatible with trust. Call this the compatibility claim. I argue for the evi-
dentialist constraint first. Then I defend the compatibility claim against objection.
3. When Trust Follows the Evidence

Recall Lief ’s preparation to trek to the South Pole, in Antarctic Resupply. He is very
keen both to achieve his goal and to survive the expedition, for which he needs
Katherine’s help. I propose that Lief ’s trust is rational only if he has good evidence that
she will follow the resupply plan.
Why so? The claim turns on the practical status of Lief ’s desire to survive the
expedition. To elucidate that, let me adapt Joseph Raz’s (1999) terminology of practical
reasons. Raz is interested in the variety of conflicts that an agent may have between her
reasons for and against φ-ing. In the simple case, the stronger reason that she has to φ
will override the weaker reason that she has not to φ. The balance of reasons that an
agent has for and against φ-ing guide her to a judgement about what she should do.
There is a more complex kind of case, however, in which an agent may possess
reasons not to act in accordance with what she takes her balance of reasons to be. One
of Raz’s examples is of a soldier, Jeremy, who is ordered by his commanding officer to
appropriate a tradesman’s van for an operation. He thereby has a reason to appropriate
the van. But a friend urges him to disobey the order, pointing to the weighty reasons he
has not to. Without judging what Jeremy ought to do, Raz proposes that ‘the order is a
reason for doing what you were ordered regardless of the balance of reasons’ (1999:
38). Raz’s claim is about the structure of reasons for action: they can be distinguished
into first-order and second-order reasons. Second-order reasons for action are reasons
for reasons; specifically, they are reasons for allowing or disallowing certain other
reasons to count. Second-order reasons count as ‘exclusionary’ when they give reason
for one’s action not to be guided by particular reasons. Exclusionary reasons have
scope restrictions, specifying what kinds of reasons do not count. Jeremy’s being a sol-
dier and therefore under the authority of his commanding officer provides him with an
exclusionary reason to disregard his own or his friend’s judgement about the balance
of reasons weighing against the appropriation. This is so at least insofar as the scope
restriction is satisfied: the officer is his commanding officer, and the order concerns a
military operation that Jeremy is liable to perform.
In Antarctic Resupply, Lief has an exclusionary reason for following the evidence
about whether Katherine will execute the plan. Given that he wants to survive, his con-
cern is to be in the best possible epistemic situation that he can with regard to whether
Katherine will make the drop; that is, he is concerned for truth. This concern for truth
excludes from consideration some of the reasons that he might have which could bear
on the judgement as to whether she will prove trustworthy, and thus whether he should
trust her. Suppose Katherine is his cousin. Family loyalty may provide Lief with a
reason to trust her with the resupply. But given that he wishes to live, it is not, of itself, a
reason for him. It is the wrong kind of reason for supposing that she will be trustworthy
in dropping the supplies. Of course, if family loyalty renders her more likely to make
the drop, then it should count for him. But plenty of folk don’t give a fig for their
cousins, and she may be forgetful and incompetent. Of itself, their being cousins pro-
vides no reason for him to trust her. In other situations, no doubt, their being family
may give Lief reason to trust her. But not here. Lief ’s desire for survival gives him an
exclusionary reason which requires that, in coming to the judgement that it is suffi-
ciently likely that Katherine will perform the resupply, he refrain from considering any
reason that does not bear on the likelihood of whether she will perform the resupply;
that is, any reason which is not epistemic.4
4
My use of the terminology extends Raz’s, which is explicitly concerned with practical reasons to φ
only. Nonetheless, while he takes reasons for belief and action to be the fundamental types of reasons,
reasons for other attitudes may be derived from or dependent on them (1999: 15). So the extension is not
inconsistent.
186 Thomas Simpson
What is the significance of the practical status of Lief ’s desire for survival? It is this
desire that generates the exclusionary reason. There are two construals of that desire,
according to its deontic force. Lief ’s desire to survive in Antarctic Resupply, we may
presume, is at least permissible. Granting this, and that his survival depends on his
placing trust well, it is likewise permissible that his trust follows the evidence.
Alternatively, fill out the story as follows.
Family Man. As in Antarctic Resupply, but Lief is married with five young children
dependent on him, not least his income. Marissa, his wife, has agreed to support the
expedition only because the desire to attempt the record has consumed his soul,
making him an unhappy person to be around.
Take it that, in Family Man, while Lief ’s expedition is permissible, he also has an obligation
to do his utmost to survive. Lief is morally obliged to ensure that his trust follows the
evidence. The two cases thus have a structural feature that allows generalization about
the conditions under which trust should follow the evidence: when A is appropriately
concerned about B’s actually φ-ing. The appropriateness of A’s concern for B’s
fulfilment of the trust could render it either permissible or obligatory for her to be
concerned only about the outcome. When A is appropriately concerned about whether
B will actually φ, then she should be responsive only to those reasons which bear on
whether he will do so; that is, her epistemic reasons for believing that he will φ. Under
these conditions, her trust is rational only if it follows the evidence.
Appropriate concern for fulfillment of trust occurs widely in everyday life. I do not
propose an account in terms of necessary and sufficient conditions for appropriate-
ness. But there are different categories of reasons that may render the concern for
fulfillment appropriate, and we recognize these easily enough. Practical reasons for
concern for fulfillment are the most obvious. High-stakes cases such as Antarctic
Resupply make the point compellingly. Special obligations may also ensure that my
decision about who to trust should be based only on the evidence for their trustworthi-
ness. Think of deciding upon a babysitter, for instance; the first consideration that
must be satisfied is the trustworthiness of the prospective caregiver. Trusting beyond
the evidence is here criticizable. Special obligations may also arise when the decision to
trust is exercised on behalf of someone else, who has not delegated the freedom to take
account of non-epistemic reasons. Consider civil servants, making decisions about
who to entrust contracts for large public works to. In the first instance, they must sat-
isfy themselves that there is sufficient evidence for a contractors’ likely fulfilment, and
are not entitled to count as reasons for this judgement anything which does not bear on
the likelihood of bidders’ fulfilling the contract.
In other practical cases, my appropriate concern for fulfillment is merely permissible.
In the marketplace, I am free to contract with whomever I choose. While my action is
bounded by the negative constraints of law, I have no positive obligations and am
entitled to pursue my own ends. Where these involve trust—two entrepreneurs start-
ing a business, for instance—I am entitled to consider only whether the other party
actually will fulfil his part of the deal. In doing so, I permissibly exclude other reasons
from consideration.
Epistemic reasons may also ground an appropriate concern for fulfilment. In testi-
mony, I am invited to trust someone’s word for the truth of what they tell me. Assuming
that the standard to which belief is held is that of the truth, I have an exclusionary rea-
son for not counting as a reason for belief anything that is not truth-connected, that is,
which is not evidence.
As it happens, Jill’s trust in Jack in the stock example would likely be marked by a
permissibly overriding concern for fulfilment, but grounded in aesthetic reasons.
One of the peculiarities of the first edition of Tristram Shandy was a hand-printed
marbled page, which was thus unique to each copy. Each individual volume has an
artistic value that, while it could be compensated for financially in case of loss, could
not be replaced. This artistic value is plausibly great enough to override, for instance,
any demands of friendship to assume the best. Jill’s trust would be permissibly subject
to the evidentialist constraint. While I have proposed that practical, epistemic, and
aesthetic reasons may ground an appropriate concern for fulfilment, this is unlikely
to be exhaustive.
4. Following all the Evidence

Lief ’s desire to survive his expedition to the Antarctic does not just provide him with
an exclusionary reason; it also provides him with an ‘inclusionary’ reason. This inclu-
sionary reason explains why cognitive non-evidentialism does not challenge the
scope-restricted evidentialist constraint.
An ‘inclusionary’ reason is a second-order reason to consider some reason in delib-
erating on action. (The term is mine. The only type of second-order reason Raz con-
cerns himself with is exclusionary; 1999: 40.) The function of an inclusionary reason is
to override any exclusionary reason, ruling in to the reasons which count for judge-
ment those first-order reasons which would otherwise be excluded. Lief ’s desire for
survival, I argue, means that in deliberating over whether to trust Katherine, and in
doing so coming to a judgement about whether she is likely to fulfil the trust, he should
be concerned with his total evidence, and not just some subset of it.
To see this, consider its denial. The cognitive non-evidentialist proposes that one’s
trust should be based on evidence, construed in a broad sense, but restricts the evi-
dence that counts to a particular subset. Take Hieronymi’s account as an example. On
her view, there is a kind of fully fledged trust that is a form of belief. It is essential to
trust that it involves vulnerability to betrayal; anything less is mere reliance. But treat-
ing someone’s utterance ‘as reliable evidence, calculating the likelihood of her veracity,
[is] precisely not trusting her. You are instead treating her like a good thermometer’
(2008: 222). She proposes instead that the trustor should form a belief about the trust-
ed’s future action ‘on the basis of her practical reasons, given a background assumption
of her trustworthiness, in something like the way that one might form a belief about
188 Thomas Simpson
one’s own future on the basis of one’s own practical reasons, given a background
assumption about one’s own reasonableness and competency’ (2008: 226; she disarm-
ingly offers the proposal as ‘very sketchy’). In so doing, I treat you as responsible, in the
way that I take myself to be responsible. On Hieronymi’s account, I am exposed to the
vulnerability of betrayal that is characteristic of trust only if certain reasons for trust
are excluded, namely those which involve treating someone as a thermometer.
Consider the possible implications of Hieronymi’s suggestion, as applied to Antarctic
Resupply. In deliberating whether to trust Katherine, and to do so towards adopting the
trusting belief that Hieronymi commends, Lief should weigh the practical reasons that
Katherine will have when it comes to following through on the plan. These include all
the excellent reasons that Lief has to continue living, as well as reasons specific to
Katherine, such as the value of building her business. Suppose also that the last three
expeditions she supported all ended in disaster, with trekkers starving on the ice, and
that while there were plausible excuses, those excuses have not convinced everyone. The
background assumption of trustworthiness is still plausible. According to Hieronymi’s
suggestion, in deliberating about whether to trust, Lief should not take account of the
history. His treating her as responsible means that his deliberation is solely prospective,
in effect a matter of asking himself what he would do in her situation. But this is puz-
zling. In deliberating whether to trust Katherine, Lief should learn all that he can from
the history and decide on the likelihood of her fulfilling the proposed plan from his best
possible epistemic position. If treating her as a good thermometer—say he has read the
transcripts from the police investigation where she was interviewed under a lie-
detector—would put him in a better position, then he should do so. Hieronymi’s sug-
gestion excludes such evidence from consideration; her proposal is that such reasons
for trust are ruled out. Lief ’s desire for survival, however, rules them in. He has an
inclusionary reason that overrides the proposed exclusionary reason.5
Hieronymi’s suggestion is important here because of its structure, not its details. All
cognitive non-evidentialist views propose that some evidence be excluded from con-
sideration. Other views may give epistemic reasons for excluding certain evidence,
because exclusion of that evidence is likely better to align the knower with her epi-
stemic reasons. (This is the case with higher-order and lower-order evidence. Linda
Zagzebski’s (2012) account of epistemic authority is another instance, in which she
adapts Raz’s service conception of authority.) But the cognitive non-evidentialist
excludes evidence from consideration for non-epistemic reasons. In doing so, any
such account is vulnerable to examples which parallel that above, where the excluded
5
While Hieronymi does not use the terminology of exclusionary reasons, it accurately reflects her
proposed structure of practical reasons. The argument she gives for excluding evidence that does not arise
from treating the trusted as a responsible agent far from decisively establishes any such exclusionary
reason. She proposes that trust essentially involves vulnerability to betrayal, and that trusting beliefs based
on all the evidence do not incur this vulnerability. Even granting both, the possibility of a conjunctive
account of trust remains open. If a non-belief conjunct can sustain the vulnerability to betrayal, then her
condition can be met, and without the need to exclude any evidence. The discussion is longer than presently
relevant, however, as she does not favour a conjunctive account. See Hieronymi (2008: 222, n. 16).
evidence would put the trustor in a better epistemic position regarding the likelihood
of the putatively trusted fulfilling his trust than she would be without it. When the
trustor is appropriately concerned about fulfilment, she should take account of that
evidence. In Antarctic Resupply, Lief ’s desire to survive means that, in coming to a
judgement about Katherine’s likely fulfilment, not only should he pay attention only to
the evidence that bears on this question, he should pay attention to all the evidence.
5. Is Antarctic Resupply Robust?

It may be objected that Antarctic Resupply fails to establish the case. It is an example of
trust precisely because Lief is going beyond the evidence. The stakes involved, namely
Lief ’s life, are such that his evidence could not fully suffice for rationally going on the
expedition. Going on the expedition is a decision to trust Katherine because his evi-
dence is insufficient. The evidential constraint is thus violated. If it were not violated, it
would merely be a decision to rely. Bets on probabilities are just that, and not real trust.
The objector concludes that, in endorsing the evidential constraint and equating trust
with a bet, James Coleman (1990: 99) may be commended for his clarity. But it is a
mistake.6
The objection is false. It trades on an illicit assumption about the value of Lief ’s life.
The contrast between Family Man and Antarctic Resupply makes the point. While I
stipulated, in Family Man, that Lief ’s expedition was permissible, many will take the
contrary view. Because of his family obligations, it is impermissible for Lief to under-
take the risks involved in trekking to the South Pole. These charges are made fre-
quently enough against climbers and explorers who die and leave children behind. In
Family Man, it is plausible that the evidential constraint is violated. Nonetheless, as a
general claim, it is plainly false that it is never rational to undertake additional risk,
including to one’s life, for the sake of some non-fundamental goal. People with no
special obligations to others may do so for relatively modest goals. In Antarctic
Resupply, this is explicit. Given this, there is no reason to think that the evidential
constraint is violated here.
I have made the positive argument for the evidentialist constraint in its scope-
restricted variant. I now turn to defend it against objection. This targets the compati-
bility claim, arguing that trust and evidence-following are contraries.
6. Trust and Evidence Gathering

Imagine that a husband is uncertain about his wife’s fidelity. For him to trust his wife,
he must not hire private detectives to check up on her movements; not trawl through
her email; and so forth. Katherine Hawley remarks, surely rightly, ‘If I claim to trust
you in some respect, but continue to seek evidence about your behaviour, you would
6
I am grateful to an anonymous reviewer for this objection.
190 Thomas Simpson
be justified in concluding that I do not really trust you’ (2014b: 2031; also Keren 2014).
Suppose, for the sake of argument, that if the faithfulness of one’s spouse is a concern,
then one’s concern is appropriately with the truth; that is, with whether she actually is
faithful, or is likely to be. On my account, this is an instance of trust that is subject
to the evidentialist constraint. As such, one ought to put oneself in the best possible
epistemic position. But putting oneself in the best possible epistemic position results
in evidence-gathering behaviour, and such behaviour is incompatible with trust. If one
is following the evidence, then one is not trusting. The objection concludes: so trust
and evidence-following are incompatible.
I reply: the objection equivocates on ‘following’. Recall the distinction between
presently and possibly available total evidence. The total evidence condition in the
evidential constraint, in both scope-restricted and unrestricted versions, is con-
strued synchronically. One ‘follows the evidence’ construed synchronically just if,
on one’s presently available total evidence, it is likely that B will φ. Of course, there is
a natural reading of ‘following the evidence’ on which it is read diachronically, as
when between t1 and t2 I take the actions required to acquire all of my possibly avail-
able evidence. It is following the evidence diachronically—call this ‘evidence-
gathering’, to avoid ambiguity—which is incompatible with trust, not following the
evidence synchronically.
How could the mandate to follow the evidence synchronically not result in the dia-
chronic (practical) mandate to go evidence-gathering? Consider, again, Antarctic
Resupply. Suppose Lief has learnt of the three previous disasters and so gathers evidence
as to Katherine’s reliability. Further, this evidence satisfies him that her trustworthiness
is not impugned. Prior to the satisfaction of the evidential constraint, he does not trust
her; the incompatibility claim yields the right verdict. Once he has gathered this evi-
dence, his trust satisfies the evidential constraint and may be rational. He sets out.
Suppose Lief now starts calling his friends at base camp on the satellite phone, to find
out if Katherine is following his progress sufficiently closely by logging his latest pos-
ition on a map, and so on. As before, this would surely indicate a lack of trust. But it is no
implication of the evidentialist constraint that Lief should start calling his friends back
at base, and there are a variety of explanations for why it would be rational for him not to
do so. (The following draws on Buchak 2012.) Principally, the evidence that satisfied
him that his trust would be rational—namely, her commitment to follow the plan—was
of sufficient strength to constitute a pro tanto conclusive reason in favour of her trust-
worthiness, and there simply is no need to re-evaluate it. I take this to be the typical case.
Given this pro tanto conclusive reason, there is no need to incur the costs in time, effort,
and anxiety involved in revisiting the evaluation of her trustworthiness. Further, there
are often opportunity costs involved in postponement of a decision, which may make
trust rational once the evidence has reached the threshold of likelihood.
The incompatibility of evidence-gathering and trust does not show that the eviden-
tialist constraint is false. It just shows that the constraint is not all there is to learn
about trust.
7. Trust and Giving One’s Word

In motivating his cognitive non-evidentialist account, Keren has posed a testimonial
example targeting the claim that, in coming to trust, I should consider all the evidence.
He invites us to consider Sandy.
[Sandy] addresses us and says: ‘take my word for it, p.’ And suppose that we form the belief that
p, and that we do so without weighing other relevant evidence available to us. It seems that
Sandy cannot now turn to us and say: ‘You are correct. But you should have considered all the
evidence available to you before reaching that conclusion.’ Criticizing us in this way appears
incompatible with her expectation of us that we take her word for it. If she were to criticize us
in this way, she would seem to be withdrawing her original appeal to us that we take her word
for it. (2014: 2599)
He contrasts Sandy with the case of Evelyn, who does not give her word but intends
only to provide very good evidence for p. (To dramatize the example, say she offers us a
photo which supports p.) In contrast, Evelyn can criticize us for not weighing all the
available evidence. For the sake of argument, accept that the contrast exists. It would be
inappropriate for Sandy to criticize us for not having considered all the evidence. The
objection is again an incompatibility one. Often enough, perhaps paradigmatically,
trust is based on someone having given their word. But it is incompatible with trust on
the basis of someone’s word that one has considered all the evidence.
The lesson that Keren draws from the contrast is that the phenomenon of giving
one’s word gives a hearer a different kind of reason for believing that p than that derived
from weighing all the evidence. It gives a hearer a second-order, preemptive reason not
to weigh all the evidence (in Raz’s terminology, again). Weighing all the evidence con-
sists in an evaluation of the relative strength of all the (first-order) epistemic reasons
for and against p.7
Keren’s argument is an explanatory one. He proposes that we ‘can’ explain the
difference between Sandy and Evelyn by appealing to a preemptive reasons structure
for trust, and that ‘no account of trust can be complete’ without doing so (2014: 2600,
2604). But this is not so. I reply: there is a pragmatic explanation of the contrast
between Sandy and Evelyn which appeals to the infelicity of Sandy’s criticising us for
taking account of all the evidence.
The problem is akin to that of Moore-paradoxical sentences, such as ‘p, and
I believe ~p’. Sandy’s utterances are individually felicitous, but jointly infelicitous.
‘Take my word for it, p’ is felicitous. So too is ‘You should consider all the evidence
7
Although it is not explained why the second-order, preemptive reason excludes some epistemic
r easons from consideration, thus allowing a subset of one’s epistemic reasons to count for the belief that p,
that reason cannot be epistemic. For acknowledges that it may result in trusters being in a worse epistemic
position than they otherwise would be. This is in virtue of precautions being excluded, some of which may
be epistemic (2014: 2604–6). In contrast, I argue that any such exclusion of epistemic reasons can be for
epistemic reasons only. On my account, when the evidentialist constraint applies, there are no epistemic
reasons that are excluded by non-epistemic reasons.
192 Thomas Simpson
before concluding that p’. But the following is not: ‘Take my word for it, p. And you
should consider all the evidence before concluding that p’. As with Moore-paradoxical
sentences, recasting them in the third person dissolves the tension: ‘She’s given her word
that p. And you should consider all the evidence before concluding that p’. Why does the
tension arise? To give your word is to commit to your reliability: of the truth of p in
testimonial cases, or the certainty of your performance, where it is action that is pledged.
It is to present someone with (what you take to be) a conclusive reason in favour of its
being the case that p, or that you will fulfil the trust; that is, a presentation of a reason
that is not overridden by any other. In doing so, you present yourself as having
conclusive reasons in favour of p. To then say ‘you should consider all the evidence’ is
to implicate that the reasons given in the conversation so far are not conclusive. Sandy
pragmatically implicates herself as both giving and not giving conclusive reasons in
favour of p.
The tension is not dissolved by the past tense construal of Keren’s actual example,
‘you should have considered all the evidence available to you’. So long as Sandy has not
withdrawn her word for the truth of p, she cannot without self-contradiction criticize
others for not having taken account of all the evidence. In reply to Keren, then, his
preemptive reasons account is not required for adequate explanation of the contrast.
While Sandy cannot felicitously utter both sentences, this constitutes no evidence that
the two cannot both be true. Giving your word is compatible with your hearers consid-
ering all the evidence.
8. Conclusion
The dissonance noted above between Sandy’s giving her word, and of her hearers con-
sidering all the evidence, is not restricted to the self-contradiction involved in her
expressing both propositions. It is widely remarked that a hearer’s refusal to believe a
speaker occasions offence; so too for occasions when trust is withheld. Indeed, it is
something of a trope here to quote Elizabeth Anscombe’s pregnant remark, ‘It is an
insult and it may be an injury not to be believed’ (1979: 150; Wanderer 2012: 148;
Govier 2014: 131). Richard Moran extends the point: ‘the offence remains even when
the speaker’s audience takes his having made the statement to count as evidence for its
truth’ (2006: 301; also Zagzebski 2012: 124). The observation is taken to show that the
evidentialist constraint does not apply to instances of testimony, and by extension, to
trust generally. Rather, the reason for belief offered by giving one’s assurance is a moral
one; correspondingly, belief can be morally demanded of the hearer.
It is an implication of my argument that this is false. For instances of trust where the
evidentialist constraint applies, there is no legitimate demand that the truster not con-
sider all the evidence. What is my response to the observation, then? It is undoubtedly
unpleasant to learn that you have not been believed. But does disbelief wrong you? I
doubt it. Nor, indeed, did Anscombe take her observation to imply so. For she immedi-
ately qualifies her statement about where the insult lies in rejecting testimony. ‘At least
it is an insult if one is oneself made aware of the refusal, and it may be an injury if others
are’ (1979: 150, italics added). The insult lies in the telling a hearer that he has not been
believed, not in the not believing. The injury lies in the telling others that the hearer
was not believed, and by implication, should not be. Regarding whether or not one
should be trusted, when a truster is appropriately concerned for whether the putatively
trusted has or will fulfil their trust, she should consider all and only her evidence.8
References
Anscombe, Elizabeth (1979) What Is It to Believe Someone? In C. F. D. Delaney (ed.),
Rationality and Religious Belief, 141–51. Notre Dame, IN: University of Notre Dame Press.
Baker, Judith (1987) Trust and Rationality, Pacific Philosophical Quarterly 68: 1–13.
Baier, Annette C. (1994) Moral Prejudices: Essays on Ethics. Cambridge, MA: Harvard University
Press.
Becker, Lawrence C. (1996) Trust as Noncognitive Security about Motives, Ethics 107: 43–61.
Bernstein, J. M. (2011) Trust: On the Real but Almost Always Unnoticed Ever-Changing
Foundation of Ethical Life, Metaphilosophy 42: 395–416.
Blackburn, Simon (2010) Practical Tortoise Raising and other Philosophical Essays. Oxford:
Buchak, Lara (2012) Can it be Rational to Have Faith? in J. Chandler and V. S. Harrison (eds.),
Probability in the Philosophy of Religion, 225–47. Oxford: Oxford University Press.
Coleman, James S. (1990) Foundations of Social Theory. Cambridge, MA: Belknap Press.
Darwall, Stephen (2006) The Second-Person Standpoint: Morality, Respect and Accountability.
Dasgupta, Partha (1988) Trust as a Commodity, in Diego Gambetta (ed.), Trust: Making and
Breaking Cooperative Relations, 49–72. Oxford: Basil Blackwell.
Faulkner, Paul (2007) On Telling and Trusting, Mind 116: 875–902.
Faulkner, Paul (2011) Knowledge on Trust. Oxford: Oxford University Press.
Good, David (1988) Individuals, Interpersonal Relations, Trust, in Diego Gambetta (ed.),
Trust: Making and Breaking Cooperative Relations, 31–48. Oxford: Basil Blackwell.
Govier, Trudy (2014) Victims and Victimhood. Peterborough, ON: Broadview Press.
Hardin, Russell (2002) Trust and Trustworthiness. New York, NY: Russell Sage Foundation.
Hawley, Katherine (2014a) Trust, Distrust and Commitment, Noûs 48: 1–20.
Hawley, Katherine (2014b) Partiality and Prejudice in Trusting, Synthese 191: 2029–45.
Hieronymi, Pamela (2008) The Reasons of Trust, Australasian Journal of Philosophy 86: 213–36.
Holton, Richard (1994) Deciding to Trust, Coming to Believe, Australasian Journal of Philosophy
72: 63–76.
Horsburgh, H. J. N. (1960) The Ethics of Trust, Philosophical Quarterly 10: 343–54.
Hugo, Victor (1862) Les Misérables. Paris: A. Lacroix, Verboeckhoven & Cie.
Hurka, Thomas (2001) Virtue, Vice and Value. Oxford: Oxford University Press.
Jones, Karen (1996) Trust as an Affective Attitude, Ethics 107: 4–25.
Keren, Arnon (2014) Trust and Belief: A Preemptive Reasons Account, Synthese 191: 2593–615.
8
For comments and criticism, I am grateful to audiences in Oxford and Cambridge, Tim Crane, Paul
Faulkner, Jeroen Van den Hoven, Hallvard Lillehammer, Alex Oliver, and an anonymous referee.
194 Thomas Simpson
Lähno, Bernd (2001) On the Emotional Character of Trust, Ethical Theory and Moral Practice
4: 171–89.
McGeer, Victoria (2008) Trust, Hope and Empowerment, Australasian Journal of Philosophy
86: 237–54.
McMyler, Benjanmin (2011) Testimony, Trust, and Authority. Oxford: Oxford University Press.
Moran, Richard (2006) Getting Told and Being Believed, in Jennifer Lackey and Ernest Sosa
(eds.), The Epistemology of Testimony, 272–306. Oxford: Oxford University Press.
Nickel, Philip J. (2007) Trust and Obligation-Ascription, Ethical Theory and Moral Practice 10:
309–19.
Raz, Joseph (1999) Practical Reason and Norms, 3rd edn. Oxford: Oxford University Press.
Simpson, Evans (2013) Reasonable Trust, European Journal of Philosophy 21: 402–23.
Simpson, Thomas (2012) What is Trust? Pacific Philosophical Quarterly 93: 550–69.
Strawson, P. F. (1974) Freedom and Resentment, in Freedom and Resentment and Other Essays,
1–25. London: Methuen.
Stroud, Sarah (2006) Epistemic Partiality in Friendship, Ethics 116: 498–524.
Thomas, D. O. (1979) The Duty to Trust, Proceedings of the Aristotelian Society 79: 89–101.
Wanderer, Jeremy (2012) Addressing Testimonial Injustice: Being Ignored and Being Rejected,
Philosophical Quarterly 62: 148–69.
Zagzebski, Linda Trinkhaus (2012) Epistemic Authority: A Theory of Trust, Authority, and
Autonomy in Belief. Oxford: Oxford University Press.
12
Being Pragmatic about Trust
Philip J. Nickel
1. Introduction
Despite substantial attention to conceptual questions about trust from philosophers
and non-philosophers alike, trust remains an ambiguous and contested concept.
Scholars wrestle with two problems: the familiar philosophical problem that we can-
not find a suitable analysis of trust on which everybody agrees in their intuitions (by
analogy to the problem of the analysis of knowledge, say),1 and the broader problem
that the concept of ‘genuine trust’ investigated by epistemologists and moral philo-
sophers is not the same one used in many explanations of social phenomena on the
basis of individual behaviour. In this chapter I will argue that the second, broader dis-
agreement has implications for the disagreement among philosophical accounts,
favouring accounts that are less restrictive and demanding, and that mark genuine
explanatory categories.
A so-called Trust Game (early versions in Camerer and Weigelt 1988; Kreps 1990:
p. 100; Berg, Dickhaut and McCabe 1995) brings out the conceptual discrepancy I have
in mind. Here is one version of the game: suppose if Francesca gives George $20,
George will use it to earn $80, an opportunity not available to Francesca by herself.
However, it is completely up to George whether he returns any part of the money to
Francesca. Suppose Francesca is disposed to give George $20 in this situation. As the
name of the game suggests, explanations of cooperation (or non-cooperation) in terms
of individuals’ needs and interests treat Francesca’s disposition as trusting without
regard to the particular reasons or motivations she might have. Suppose that on this
particular occasion of the Trust Game Francesca guesses that George will give her back
a portion of the money because she thinks he believes she is his enemy, and wants to keep
his enemies close. She attributes a cunning motive to him, and for present purposes she
is willing to use that to further her own ends. From the point of view of explaining
cooperation, there is no reason not to regard this as a potential motive for trust.
1
See Simpson (2012) for a description and diagnosis of the first problem. My approach differs from
Simpson’s in emphasizing the possibility of pragmatic arguments that may help to resolve these two
problems.
196 Philip J. Nickel
Strategic, calculative reasons are an important class of reasons for cooperation and can
take many forms. They can consist of a general expectation that the two interacting
parties will encounter each other many times in the future (so that it makes sense for
them to cooperate now); a sheer lack of better options; a desire to protect one’s reputa-
tion; or they can consist of particular beliefs such as Francesca’s.
For most philosophers, such reasons are unsatisfactory as reasons for trust.
Intuitively, not just any reason or motivation disposing Francesca to ‘invest’ in George
in such a situation is compatible with genuine trust. Intuitively, Francesca’s disposition
is not trusting even though she has sufficient confidence that George will return some
of the money that she is willing to give him. More generally, dispositions towards reli-
ance based on this sort of strategic expectation must be different from dispositions
based on trust. As a result, the concept as used in explanations of social phenomena
and the philosophical concept clash in what they treat as trust.
To spell out the narrower, intuitive concept, philosophers often introduce concep-
tual restrictions on the allowable motivations and reasons embodied in a person’s
disposition to rely on another person, if it is to count as trust. On various accounts, a
trusting person must have:
• Optimism that the person relied upon will act with competence and goodwill
(Baier 1994: p. 98; Jones 1996) or moral integrity (McLeod 2002);
• An affective expectation that the person will be responsive to the fact that one is
depending on her (Faulkner 2007);
• A normative or moral expectation that the person will perform in a certain way
(Nickel 2009), for which we hold them responsible (Walker 2006: p. 78);
• A belief that the person relied upon is trustworthy (Hieronymi 2008).2
These restrictions imply, more or less, that Francesca’s strategic expectation of reli-
ability does not count as genuine trust. Her disposition to invest money in George is
not based on optimism about George’s goodwill or integrity, nor on an affective expect-
ation that George will be responsive to her reliance, nor on a moral expectation, nor
does it involve a belief that he is trustworthy (understood the way these notions are
intended). I will come back to some of the details later, but what I want to emphasize at
first is what these views share—their focus on conceptually restricting the motives and
reasons of trust. I will label them restrictive views of trust, and the alternative the unre-
strictive view. I will also speak of less restrictive and more restrictive views.
Considerable philosophical motivation for various restrictive views has been pro-
vided in some of the articles cited above. In this chapter I explore what might be said on
behalf of unrestrictive and less restrictive views on which the concept of trust is left
open, potentially covering all manner of dispositions to rely on another person. Russell
Hardin (2006) formulates the point I want to develop, suggesting an argument from
2
‘One person trusts another to do something only to the extent that the one trustingly believes that the
other will do that thing’ (Hieronymi 2008: p. 214).
being pragmatic about trust 197
explanatory potential according to which less restrictive views of trust are more useful
because they can provide a more adequate explanation of the origin of cooperative
behaviours and social institutions than more restrictive views.
I begin by presenting the argument from explanatory potential. I contend that the
degree to which restrictive views are undercut by this argument depends on exactly what
they require of the trustor. I consider in somewhat more detail how two of the more
promising restrictive accounts fare. At the end of the chapter, I briefly consider the reasons
for thinking that such an argument actually might partly determine our concept of trust.
2. Trust in an Explanatory Role

In this section I develop the argument from explanatory potential and apply it to
restrictive and unrestrictive theories of trust. The basis of the argument is a methodo-
logical constraint on theorizing about trust: trust should (a) be explained as the out-
come of central concerns or interests of the relevant actors, and (b) explain the
emergence and sustenance of cooperative practices and social institutions (Hardin
2006: p. 16).3 In what follows I will call this the Explanatory Constraint, and I will call
(a) the Input Condition and (b) the Output Condition. It is explicitly a constraint about
social explanation, and is meant to allow for empirical research on trust that explores
the particular mechanisms and types of causation involved. It is neutral about other
possible explanatory aims (e.g. investigating the developmental significance of a trust-
ing orientation for the individual, or philosophically motivating the linkage between
trust and certain moral attitudes).
In order to see why the Explanatory Constraint is attractive, it is worth considering
the situation that would obtain if it were not satisfied. Trust is widely regarded as an
important concept for understanding social phenomena. But if trust is not used to
describe how central concerns or interests of interacting parties lead to social behav-
iours, practices, and institutions, then it cannot be particularly important for under-
standing social phenomena. For example, if a person’s needs for food and shelter, and
3
Hardin’s formulation is that it should ‘be explained … as the outcome of behaviors guided by some
central concern or motivation of the relevant actors’ (Input Condition) and should yield ‘explanations of
behavior and social institutions’ (Hardin 2006: p. 16) (Output Condition). In my formulation I have
dropped the idea that trust should be the outcome of behaviours, since it seems that it might also be the
simple result of the concerns and interests themselves. Hardin does not develop the argument in any detail.
The claim that trust explains the emergence and sustenance of cooperative practices and social institu-
tions is not meant to imply that it is the only factor that explains the development of cooperative practices,
nor that cooperative practices conceptually require trust. In fact, Hardin and his colleagues explicitly
restrict the concept of trust to dyadic relationships in which one has specific beliefs about the other’s
motive, and at the same time try to ‘make sense of a wide array of devices for organizing cooperative behav-
ior in the absence of trust or in the presence of very weak trust’ (Cook, Hardin, and Levi 2005: p. 7). Many
other factors also play a role on their account. But if the Explanatory Constraint is true, trust has an
important role, not only when it is ‘successful’ in the sense that the person one trusts actually performs,
leading to a pattern of cooperation or a relationship of reciprocity, but also when it fails and leads to other
strategies for securing cooperation, as it sometimes appears to.
the social opportunities for the realization of these needs, do not normally lead to a fur-
ther state described in terms of trust—whether fulfilled by the performance of the other,
or not—that mediates the formation of cooperative practices and social institutions,
then trust cannot play a significant role in explaining how social phenomena emerge
from basic human needs. In that case, trust will not be especially important in under-
standing social phenomena. This argument shifts the burden to those who do not
evidently satisfy the Explanatory Constraint, to show why trust nonetheless deserves a
place in social philosophy, or to explain why its apparent importance is illusory.
One of the foremost advocates of the unrestrictive view of trust is the sociologist
James Coleman. On Coleman’s version of the view, trust is simply a disposition to rely
on another person in order to satisfy substantial needs or interests, so that the expected
gains justify the losses (1990: p. 99). Trust is explained in terms of the needs and inter-
ests of individuals (or sub-social groups), together with the opportunities that reliance
on others can provide towards fulfilling those needs and interests. The trusting
disposition consists of a posited mental state that takes these factors into account and
resolves action in light of them. In turn, the successes and failures of actions taken in
accordance with such dispositions explain (i) the emergence of stable patterns or prac-
tices of cooperation; and (ii) the evolution and design of norms and institutions that
help change the balance of incentives so that failures of trust can be avoided in the
future to a greater degree. (In what follows I will sometimes refer to (i) and (ii) together
as ‘practices and institutions’.)
Coleman’s theory emphasizes the strategic rationality of individuals when it comes
to explaining why practices and institutions emerge (see Coleman 1990: pp. 13–19).
His attribution of self-interested, strategic rationality to individuals or sub-social agents
is an interpretive posit that makes general sense of human behaviour for explanatory
purposes.4 It is elastic in what it treats as rational, allowing rational agents to take, for
example, internalized social norms as reasons (Coleman 1990: pp. 292–3). The posit-
ing of rational choice is not meant as a conceptual requirement that distinguishes
between trusting reliance and non-trusting reliance: the notion of rational choice does
not restrict what counts as falling under the concept of trust in the same way that the
criteria from restrictive views are meant to. In particular, the unrestrictive view of trust
is not to be thought of as a conceptual view on which a trusting person always tries to
satisfy her individual, self-interested preferences. Such a view would encounter the
empirical problem that in one-off variants of the Trust Game, experimental partici-
pants often exhibit a disposition to cooperate (Johnson and Mislin 2011). This cannot
easily be explained in terms of individual, self-interested preferences.
A possible worry one might have about such a view is that it is too broad to allow for
empirical explanation. It is not empirically explanatory to say that trust is simply
whatever disposition leads one to rely on others: this is like saying that desire is simply
whatever it is that leads one to intentional action, or that food is simply whatever
4
For more on this idea, see Buchak (2016) for her notion of ‘interpretive decision theory’.
needs/
interests success evolve stable
T cooperation
opportunities failure
for reliance
norms and institutions
Figure 12.1 Trust and cooperation
nourishes us (in Aristotelian fashion). To say this is merely to label a phenomenon,

rather than to explain it. However, the unrestrictive view does not aim to make a def-
initional statement that labels all possible grounds of attempts at reliance on others as
instances of trust. Rather, the point is to take a pragmatic view, leaving open what
might count as instances of trust so that a range of possible motivations can poten-
tially fit, instead of defining them away analytically a priori.5
Under a single concept, so to speak, the unrestrictive view allows for the emergence
of stable patterns of cooperation from both strategic and non-strategic dispositions
towards reliance. For example, Francesca’s strategic reliance, if successful, could lead
towards a stable cooperative relationship between George and her; or if it fails, it could
lead Francesca to seek to adopt norms and/or institutional controls that would give
George (or others like him) more reason to perform (see Figure 12.1). If one of the
interacting parties happens to have non-strategic reasons for reliance on the other (e.g.
moral reasons), this also counts as trust.
As this example illustrates, the Explanatory Constraint seems to favour an unre-
strictive view of trust. If a trusting disposition towards reliance on others can be held
prior to cooperative practices and institutions, and is compatible with a broad range of
cooperation that takes place within those practices and institutions, then it can explain
the emergence and sustenance of those practices and institutions in terms of trust,
satisfying the Output Condition. An unrestrictive account has an easier time doing
this than a restrictive view. Because of higher uncertainty about the parties on whom
one could rely, and the absence of practices and institutions for guidance in decisions
about reliance, the dispositions towards reliance that one finds in situations prior to
practices and institutions often include purely strategic expectations of reliability. In
addition, it is hard to see how optimism about the other’s goodwill, judgements about
shared values, and so on—conditions that are constitutive of trust on the restrictive
view—could be justified in such conditions.
A different kind of example will help develop the argument further. Consider a
situation that approaches a ‘state of nature’: one of the first encounters of European
5
Thanks to Tom Simpson for prompting me to clarify this. On Simpson’s own view, Ur-trust is a bit like
this (2012). Simpson emphasizes the human need for some such notion in order to ground cooperation
and shared activity, and also doubts the prospects of a sharper philosophical analysis.
explorers of North America with Native American tribes. The French explorer La
Salle and his party approached an Illinois settlement by canoe:6
At nine o’ clock, doubling a point, [La Salle] saw about eighty Illinois wigwams, on both sides
of the river. He instantly ordered the eight canoes to be ranged in line, abreast, across the
stream . . . The men laid down their paddles and seized their weapons; while, in this warlike
guise, the current bore them swiftly into the midst of the surprised and astounded [Illinois
people]. The camps were in a panic. Warriors whooped and howled; squaws and children
screeched in chorus; some ran in terror, and, in the midst of the hubbub, La Salle leaped ashore,
followed by his men. None knew better how to deal with Indians; and he made no sign of
friendship, knowing that it might be construed as a token of fear. His little knot of Frenchmen
stood, gun in hand, passive, yet prepared for battle. The Indians, on their part, rallying a little
from their fright, made all haste to proffer peace. Two of their chiefs came forward, holding out
the calumet, while another began a loud harangue, to check the young warriors who were aim-
ing their arrows from the farther bank. La Salle, responding to these friendly overtures, dis-
played another calumet; while Hennepin caught several scared children and soothed them
with winning blandishments. The uproar was quelled, and the strangers were presently seated
in the midst of the camp . . . Food was placed before them; and, as the Illinois code of courtesy
enjoined, their entertainers conveyed the morsels with their own hands to the lips of [La Salle’s
party]. (Parkman 1983 [1878]: pp. 836–7)
The unrestrictive view clearly counts the Illinois’ willingness to allow La Salle into their
camp, and La Salle’s acceptance of food from the Illinois, as instances of trust, whereas
the restrictive view has a harder time doing so. Suppose that as a result of this initial
contact, a cooperative practice of repeated interaction results, such as a continued pat-
tern of interaction in which each party performs in a way that is useful to the other. Or
counterfactually, suppose that after the failure of the first interaction an external condi-
tion is imposed in order to change the terms of future interactions, such as the ‘friendly’
taking of hostages by both sides (a common practice at the time), or a peace treaty
between the French and the Illinois carrying clear sanctions for violations of its terms.
The unrestrictive view explains these outcomes in terms of trust, and the restrictive
view cannot do so. On restrictive views, both the initial interaction and the patterns of
cooperation that result are in all likelihood too strategic to count as instances of trust.
Hence trust does not play a role in explaining whatever cooperative or non-cooperative
outcomes result.
We can pose this as a dilemma for restrictive theories of trust. The first horn is to
hold that these reliant dispositions prior to applicable practices and institutions do not
count as instances of trust. If this horn is taken, then the restrictive notion of trust fails
the Explanatory Constraint because trust does not play a role in explaining the emer-
gence of cooperative practices and institutions in such situations. There will be other
contexts where it does play such a role, but these will be limited to situations in which
6
Hobbes claims that Native Americans in his time lived in a state of nature (Leviathan I, 13: Hobbes 1968
[1651]: p. 187). My focus is instead on a context of interaction that lies outside shared practices, norms, and
institutions. I assume that both parties are internally governed by such practices, norms, and institutions.
there is already a background of shared values or a history of interaction that can pro-
vide additional reasons for trust. To accept this horn does not, of course, imply that
there is no explanation of the emergence of cooperative practices and institutions in
terms of dispositions to rely. It just means that trust does not figure in the explanation.
The reason this is a problem is that the underlying process by which cooperative prac-
tices and institutions form seems indifferent to whether the conceptual restrictions are
fulfilled or not. Whether a person with a disposition to rely on another person is opti-
mistic about their competence and goodwill, for example, seems to make no difference
to whether their reliance leads to stable practices of cooperation or to the evolution
and/or adoption of norms and institutions that bolster cooperation. It makes no prac-
tical difference, in the sense of drawing an important distinction between two different
explanatory situations. Hence these restrictions are unmotivated from an explanatory
perspective. There is just one phenomenon here, and it appears to be the broader one
referred to by the unrestrictive view. (We will have reason to revisit this below when we
consider some specific restrictive accounts.)
The second horn is to hold that despite all appearances, people such as La Salle and the
Illinois tribespeople often do trust (in the restrictive sense) in such stark situations, prior
to practices and institutions. Here we can still explain the emergence of practices and
institutions in terms of trust, but only at the cost of holding the implausible view that
people in such unpredictable situations have the special motives that restrictive views
require for trust. The second horn creates two serious problems for restrictive views. The
first is that it is not plausible on the face of it that the Illinois tribespeople, La Salle and his
party, or many others in comparable situations, satisfy the restrictive conditions on trust
(by having optimism about the goodwill of the other, belief in their trustworthiness,
affective expectations of their responsiveness, or moral expectations of them). This is not
how their attitudes are described in the account. Hence the explanation involves an
implausible attitude ascription.
The second problem with the second horn is that even if this attitude ascription
were correct it would require a basic sort of irrationality at the heart of a trust-based
explanation of practices and institutions, because there really is not sufficient reason
to justify such attitudes. Such situations offer only a little by way of moral assurances
or common norms that would ground a strong judgement that the other will be trust-
worthy or take the interests of the other party as intrinsically important. In the
Illinois–La Salle meeting, although the two parties have never encountered one
another before and share almost no common practices and institutions, a means is
found of communicating a sign of reliability: the calumet or ‘peace pipe’, a ceremonial
tobacco pipe the display of which is used as a symbol of non-hostility. Each party pos-
sesses such an item, and according to Parkman’s description it is instrumental in
establishing initial cooperation. This could perhaps be seen as a fragile application of
a social practice to facilitate cooperation. However, although such a symbol may pro-
vide some small evidence of goodwill or trustworthiness or shared values, it does not
constitute an adequate (still less conclusive) epistemic reason for any of these things,
and its use on one very new occasion does not yet constitute a practice here. Admittedly,
each party recognizes that the other is aligned with a larger group of people (the
French settlers, and the various subgroups of the Illinois) who might eventually react
or retaliate based on how the interaction proceeds. Each party also realizes that it may
have many occasions in the future to rely on the other in which it could be useful to
cooperate (a factor emphasized by Hardin (2006)). These facts certainly give some
reason to rely on the other party, but not of the kind that seems to be assumed by the
restrictive views of trust.
This argument does not apply equally to all restrictive views. Different restrictive
views are more or less compatible with the Explanatory Constraint, depending on the
extent to which they require that a trusting person has very particular emotional or
cognitive states. A view like Baier’s, on which the trusting person must be optimistic
about the competence and goodwill of the other, is less restrictive than a view like
McLeod’s (2002) on which the trusting person must be optimistic that the other has
moral integrity.7 In order to consider this matter in more detail, in the next two sec-
tions I consider two restrictive views: Baier’s and Faulkner’s. I will not consider views,
such as Hieronymi’s (2008), that link trust with full belief in the trustworthiness or
reliability of the other, nor will I consider McLeod’s (2002) view. In my view their
requirements are simply too stringent on any reading to satisfy the Explanatory
Constraint for the types of cases we are interested in.
3. Baier’s View and the Explanatory

Rationale for Restrictions
In defence of her own restrictive view of trust, Baier has argued that the Hobbesian
conception of trust, taking interactions between strangers as paradigmatic, ignores the
experiences of trust within family and intimate relationships. She takes the side of
Hume against Hobbes, echoing Hume’s famous point against social contract theory
that there is no such thing as interaction not yet conditioned by the experience of being
raised in a ‘family-society’ (Hume 1998 [1751]: p. 88).
There are two claims here. The first is that interactions between strangers, like the La
Salle–Illinois encounter, are not paradigmatic for trust. Apart from the fact that many
have found these sorts of encounters to be paradigmatic of trust, this first point does
not threaten the idea that there are contexts of interaction where the emergence of
practices and institutions occurs for the first time, not merely as an extension of exist-
ing practices and institutions. We are often interested in explaining the emergence of
7
McLeod situates this claim within a prototype theory of trust on which ‘our concepts are more malle-
able than traditional analytic philosophy makes them out to be’ (2002: p. 14) but where a certain number
of prototypical features must be present in order for something to count as trust to a particular degree. In
such a view there may be room for trust in a wide sense to be socially explanatory.
specific cooperative behaviours and institutions in concrete contexts.8 We can easily

find situations that exemplify this within a family society, for example in questions
over how property of a deceased person should be divided among family members
where custom and the law are unclear or nonexistent.
The second point, more important for us, is that the motives of trust are conditioned
by our genetic and developmental inheritance, including the experiences of trust and
trustworthiness of early childhood. In the context of child development, optimism
about the goodwill and competence of another (often the parent) is normal. What
Baier calls ‘infant trust’, the instinctual reliance of young children on their parents to
care for them, has a kind of psychological priority whose influence endures until adult-
hood and can be expected to influence people’s behaviour in situations such as the
Trust Game.
That infant trust normally does not need to be won but is there unless and until it is destroyed
is important for an understanding of the possibility of trust. Trust is much easier to maintain
than it is to get started and is never hard to destroy. Unless some form of it were innate, and
unless that form could pave the way for new forms, it would appear a miracle that trust ever
occurs. (Baier 1994: p. 107)
Baier counts infant trust as a genuine case of trust even on her restrictive account. She
argues that her account of trust does not require advanced concepts or abilities beyond
what young children already possess: ‘One constraint on an account of trust which
postulates infant trust as its essential seed is that it not make essential to trusting the
use of concepts or abilities which a child cannot be reasonably believed to possess’
(1994: p. 110). We might ask whether Baier’s own account, on which competence and
goodwill are ascribed when trusting, satisfies this constraint. Since accidental failures
are linked with incompetence, and purposeful failures with ill-will, perhaps the most
important sign that one possesses concepts of competence and goodwill is that one is
able to distinguish accidental from purposeful failures. It appears that the ability to
distinguish these two kinds of failures is acquired in the second or third year of life
(see, e.g., Olineck and Poulin-Dubois 2005).
Here, then, is a way that at least some restrictive views of trust can help explain the
emergence of practices and institutions: by identifying a form of trust that exists in
individuals in an early, perhaps simple form in a way that is psychologically prior to
those practices and institutions. This is particularly useful in explaining the emergence
of stable patterns or practices of cooperation. Furthermore, although innate trust in
one’s kin cannot be explained as the outcome of central concerns or interests of the
relevant actors in an agential sense (assuming the infant is not an agent), it can be so
8
Empirically studied examples include the formation of institutions for administering American col-
lege entrance examinations in terms of the needs of elite colleges and universities that began to search for
students from a much broader, national pool (Coleman 1990: pp. 647–8), and the design and implementa-
tion of reputational feedback mechanisms on eBay (Utz, Matzat, and Snijders 2009).
explained in an evolutionary sense.9 Baier’s account does well on this criterion,

although this is not a unique feature of her view: the unrestrictive view, and several
other restrictive views, will also count infant trust as genuine trust.
In order to develop a positive argument for Baier’s view, we need to show that the
conception of trust as optimism about the goodwill and competence of the trusted
draws a distinction that has important explanatory value in its own right. There is a
way of trying to do this. Some psychologists and sociologists draw a distinction
between a disposition based on the predictive expectation of reliability, which they
sometimes call ‘confidence’, and a trusting disposition, which is not a prediction, but
rather an evaluation of the social situation and the person on whom one is in a position
to rely. The reason for this distinction is not primarily that it is an intuitive distinction,
but that it tracks reliance based on two different kinds of information processed
through different psychological channels, forming two measurably different compo-
nents of social cognition. As Midden and Huijts explain the idea, ‘trust becomes the
basis of decisions at the point when other assurances are not sufficiently available and
(experience-based) confidence is lacking’ (2009: p. 744).
From this perspective, what is characteristic of the Trust Game, and indeed of many
of the situations where we want to explain the emergence of cooperative practices and
institutions in terms of trust, is that one does not have a track record of interaction that
gives one access to experience-based information about the prior performance of the
other party (or at least not a track record that is applicable in this new situation). Of
course we can imagine variations of the Trust Game where we do have such informa-
tion, but these will be somewhat uncharacteristic for cases where cooperative practices
and institutions emerge. What we want to know for explanatory purposes is how
people deal with reliance on the other in the absence of any kind of experience-based,
mechanical, or causal certainty—that is, confidence—about how they will act. This
suggests that the Explanatory Constraint may be compatible with drawing a concep-
tual distinction between trust and confidence. Restrictive accounts of trust can satisfy
the Explanatory Constraint insofar as they (a) capture this psychological distinction
between trust and confidence, (b) are not otherwise too restrictive about what counts
as trust, and (c) suggest a distinctive mechanism by which trust brings about coopera-
tive practices and institutions.
The aspect of Baier’s account that best captures this distinction is the idea of opti-
mism, a kind of affective slant to how one perceives opportunities in a situation. Karen
Jones links this with the distinction just introduced:
We can be justified in trusting even when we would not be justified in predicting a favourable
action on the part of the one trusted. Our evidence for trusting need not be as great as the
9
There are limits to this. Failures of infant trust do not do much to explain the design or evolution of
norms and institutions that bolster trustworthiness, in the way that failures of ‘adult’ trust prior to such
norms and institutions can help explain their emergence.
evidence required for a corresponding justified prediction. In this respect trusting is more like
hoping than like predicting. (1996: p. 15, quoted in McGeer 2006: p. 241)
This fits with the idea that we have a distinctive ground for our other-regarding atti-
tudes in situations where the evidence of their reliability is not conclusive.
However, without providing a further basis for such reliance, this story is unsatisfac-
tory for two main reasons. First, it remains unclear why there does not need to be as
much evidence for trust as for a prediction. Even if it is plausible that one is entitled on
rational or moral grounds to trust others, because doing so is a matter of respect or
decency (as in Ross 1986), it is hard to see this alone as an adequate epistemic or
practical reason for doing so. Second, its explanatory value is limited. Optimism is like
a wind that blows in the direction of reliance on others. When this wind is blowing, as
it were, it helps explain why people tend to rely on others more than the evidence may
strictly seem to warrant. However, there are other motivations, more closely connected
with warrant, justification, and practical reason, that bring about reliance even when
the wind is still. These should be part of the account of trust, according to the
Explanatory Constraint. In light of these problems, I will now turn to a view that
articulates the distinctive reasons associated with trust: what I will call the ‘dependence-
responsiveness’ view. This view can be seen, in a way, as grounding the optimism of
trust in a special kind of epistemic and practical reason, and thereby responding to
these two objections.
4. Dependence-Responsiveness Reconsidered
Several philosophers have endorsed as central to trust the idea that the person trusted
will be responsive to the act of reliance (Pettit 1995; McGeer 2006; Faulkner 2007, 2014).
I call this aspect of trust ‘dependence-responsiveness’ (Nickel 2012).10 It can be con-
ceived of as a necessary component of the trust attitude itself, albeit implicit. Suppose
Britta lists Leif as a reference on her job application. In doing so she counts on him to
be responsive to her dependence on him, so that if the hiring committee contacts him
he will provide a suitable recommendation. This responsiveness to reliance is sup-
posed to lie at the heart of trust.
One argument for the centrality of dependence-responsiveness to trust is that it
explains the distinctive wrong of betrayal. Betrayal occurs when a person signals that
he will be suitably responsive to one’s reliance, and then fails to respond when the need
arises. If Leif does nothing to disabuse Britta of the idea that he will provide her with a
timely and supportive reference, and then (intentionally) fails to do so, this is an
instance of a distinctively manipulative wrong. On the assumption that betrayal is the
10
Pettit (1995: p. 203) calls this ‘trust-responsiveness’ but I prefer ‘dependence-responsiveness’ for sev-
eral reasons: the responsiveness doesn’t seem to distinguish between reliance and trust, nor does the
ascription of it to the person on whom one relies. Also, if the point is to define and rationalize trust in terms
of this feature, it seems better not to use trust to define the feature.
distinctive wrong associated with broken trust, this appears to support the link
between trust and dependence-responsiveness (Faulkner 2007).
Elsewhere I have criticized this view of trust on intuitive grounds, arguing that it is
not a necessary condition for trust (Nickel 2012). Although I still maintain this objec-
tion, here I would like to argue that the dependence-responsiveness view does a good
job satisfying the Explanatory Constraint. As argued in the previous section, restrictive
accounts of trust can satisfy the Explanatory Constraint insofar as they (a) capture the
psychological distinction between trust and confidence, (b) are not overly restrictive
about what counts as trust, and (c) suggest a distinctive mechanism by which trust
brings about cooperative practices and institutions. The dependence-responsiveness
view, at least when interpreted broadly, does admirably on all three criteria.
First, it captures the psychological distinction between trust and confidence. The
trust-confidence distinction, broadly speaking, is the distinction between a disposition
towards reliance based on reliable information that the other will behave in a certain
way, or has a certain likelihood of behaving in a certain way, and a disposition towards
reliance where this solid information is lacking, so that one cannot strictly estimate
the reliability of the other’s performance but must place oneself in their hands on
some other basis. Confidence is not essentially an attitude that is directed towards
persons. It could hold of institutions or even physical objects. But it can hold of per-
sons, for example when an institution is set up such that if one particular employee
does not carry out certain tasks for those who rely on them, an appeal can be made to
another employee (a supervisor) who will do so. The function of the institution pro-
vides reliable guidance to one’s expectations, hence this kind of reliance on another is
confidence rather than trust. This leaves unspecified what kinds of grounds might
count as evidence for trust, which is where the dependence-responsiveness account is
useful. Trust, on this view, is distinguished from confidence by the fact that one’s
placing oneself in the hands of another is expected to make the other more reliable
than she would otherwise be. This is clearly not the case in purely bureaucratic insti-
tutions. This makes it clearer what the distinctive ground of trust is, as against other
kinds of reliance on persons.
Second, this distinctive ground for trust, at least when taken broadly enough, is not
too restrictive to be plausibly present in most one-on-one trust relationships and
exchanges. It merely requires that the one relied upon will be responsive to the fact of
reliance and will adjust her behaviour to better meet the expectations of the one who
relies. It is even plausible that something like this is present in the La Salle–Illinois
exchange. The fact that in this exchange it would be appropriate for either side to feel,
in a certain sense, betrayed if the other side suddenly attacked, or poisoned the food
being served, is some indication of this.11 Some expectation of dependence-
responsiveness is compatible with low levels of overall assurance or confidence about
the other party. This means that the dependence-responsiveness account can better
11
It is interesting to consider whether such a feeling of betrayal must imply moral blame.
satisfy the Explanatory Constraint, because it holds of attitudes and interactions prior
to cooperative practices and institutions, and can therefore do something to explain
the emergence of those practices and institutions.
Finally, the dependence-responsiveness account links with a naturalistically well-
grounded mechanism by which cooperative practices and institutions emerge. The
mechanism is particularly strong when the idea of dependence-responsiveness is
coupled with a means of signalling that one is trustworthy. La Salle and the Illinois
each show the calumet to the other as a means of inviting reliance, signalling that if
relied upon not to do violence to the other, they will behave peacefully. Philosophers
with a knack for modelling have used such signalling to explain the success or failure
of cooperative interaction in multi-agent simulations (Skyrms 2010). This explan-
ation can be extended if we include the idea of somebody’s having a visible ‘reputa-
tion’ that is affected by how well they perform. A person who knows that his
reputation will be affected by his performance, such as Leif in relation to Britta, will
tend to be more reliable (Pettit 1995). This is then an extrinsic reason for being
dependence-responsive, which is used in the explanation of the emergence or non-
emergence of cooperation. Signals and reputational information then figure as
important ways of filling in what the distinctive kind of information consists of, on
which trust is based.
In order to have these advantages, however, the theory must be interpreted broadly
enough to count various reasons or motivations, including such extrinsic motivations,
as instances of dependence-responsiveness. Possible dependence-responsive motiv-
ations should include worrying about the possibility of sanctions, trying to preserve
one’s relationship, and perhaps even strategic motives such as the cunning one that
Francesca attributes to George when she plays the Trust Game. Some proponents of
the theory seem to distance themselves from this broad interpretation of dependence-
responsiveness. In a recent paper, Faulkner argues that it is a mistake to see trust as
depending on the kinds of reasons specified within the framework of purely instru-
mental (what he calls ‘Humean’) rationality (Faulkner 2014). The worry is that this
rules out strategic motivations. Faulkner argues that when we trust, we do not para-
digmatically think of the performance of the trusted person as a means to the achieve-
ment of something that we want, such as the continuation of our relationship with
them. At one point, he indicates that in the norm, the trusted person’s reason for per-
formance must match the expectations of the person who trusts: ‘A trusted party S is
trustworthy, in a circumstance defined by A’s (affectively) trusting S to φ, if and only if
S sees A’s depending on his, S’s, φ-ing as a reason to φ and φs for this reason’ (Faulkner
2014: p. 1982). In order to reap the explanatory advantages of the dependence-
responsiveness account, we must take seriously that this further specification (like
the rational choice norm in Coleman’s account) is not intended to settle the question
of what counts as trust; and thereby allow that in some cases the trusting person sim-
ply leaves it open to the kinds of reasons to which she expects the trusted person to be
responsive.
Some other views of trust can meet the Explanatory Constraint in a similar way. The
theory of trust I favour is one on which a person who trusts normatively expects the
trusted party to perform a certain way, where this does not require dependence-
responsiveness in every case. The view differs subtly from dependence-responsiveness
in that the norm can—and must, sometimes—be applied even in circumstances where
it is not expected to have a knock-on effect (Nickel 2012). Correspondingly, the view
does not require that the trusted person responds to the very norm of behaviour on
which the trustor’s normative expectation is based. However, it allows for that kind of
dependence-responsiveness as a normal case. This has explanatory potential insofar
as the idea of ascribing and applying norms to others is a fundamental mechanism
among humans for changing the balance of reasons and improving compliance with
expectations in cooperative exchanges; and also because when these normative
expectations fail, we sometimes take other steps to encourage better compliance in
the future, such as the establishment of sanctions and institutions.
5. The Concept of Trust: Natural, Social, or Political?

The conclusion of the preceding sections is that some (less) restrictive theories of
trust are favoured by the Explanatory Constraint on non-intuitive grounds. What
does this show? At best it shows that a theory of trust is better if it allows for explana-
tory aims, not that this determines the referent of the ‘folk’ concept of trust, which is
presumably also the one that philosophers investigate via intuitions, phenomenology,
and/or rational reconstruction. Additional argument is needed to link the explana-
tory aims with the concept of trust with which philosophers have been concerned,
showing that this concept, contrary to appearances, is settled by the Explanatory
Constraint rather than by these other methods. In this section I consider a possible
argument from scientific essentialism. I claim that this argument is not decisive, and
that the explanatory advantages of a less restrictive account of trust must be balanced
against other factors.
A preliminary move in trying to establish that scientific aims might determine the
concept of trust is to relativize philosophical intuitions. The folk concept of trust is
messy. It is common enough to speak of domesticated animals trusting and being
trusted, and also to speak of people trusting computer systems, elements of the built
environment, and even natural phenomena like the weather. Philosophers often
respond to this fact by drawing an intuitive distinction between trust and ‘mere reli-
ance’, or between two kinds of trust (Faulkner 2014). However, without further argu-
ment of the kinds we have offered in the previous sections, the appeal to intuition begs
the question against the unrestrictive view. To emphasize this point, it is useful to point
out that the folk concept, and in particular the distinction between trust and reliance,
do not translate identically into different natural languages. Some languages do not
draw as sharp a distinction.12 Our intuitive sense of what genuine trust is, is perhaps
not shared universally.
This prepares the way for a further argument for the claim that our concept is deter-
mined pragmatically by our scientific explanatory aims, rather than by our intuitions.
According to anti-individualism about thoughts, the identity of thoughts embedding
certain concepts is not fully determined by that to which the thinker has subjective
access. Instead, which concept one is thinking about is partly determined by external
facts to which the thinker might have imperfect access: either facts about scientific
essences or natural kinds which the concepts pick out, or facts about the social
determination of the concepts (Burge 2007). For example, a person thinking about
arthritis is thinking about a rheumatic disease of the joints. Even if her subjective grasp
is insufficient for her to have an opinion about whether arthritis can occur in the thigh,
she can nonetheless have thoughts that embed this concept of arthritis, according to
which its occurrence in the thigh is excluded. There are two grounds for explaining
this fact: an essentialist ground and a social ground. Arthritis may be a natural kind
with an essence, or it may have been settled by the community of scientists who work
on arthritis that arthritis only concerns the joints (although perhaps it might have
been settled differently). In the case of arthritis, it is not obvious which factor plays the
decisive role: is it that medicine has been given the task of defining medical concepts,
or the fact that joint disease is a natural kind?
One might think that philosophical concepts are not susceptible to anti-individualist
argument. But in fact we find both the essentialist and the social variant of the argu-
ment applied by some philosophers to the concept of knowledge. Hilary Kornblith
(2014) argues for the claim that knowledge is a natural kind, and that intuition-based
investigations of the concept of knowledge are poor guides to the concept. According
to Kornblith, the idea of a representational state that correctly corresponds to some
external state of affairs has such great explanatory value in describing animal behav-
iour, and evolutionary success, that it is an essential biological concept:
When we seek to explain the presence of certain cognitive capacities in a species . . . knowledge
enters the picture. The environment makes certain informational demands on a species, and
cognitive capacities answer to those demands. . . . if we want to know why some individual has
a certain cognitive capacity, we will need to advert to the evolutionary explanation . . . The cat-
egory of beliefs for which these capacities were selected—reliably produced beliefs that are
also true—is thus important to ethologists. As I see it, the explanatory importance of the cat-
egory and its theoretical unity provide reasons for viewing it as a natural kind.
(Kornblith 2014: pp. 176–7)
12
In Dutch, ‘I rely on x’ and ‘I trust x’ can both be naturally translated as ik vertrouw [in, op] x, and ‘reli-
ability’ and ‘trustworthiness’ can both be naturally translated as betrouwbaarheid. In French, the respective
pairs are both translated as faire confiance en and fiable. Of course a distinction can be drawn, but it lies
further from the surface than in English.
The most straightforward version of an anti-individualist argument applied to trust

would claim that trust is a natural kind. Our thoughts about it are externally deter-
mined in the same way that our thoughts about water are externally determined by the
substance H2O. In the case of water, even before modern chemistry described the sub-
stance, H2O was what we were thinking about. Those who have theories of water based
on intuitions about their experiences of water, perhaps even denying the existence of
H2O, are nonetheless thinking about H2O.
Is there a case to be made that trust is a natural kind like this, and that intuitions are
a poor guide to the concept? Philosophers since Aristotle have made serious com-
parisons between the cooperative, social behaviours of humans and other animals.
The social disposition of humans to rely on other members of their species might be
considered a natural kind, particularly if we understand this disposition in the sim-
plest possible way, perhaps in the style of Baier’s ‘infant trust’, so that other social ani-
mals partake of it. Baier makes the point that infant trust does not even require that
the child has a fully developed capacity of choice. The ability to distinguish inten-
tional from accidental non-reliability is also shared to a degree by non-human ani-
mals. What is distinctive about human trust, on this view, is that unlike other animals
we are also able to engineer the environment by creating complex signals, social
norms, and institutions that change the balance of motivation for those relied upon in
situations of uncertainty, rendering social interaction within societies more stable
and reliable.
However, it appears from the argument of the previous sections that a cognitively
richer view of trust may also offer a compelling explanation of cooperative prac-
tices and institutions as a social phenomenon. The dependence-responsiveness view
requires that the trustor attribute a capacity of choice or decision to the one trusted. On
this view, the trustor expects the trusted to be aware of the trustor’s reliance and, in
virtue of this, to become more responsive in fulfilling their expectations. This is a com-
plex attribution of intention and choice, requiring a complex cognitive apparatus not
likely to be shared by many other animals. Although such a trust disposition might
also conceivably be a natural kind, the style of argument that Kornblith deploys to
show this in the case of knowledge would not apply. His argument turns on the claim
that knowledge is a category that we need at many points to explain evolutionary suc-
cess across many species. If trust as dependence-responsiveness is a natural kind, it is
anthropogenic.13
Burge’s original argument for anti-individualism emphasizes the idea that the con-
cept embedded in our thoughts is a matter of use within our language community, the
‘social environment’, rather than Kornblith’s idea that natural kinds directly fix the
13
Kornblith appears to limit himself to biological kinds when arguing that knowledge is a natural kind.
When criticizing Edward Craig’s (1990) view that knowledge is a concept that humans (including early
humans) needed in order to identify good sources of information, he never considers the possibility that
an anthropogenic kind could also be a natural kind (Kornblith 2014: Ch. 12).
contents of our thoughts.14 At least in some cases, it is because of the authority socially
accorded to scientists that scientific explanatory values determine which concept is
embedded in our thoughts. This point does not require that there be extralinguistic,
metaphysical truths involving biological or other kinds. It is only required that the
actual concept of arthritis, for example, is fixed the way it is because scientists have
found this concept useful, and because the broader linguistic community gives them
authority over the concept. The fact that the authority traces back to the community
does not imply that an individual can deploy an idiosyncratic, non-standard concept,
any more than the fact that laws trace their authority back to the community implies
that a person can opt for an alternative law.
As applied to trust, however, there are two serious problems with this idea. The first
is that there is little interdisciplinary agreement about trust. That medical scientists’
understanding of arthritis should determine the boundaries of the concept is plaus-
ible, but it is not similarly plausible that biologists, rather than psychologists, political
scientists, economists, or philosophers, should determine the boundaries of the con-
cept of trust. The second problem is that trust is a concept with heavy social and polit-
ical significance in modern times (see, e.g. Baberowski 2014). Concepts like this, such
as legitimacy and democracy, are public property, maybe even contested territory. The
boundaries of such concepts are not dictated by experts alone. It appears, then, that
there is no decisive reason to regard the concept of trust as being determined by a nat-
ural kind, nor by social facts that give experts authority over the concept. Perhaps
philosophers even have a special role in mediating public and expert discourse about
trust, giving them a kind of special interpretive authority over the concept. This sug-
gests that the argument from explanatory potential is not a knock-down argument, but
that it must be taken into account alongside other considerations.
6. Conclusion
Philosophical accounts of trust based on intuitions and the phenomenology of trust
have strong prima facie appeal, and the argument from explanatory potential does not
bypass our ordinary intuitions in the way that an analogous argument might in the
case of the concept of arthritis or H2O. My conclusion is pragmatic: philosophers
should take the argument from explanatory potential on board as a way of grading
accounts of trust. Other things equal, it is better to have minimal restrictions on
motives, to interpret these restrictions psychologically in such a way that permits them
to be ascribed broadly to many agents in many situations, and to mark genuinely
explanatory distinctions with the concept, identifying features that make a difference
to cooperative behaviour and institutions. I have argued here that the argument from
14
In a postscript to ‘Individualism and the Mental’, Burge stresses that the argument in the original
paper concerns the role of the social environment in fixing our concepts, but states that on his current view
both the social and the physical environment play a role in fixing the concept (Burge 2007: p. 152).
explanatory potential may favour a view of trust on which it is strongly linked with
dependence-responsiveness: the feature that when one person trusts another person,
the first person assumes that the second person, aware of the reliance of the first, will
(be more likely to) choose to be reliable. In addition, I have argued that this notion of
dependence-responsiveness should be interpreted broadly to include responsiveness
to reputational considerations and other strategic motives.15
References
Baberowski, J., ed., 2014. Was ist Vertrauen? Ein interdisziplinäres Gespräch. Frankfurt: Campus.
Baier, A., 1994. Moral Prejudices: Essays on Ethics. Cambridge, MA: Harvard University Press.
Berg, J., J. Dickhaut, and K. McCabe, 1995. Trust, Reciprocity, and Social-History. Games and
Economic Behavior 10(1): pp. 122–42.
Buchak, L., 2016. Decision Theory. In A. Hájek and C. Hitchcock, eds., Oxford Handbook of
Probability and Philosophy. New York: Oxford University Press.
Burge, T., 2007. Foundations of Mind: Philosophical Essays, Volume 2. New York: Oxford University
Press.
Camerer, C. and K. Weigelt, 1988. Experimental Tests of a Sequential Reputation Model.
Econometrica 56(1): pp. 1–36.
Coleman, J. S., 1990. Foundations of Social Theory. Cambridge, MA: Harvard University Press.
Cook, K. S., R. Hardin, and M. Levi, 2005. Cooperation without Trust. New York: Russell Sage
Foundation.
Craig, E., 1990. Knowledge and the State of Nature. New York: Oxford University Press.
Faulkner, P., 2007. On Telling and Trusting. Mind 116: pp. 875–902.
Faulkner, P., 2014. The Practical Rationality of Trust. Synthese 191: pp. 1975–89.
Hardin, R., 2006. Trust. Cambridge: Polity Press.
Hieronymi, P., 2008. The Reasons of Trust. Australasian Journal of Philosophy 86: pp. 213–36.
Hobbes, T., 1968 [1651]. Leviathan. Ed. C. B. Macpherson. New York: Penguin Books.
Hume, D., 1998 [1751]. An Enquiry Concerning the Principles of Morals. Ed. T. L. Beauchamp.
New York: Oxford University Press.
Johnson, N. D. and A. A. Mislin, 2011. Trust Games: A Meta-Analysis. Journal of Economic
Psychology 32(5): pp. 865–89.
Jones, K., 1996. Trust as an Affective Attitude. Ethics 107: pp. 4–25.
Kornblith, H., 2014. A Naturalistic Epistemology: Selected Papers. New York: Oxford University
Press.
Kreps, D. M., 1990. Corporate Culture and Economic Theory. In J. E. Alt and K. A. Shepsle, eds.
Perspectives on Positive Political Economy. Cambridge: Cambridge University Press: pp. 90–143.
McGeer, V., 2006. Trust, Hope and Empowerment. Australian Journal of Philosophy 86(2):
pp. 237–54.
McLeod, C., 2002. Self-Trust and Reproductive Autonomy. Cambridge, MA: MIT Press.
15
Thanks to the editors of this volume, and to participants at an Oxford workshop in 2014, for com-
ments on earlier versions of this chapter. This research has been funded in part by a grant from the Socially
Responsible Innovation (MVI) programme of the NWO, for the project ‘Medical Trust Beyond Clinical
Walls’.
Midden, C. J. H. and N. M. A. Huijts, 2009. The Role of Trust in the Affective Evaluation of
Novel Risks: The Case of CO2 Storage. Risk Analysis 29(5): pp. 743–51.
Nickel, P., 2007. Trust and Obligation-Ascription. Ethical Theory and Moral Practice 10:
pp. 309–19.
Nickel, P., 2009. Trust, Staking, and Expectations. Journal for the Theory of Social Behaviour 39:
pp. 345–62.
Nickel, P., 2012. Trust and Testimony. Pacific Philosophical Quarterly 93(3): pp. 301–16.
Olineck, K. M. and D. Poulin-Dubois, 2005. Infants’ Ability to Distinguish Between Intentional
and Accidental Actions and Its Relation to Internal State Language. Infancy 8(1): pp. 91–100.
Parkman, F., 1983 [1878]. La Salle and the Discovery of the Great West. In Francis Parkman:
France and England in North America: volume I. Ed. D. Levin. New York: Library of America:
pp. 713–1054.
Pettit, P., 1995. The Cunning of Trust. Philosophy and Public Affairs 24: pp. 202–25.
Ross, A., 1986. Why Do We Believe What We Are Told? Ratio 1: pp. 69–88.
Simpson, T., 2012. What is Trust? Pacific Philosophical Quarterly 93(4): pp. 551–69.
Skyrms, B., 2010. Signals: Evolution, Learning and Information. New York: Oxford University
Press.
Utz, S., U. Matzat, and C. Snijders, 2009. On-Line Reputation Systems: The Effects of Feedback
Comments and Reactions on Building and Rebuilding Trust in On-Line Auctions.
International Journal of Electronic Commerce 13(3): pp. 95–118.
Walker, M. U., 2006. Moral Repair: Reconstructing Moral Relations After Wrongdoing. Cambridge:
13
Trusting a Promise and
Other Things
David Owens
We are social animals and the form of our social life depends on features of our psych-
ology. Some social theorists fix on a specific attitude as being crucial. For Hume the
psychological cement of human society was sympathy; for Hobbes it was fear. For
many recent writers, both popular and philosophical, trust plays this foundational
role. According to the latter there is a special attitude we adopt towards other people
when we trust them and that attitude is a sine qua non of our social institutions. So
long as we trust one another, basic forms of cooperation will survive; where that atti-
tude weakens or decays, things quickly fall apart.
I agree that society could not exist in its present form unless people both trusted one
another and showed themselves worthy of that trust. Ubiquitous distrust would indeed
be a social solvent (Reid 2010: 334–5). What I doubt is that there is a distinctive attitude
of trust that serves as a psychological underpinning of our sociality; rather different
forms of trust are appropriate to different sorts of object. All of these phenomena may
be called forms of trust but to uncover their psychological character we must look not to
the nature of trust (or distrust) but to the disparate character of the objects of trust.
1. A Thin Theory of Trust

Richard Holton maintains that much as we believe or intend that p, we may also trust
that p:
Trust, I will suggest, is a distinctive kind of attitude involving a distinctive state of mind. My
project is to look at the ways in which it is distinctive, and the ways in which it interacts with
belief and with the will. (Holton 1994: 63)
For Holton trust, like belief and intention, is a propositional attitude with a specific
psychological nature.1 Given the ubiquity of trust, were Holton’s assumption correct,
1
‘Trust is an important and distinctive kind of psychological state’ (McMyler 2011: 113).
trusting a promise and other things 215
we would expect the psychology of trust to be a fundamental element of our social

theory and we’d also expect the value of trust to ground the authority of various social
norms including those that require us to keep our promises.
Annette Baier shares these expectations and in a paper published some thirty years
ago she expressed surprise at the ‘strange silence’ to be found on the topic of trust:
It seems fairly obvious that any form of cooperative activity, including the division of labor,
requires the cooperators to trust one another to do their bit. . . . But when we turn to the great
moral philosophers, in our tradition, what we find can scarcely be said to be even a sketch of a
moral theory of trust. (Baier 1986: 232)2
More recent writers have sought to fill the gap, proposing accounts of the general
nature and value of interpersonal trust. To complete their project these writers need
not offer us an analysis of trust nor decompose the attitude of trust into other attitudes
like belief and desire; trust may be indefinable, a psychological primitive. But if trust is
a propositional attitude with a definite psychological character then it must be possible
to answer some basic questions about it.
First, what kind of thing is trust? Is it in fact a propositional attitude or is it more like
a speech act? If an attitude is it cognitive (like belief) or conative (like intention) or
perhaps something in between like an emotion (Jones 1996)? I’ll call this the modal
question. Second, what is the characteristic content of trust? What must the trusted
object be like for your trust in it not to be misplaced or mistaken? In the context of the
recent philosophical literature on trust this question of content becomes what I’ll call
the motivational question. Third, what kinds of considerations justify trust? Is trust
justified by evidence for the truth of some proposition or by considerations of a rather
different sort or by some combination of the two? This is the question of justification.
All of these questions have been raised in the literature on trust and none have been
resolved to general satisfaction. Faced with this impasse, one may wonder whether ‘the
great moral philosophers’ were not wise to refuse trust a foundational role in their eth-
ical and social theories.
I maintain there is no such thing as the value of trust. Rather various objects of trust
have their distinctive form of value and for X to trust Y is for X to engage Y in a way that
will realize the value distinctive of Y-things (provided this thing is a good Y). Call this a
thin theory of trust. On this view, trust has no distinctive value; rather its role is simply
to realize the value of the object of trust.3 Furthermore, trust has no distinctive psycho-
logical nature for the attitudes that would realize the value of the various objects of
trust are themselves very various. In some cases the only propositional attitude
2
Most of what we do find is discussion of trust in a promise and, as Baier rightly says, that is just one
form of trust.
3
Contrast trust with pleasure. Even if the value of enjoyment is conditional on the value of the object
enjoyed, there is a distinctive value in enjoying things: we should seek out (worthy) things to enjoy so that
we have enough pleasure in our lives. We shouldn’t in the same way seek out things to trust so that we have
enough trust in our lives. The point of trust is simply to realize the value of the trusted thing, not to
enhance that value by making us feel trusting towards it.
216 David Owens
involved in trusting Y may be an intention to behave towards Y in a certain way, though

in other cases some further attitude (perhaps some motive for forming the relevant
intention) might be needed to realize Y’s distinctive value. And we can acknowledge all
this without multiplying senses of ‘trust’. For X to trust Y is always for X to engage Y in
a way that satisfies the above specification.
Let’s leave the realm of interpersonal trust for a moment. I trust my car when I am
prepared to drive it around (not merely sit in it), an attitude that enables me to realize
the distinctive value of a car. I trust an apple when I am prepared to eat it (rather than
use it as a football), an attitude that enables me to realize the distinctive value of an
apple. A specific apple and a specific car are worthy of this trust when they actually
possess the value distinctive of their kind.4 In the absence of trust, the value of a (trust-
worthy) apple and of a (trustworthy) car will be wasted. There is some leeway and a
potential source of disagreement in the notion of a thing’s ‘distinctive value’: am I trust-
ing my apples if I’m only prepared to use them as compost? I shall offer no account of
what might make edibility the distinctive value of an apple. I note only that disputes on
this score are not to be settled by investigating the nature or value of the attitude of
trust; rather it is the nature and value of apples that requires our attention.5
Trust is one of a range of psychological phenomena (including respect and appreci-
ation6) the function of which is to realize the value of their object and whose nature
and value is consequently as various as the nature and value of their objects. My main
concern is to place trust in this class—it is less important to me how exactly we distin-
guish trust from its other members—but I’ll say something tentative about what differ-
entiates trust from respect, appreciation, and so forth.
One can realize the value of something in various ways without engaging that value
in the sense I have in mind. For example, I might show respect for the value of the life of
Joe the Hermit by deliberately avoiding any contact with Joe and I might show my
appreciation of the value of a joke by laughing at it, or of a sunset by looking at it.7 Here
my attitudes of respect and appreciation help realize the value of their objects, a value
which might otherwise be wasted but it would be strange to describe me as trusting the
hermit, the sunset, or the joke. Trust implies a more positive engagement with its
object, involving the expenditure of energy or one’s undergoing a significant change
that risks being misdirected unless the trust is well founded. Activity is required to
realize the distinctive value of food, of machinery, of books, and so forth; even simply
believing what someone tells you involves undergoing a complex psychological trans-
formation with various emotional and behavioural ramifications. Boundaries are hard
4
Trusting Y may or may not involve representing Y as trustworthy. That will also depend on whether
such an attitude is required to realize the distinctive value of a trustworthy Y.
5
There may be objects that are valued by us but which have no distinctive value; they are just valued in
different ways by different people. Perhaps the moon is such an object, valued by some as a celestial adorn-
ment and by others for its gravitational pull. In that case there would be no such thing as trusting the moon.
We may trust that tonight will bring a full moon but we won’t thereby be trusting the moon.
6
Raz treats respect and appreciation as value-realizers (Raz 2001: 154–7).
7
Thanks to Alison Hills for the hermit and Jessica Moss for the joke and the sunset.
to draw—if I strain to hear your jokes then at some point I am trusting your sense of
humour, not merely appreciating it—but the point remains that trust is like respect
and appreciation in that it realizes the value of its object.
How is this conception of trust to be applied to interpersonal trust? It is doubtful
whether people as such have a distinctive value, one that can be realized by trust.
Some might conclude from this that trust in things and trust in people are just two
different phenomena. I instead propose that trust in a person be understood as short-
hand for trust in something that a person is or does.8 We think of various actions and
character traits as having a distinctive value and this value is often wasted unless we
respond to them with trust. You realize the value of my promises and my assertions by
trusting them.9 Character traits like good judgement, technical skills, and aspects of
temperament (e.g. equanimity, self-confidence, self-discipline) are a similar case. You
benefit from my driving skills or my courage (in the relevant way) by trusting them.10
Here I’ll be focusing on the relation ‘X trusts Y’. There are other idioms like ‘X trusts
that Y will A’ which are often used to mean little more than ‘X is relying on the fact that
Y will do A’. For example trusting a promise is a different matter from trusting that a
promise will be kept (perhaps because the police will force the promisor to keep it).
One trusts that a promise will be kept whenever one extracts some value or other from
the promise by relying on its being kept. To trust the promise (and thus the promisor)
is, I shall propose, to engage with it in a way that realizes its distinctive value qua prom-
ise, a trust that (as we shall see) involves more (and perhaps also less) than simply rely-
ing on its being kept.11
I’ll offer no general account of what makes V the distinctive value of Y but I shall say
something about those objects of trust which (unlike apples) are governed by norms.
Such an object’s distinctive value explains the norms that govern it, so the distinctive
value of a speech act (like a promise or an assertion) is the value which accounts for the
characteristic content of the norms governing the speech act (e.g. those norms that
distinguish promises from assertions and vice versa) and which explains why we
should take those norms seriously.12 For example, suppose the characteristic point of
an assertion is to provide its audience with a way of knowing the proposition asserted.
That will explain why assertions should be made on the basis of evidence for the
8
So you want to be able to trust those around you because you want to be able to trust their assertions,
their promises, their judgement, and so forth.
9
You may be prepared to trust someone’s promises but not their assertions; their honesty but not their
courage. My account suggests why a refusal to trust people is often problematic.
10
To maliciously expose me to danger is to exploit my courage rather than to trust it because it is to use
my courage in a way that does not realize the distinctive value of courage. Similarly for exploiting my good-
will to trick me and the like.
11
Thus the distinction between trust and reliance that figures so prominently in the philosophical litera-
ture is a distinction with application well beyond the interpersonal realm. Theories of trust that distinguish
trust from reliance by giving an essential role to interpersonal attitudes like goodwill or resentment are
being too restrictive.
12
For applications of this methodology to promises with rather different results, see Scanlon (2003:
282–3) and Owens (2012: Chapter 6).
218 David Owens
roposition asserted and it also explains why you can trust an assertion only by believ-
p
ing it, for you can learn the truth of what you are being told only by believing it.
In this chapter I focus on promises.13 Some writers have tried to explain the nature
and value of a promise by treating a promise as ‘an invitation to trust’ the promisor.
On this view, the normative significance of a promise depends on the normative sig-
nificance of the attitude of trust that it invites.14 Though a normal promise may well
involve an invitation to trust, I doubt this observation explains the binding force of a
promise.15 To account for that we must move in the opposite direction: first identify
the distinctive value of a promise and then deduce what trust in a promise must
involve. We can agree that promises generally invite trust (and would be pointless
unless they were sometimes trusted) without seeking to explain the point of a prom-
ise by reference to some promise-independent notion of trust, for it is impossible to
understand what trust in a promise amounts to other than by reference to the dis-
tinctive value of a promise.
In the philosophical literature we find no consensus about what is involved in trust-
ing either a promise or an assertion and I suspect these disagreements can be traced to
differing views of the distinctive value these speech acts.16 I shall rehearse several
hypotheses about the distinctive value of promise but I’m not out to decide between
them, only to use them to explain disagreements about trust in a promise. It may be
that one hypothesis is correct and the others are wrong. It may also be that promises
have more than one form of distinctive value and so there is more than one way of
really trusting them. The point remains: our understanding of the nature and value of
trust in a promise derives from our understanding of the distinctive value of a promise
rather than that of trust.
2. Trust in a Promise: Social Coordination

A promise is made when the promisor communicates the intention of hereby under-
taking an obligation to perform, a mouthful I’ll abbreviate by saying that promises are
13
Owens (2012) offers a theory of promissory obligation that makes no mention of trust in a promise.
In this chapter I’m seeking to repair that omission whilst also giving the reason for it.
14
For two accounts along these lines, see Friedrich and Southwood (2009: 282) and Pink (2009: 408).
The explanatory burden here is thrown on the notion of trust and so one looks for an account of the nature
of interpersonal trust that can reveal the basis of a specifically promissory obligation. Friedrich and
Southwood (2009: 280) tell us that trust in a promise involves ‘a certain faith or optimism in the promisor’s
character that the promisor will perform some action that is of importance to the promisee’. This formula-
tion raises all of the questions to be addressed in what follows: Does ‘faith or optimism’ involve belief in or
reliance on future performance? Can it be based solely on evidence of future performance? And what
aspects of the promisor’s character and motivational psychology constitute suitable objects of trust?
15
There are (binding) promises which do not invite trust in themselves but this involves adopting a per-
verse attitude to the value of the promise. See Owens (2012: 201–2).
16
Were there a consensus both about the value of a promise and about the value of an assertion, this
methodological point could instead be made by contrasting what is involved in trusting a promise with
what is involved in trusting an assertion.
made by declaration. The promisor’s declared intention to bind themselves is not

enough to create the promissory bond, for the potential promisee must both under-
stand what is happening and react to the promisor’s offer—they must communicate
the intention of hereby accepting the offer (Reid 2010: 336). In contexts where there is
a default assumption that such offers will be accepted, silence is enough to effect this
further declaration. In other contexts, the promisee may need to be more explicit.
None of these claims about promise are uncontroversial, but for my purposes their
truth is not crucial. I am using them to illustrate a certain methodological moral
about the connection between the value of a promise and trust in a promise, a moral
that does not depend on the particular model of promissory obligation we adopt.17
Turning now to trust in a promise, this also requires some form of uptake on the part
of the promisee.18 There are at least three possibilities:
(1) Trust is Acceptance: You trust a promise just by accepting it (a speech act),
thereby ensuring that it binds the promisor.
(2) Trust is Reliance: You trust a promise by relying on the promise, by (at least)
intending to behave as if it will be kept.
(3) Trust is Expectation: You trust a promise where you believe that it will be kept.
These three forms of uptake, though distinct, often become entwined in practice but
my present interest is in the differences between them.
My brother is in a generous mood and promises me an expensive birthday present.
I wouldn’t mind the present so I accept his offer whilst strongly suspecting that the
present will fail to materialize. Here we have acceptance without either reliance on or
expectation of performance. Acceptance can also be joined to the former without the
latter. Though much of the time I rely on promises because I believe they will be kept,
I sometimes have reason to rely on someone’s promise even when I’m far from con-
vinced that they will perform. Perhaps there is some reason to think they’ll come
through, enough to justify behaving as if they will (the stakes are low) but not enough
to convince me that they will. Perhaps I have some reason to rely on their promise
other than my conviction that they’ll perform (e.g. as a display of confidence in them).
Here we have reliance without expectation of performance.19
So what is involved in trusting a promise: must you believe that the promisor will
perform, must you at least be prepared to rely on their promise or is it enough to simply
accept their promise, at least where acceptance alone ensures that the promise binds?
17
These assumptions about promissory obligation are in fact quite widely shared and are defended in
Owens (2012). An expectations theorist like Scanlon would reject them but I believe the expectations the-
ory could also be used to illustrate our methodological morals.
18
Austin says that an illocutionary act like a promise secures uptake whenever it brings about ‘the
understanding of the meaning and of the force of the locution’ (Austin 1975: 117). (1)–(3) all involve
uptake in that minimal sense but all three go beyond it.
19
Perhaps one can also expect performance without being willing to rely. When the stakes are high
enough I don’t feel I can expose myself to the risk of your not performing even though I do believe that you
will perform (Jones 1996: 24).
220 David Owens
This is the modal question, it being a question about the modality of trust.20 Is trust a
speech act or an attitude and, if the latter, is that attitude cognitive like belief or cona-
tive like intention? On my methodological hypothesis, we can address this issue (and
others that will soon arise) only by asking ourselves what the distinctive value of a
promise is.
In the eyes of many (Hume for example) the function of a promise is to facilitate
social coordination by giving us all a way of getting other people to rely on us. I want
help with my harvest and so I promise to help you with yours if you help me with mine
(Hume 1978: 520–1). According to Hume’s social coordination hypothesis, the key to
realizing the coordinative value of a promise and thus to trust in a promise is reliance.
Provided both parties rely on the compact between us, social coordination will be
achieved regardless of whether each believes the other will perform. On this view mere
acceptance need involve no trust in a promise since it need involve no tendency to act
in reliance on the promise and so no tendency to coordinate my behaviour with the
promisor’s in a beneficial fashion. I may accept my brother’s promise of a fine birthday
present but I do not trust it. Thus the modal question is resolved: to trust a promise is to
rely on it (i.e. to adopt a certain policy towards it) and I can rely on you to help me with
my harvest without being convinced that you will. Here I trust your promise because
I (intend to) behave in a way that realizes the distinctive coordinative value of the
promise. Provided you adopt the same attitude to my promise, successful coordination
will be achieved.21
Some writers do not agree that trust in a promise turns on reliance. They complain
that simply to behave as if someone will perform without believing that they will is not
really to trust their promise. On the other hand we do say ‘You’ll have to trust me, you
have no alternative’ or ‘Being unsure whether you’d do it, I just had to trust you’ and so
forth. One might close this discussion down by multiplying senses of ‘trust’ or else
conclude that there is no real disagreement here because a promise can be used either
as an object of reliance or as a source of knowledge about what others are going to do
and so trust in a promise can take more than one form. Each of these responses con-
cedes that the dispute is verbal but, if the social coordination hypothesis is correct, we
can resolve the matter in a theoretically motivated fashion: reliance is the attitude that
realizes the distinctive value of a promise.
Let’s now turn to a second question about the content of the attitude of trust. When
trust in a promise is at issue, this becomes a question about what the motivational
psychology of the trusted party must be like for the trust in them to be well placed;
20
For opposing views on this question, see Holton (1994: 68–9), Hieronymi (2008: 216–19 and 227–31)
and McMyler (2011: 131–41). On the whole these authors (and those cited in the notes that follow) do not
focus on promises in particular. Rather they make general claims about the nature of trust, claims that I am
applying to the case of trust in a promise.
21
The possibility of relying on a promise without being convinced that it will be fulfilled ensures that we
can decide to trust a promise; ‘I wasn’t sure you’d perform but I decided to trust you anyway’ makes perfect
sense where trust involves only reliance. By contrast it is doubtful whether we can decide to believe that the
promisor will perform.
hence I dub this the motivational question. Promises are usually extracted where p eople
feel they can’t otherwise rely on the promisor to do the thing they are promising to do.
If you know that the guy will mow your lawn anyway out of personal affection or
because he loves mowing lawns, you probably won’t seek a promise from him: promise
is redundant, at least as a device of social coordination. Given that the promisor binds
themselves by declaration, it is natural to assume that the function of a promise is to
achieve social coordination in a special way, namely by imposing an obligation to per-
form on the promisor (Hume 1978: 518). If so, this fact should be reflected in our
account of what it is to trust a promise, of how one must react to the promise in order to
realize its distinctive value.
Here is one way of answering our motivational question: to trust a promise involves
relying on the promise because one assumes that the promisor knows they are obliged
to perform and is more likely to do so in the light of this fact. To trust a promise is not
just to rely on the promise’s being fulfilled for whatever reason; it is to rely on the prom-
isor’s conscientiousness in particular, making the assumption that conscientiousness
is available at least as a back-up motive.22 If one thinks the promisor will perform only
from some unrelated motive like fear or favour, one is not trusting their promise (as
opposed to trusting that they will do what they promised to do) because one is not
reacting to the promise in a way that realizes its distinctive value, namely as an indica-
tor of dutiful performance. And when the promisor breaches their promise they are
not merely letting you down (like your car), they are betraying your trust by disregard-
ing their obligations towards you.23
Not everyone agrees with Hume that the trusting promisee trusts the promisor to
keep their promise out of fidelity to their promise. Some writers employ an obligation-
independent notion of trust and argue that a promise involves an invitation to trust in
that sense. On this conception of trust, neither giving one’s own word nor accepting
the word of another need involve thoughts of obligation. True, a promise usually binds
the promisor to performance but on this view the obligation to keep a promise is a by-
product of the relationship of trust:
The question when do gratuitous promises oblige comes down to this: under what conditions
does inviting someone to trust you to do something oblige you to do it? And restating the
22
Suppose the promisor regards their promise as a kind of vow—one they owe it to themselves to
keep—rather than as a commitment, breach of which would wrong the promisee. Could the promisee trust
this promise on the grounds that the promisor feels obliged to keep it?
23
Pink observes that when someone breaches a promise, the promisee doesn’t just resent being deprived
of what they are owed, they also feel hurt at having been ‘trifled with’ (Pink 2009: 393 and 412). I resent it
when, on a train, I leave some valuable exposed whilst using the restroom and it is gone when I return.
Here I trusted my fellow passengers and my trust was betrayed but since none of them invited my trust,
they did not abuse my trust and Pink’s special sense of hurt is out of place (O’Neill 2012: 318–25). My trust
would have been abused had one of them promised to take care of the valuable for me. A normal promise
involves an invitation to trust and so breach of a promise normally involves an abuse of trust. This is so
even when the promisee places no reliance on the promise provided acceptance alone can be a form of trust
(see Section 3).
222 David Owens
question in these terms really is explanatory. . . . For the idea of trustworthiness gives us inde-
pendent purchase on when gratuitous promises give rise to obligations and when they do not.
(Pink 2009: 411)
So what is this obligation-independent notion of trust?

It has been said that trusting someone to do something involves relying on them to
be motivated to do the relevant thing at least in part because they know that you are
relying on them to do it (Jones 1996: 8; Pettit 1995: 205–6). On this view an invitation
to trust is an invitation to rely on me because I will be responsive to your reliance. Let’s
call this interactive reliance.24 In the definition of interactive reliance, there is no men-
tion of why the trustor supposes that their reliance on the trustee will move the trustee
to perform. They might be assuming that the trustee will feel obliged to perform, but
equally they might be assuming that the trustee feels a certain goodwill towards them
or else wants to cultivate a reputation for reliability. Thus two people can engage in
successful interactive reliance without either party imagining that obligation is
involved.25 Can we construct an obligation-independent notion of trust in a promise
by understanding an invitation to trust as an invitation to interactive reliance?
Conceiving of trust in a promise as involving interactive reliance might seem a nat-
ural concomitant of the idea that a promise is a device of social coordination but I
doubt an adherent of the social coordination hypothesis should endorse it. One can
have successful interactive reliance (as I have defined it) in a situation where there is no
promise at all and so talk of interactive reliance does not identify the special way in
which a promise achieves social coordination. Suppose I ask you whether you are going
to a certain conference with a view to deciding whether to go myself. You reply ‘Well, as
things now stand, I predict that I’ll go but I can’t promise you that I will go’. Perhaps you
are more inclined to go precisely because we have had this conversation (you don’t like
to disappoint) and I may rely on that fact when I decide whether to go. Knowing this
you might also feel obliged to warn me should you change your mind and so we may
end up coordinating our plans in part by means of interactive reliance. But since I can’t
sensibly imagine that our conversation obliges you to actually go, I can’t trust you to go
out of fidelity to your word.26 Here social coordination isn’t achieved in the special way
that a promise achieves social coordination.
I conclude that an obligation-independent conception of trust can’t help to elucidate
trust in a promise, but we might modify our definition of interactive reliance to include
the idea that the trustee feels obliged to fulfil their promise only because the trustor is
relying on them. Does trust in a promise involve interactive reliance so understood?
24
This phrase comes from Pettit but he may mean something more by it than what is contained in my
definition. Faulkner (2007: 881) argues that trust involves something like interactive reliance, though he
adds that resentment would be apt should the trust be misplaced.
25
This may be what is going on in Pink’s doctor example (Pink 2009: 394–5).
26
‘A purpose is no contract, even when it is declared to the person for whose benefit it is intended.
I may say to a man that I intend to do such a thing for your benefit, but I come under no engagement’
(Reid 2010: 336).
I’m unsure but I suspect not. Clearly you can rely on someone to do something out of
fidelity to their promise without imagining that your reliance will move them to keep
the promise: perhaps they don’t know that you are relying on them and don’t need to
know in order to feel bound to keep the promise, or perhaps they suspect you don’t
trust them and are seeking to demonstrate their conscientiousness. In any case, the
question for us is whether such reliance amounts to trusting the promise.27 Since
the social coordination hypothesis does not obviously settle the matter, the issue may
be verbal. We’ll revisit this point in the next section but for now I’ll proceed on the
assumption that you can trust someone’s promise without assuming that they will
be moved by your trust.
We can sum up our first conception of trust in a promise, based on Hume’s social
coordination hypothesis, as follows:
(A) You trust someone to keep their promise iff you rely on them to perform
because they know they are obliged to perform.
Note that, on this conception, there is no concern with how the promisee comes to rely
on the promisor’s conscientiousness. For all (A) says the promisee might credit the
promisor’s assurances only because they have conducted a careful analysis of the
promisor’s past behaviour and, on that basis, have concluded that the promisor is con-
scientious. This observation raises a third and final issue about trust in a promise that
I’ve dubbed the question of justification.
This question concerns how the promisee’s reliance on the promisor’s conscien-
tiousness must be justified for that reliance to count as trust in their promise. Those
who think that trust involves belief in performance must insist that reasons for trust
include whatever evidential (or other) reasons are required to justify the belief that the
promisor will perform. Having allowed that a promise can serve its characteristic pur-
pose without convincing the promisee that the promisor will perform, we can afford to
be more liberal on the point. An advocate of the social coordination hypothesis should
permit my trust in someone’s promise to be based on anything that makes reliance
reasonable; that includes both evidence of the promisor’s conscientiousness and also
the non-evidential value of reliance. Thus I might rely on my brother’s conscientious-
ness, even though I do not myself regard him as conscientious simply because my
mother has asked me to. On the social coordination hypothesis I am here trusting my
brother even though I do not see him as trustworthy because I am adopting an attitude
towards him which will realize the distinctive value of his promise should he turn out
to be conscientious.
27
O’Neill (2012: 318–25) might describe this as a case in which the promisor is trusted not to betray
trust rather than not to abuse it. Nickel (2012: 307–9) makes the point that you trust someone in relying
on their conscientiousness even if you don’t suppose that they will come through because of your trust,
but Nickel appears to deny that this can happen in the case of a promissory obligation. I agree that a
promise generally invites trust but the trust it invites may just involve reliance on the conscientiousness
of the promisor.
224 David Owens
There are now two answers to the question of justification on the table but we are not
yet done. For some writers, to insist on having evidence sufficient to justify the belief
that John will perform would be to manifest a lack of trust in John: trust actually
excludes such confidence. On this third view, a paradigm instance of trust is where a
prospective reformer is asked to employ a man convicted of theft in their shop
(Holton 1994: 63).28 Here the reformer is in no position to know whether the convict
will steal but might well decide to trust him nevertheless. As Faulkner puts it:
if the reformer is willing to depend only given this belief and good reason . . . then she does not
really trust. Too thorough an assessment of the risk is inimical to trust. (Faulkner 2007: 897)29
Again there is a temptation to respond to such examples by multiplying senses or else by

selectively highlighting those bits of ordinary talk about trust that appear to favour one
usage of ‘trust’ over another. Again the issue acquires theoretical significance only once
we link it to developed views about the characteristic value of a specific object of trust like
a promise. An advocate of the social coordination hypothesis will reject Faulkner’s obser-
vation but, as we’ll see in the next section, a different account of the value of a promise
delivers a rather different answer to some and perhaps to all of our three questions.
3. Trust in a Promise: Cultivating Relationships

On the social coordination hypothesis, the value both of a promise and of trust in a
promise are fundamentally instrumental; they render the future behaviour of others
predictable and such predictability yields many further benefits. Alternatively it has
been proposed that trust may be valued for its own sake because trusting relationships
are to be valued for their own sake and not merely for the further benefits they provide.
This opens up the possibility that the distinctive value of a promise is something to do
with the non-instrumental value of these relationships.
Let’s leave trust in a promise behind for a moment and consider the following
example:
Suppose we are rock climbing together. I have a choice between taking your hand, or taking the
rope. I might think each equally reliable; but I can have a reason for taking your hand that I do
not have for taking the rope. In taking your hand, I trust you; in so doing our relationship moves
a little further forward. This can itself be something I value. We need not imagine that you would
be hurt if I chose the rope over your hand; you might be perfectly understanding of the needs of
the neophyte climber. But our relationship would not progress. (Holton 1994: 69)
Here Holton takes the hand of his companion because he values having a climbing
buddy and not just a source of physical support. Were the latter all that mattered to
28
In other cases (close relatives for example) one may have conclusive evidence of reliability but one’s
trust in them cannot, on this third view, depend upon it.
29
For an opposing view, see Pettit (1995: 207–8).
him, the rope might be a better buddy (and one that demands no return). Holton
clearly expects that taking the hand rather than the rope will help to bring that valuable
relationship into existence.
We might use the notion of interactive reliance to understand what is going on here.
Perhaps what Holton values for its own sake is someone’s holding him in part because
Holton is relying on them to hold him. That does sound like a human connection that
one might value for its own sake and one engendered by reliance. Furthermore this
sort of reliance often creates obligations. By accepting his companion’s hand, Holton
ensures that his prospective climbing buddy is obliged to hold him and would be
wronging him by letting him go (even if this would not lead to a serious fall). But, it
might be thought, such obligations are a by-product of the real source of the relation-
ship’s value, namely interactive reliance.
At the outset, I characterized trust as something that would realize the distinctive
value of the object of trust. Suppose that interactive reliance on another person is
valuable for its own sake and constitutes the distinctive value of climbing buddy rela-
tionships. Given that trust realizes this value, we can answer the questions of justifica-
tion and motivation as applied to Holton’s example. The latter question concerned the
trustee’s motivation and in particular whether they should be moved to perform by
the fact that the trustor is relying on them to perform and the answer is now a definite
‘yes’. As to the justification for such trust, on the present view to be motivated to trust
someone’s offer of a hand simply by evidence of their reliability is not to be responsive
to the distinctive value of what is on offer; rather it is to be responsive only to its value
as an indicator of future performance and not to the non-instrumental value of inter-
active reliance. It is to treat the climbing buddy relationship as having purely instru-
mental value. That is why one might (like Faulkner) regard such caution as untrusting.
Once more, we see how differing views about the characteristic value (instrumental
or non-instrumental) of relationships like being someone’s climbing buddy generate
different views about what constitutes a trusting response to your companion’s offer
of a hand.
Holton’s example need involve no promise. The parties may realize the normative
significance of the offer and acceptance of a hand (i.e. the obligations it will entail) but
they need not intend to change the normative situation, for they might have no interest
in who is obliged to do what. How can we generalize from Holton’s example to the case
of promise where the focus is on the creation of an obligation? In the last section,
I argued that interactive reliance is insufficient for trust in a promise: the trusting
promisee must also assume that the motive of duty is in play. Given this, we could
adopt something like the following conception of trust in a promise:
(B) You trust someone to keep their promise iff (i) you rely on them to perform
because they know they are obliged to perform and (ii) you so rely on them in part
because of the non-instrumental value of the obligation to perform.
226 David Owens
On this conception, part of the value of a trusting relationship lies in the very obliga-
tions that constitute such a relationship. Call this the normative interest hypothesis. (B)
uses the notion of reliance to formulate that hypothesis.
Trusting relationships are human connections with a certain normative character,
namely connections that involve a network of reciprocal rights and obligations.
Friends, neighbours, family members, and climbing buddies are embroiled in such
relationships. I suggest that the rights and obligations constitutive of these relation-
ships are not mere by-products of the features that give those relationships their value.
Rather, they contribute a deontic element to that value. Furthermore, this deontic
value is at least part of the distinctive value of such a relationship, of the value that
explains the binding force of its constituent obligations. Friends rightly value the
bonds of loyalty for their own sake and they are bound to one another in part because
being so bound makes their lives go better (Owens 2012: Chapter 4). To generalize, the
obligations that constitute relationships of trust are (a) valuable for their own sake at
least in the context of the relevant relationship and (b) bind the parties in part because
they are valuable for their own sake.
The offer and acceptance of a hand in Holton’s example probably aren’t declarations
that constitute the making and acceptance of a promise but, on our new conception of
the value of a trusting relationship, they may have much the same rationale and effect as
clear cases of promise. Here is such a case. Someone has moved in next door and I wish to
cultivate a neighbourly relationship with them, so I accept their offer to water my plants
while I am on holiday, knowing that I must make a similar offer when they go away. I do
this even though I could easily move the plants to my son’s house for the duration because
I wish to build a network of rights and obligations constitutive of a relationship with a
certain sort of non-instrumental value. Where the conventions of neighbourly behav-
iour don’t already tell us how to behave, the way to do this is via explicit agreements
(Raz 1986: 173–6). Note that if our relationship develops as envisaged, these neighbourly
obligations will soon cease to be obligations based on a dateable promise and will become
part of how we expect our neighbour to behave towards us in virtue of the nature of our
relationship (Owens 2012: 106–7), but in the meantime we use our power of promise to
create obligations that we value (or envisage coming to value) for their own sake.30
(B) tells us that trust in a promise is reliance motivated at least in part by the non-
instrumental value of the promissory obligation. This answers the question of justifica-
tion. Someone who relies on the promise simply because there is strong evidence that
it will be kept is treating the promise as having purely instrumental (i.e. evidential)
value and so is not trusting it.31 What of the modal question? Here we encounter a
30
Here I’m focused on promises made in the context of a developing relationship whose value depends
on that context but I don’t mean to imply that obligation can’t matter to people (and for its own sake) out-
side that context. We can both make binding promises and trust them whenever it makes sense to value an
obligation (and the power to impose it). See Owens (2012: Chapter 6).
31
There are other ways of treating the promissory obligation as having purely instrumental value. I
might rely on your promise out of politeness or because my mother told me to and without considering
potential problem. If valuing the obligation is the essence of trusting a promise (rather
than valuing the promised act) it looks as if the basic form of trust should be accept-
ance of the promise rather than reliance on the promisor’s behaving as promised, for
the promise binds provided it is accepted and regardless of whether the promisee relies
on it. If so, the normative interest hypothesis should be formulated using the notion of
acceptance rather than reliance.
But can simply accepting a promise really be a way of trusting the promise? There
are cases and cases. Obligations and the promises that create them are frequently val-
ued in the context of a developing relationship and where this is so I can often realize
the value in question only by relying on the relevant promises. I value my obligation to
water my neighbour’s plants as one aspect of an ongoing relationship with my neigh-
bour, as part of a network of rights and obligations (and of the habits and customs
which embody them). That network will not come into existence unless the promises
we make to one another are not merely accepted but also relied upon, for our relation-
ship will not develop in the envisaged way if, having accepted the neighbour’s promise,
I panic and transfer the plants to my son’s house. Here my neighbour is no longer
obliged to water my plants.32 In failing to rely, I decline to make myself vulnerable to
being wronged by my neighbour (in respect of non-performance) and so I fail to create
the debt that will be the occasion of my later promising to water their plants whilst they
are away. The desired connection between us never materializes and the promissory
obligation loses its rationale.
But in other cases (e.g. my brother’s promise of a present) there may be nothing I can
do or fail to do that would count as reliance on the promise beyond simply accepting it.
Here acceptance makes me vulnerable to being wronged and such acceptance all by
itself constitutes trust in a promise (provided that acceptance is motivated by my valu-
ing such vulnerability for its own sake). If I accept your promise just to please my
mother and without any real interest in whether you are bound to me or not, I am still
wronged by your breach but it would be odd to describe you as betraying my trust or
even as letting me down. On the other hand, if I accept your promise because I want to
create a relationship of mutual vulnerability between us, because I want to make room
for the possibility of fidelity and betrayal, then acceptance even in the absence of reli-
ance on or expectation of performance can indeed amount to trusting your promise.
I’ve rather fallen out with a friend who has stood me up once too often. They sol-
emnly promise to help me move house and I accept their promise as a way of giving
them the opportunity to restore relations between us, even though I have an open
mind about whether they’ll show up, and things are arranged so that all will go
smoothly whether or not they appear. On the present view of promise, I do trust them
whether your promise is a good predicator of performance. In these cases also I am relying on the promise
without trusting it (on the present view) because my reliance is not based on my valuing the promissory
obligation for its own sake.
32
This is so in virtue of the impossibility of performance and regardless of whether I have explicitly
released the neighbour from their promise.
228 David Owens
and they can betray my trust even though I place little credence in or reliance upon
what they say. After all, I’ve opened myself up to being wronged. Here the fact that they
would wrong me by not showing up has a significance of its own and I accept their
promise because I wish to give their behaviour that significance, to make myself vul-
nerable to them in that way.33
Trust in a promise is an attitude that realizes the distinctive value of the promise,
a value that, on the normative interest hypothesis, is the (non-instrumental) value
of the promissory obligation. Acceptance and reliance can each play this role on
different occasions and so both can constitute trust in a promise. Thus the normative
interest hypothesis and the social coordination hypothesis differ in their answers to
both the modal question and the justificational question. Our conception of trust in a
promise once more depends on our conception of the value of trust’s object, namely
the promise.
4. Conclusion
The recent debate about the nature of trust is intricate and inconclusive because its
participants see themselves as analysing a distinctive psychological attitude, an atti-
tude about whose nature and value they are all disagreeing. I suggest that the tangle
makes more sense if we regard it from the other end, starting with the objects of trust.
With so many different objects of trust in play, and so much disagreement about their
nature and value, an inconclusive debate is only to be expected. We made this point by
focusing on trust in a promise and I argued that one’s view of that should depend on
one’s conception of the distinctive value of a promise.
I did not seek to establish the correctness of any particular conception of the value of
a promise. Indeed we didn’t even consider one of the most influential accounts of this
matter, namely the expectations theory (Scanlon 2003). According to that theory, the
point of a promise is to provide the promisee with knowledge of what the promisor is
going to do. For Scanlon, trust in a promise involves the belief that the promisor will
perform, a belief for which the promisee must have sufficient evidence. On this con-
ception of the value of a promise our answer to the modal and justificational questions
for promise will likely be similar to those applicable to assertion, whilst on both the
social coordination and the normative interest hypothesis these answers are going to
be very different.
Elsewhere I argue that neither the social coordination hypothesis nor the expecta-
tional account provides an adequate conception of the distinctive value of a promise
and I endorse the rival normative interest hypothesis (Owens 2012). Nothing I say here
33
Here the obligation has value only in the context of the restored friendship and the friendship will be
restored only if the former friend fulfils their promise. So they are worthy of my trust only if they turn out
to be reliable but trusting them involves no more than accepting their promise. Thanks to Collin O’Neil for
forcing me to clarify this.
rests on that claim. What this chapter seeks to establish is that none of our three ques-
tions about trust can be settled by applying some general theory of trust to the case of
promise; rather they must be approached from the opposite direction by first formulat-
ing a view of the value of a promise. Trust in someone’s assertions or trust in their cour-
age should be tackled in the same fashion. There is no general attitude of trust fitted to
play a foundational role in our social theory. Trust is as various as the objects of trust.34
References
Austin, J. (1975) How To Do Things With Words, 2nd edn. (Cambridge, MA: Harvard University
Press).
Baier, A. (1986) Trust and Anti-Trust, Ethics 96: 231–60.
Faulkner, P. (2007) On Telling and Trusting, Mind 116: 875–902.
Friedrich, D. and N. Southwood (2009) ‘Promises and Trust’, in Promises and Agreements,
ed. H. Sheinman (Oxford: Oxford University Press): 277–94.
Hieronymi, P. (2008) The Reasons of Trust, Australasian Journal of Philosophy 86: 213–36.
Holton, R. (1994) Deciding to Trust, Coming to Believe, Australasian Journal of Philosophy 72:
63–76.
Hume, D. (1978) Treatise on Human Nature (Oxford: Oxford University Press).
Jones, K. (1996) Trust as an Affective Attitude, Ethics 107: 4–25.
McMyler, B. (2011) Testimony, Trust, and Authority (Oxford: Oxford University Press).
Nickel, P. (2012) Trust and Testimony, Pacific Philosophical Quarterly 93: 301–16.
O’Neil, C. (2012) Lying, Trust and Gratitude, Philosophy and Public Affairs 40: 301–33.
Owens, D. (2012) Shaping the Normative Landscape (Oxford: Oxford University Press).
Pettit, P. (1995) The Cunning of Trust, Philosophy and Public Affairs 24: 202–25.
Pink, T. (2009) Promising and Obligation, Philosophical Perspectives 23: 389–420.
Raz, J. (1986) The Morality of Freedom (Oxford: Oxford University Press).
Raz, J. (2001) Value, Respect, and Attachment (Cambridge: Cambridge University Press).
Reid, T. (2010) Essays on the Active Powers of Man, ed. K. Haakonssen and J. Harris (Edinburgh:
Edinburgh University Press).
Scanlon, T. (2003) Thickness and Theory, Journal of Philosophy 100: 275–87.
34
Many thanks to Robert Stern, Alison Hills, Paul Faulkner, Tom Simpson, Jessica Moss, Sharon Street,
David Velleman, Matty Silverstein, Jorah Dannenberg, Nic Bommarito, and especially to Collin O’Neil for
comments.
14
Trustworthy Groups
and Organizations
Katherine Hawley
1. The Significance of Group Trustworthiness

We value trustworthiness where we find it, in friends, colleagues, and strangers. This is
because it is easier to cooperate with trustworthy individuals than with untrustworthy
ones, but also because trustworthiness is appreciated for its own sake. Conversely, we
resent untrustworthiness when we encounter it, and try to minimize interaction with
untrustworthy individuals. This preference for trustworthiness over untrustworthi-
ness is not limited to our interactions with individuals. Who wouldn’t prefer to work
for a trustworthy company, to be treated at a trustworthy hospital, and to vote for a
trustworthy political party?
The trustworthiness of collective entities is central to much public discourse
around trust and distrust. A collapse of trust in bankers, say, is not merely a collapse
of trust in individual, identifiable bankers, but takes in the profession as a whole;
moreover, a collapse of trust in bankers goes hand-in-hand with a collapse of trust in
the banks (de Bruin 2015). Likewise, discussion of trust or distrust in corporations,
governments, news organizations, or even brands seems to concern collective entities
of various kinds.
Unsurprisingly, then, trust and trustworthiness in collective contexts are widely
discussed by social scientists. For example, in international relations there is debate
about trust and distrust between states, other organizations or groups, and individual
representatives of those collectives (e.g. Booth and Wheeler 2008). And ‘organiza-
tional trust’ is a recognized research topic in management studies, taking in not only
corporations, but also non-governmental organizations and other complex group
agents. Such research investigates both the influence of organizational contexts on
our trust in individuals, and the ways in which trust is invested in organizations
themselves (e.g. Saunders, Skinner, et al. 2010).
If philosophers are to contribute to public discussions or to interdisciplinary dia-
logue, we need to understand how our own accounts of trust and trustworthiness can
trustworthy groups and organizations 231
be applied in such contexts (Mäkelä and Townley 2013 make a related point).
Moreover, an investigation of trust and trustworthiness in collective contexts prom-
ises to improve our understanding of collective agency, intentionality, responsibility,
belief, and knowledge; issues of trust span the theoretical–practical divide, often in
distinctive and interesting ways.
But philosophers’ accounts are typically focused on issues of trust in individuals,
aiming to capture the nature of this attitude, the circumstances under which it may be
rational, and the notion of trustworthiness which corresponds to this attitude of indi-
vidual trust. (There are important exceptions to this individualistic rule; I return to
some of these below.) In particular, philosophers often distinguish trusting someone
from merely relying upon her, then theorize this distinction. Reliance is an attitude we
may adopt towards inanimate objects: we rely upon the tent to keep us warm and dry
overnight. Trust is usually taken to include reliance plus some further factor. This
respects Annette Baier’s (1986) insight that trust is distinctively connected with
the possibility of betrayal and resentment, unlike ‘mere’ reliance: if the tent leaks, we
should not resent the tent itself, though we might resent the person who promised to
mend the tent before we set out. In Hawley (2014) I argued that we should also distin-
guish distrust from mere absence of reliance, and moreover that we should not neglect
the many interactions with other people in which either reliance or non-reliance is
appropriate, but trust and distrust are inappropriate.
What is the magic ingredient which distinguishes (dis)trust from mere (non-)
reliance? Philosophers, inevitably, disagree: maybe a truster imputes appropriate
motives to the trustee, including perhaps a concern with or responsiveness to the
needs, desires, or indeed trust of the truster; or maybe a truster sees a trustee as mor-
ally obligated, committed, or accountable in appropriate ways (Simon 2013 is an
excellent bibliographical guide). Across this variety of accounts, trust is understood
to be directed at agents, or at persons, or in a distinctively second-personal fashion,
or as an aspect of the participant stance: it is always a mistake to trust a tent, a trout,
or a turkey, although we may rely upon such things. This emphasis on the interper-
sonal flows from the way in which the trust–reliance distinction is identified via the
connection between trust and the reactive attitudes around betrayal. Not every
interpersonal interaction is characterized by either trust or distrust, but every proper
attitude of trust or distrust is directed interpersonally. Or so it seems.
Thus it is no accident that within this philosophical paradigm we rarely discuss trust
or distrust as directed towards collective entities, since even the most inflationary
accounts of collective agency hold back from treating such entities as fully-fledged
persons in every respect, on a par with individual human persons. (To date, not even
Mitt Romney has advocated corporate suffrage.) Given that we need a philosophical
understanding of trust in collectives, how then should we proceed?
One strategy would be to consider various accounts of individual trust and trust-
worthiness pairwise with accounts of collective agency, responsibility, and so on, look-
ing for rewarding and plausible combinations. This is a valuable project, which I do not
232 Katherine Hawley
pursue here. Another strategy would be to address collective issues on their own terms,
without expecting to find much commonality between interpersonal trust and trust
at the collective level. Baier (2013: 175–6), for example, is explicit that her account of
interpersonal trust, in terms of vulnerability, competence, and goodwill, cannot easily
encompass trust in organizations. We might hazard that accounts of trust which
emphasize mutuality, emotional affect, and complex relationships will in general not
apply to more distant forms of trust, including trust in collective entities; advocates of
such accounts might see this as a strength, arguing that it would be a mistake to col-
lapse such different attitudes.
In this chapter, I adopt a different strategy, one which may complement either of the
first two projects: I examine the costs of abandoning the trust–reliance distinction in
collective contexts. Vindicating this distinction has been regarded as an essential cri-
terion of success for accounts of trust in individuals. But, I will argue, we can explain
and justify much of our practices around groups without using this distinction; the
costs of abandoning it are low, as compared to the individual case.
Supposing my arguments are persuasive, what then? The fact that we can manage
without this distinction does not entail that there is in reality no distinction between
group trustworthiness and group reliability. For example, we might on closer investi-
gation decide to adopt accounts of individual trust, and of collective belief, intention,
and responsibility which together entail that groups can after all be trustworthy in the
same way that individuals can be trustworthy; nothing I say in this chapter will rule out
that possibility.
However, if the costs of doing without the distinction are low, then we cannot use
the supposed importance of the distinction as a consideration when deciding between
different accounts of groups. Thus it is not a constraint on an adequate theory of groups
that groups be capable of trustworthiness as opposed to mere reliability. A strong read-
ing of my project is that we should stop thinking in terms of either group trustworthi-
ness or appropriate trust in groups. This might seem to spell disaster for the project of
philosophical engagement with public and cross-disciplinary discussion of organiza-
tional trust. I reflect upon this challenge at the end of the chapter and gesture at a more
optimistic way forward.
2. Groups, Organizations, and Institutions

First, however, a note on terminology. Within philosophy, the term ‘group’ is a catch-
all label for a great variety of candidate collective agents. For example, Frederick
Schmitt’s (1994) ‘The Justification of Group Belief ’ begins with examples such as ‘the
Engineering Division of the Ford Motor Corporation’, ‘the crowd that had assembled
on the square’, and ‘This Court’. Deborah Tollefsen’s (2004) ‘Group Deliberation . . .’
focuses on small teams of scientists. Kay Mathieson’s (2006) ‘The Epistemic Features of
Group Belief ’ focuses on groups identified by listing their members. In her ‘Group
Testimony . . .’, Miranda Fricker (2009: 272) mentions committees, news teams, gov-
ernments, research groups, and consultancies. The cover blurb for List and Pettit’s
(2011) Group Agency begins with companies, churches, and states. All of these philo-
sophers are sensitive to the differences between various types of collective entity, but
use ‘group’ as the general term.
The situation is different within the social sciences. For example, Piotr Sztompka’s
(1999: ch. 3) influential Trust: A Sociological Theory distinguishes ‘social categories’,
such as gender, age, and race, from ‘social groups’ such as the football club Real Madrid,
a class of students, or an army platoon, and from ‘institutions and organizations’ such
as the university, the army, the courts, and the banks. Sztompka goes on to discuss
technological systems, food products, and general social systems such as democracy as
objects of trust or distrust. It is hard for a novice to disentangle the relevant termin-
ology, and harder still to map these distinctions onto those which typically interest
philosophers. (I take consolation from economist Geoffrey Hodgson’s remark: ‘. . .
endless disputes over the definitions of key terms such as institution and organization
have led some writers to give up matters of definition and to propose getting down
somehow to practical matters instead’ (2006: 1)).
I will stick with the philosophical usage of ‘group’, as a general term for all sorts of
candidate collective agents. But we should bear in mind the variety of social entities
which might be considered the object of trust or distrust, and the potential obstacles to
communication across disciplinary divides, given this terminological choice.
3. Trust, Trustworthiness, and Reliability

So, do we need a distinction between trust and reliance with respect to groups? I will
approach this via a related question: do we need to distinguish group trustworthiness
from group reliability? Trust and trustworthiness are closely entwined, of course: very
roughly, to be trustworthy is to merit the attitude of trust, and to trust is to regard as
trustworthy. But it is useful to begin with trustworthiness, for several reasons. In my
opinion, trustworthiness is the primary site of moral evaluation in this area, though
I cannot defend that position here. More pragmatically, there is very likely some confu-
sion in our actual attitudes of trust and distrust towards groups. To understand whether
our attitudes of trust and distrust are coherent, it is helpful to understand the features
of groups—trustworthiness and untrustworthiness—which the attitudes target. Jones
(2012) demonstrates the fruitfulness of exploring trustworthiness first, and I hope that
the viability of this approach is further demonstrated in the present chapter.
This trustworthiness-first approach does not assume that only trustworthiness can
justify trusting (hence my ‘very roughly’ above). There can be many reasons to trust in
the absence of trustworthiness, including therapeutic trust aimed at cultivating trust-
worthiness, psychological self-protection, efficiency in low-risk situations, and so on.
Nevertheless, even in such cases, trusting involves behaving as if the recipient were
trustworthy. Moreover, we can approach theorizing trustworthiness-first without

making substantive commitments about how best to cultivate stronger trust relation-
ships in society (Baier 2013; O’Neill 2013).
Trustworthiness is to be distinguished from mere reliability. Circumstances under
which either trust or distrust is appropriate, as opposed to mere reliance or non-
reliance, are circumstances under which the actor’s trustworthiness is tested. Many
intentional actions fall outside the scope of trustworthiness or untrustworthiness,
especially where they seem independent of obligations to others. The head teacher at
the school next door regularly releases noisy children into the yard at midday, and I
rely upon this to remind me to take a lunch break. But the head teacher’s reliability in
this matter is not an issue of trust or distrust for me, and if she decides to keep the chil-
dren indoors one day, this has no implications for her trustworthiness.
We do not mark the trust–reliance distinction sharply in everyday language, and
nor do we mark the trustworthiness–reliability distinction consistently. This everyday
loose talk presents a problem to the theorist who hopes to explain the underlying
differences between trustworthiness and reliability. Which cases are which? Exactly
what are we trying to account for?
Despite these difficulties, the distinction between trustworthiness and mere
reliability for individuals is worth investigating because it is morally significant.
Trustworthiness is an admirable character trait, something to aspire to and inculcate
in one’s children, whereas mere reliability may have practical value but does not seem
valuable for its own sake. As individuals, it is important for us to understand the
difference between circumstances under which our behaviour is a test of our trust-
worthiness, and circumstances under which it is merely a sign of our predictability or
reliability. There are indeed plenty of borderline cases, but there are also plenty of clear
cases to work with.
The problem of distinguishing trustworthiness and reliability is all the more difficult
in collective cases. Recall the variety of social entities which are spoken about in terms
of trust and distrust: is a trustworthy corporation or political system trustworthy in
anything like the same way that a spouse or friend can be trustworthy, or is such imper-
sonal trustworthiness more like the reliability of a well-made car? Regarding individ-
uals, it is worth persevering with the trustworthiness–reliability distinction. But what
about groups? Do we need a morally laden notion of trustworthiness for collective
entities, or can we get by with mere reliability and unreliability?
4. Group Reliability?
Ideally, I would now propose a detailed account of group reliability and then investi-
gate whether any additional notion of group trustworthiness is both coherent and
valuable. Less than ideally, I will simply take it that we have some grasp of reliability
and unreliability as these apply to groups. But a few points are worth making.
Both reliance and reliability can be differentiated with reference to tasks or

domains. I rely on my alarm clock to wake me up, but I do not rely on it to pass
unqueried through airport security. I rely upon my dog to behave sensibly in public
(in a somewhat unspecified way), but I do not rely upon her to stay off the furniture.
At the limit, reliability in a very specific respect can be identified with success in a
one-off task. I relied upon my alarm clock to wake me up today at 7 a.m.; it did so, and
it was thus reliable in that particular respect, even if, over the course of several days, it
proves to be more generally unreliable and thus useless as an alarm clock. This gener-
alizing move can be understood statistically, or in terms of dispositions, or ceteris
paribus clauses. As is familiar from the epistemological literature on the ‘generality
problem’ for reliabilism, the range or comparison class can make a big difference to
judgements of reliability (e.g. Bonjour 2002). As is obvious from everyday life, reli-
ability can come in degrees.
What about group reliability or unreliability? Sometimes when we talk about a
group of people as reliable, we simply mean that all or most people in that group are
reliable in the relevant respect, or perhaps, via a generic, that ‘normal’ members of
the group are reliable. For example, my students can be relied upon to bring their
textbooks to class and I plan our activities on that basis; the group is reliable in this
respect because enough individual group members are reliable. Sometimes, such
judgements about individuals are based on their group membership, for better or
for worse.
In this chapter, I will set aside this type of group reliability, focusing instead upon
situations in which the group as a whole acts, and is reliable or unreliable in doing so.
We might see bringing-their-textbooks-to-class as a kind of group action: this is some-
thing the students do together, simply in virtue of each individual bringing his or her
textbook to class. But many interesting group actions stand in more complex relation-
ships to the actions of individual group members. For example, the students create a
relaxed environment in class, occupy all the seats, and disagree about the nature of free
will. I will not attempt to specify the range of possible group actions, nor to engage with
the rich, extensive literature on collective or group agency (e.g. List and Pettit 2011).
My task in this chapter is to investigate whether we need to think of groups as trust-
worthy actors, or merely as reliable actors. Thus I need to assume that groups can act,
reliably or unreliably.
Trustworthiness can be discussed both in the context of practical action and in the
context of testimony: we trust people to do stuff and we trust people to speak truth-
fully. Such discussions are enriched when they are integrated with one another.
Speaking truthfully is a kind of action, and moreover trusting people to do stuff often
involves trusting them to follow through on their words. Nevertheless, trust in testi-
mony does have some distinctive features which merit special attention. Moreover,
there is a small but valuable recent literature on group testimony, so I will begin my
discussion with testimony.
5. Trustworthiness and Individual Testimony

There is much we might hope for when others speak. Embarrassingly often, we just
hope to be entertained, and on this front we can distinguish someone who is reliably
entertaining from someone who can be trusted to be entertaining. For example, some-
one who is often unintentionally entertaining is reliable in this respect, but not
trustworthy.
More soberly, we might hope to learn something when others speak—appreciating
what others say is a key source of true beliefs. An individual can serve as a reliable
guide to the truth in all sorts of ways. Someone with an open countenance reliably
reveals her thoughts via blushing or other ‘tells’, whilst someone else’s eye-bags provide
information about her recent lifestyle. Verbal behaviour can also reliably indicate the
truth. For example, the questions I ask reveal my preoccupations; my strenuous denial
of interest in some topic suggests quite the opposite; and what I say in my sleep indi-
cates my true fears.
This type of reliable openness or readability does not amount to trustworthiness.
Why not? A full answer would involve a substantive, and therefore controversial,
account of the difference between trust and reliance. But the short answer is that unre-
liability in these respects does not constitute a betrayal; we do not in general owe such
reliability to others and others are not entitled to resent us if we lack such reliability.
That is, one can be unreliable in these ways without being untrustworthy. If I can dance
the night away and still look fresh in the morning, this does not make me untrust-
worthy; likewise if I rarely blush, talk only nonsense in my sleep, and ask questions in a
neutral manner. Such traits help me to mislead others if I wish to, and can thereby
facilitate untrustworthiness. But unreadability does not constitute untrustworthiness
in its own right.
In contrast, in core cases of testimony where an individual asserts something to an
audience—or, if you prefer, where an individual tells an audience something—reliability
with regard to truth-indication does constitute trustworthiness. As elsewhere, the notion
of reliability here is very rough and ready. Reliability comes in degrees; moreover we
should distinguish between reliability on a given occasion, reliability across a subject
matter, and reliability quite generally. My main purpose here is to contrast reliability with
trustworthiness, and individual with group cases: we do not need a full understanding of
reliability in order to make some comparative judgements.
But however we understand reliability, reliability in assertion (or telling) is a matter
of trustworthiness, because assertion involves undertaking responsibility for what is
said. In contrast, speaking truthfully or falsely in one’s sleep does not contribute to
either trustworthiness or untrustworthiness, because we do not hold people respon-
sible for sleep talk. This aspect of assertion is acknowledged by a wide range of philo-
sophers who otherwise take quite different views of what assertion involves. For Peirce,
‘to assert a proposition is to make oneself responsible for its truth’ (1932: 384). For
Searle an assertion of p ‘counts as an undertaking to the effect that p represents an
actual state of affairs’ (1969: 66). And for Williamson, ‘To make an assertion is to con-
fer a responsibility (on oneself) for the truth of its content’ (2000: 268–9).
More ambitiously, Brandom (1983) argues that asserting that p involves commit-
ment to withdrawing the assertion if it proves to be mistaken, and/or to defending the
assertion against reasonable challenges. Such commitments last beyond the moment
of assertion itself. But one can recognize that assertion involves an undertaking of
responsibility without recognizing any such longer-lasting commitments: this is illus-
trated by the quotation from Williamson above, which comes just a few pages after he
explicitly rejects Brandom’s account. Williamson goes on to say that one discharges
this responsibility ‘by epistemically ensuring the truth of the content’ (2000: 269), that
is by satisfying the condition that one knows what one asserts, at the moment of asser-
tion. This encapsulates Williamson’s preferred ‘knowledge norm’ on assertion, but
those who prefer other norms—perhaps a truth norm, or a justified-to-believe norm—
can also cast these in terms of discharging one’s responsibility as an assertor.
So I will take it that, in the individual case, reliability or unreliability in the provision
of (apparent) information becomes a matter of trustworthiness or untrustworthiness
when the individual makes an assertion, or tells an audience something, thus somehow
taking responsibility for what is said. Can we apply these ideas to collective entities?
6. What is Group Testimony?

What is it for a group to be reliable as a guide to the truth? We can often obtain infor-
mation by observing the behaviour of groups of people: if the crowd is gathered around
laughing, I infer that there is something funny going on. Sometimes we obtain infor-
mation from the verbal behaviour of a group. Surowiecki (2004) illustrates the ‘wis-
dom of crowds’ with a phenomenon noted by Francis Galton: when a large number of
people at a country fair were asked to estimate the weight of an ox, the average estimate
was very accurate, indeed more accurate than the estimates of individual experts.
Interesting and useful though this phenomenon is, it does not involve group asser-
tion, or anything which involves trustworthiness as opposed to mere reliability.
Neither the crowd collectively nor any individual member of the crowd can sensibly be
held accountable for the accuracy of the average estimate; neither the crowd nor any
individual takes responsibility for the truth of the average judgement.
In other circumstances, however, a group or organization produces something
which looks much more like assertion or testimony. Seemingly just as an individual
may express herself by employing a spokesperson, issuing a written statement, or
speaking in the ordinary way, a group may issue a statement via a spokesperson, or
publish a report with the group’s imprimatur. Not all groups are capable of doing this:
issuing a statement or report seems to require a certain degree of internal structure and
organization. Moreover it is sometimes unclear whether an individual is authorized to
speak on behalf of a group: consider how journalists turn to informal ‘community
leaders’ to represent the collective view.
Nevertheless, we often understand groups or collective entities as issuing statements

in these ways. This might suggest that we need to distinguish the trustworthiness of a
group, with respect to its assertions or testimony, from the mere reliability we see in
‘wisdom of crowds’ cases. Does this provide a clear case of group trustworthiness as
something over and above group reliability?
Jennifer Lackey (2014) argues that, when a group ‘speaks through’ a spokesperson,
the group’s testimony simply is the spokesperson’s testimony, and thus the epistemic
significance of the group’s testimony just is the epistemic significance of the individual
testimony. Of course, the epistemic significance of the spokesperson’s statement—
whether it is reliable, for example—will depend in large part upon the way in which
the content of the statement was generated by the group and provided to her. But, as
Lackey points out, this is often true for individual testimony too. The epistemic signifi-
cance of what I say to you in my personal capacity depends in large part upon the epi-
stemic standing of the people from whom I acquired my ‘information’.
Suppose we accept Lackey’s deflationary account of the epistemic significance of
group testimony. Does this mean that group trustworthiness in testimony is to be
identified with the individual trustworthiness of the spokesperson? No: the spokes-
person is not making an assertion, and does not undertake responsibility for the truth
of what she says. The spokesperson is not like an embedded journalist: she does not, in
general, make an assertion about the group. Instead, she speaks on behalf of the group,
about the relevant subject matter. This distinction between reporting views and
expressing them is familiar from ordinary individual assertions: in saying that p, I do
not report that I believe that p, but instead I purport to express my belief that p. In mak-
ing my assertion, I undertake responsibility for the truth of p, not merely for the truth
of the claim that I believe that p.
Does the spokesperson undertake responsibility for the truth of what she says on
behalf of the group? No. Her responsibility is to execute her duties as a spokesperson,
and we can judge her as trustworthy or untrustworthy in that capacity, but this is not to
judge her trustworthiness as an assertor, even in this circumscribed setting. Indeed,
Lackey’s view is that the spokesperson is not governed by standard norms of assertion,
but rather by norms specific to her status as spokesperson (Lackey forthcoming). So
although there is a sense in which what the spokesperson says is what the group says,
this does not mean that the spokesperson makes an assertion which is the group’s
assertion.
We can judge the spokesperson as reliable or unreliable with regard to the truth in
this matter—and Lackey may well be correct about the epistemic significance of what
the spokesperson says—but this is not to evaluate the spokesperson’s trustworthiness
with regard to this subject matter. If any entity is taking responsibility for the truth of
what is said, it is the group rather than the spokesperson.
Does the group take responsibility for what is said on its behalf? One approach to
this question would be to take up general issues about collective responsibility, to
understand first what it is for a group to be responsible for something, then what it
is for a group to take responsibility, and specifically to take responsibility for the
truth of a statement. But I will adopt a different approach, asking not what it would
be for a group to take responsibility for the truth of what is said, but whether we can
manage without that notion: what costs would we incur if we did not think of
groups in this way?
We saw in the individual case that the distinctions between trust and reliance, and
between trustworthiness and reliability, seem to be of central importance in our deal-
ings with one another. The notion of reliance doesn’t seem sufficient to account for our
interpersonal interactions and attitudes. But what about the group case: can we man-
age without the notions of group trustworthiness, group responsibility for the truth,
and group assertion? I will investigate this by looking first at the epistemic aspects of
group testimony and then at the ethical aspects of group testimony. I will argue that in
each case we can vindicate our practices without resort to a distinction between trust-
worthiness and reliability; this does not settle the question of whether groups can be
trustworthy, but it does shift the burden of proof.
7. Group Trustworthiness in Testimony:

Epistemic Aspects
A number of authors, including Hinchman (2005), Moran (2006), McMyler (2007),
and Faulkner (2011), have argued that trust, as opposed to mere reliance, plays an
important role in the epistemic significance of individual testimony. There are signifi-
cant differences amongst these various authors, but some common themes emerge.
On this picture, the speaker’s undertaking of responsibility for what is said provides
the intended audience with a distinctive epistemic reason to believe what is said.
Conversely, the audience’s trust, not mere reliance, in the speaker plays a key role in
justifying beliefs acquired through testimony. The speaker offers his or her assurance
to the audience and, when circumstances are favourable, this assurance gives the audi-
ence reason to trust and reason to believe.
If assurance plays a key epistemic role in our acquisition of knowledge from testi-
mony, then so does the trust–reliance distinction and thus also the trustworthiness–
reliability distinction. When we learn from what someone tells us, complete with
assurance, we regard her as trustworthy in this matter. According to the assurance
view, we are then in a better epistemic position than when we merely learn from some-
one’s verbal behaviour, regarding her as merely reliable.
Thus an assurance account of the epistemic significance of group testimony offers
the best prospect for vindicating a trustworthiness–reliability distinction in that
domain. Assurance accounts of individual testimony emerged in reaction to more
standard views and they have been criticized on various counts (e.g. Lackey 2008); I
will not try to adjudicate this debate in the individual case. Rather, I will explore
what non-assurance and assurance accounts of the epistemic significance of group
testimony might look like, before arguing that assurance accounts are less motivated
in the group case than in the individual case.
Deborah Tollefsen (2007) advocates a non-assurance reductionism about the justi-
fication of the beliefs we form on the basis of group testimony. Such beliefs are justified
insofar as we can monitor the reliability of the group’s testimony more generally; the
view is reductionist in that the justification of testimonial beliefs is reducible to justifi-
cation gained from other sources. On this view, group pronouncements are treated as
just more evidence, in the mix with group behaviours and features of all sorts; there is
no epistemic role here for group trustworthiness over and above group reliability.
Similarly, Lackey’s identification of group testimony with spokesperson testimony
allows us to assess the epistemic merits of group testimony by assessing the testimony
of the spokesperson. Lackey (2008) adopts a non-assurance view of the epistemic sig-
nificance of individual testimony, meaning that, on this view, there is no epistemic role
for assurance in group testimony.
What would an assurance account of the epistemic significance of group testimony
look like? In a later paper, Tollefsen advocates a more demanding notion of what it is
for a group to testify: ‘. . . the fact that groups issue intelligible statements either in writ-
ing or via a spokesperson seems to me now not sufficient to say that they, themselves,
are testifiers’ (2011: 12). Tollefsen turns to assurance accounts, seeing testimony as
essentially interpersonal, ‘deeply tied to epistemic responsibility’ (2011: 15). For
groups to testify, on this picture, they must be able to acknowledge their epistemic
responsibilities; this echoes the connections I have highlighted between trustworthi-
ness and fulfilling responsibilities.
One of Tollefsen’s real-life examples involves a panel of scientists, tasked to issue a
definitive statement about the genetic hazards of radiation. Although the scientists had
underlying disagreements about the facts, Tollefsen argues that ‘group members real-
ized the fact that others were depending on the group to speak its mind and that they
had a responsibility to say something definitive and take collective responsibility for
what was said’ (2011: 17). It is striking that Tollefsen refers to the individual group
members as acknowledging responsibility for getting the group as a whole to take
responsibility for what is said.
Miranda Fricker asks what is required for a group to be able to offer assurance
in its testimony, and suggests that ‘any group partly constituted by way of a joint
commitment to trustworthiness (regarding some relevant range of questions) is
pre-eminently suited to enter into the second-personal relations of trust that charac-
terize testimony’ (2012: 272). The notion of joint commitment here is based on that
developed by Margaret Gilbert across many works (e.g. Gilbert 2006). According to
Fricker, group members make—initially to one another—commitments to be jointly
trustworthy. The resulting joint commitment helps constitute a group which can then
itself offer assurance to its audience. Much like Tollefsen, Fricker gives a key role to
individual commitments, within the collective context, in generating assurance on
behalf of the group.
Fricker’s picture has a certain appeal when we think about the ‘good’ cases, in which
well-intentioned people get together in the hope of forming a trustworthy group.
Fricker shows how it can accommodate cases in which such a group becomes untrust-
worthy because members do not fulfil their commitments to joint trustworthiness,
perhaps because of personal corruption.
However, other sorts of cases do not fit so well. Imagine a nefarious group, formed
by members who make a mutual commitment to jointly deceive the public (insert your
own example here). Such groups are constituted by the individuals’ commitments to
deception, not by joint commitments to trustworthiness, so do not fall under Fricker’s
account. Are these groups ‘suited to enter into the second-personal relations of trust
that characterize testimony’? Not if we construe the question literally: nobody should
trust such groups. But if some groups are trustworthy, then surely these nefarious
groups qualify as untrustworthy, not merely unreliable (Lackey makes a related point
in her draft ‘Group Lies’, n. 9). Such groups seem as suited to enter into second-personal
relations of distrust, resentment, and betrayal, as more admirable groups are suited to
enter into second-personal relations of trust.
Moreover Fricker’s account does not allow for the possibility of trustworthy organ-
izations whose members—for example, employees—follow appropriate procedures,
but are not motivated by a desire for or commitment to group trustworthiness. Perhaps
Fricker would be happy to conclude that such a group does not produce genuinely
assurance-based testimony, but is in practice more similar to an information-
producing machine, or a source of information, rather than an informant, in the terms
she shares with Craig (1990).
Despite these concerns, some groups do seem to operate in the ways Fricker
describes. But what is the epistemic significance of this? Fricker herself argues that, in
the individual case, assurance provides a different, but not essentially stronger reason
to believe what is said than does merely overheard testimony. She writes that,
an addressee [who receives assurance] and an eavesdropper [who does not] could have
exactly the same background reasons to trust [an individual] testifier’s word; but that in
telling his addressee that p a testifier offers her a second-personal trust-based reason to
believe his word that p; whereas he (wittingly or unwittingly) offers the eavesdropper a
third-personal trust-based reason to believe that p. These are both epistemic reasons, for
they both bear on the likely truth of p, and they may deliver the same strength of warrant;
but I have tried to vindicate the idea that they are subtly different sorts of epistemic reason.
(2012: 268–9)
Less sympathetically, Lackey (2008: 249) suggests that the difference between trust and
reliance ‘may very well be psychologically, morally, or even pragmatically relevant’,
whilst denying that it has epistemic significance.
One can recognize the significance of assurance and trustworthiness in the individ-
ual case without downgrading the epistemic significance of the ways in which we can
learn from others’ words without being mediated by assurance. Moreover, the various
considerations which may make assurance views of epistemic significance attractive in

the individual case do not easily transfer to the group case.
For example, Moran argues that the fact that individual speakers freely choose what
to say is what makes testimony epistemically valuable to the audience, indeed more
valuable than a hypothetical opportunity to directly inspect the speaker’s beliefs. But
this is puzzling given a non-assurance view of testimony which regards speech as mere
evidence of the speaker’s beliefs:
If speech is seen as a form of evidence [of the speaker’s beliefs], then once its intentional char-
acter is recognized (that is, not just as intentional behavior, but intentional with respect to
inducing a particular belief) we need an account of how it could count as anything more than
doctored evidence. (2006: 278)
On the assurance account, argues Moran, the fact that the speaker opts to ‘stand
behind’ her words, actively offering them as reasons to believe, is an advantage, not a
disadvantage to the audience. (Keren 2012 challenges Moran on this front.)
The analogous ideas are much less compelling in the group case. First, it is not obvi-
ous that non-assurance accounts of the epistemic significance of group statements
must see them as evidence of group belief, especially since the very notion of group
belief is disputed. Second, group statements do not obviously have the status of freely
chosen words, in any relevant sense. Instead, we can see groups as producing state-
ments via the functioning of various internal mechanisms; we may then consider
whether those mechanisms lend themselves to the production of truth or false state-
ments. Non-assurance accounts of group testimony are not challenged by the consid-
erations Moran raises for the individual case.
Advocates of assurance-style accounts of individual testimony sometimes lean on
ideas about respect for others as interlocutors. For example Fricker (2007: ch. 6) argues
that, under some circumstances, regarding others as mere sources of information
rather than as informants involves an ethically dubious brand of objectification. And
Hinchman (2005) motivates his account with reference to the insulting ‘slights’ and
‘rebuffs’ we create when we do not accept what we are told. But these considerations
bear much less weight in group cases: objectifying group testifiers is problematic or
insulting only to the extent that it in fact amounts to objectifying or insulting individ-
ual members of the group, perhaps on the basis of their group membership.
Tollefsen (2007) argues that Burge-style anti-reductionism about the epistemology of
testimony does not fit well with group statements: there is no default of accepting what
groups say in the absence of evidence of their reliability. Tollefsen sees this as favouring a
reductionist rather than an anti-reductionist approach to the evidence offered by testi-
mony, considering an assurance view only in her later work. But I think similar consider-
ations tell against an assurance account of group testimony: even if we needed the
epistemic boost provided by assurance to get the practice of testimony going in the first
place, the resources thereby provided through individual testimony seem sufficient to
underpin an evidential account of the epistemic significance of group testimony.
To summarize: we can acknowledge the epistemic significance of group testimony

without requiring a notion of group assurance, responsibility, or trustworthiness;
moreover, the arguments used to motivate assurance accounts of the epistemic signifi-
cance of individual testimony do not easily transfer to the group case. This does not
demonstrate that assurance views are false in the group case, but it does place a heavy
burden of proof upon their supporters, over and above the burden of proving the
strength of assurance accounts in the individual case.
8. Group Trustworthiness in Testimony:

Ethical Aspects
I have argued that we do not need the distinction between group trustworthiness and
mere group reliability in order to make sense of our epistemic practices around group
testimony. What about our ethical practices? The epistemic significance of assurance is
contentious even in the individual case. But no one doubts the ethical significance of
trustworthiness for individuals; lying is a paradigmatic untrustworthy act, with para-
digmatic ethical significance, and reckless testimony which pays no regard to truth or
falsity is likewise ethically culpable. So it might seem that, if we could not attribute lies
(as opposed to mere false statements) to a group, nor regard groups as either trust-
worthy or untrustworthy, then we would lose an important aspect of accountability
both in law and in morals.
The issue of corporate or collective moral responsibility is complex and contested
(e.g. Isaacs 2011; van de Poel, Royakkers, and Zwart 2015), and I cannot hope to engage
it here in any depth. Instead, I will focus on an issue which emerges more directly from
the literature on trust and trustworthiness: reactive attitudes. Following Baier, philo-
sophers distinguish trust and reliance as directed at individuals because of the intimate
connections between trust and certain reactive attitudes, connections which are miss-
ing in the case of reliance. And certainly some of us seem ready to react to groups and
organizations with attitudes like loyalty, gratitude, resentment, and a sense of betrayal.
If groups are not genuinely trustworthy or untrustworthy with regard to their testi-
mony, merely reliable or unreliable, then such reactions would seem to be mistaken.
I think many of us do direct trust- and distrust-related reactive attitudes towards
groups, and that these are indeed misplaced if groups can be neither trustworthy nor
untrustworthy (Thompson (forthcoming) discusses reactive attitudes in connection
with groups). Nevertheless, there are two related reactions which can be appropriate,
even without group trustworthiness or untrustworthiness. First: individual people are
liable to praise and blame, and to gratitude and resentment, for their personal trust-
worthiness and untrustworthiness in helping to generate group statements. Second: it
is appropriate to react positively (negatively) to living in a society in which important
groups and institutions are reliable (unreliable) producers of true statements. I will
examine these ideas in turn.
Imagine a situation in which a number of people undertake to help ensure that a

group issues reliable statements. This may include members of the group, who can fill
different roles in their organization, but may also include non-members, for example
people who founded or designed the group, or people now tasked with regulating or
overseeing its activities. Such individuals may be more or less trustworthy in playing
their various roles, and can sensibly be judged, praised, or resented for their actions in
that capacity.
This picture is reminiscent of Fricker’s joint commitment account of group assur-
ance, but I draw different conclusions from it. Each individual takes responsibility for
helping to ensure that the group produces a true statement, but in line with my argu-
ments of the previous section, none need take responsibility for the truth of the group’s
statement in the way required for assertion, and nor does the group itself take such
responsibility. Even if we endorse a Williamsonian knowledge norm on assertion, for
example, we need see nothing problematic in a situation where none of the group
members knows the truth of the group’s statement. On this picture, a group statement
is like a reading provided by a complex machine: we can hold the designers, operators,
and maintainers responsible for ensuring that the machine provides accurate readings,
without regarding any individual as asserting the content of the reading.
So reactive attitudes connected to ‘trusting’ a group can sensibly be directed at indi-
viduals connected to the group. But is this really feasible where the audience does not
know who these individuals are? Yes: you easily resent the person, whoever it was, who
wrote graffiti on your front door, and your feelings are quite different about the wind
which inconveniently blew litter into your garden. Likewise, you can resent the indi-
viduals who contributed to the publication of a misleading report, even if you do not
know who those individuals are.
So far, I have focused on situations in which we can regard individuals as trust-
worthy to the extent that they help to ensure the reliability of a group’s statements. But,
as I suggested in response to Fricker, other situations do not fit this pattern. There
could be situations in which the activity of various people leads to a reliable group
‘statement’, although the individuals themselves are merely reliable or unreliable,
rather than trustworthy or untrustworthy. Perhaps the ‘wisdom of crowds’ cases are
like this, at least where crowd members have no particular responsibility to be careful
or accurate. But, appropriately, if we appreciate that we face such a situation, we are not
tempted to adopt reactive attitudes towards the crowd.
Less innocuously, there seem to be situations in which individuals perform well in
their roles exactly to the extent that a group produces unreliable statements, precisely
because that is the group’s purpose. Perhaps this is true of a public relations agency, a
lobbying group, or a propaganda ‘machine’. It is natural for an audience to feel resentful
if they discover the unreliability of their sources, yet there may be no individual who is
untrustworthy in the sense of not doing her job properly. Is this a situation in which we
must either take up reactive attitudes towards the group itself, or else abandon the par-
ticipant stance?
No. In many such circumstances, there are individuals who may legitimately be
resented for creating or perpetuating an agency whose main function is to mislead.
Whether this involves a failure of trustworthiness, rather than some other sort of
moral or political failure, may depend upon which detailed account of trust and trust-
worthiness we espouse, but reactive attitudes reach beyond the domain of trustworthi-
ness and untrustworthiness in any case.
In other circumstances, there may be no individual(s) to blame, and yet still it seems
reasonable to feel resentful about societal structures: this is the kind of generalized
resentment I alluded to above. For example, we might well feel angry that we live in a
society where many media outlets fail to provide us with reliable information, even
supposing (implausibly) that there are no specific individuals who can be held account-
able for this. This raises large issues within political and social philosophy which I can-
not explore here; a natural first step would be to use Iris Marion Young’s notion of
‘structural injustice’ (e.g. Young 2011).
I have suggested that we can retain much of our ethical practice around group testi-
mony without needing to invoke notions of group trustworthiness and untrustworthi-
ness, as opposed to group reliability and unreliability. But the connection with reactive
attitudes was not the only ethical motivation for distinguishing trustworthiness and
reliability in the individual case: I also suggested that the former, unlike the latter, is a
candidate virtue.
Fricker (2012), drawing on Lahroodi (2007), offers an account of institutional epi-
stemic virtues such as fair-mindedness. She focuses primarily on the ways in which
individuals, acting as members of collectives, can display virtue and vice. She also dis-
cusses virtue and vice in institutional structures, arguing that an institution in isola-
tion cannot be virtuous or vicious, only as it is populated by individual people.
Responding to Fricker (2007), Elizabeth Anderson emphasizes the need for structural
remedies to testimonial injustice, over and above remedies involving individual action
and cultivation of virtue. She says, ‘when the members of an organization jointly com-
mit themselves to operating according to institutional principles that are designed to
achieve testimonial justice . . . this is what it is for the organisation itself to be testimoni-
ally just’ (2012: 168–9).
These are compelling thoughts: we need to address structural problems and collect-
ive contexts if we are to combat injustice and create better institutions. However, we do
not need the distinction between trustworthiness and mere reliability at the group
level in order to pursue these projects. The distinction matters at the individual level
precisely because we cannot always require others to be reliable in respects which mat-
ter to us. I cannot require that you read all my published works, and if you do not do so,
then you are unreliable in that respect, but in no sense untrustworthy. There is no gen-
eral obligation upon individuals to be reliable, which is why we need the language of
trustworthiness to highlight those particular respects in which individuals are obliged
to be reliable. But we can require of our institutions that they be reliable in the respects
that matter to us, or at least we can require this of our public institutions.
9. Trustworthiness and Group Action

I have focused on trustworthiness and reliability in the context of group testimony,
arguing that we can abandon the former notion without incurring significant costs.
Does the same go for trustworthiness in the context of group action? Some of the issues
I have considered—for example the connections between trustworthiness, epistemic
responsibility, and assertion—seem specific to the case of testimony, so the transition
will not be seamless. Matters are complicated by the fact that, even in the individual
case, different philosophical accounts of trust distinguish between trust and mere reli-
ance in disparate ways: what it takes for an issue to become a question of trust may be
to do with the perceived or actual motivations, interests, commitments, or obligations
of the actor, according to different theories.
But the ethical issues I discussed in the preceding section do seem to apply not just
to the case of testimony but to group behaviour more generally. The trust–reliance dis-
tinction is motivated by the connections between trust and certain reactive attitudes,
and also through the moral significance of trustworthiness, over and above reliability.
I suggested above that although we do in fact direct trust-related reactive attitudes
towards groups, with regard to their testimony as well as their actions more generally,
we could more appropriately direct these towards individuals, known or unknown,
group members or not, who are in various ways responsible for the functioning of the
group. To generalize this account, we would need to think carefully about what we can
reasonably expect of group behaviour, and also about the ways in which individuals
can be responsible through inaction, as well as through action.
Moreover, there are significant differences in our thinking about trustworthiness in
individuals and trustworthiness or reliability in groups. Regarding an individual as
trustworthy is typically caught up with respect for her as autonomous in some way:
someone who reliably follows a benevolent despot’s orders under duress is not display-
ing her trustworthiness. Trustworthy behaviour often, though not always, reflects a
determination to fulfil obligations or commitments which were voluntarily acquired.
Likewise, it can be important to recognize when an individual’s behaviour is not a mat-
ter of trustworthiness or untrustworthiness: if the head teacher does not release the
noisy children at noon, so I am not reminded to take my lunch break, I should recog-
nize that she has not betrayed me in any way.
Do we need to regard groups and organizations in the same way? It may depend
upon the group, its constitution and supposed purpose, including questions about
whether an organization is an element of the state or a private entity, and whether
membership of a particular group is chosen or imposed. To what extent do certain
groups have obligations to individuals, or to other groups? As these considerations
indicate, issues of trustworthiness in group action quickly involve larger questions
about collective responsibility and commitment; a fruitful line of inquiry would be to
investigate whether or not these questions can be sidestepped by thinking in terms of
reliability rather than trustworthiness.
10. Onwards and Outwards

I have considered the question of whether we need a notion of group trustworthiness,
over and above group reliability, in order to understand the epistemic and ethical sig-
nificance of group testimony, and my tentative answer was ‘no’: reliability is all that we
need. Likewise, I suggested, we do not need this distinction in order to understand our
attitudes to group behaviour more generally. My discussion was not comprehensive,
and I do not regard my arguments as conclusive. Nevertheless, I have begun to cast
doubt on the value of the trustworthiness–reliability distinction as applied to groups.
Suppose that in fact this distinction does no useful work. Does this mean we should
abandon talk of group trustworthiness? Should we regard public and interdisciplin-
ary discussion of organizational trustworthiness (e.g. Hawley 2012: ch. 8) as simply
confused?
At one level this is a strategic issue about choice of terminology, and about whether
it is worth struggling to impose quasi-technical terms onto broader discourse. A
related issue arises even in the individual case. As I noted earlier, although the trust–
reliance distinction is rightly valued by philosophers discussing interpersonal trust, it
is not consistently marked in ordinary language: we often talk of trusting or distrusting
inanimate objects, for example, or indeed trusting someone to get things wrong. One
response is to change terminology. Instead of contrasting trust and reliance, we might
follow Hollis (1998: 10) in distinguishing normative trust from predictive trust or, as
Faulkner (2007: 880) prefers, affective trust from predictive trust. Whilst these choices
may have independent philosophical merit, they do not help us avoid jargon in the
public realm.
More substantively, whatever our terminology, if the trustworthiness–reliability dis-
tinction lacks merit in the group case—or, in other terms, if there is a type of trust-
worthiness which individuals but not groups can exemplify—then this is of significance
to wider debates. Even in the individual case, many public concerns about ‘trustworthi-
ness’ are really about reliability, but they are pressing concerns nonetheless.
Moreover, many public concerns about trust and groups are in fact best construed as
(genuine, pressing) concerns about trusting individuals in a group context. I have dis-
cussed some aspects of this above, in connection with individual responsibilities for
group behaviour. But in addition distrust or trust of an individual may be caused or
rationalized by beliefs about her group membership. Fricker (2007) shows how such
judgements can reflect ethically culpable prejudices, but not all group-based trust or
distrust is problematic in this way. Indeed, if we construe ‘group membership’ very
loosely, almost any trust or distrust can be construed as group-based: I trust you
because of your good track record—your membership of the group of people who have
good track records.
This notion of group membership may seem trivial, but other groups are more
robustly unified: they are held together not just by some common feature of the
members, but by informal or formal relations and structures within the group. Such
relations and structures may genuinely help to cause, promote, or constitute the
trustworthiness of group members in relevant domains, either by generating sanc-
tions and rewards for individual members, or by enforcing trustworthiness-based
entry barriers. Professional self-regulatory bodies are supposed to fit this pattern:
membership of the body indicates professional competence. On the other hand, of
course, groups may cause, promote, or indicate the untrustworthiness of individual
members in certain respects, for example by creating incentives for members to put
group loyalty above trustworthiness to non-members, or by reducing members’
competence in certain respects.
The ways in which different group structures and identities can enhance or discour-
age individual trustworthiness in different domains are investigated empirically in
organizational studies, and by social psychologists. Consider peer pressure, or the var-
iety of phenomena which could be studied under the heading of ‘institutional’ or ‘cor-
porate’ culture or climate. For example, such local cultures can influence which
respects of trustworthiness are seen as most important. Is it a priority to defend your
team, or the company, or to ‘champion’ the client, or blow the whistle where necessary?
It is easy to get into a situation where the demands of trustworthiness point in different
directions, and different cultures may indicate different resolutions of such dilemmas.
Another dimension of variety may be the importance placed upon individual trust-
worthiness as opposed to other virtues or goals.
Finally, although I have distinguished our attitudes to groups from our attitudes to
individuals in group contexts, this distinction is not always sharply marked in practice.
We often interact with individuals as members, leaders, spokespersons, or representa-
tives of groups, and we may do so with varying degrees of trust. But it can then be
indeterminate whether our attitude of (dis)trust targets the individual, the group, or
both; if both, there is a further question as to which comes first, epistemically speaking.
Do I trust the customer services representative because I have found the company to
be reliable, or does my relationship with the individual come first? There will of course
be no single answer to such questions. In fact, this gives us another good reason to start
with the notion of trustworthiness or reliability—what does it take for an individual or
group to exhibit these features?—rather than with our often indistinct attitudes of trust
and distrust, reliance and non-reliance.
Acknowledgements
Material from this chapter was presented at the New Philosophical Perspectives on Trust con-
ference held at the Blavatnik School of Government in Oxford; at a workshop on Trust at the
University of Manchester; as a Women in Philosophy lecture at the University of Graz; at a
seminar at King’s College London; and (remotely) to the Cambridge Trust seminar. Many
thanks to the organizers and participants on all those occasions for their hospitality and philo-
sophical engagement. In addition I am grateful to Paul Faulkner, Jennifer Lackey, Tom
Simpson, Chris Thompson, Deborah Tollefsen, and Leo Townsend, for written comments and
questions. This work was supported by a Major Research Fellowship from the Leverhulme
Trust, which I very gratefully acknowledge.
References
Anderson, Elizabeth S. (2012), Epistemic Justice as a Virtue of Social Institutions, Social
Epistemology 26.2: 163–73.
Baier, Annette (1986), Trust and Antitrust, Ethics 96.2: 231–60.
Baier, Annette (2013), What is Trust? in Reading Onora O’Neill, ed. Archard, Deveaux, Manson,
and Weinstock, London: Routledge: 175–85.
Bonjour, Laurence (2002), Internalism and Externalism, in Oxford Handbook of Epistemology,
ed. Paul K. Moser, Oxford: Oxford University Press: 234–63.
Booth, Ken and Nicholas J. Wheeler (2008), The Security Dilemma: Fear, Cooperation and Trust
in World Politics, Basingstoke: Palgrave Macmillan.
Brandom, Robert (1983), Asserting, Noûs 17: 637–50.
Craig, Edward (1990), Knowledge and the State of Nature, Oxford: Oxford University Press.
De Bruin, Boudewijn (2015), Ethics and the Global Financial Crisis, Cambridge: Cambridge
University Press.
Faulkner, Paul (2007), On Telling and Trusting, Mind 116: 875–90.
Faulkner, Paul (2011), Knowledge on Trust, Oxford: Oxford University Press.
Fricker, Miranda (2007), Epistemic Injustice, Oxford: Oxford University Press.
Fricker, Miranda (2009), Can There be Institutional Virtues? in Oxford Studies in Epistemology,
vol. 3, ed. Gendler and Hawthorne, Oxford: Oxford University Press: 235–52.
Fricker, Miranda (2012), Group Testimony? The Making of a Good Informant, Philosophy and
Phenomenological Research 84.2: 249–76.
Gilbert, Margaret (2006), A Theory of Political Obligation: Membership, Commitment, and the
Bonds of Society. Oxford: Oxford University Press.
Hawley, Katherine (2012), Trust: A Very Short Introduction, Oxford: Oxford University Press.
Hawley, Katherine (2014), Trust, Distrust and Commitment, Noûs 48.1: 1–20.
Hinchman, Edward (2005), Telling as Inviting to Trust, Philosophy and Phenomenological
Research 70.3: 562–87.
Hodgson, Geoffrey (2006), What are Institutions?, Journal of Economic Issues 40.1: 1–25.
Hollis, Martin (1998), Trust Within Reason, Cambridge: Cambridge University Press.
Isaacs, Tracy (2011), Moral Responsibility in Collective Contexts, Oxford: Oxford University Press.
Jones, Karen (2012), Trustworthiness, Ethics 123.1: 611–85.
Keren, Arnon (2012), On the Alleged Perversity of the Evidential View of Testimony, Analysis
72: 700–7.
Lackey, Jennifer (2008), Learning From Words, Oxford: Oxford University Press.
Lackey, Jennifer (2014), A Deflationary Account of Group Testimony, in Lackey (ed.) Essays in
Collective Epistemology, Oxford: Oxford University Press.
Lackey, Jennifer (forthcoming), Group Assertion, Erkenntnis.
Lackey, Jennifer (draft), Group Lies.
Lahroodi, Reza (2007), Collective Epistemic Virtues, Social Epistemology 21.3: 281–97.
List, Christian and Philip Pettit (2011), Group Agency: The Possibility, Design, and Status of
Corporate Agents, Oxford: Oxford University Press.
McMyler, Ben (2007), Knowledge at Second Hand, Inquiry 50.5: 511–40.

Mäkelä, Pekka and Cynthia Townley (2013), Editors’ Introduction, Trust: Analytic and Applied
Perspectives, ed. Mäkelä and Townley, Amsterdam: Rodopi.
Mathieson, Kay (2006), The Epistemic Features of Group Belief, Episteme 2.3: 161–75.
Moran, Richard (2006), Getting Told and Being Believed, in The Epistemology of Testimony, ed.
Lackey and Sosa, Oxford: Oxford University Press: 272–306.
O’Neill, Onora (2013), Responses in Reading Onora O’Neill, ed. Archard, Deveaux, Manson,
and Weinstock, London: Routledge: 219–39.
Peirce, C. S. (1932), Collected Papers of Charles Sanders Peirce, vol. 2, ed. C. Hartshorne and
P. Weiss, Cambridge, MA: Harvard University Press.
Saunders, Mark N. K., Denise Skinner, Graham Dietz, Nicole Gillespie, and Roy J. Lewicki (eds.)
(2010), Organizational Trust: A Cultural Perspective, Cambridge: Cambridge University Press.
Schmitt, Frederick (1994), The Justification of Group Beliefs, in Socializing Epistemology, ed.
Schmitt, Lanham, MD: Rowman and Littlefield: 257–88.
Searle, John (1969), Speech Acts: An Essay in the Philosophy of Language, Cambridge: Cambridge
University Press.
Simon, Judith (2013), Trust in D. Pritchard (ed.) Oxford Bibliographies in Philosophy, New
York: Oxford University Press.
Surowiecki, James (2004), The Wisdom of Crowds: Why the Many Are Smarter than the Few,
London: Little, Brown.
Sztompka, Piotr (1999), Trust: A Sociological Theory, Cambridge: Cambridge University Press.
Thompson, Christopher (forthcoming), The Moral Responsibility of Group Agents.
Tollefsen, Deborah (2006), Group Deliberation, Group Cohesion and Scientific Teamwork: Is
there Room for Dissent? Episteme 3: 37–51.
Tollefsen, Deborah (2007), Group Testimony, Social Epistemology 21.3: 299–311.
Tollefsen, Deborah (2011), Groups as Rational Sources, in Collective Epistemology, ed. Hans
Bernhard Schmid, Daniel Sirtes, and Marcel Weber, Berlin: Ontos.
van de Poel, Ibo, Lambèr Royakkers, and Sjoerd D. Zwart (2015), Moral Responsibility and the
Problem of Many Hands, London: Routledge.
Watson, Gary (2004), Asserting and Promising, Philosophical Studies 117.1–2: 57–77.
Williamson, Timothy (2000), Knowledge and its Limits, Oxford: Oxford University Press.
Young, Iris Marion (2011), Responsibility for Justice, Oxford: Oxford University Press.
15
Faith in Kant
Guy Longworth
1. Introduction: Faith and Trust

Can it ever be reasonable to trust other people, or to rely upon them, without evidence
of their trustworthiness or reliability? According to Kant, it can. There can be practical
reasons for holding things true, including that other people are trustworthy or reliable.
I shall examine Kant’s case for allowing that practical reasons can render trust in other
people reasonable.
As limited beings, we often need to rely on the acts and attitudes of other limited
beings. For example, I might need your help to reach a high shelf; you might need
mine to open a lid. For it to be reasonable for us to rely upon others, it seems that we
must have reasons to hold true that those we rely on will be appropriately reliable.
And yet many of the circumstances in which we need to rely on others are precisely
circumstances in which we lack evidence for holding true that they will be reliable.
Thus, our practical and epistemic limitations conspire to produce a puzzle: how, if at
all, can it be reasonable for us to do what it seems that we must: rely on, or trust, other
limited beings?
Different cases might well call for different responses. One response to a case of the
puzzle would be to withhold trust: we might sacrifice the benefits to be secured by rely-
ing on others in order to preserve our claim to reasonableness. A second response
would be to rely on others whilst acknowledging that doing so is unreasonable. A third
response would be to argue that our epistemic limitations are less stark than the puzzle
makes them out to be—that, in a particular case, we have sufficient evidence to hold
that people will be reliable. A fourth response would involve attempting to show that
our reliance on others can be reasonable despite our lacking evidence, because we can
possess practical reasons for holding that others will be reliable.
I want to explore a particular version of the fourth response to the puzzle. Kant’s
aim, in the discussions on which I focus, is to secure the reasonableness of what he calls
moral faith. Moral faith, for Kant, is a form of holding something true that is grounded
in practical reason. In particular, faith is grounded in the moral demands to which
practical reason is subject. Thus, faith is grounded in practical rather than theoretical
252 Guy Longworth
reason. It is a matter of holding, on moral grounds, that we are free, that we are immor-
tal, and that God exists.
Kant understands faith in God to involve a form of trust:
Faith (simply so called) is trust in the attainment of an aim the promotion of which is a duty
but the possibility of the realization of which it is not possible for us to have insight
into . . . (Kant 1793: 5: 472)
Faith means the confidence that, so long as we have done everything possible to us, God will
supply what does not lie in our power . . . The sole object of spiritual trust is in the pure morality,
the holiness of man, and then his eternal blessedness under the conditions of morality . . . Hence,
in order that our trust may coincide with the plan of wisdom, it must be a wise trust, and
unconditional, so that we believe in general that God, in His goodness and holiness, will both
lend us His aid in regard to acting morally, and also allow us to participate in blessedness.
(Kant 1784–5: 27: 321–2)
However, my central interest in Kant’s discussion is not in his entitlement to the spe-
cific objects of faith that he emphasizes. I draw on Kant’s account of moral faith in
order to provide a general justification of the rationality of holding things true on
practical grounds—a justification that can then be applied to the special case of hold-
ing true that someone is, in relevant respects, worthy of trust. On this view, trusting
someone can require holding true that they are trustworthy, without also requiring
evidentially grounded belief in their trustworthiness. (The view thus connects with
Pamela Hieronymi’s (2008) proposal that fully trusting someone requires believing
that they are trustworthy. The view affirms that trusting someone requires holding true
that they are trustworthy, but rejects the assumption that holding true is a form of
belief. It therefore avoids the problems that Hieronymi raises for views that seek to
base belief in trustworthiness on purely practical reasons.)
I proceed as follows. In section 2 (‘Kant on Moral Faith’), I expound Kant’s case for
moral faith and highlight two principles that figure in that case: that setting an end
requires holding true that the end is in principle attainable; and that it can be rational to
hold true that one’s ends are attainable on practical grounds, in the absence of evidence.
Section 3 (‘Problematizing Faith’) raises the central difficulty to be pursued: the need to
divorce the attitude of holding true that is involved in faith from evidentially grounded
belief. Sections 4 (‘Defending Kant’s Principles’) and 5 (‘Faith and Evidence’) develop a
case for the required divorce, thus opening space for the possibility of rationally holding
things true for practical, as opposed to theoretical, reasons. In the concluding section,
I indicate how the general account derived from Kant’s case for moral faith provides the
basis for a deflationary account of the special case of trust in other people.
2. Kant on Moral Faith

Let’s begin with a sketch of Kant’s case for moral faith in God. This will enable us to see
the main moving parts of Kant’s case, and to isolate those that are of greatest relevance
faith in kant 253
to the reasonableness of extra-evidential trust. (Kant presents versions of his basic case
for moral faith in a large number of places, in particular: 1781/1787: A795/B823–
A831/B859; 1786; 1788: 5: 107–48; 1790: 5: 397–415, 5: 434–84; 1793: 6: 3–11. For
present purposes, I won’t be attending to the fine details of the texts.)
Kant’s aim is to show that it is reasonable to hold true, with certainty, that there exists
a being that is both willing and able to organize the world in accord with a moral design.
Kant takes himself to have demonstrated, in the course of the first critique, that it is
impossible for us to know that there is (or that there isn’t) such a being. It is impossible
for us to have theoretical, or evidential, grounds for holding with certainty that such a
being exists. For evidence that is accessible to us must be grounded in our intuition, and
our intuition is restricted to appearances; but if God exists, then He is super-sensible, so
evidence for or against His existence is not to be found amongst the appearances.
Reasons for holding with certainty that God exists must be non-evidential.
Practical reason provides the required non-evidential reasons. For practical reason
gives rise to the moral law. Practical reason demands, moreover, that we act from
respect for the moral law. And Kant argues that we can act rationally from respect for
the moral law only on condition that we set what he calls the highest good as our ultim-
ate end. The highest good unifies all more specific moral goods. The highest good
would be a state of the world in which all persons acted only from respect for the
moral law and, moreover, a state in which their doing so gives rise to their also being
proportionately (so maximally) happy. But the rationality of our setting the highest
good as our ultimate end depends upon its being reasonable for us to hold that the
highest good is in principle attainable. And its being reasonable for us to hold that
the highest good is in principle attainable requires that it is reasonable for us to hold
true anything on which its in principle attainability depends. In particular, it requires
that it is reasonable for us to hold two things. First, it requires that it is reasonable for
each of us to hold not only that we will act only from respect for the moral law, but in
addition that everyone else will do so too. And second, it requires that it is reasonable
for us to hold that if everyone acts only from respect for the moral law, then their
doing so will be efficacious in determining that everyone is made happy in proportion
to their moral worthiness. Holding true that everyone will act from respect for the
moral law requires reasonable faith or trust in one’s own potential morality as well as
in that of all other people (Kant 1781/1787: A810/B838; 1785: 106). Holding true that
this will lead to proportionate happiness imposes a disjunctive requirement: either it
must be reasonable to hold that nature is organized so that morality leads to propor-
tionate happiness, or it must be reasonable to hold that the hedonic efficacy of moral-
ity is enforced supernaturally, by God. But we have evidence against holding that
nature is so organized. Hence, the reasonableness of our setting the highest good as our
end depends upon the reasonableness of our holding that God will enforce the efficacy
of morality in determining happiness. Since it is reasonable for us to act from respect
for the moral law, it is reasonable for us to set the highest good as our end; and since it
is reasonable for us to set the highest good as our end, it is reasonable for us to hold
254 Guy Longworth
that God exists and will ensure that everyone’s doing their part—everyone’s acting
from respect for the moral law—will lead to their proportionate happiness.
The sketched argument requires development at a number of points. First, the claim
that it is reasonable to act from respect for the moral law (at least, pro tanto, on the
assumption that the demands imposed by reason are consistent) requires defence.
Second, the claim that acting from respect for the moral law demands that we set the
highest good as our end requires support. Kant vacillates over the precise relationship
between the moral law and the highest good and about the precise nature of the high-
est good. And his most detailed accounts of the relationship fail to make transparent
how precisely the rational demand that we act from respect for the moral law transmits
to a demand that we set the highest good as our ultimate end.
Third, it needs to be explained why it is reasonable to set an end only on condition
that it is also reasonable to hold that the end is in principle attainable. For the argument
would fail, not only if it was reasonable to set an end that one held to be unattainable,
but also if it were reasonable to set an end that one reasonably didn’t hold to be
unattainable. For the argument of the first critique precludes not only our acquiring
evidence for the existence of God, but also our acquiring evidence against His exist-
ence. So, we can have no evidential grounds for holding that God doesn’t exist, and so
no evidential grounds for holding that the highest good is unattainable. Thus, if the
requirement on rationally setting an end were only that we lack reason to hold that it is
unattainable, then we could meet the requirement independently of possessing posi-
tive reasons to hold true that God exists.
Fourth, an explanation is needed for why we shouldn’t hold that nature itself is
organized in accord with the attainability of the highest good. Even if we agree with
Kant that we currently lack evidence that nature is so organized, the absence of evi-
dence need not be due to a failure of correlation between moral worthiness and happi-
ness. It might be due, instead, to the current paucity of morally worthy people. If we
were required to hold positively that the highest good is attainable in the absence of
such evidence, our doing so would require a form of moral faith. But it would require
faith in the organization of nature, rather than a supernatural organizer.
The third and fourth issues are connected. For suppose that rationally setting an end
required only that one lacks reasons for holding that the end is unattainable and not
that one possesses reasons for holding that it is attainable. In that case, the absence of
evidence that nature is organized so as to sustain the attainment of the highest good
would preclude our setting the highest good as our end. And so even if practical reason
required us to set the highest good as our end, that wouldn’t require us to hold that
nature is organized so as to ensure the attainability of that end.
Fifth, argument is needed for the claim that the attainability of the highest good
must be secured either by the intrinsic organization of nature or by the organizational
work of God—that there is no third way.
Finally, sixth, the claim that the pro tanto practical reasonableness of our acting
from respect for the moral law transmits to our holding true that God exists, rather
faith in kant 255
than imposing on us a requirement to obtain independent theoretical grounds for

holding that God exists, wants defence. Now the requirement that we find independent
grounds might be one that we cannot in principle satisfy. In that case, and assuming
that the remainder of the argument were cogent, we would be presented with a conflict
between the demands set by practical reason and demands set by theoretical reason.
That might be unfortunate. But it’s not clear why the hope of avoiding conflict between
practical and theoretical reason should be taken to provide grounds for treating as
spurious their apparent demands. And we would anyway have to confront the further
question, why should it be the apparent demands of theoretical reason that are to be
surrendered, rather than those of practical reason?
Proper assessment of Kant’s argument would require extended labour and would
need to go to the heart of his moral theory. (An excellent attempt is made in Wood 1970.
See also Gardner 2006; Kleingeld 1998; Moore 2003; Neiman 1994; O’Neill 1996;
Timmermann 2009; Watkins 2010; Willaschek 2010.) However, for present purposes,
we can restrict attention to the third and sixth of the points just listed: the purported
demand on the rational setting of ends, according to which setting an end requires
reason to hold that the end is in principle attainable; and the claim that that demand
can serve to furnish extra-evidential reasons for holding things true. In doing so, we’ll
be pursuing the following questions:
Q1. What cognitive requirements must one meet if one is rationally to set an end?
Must one, for example, hold true that the end is in principle attainable?
Q2. What is the nature of the attitude of holding true that figures in meeting those
cognitive requirements? Is it a paradigmatic form of belief, or does it take some
other form?
Q3. To what extent can the required form of holding true be sustained rationally
by the needs of practical reason? Does the answer to that question depend on
whether the operative needs arise from an absolute demand on practical reason,
that we follow the moral law? And does it depend on whether the required form of
holding true concerns only things about which we cannot know?
3. Problematizing Faith
In order to fix ideas, it will be useful to sketch an argument that begins from Kant’s
principle governing the rational setting of ends and terminates in the conclusion that
there is an evidential requirement on setting ends that precludes the possibility of
moral faith.
The argument begins from a principle that Kant accepts (e.g. 1781/1787: A823/
B851–A832/B852):
(P1) One can rationally set an end, E, only if one rationally holds true that E is in
principle attainable.
256 Guy Longworth
Thus, for example, one can rationally set the end of winning the election if one holds
that it is possible for one to win. The attainability in principle of E typically depends,
and typically is known to depend, on two sorts of factors. First, E’s attainability depends
on what agents can do towards bringing the end about, conditional on their contingent
circumstances. That is, it depends on the range of things that are within an agent’s
power to bring about, given specific ways things outside their power are. My winning
the election depends on my campaigning. Second, E’s attainability depends upon those
aspects of the agent’s contingent circumstances that they are unable to affect. Since the
successful attainment of one’s ends is often dependent upon the activities of other
agents, the latter conditions will often include those other agents’ activities, and the
willingness to act on which those activities depend. My winning the election depends
on other people campaigning too. Let’s label the former type of necessary condition on
the attainability of E agential conditions, and the latter type of necessary conditions
environmental conditions. A consequence of (P1), then, is (P2):
(P2) One can rationally set an end, E, only if, with respect to all conditions that
one holds to be either agential conditions or environmental conditions, one ration-
ally holds true that those conditions obtain or will obtain.
That principle now interacts with principles purportedly governing what one may
rationally hold true so as to generate the putative difficulty for Kant’s account. So, sup-
pose that, in addition to (P2), one held (P3) and (P4):
(P3) One can rationally hold true that a condition will obtain only if one can
rationally believe that the condition will obtain.
(P4) One can rationally believe that a condition will obtain only if one possesses
evidentiary support for the obtaining of the condition.
(P3) and (P4) might be motivated by appeal to a conception of the constitutive aims of
holding true on which the constitutive aim of holding something true is truth or know-
ledge. Plausibly, such a conception of holding true would align the constitutive aim of
holding true with the constitutive aim of belief, so sustaining (P3). And it’s plausible that
the conception would, thus, sustain (P4) by sustaining the evidentiary requirement in
(P4). On the basis of those principles, we could derive an evidentiary requirement on
the setting of ends to the effect that one can rationally set an end only if one possesses
evidentiary support for the obtaining of agential and environmental conditions on the
attainability of the end.
That requirement on the rational setting of ends might appear too strong. For at
least from an agent’s own perspective, answering the question whether agential con-
ditions will obtain seems to depend, not on the acquisition of evidence, but rather on
making up one’s mind to pursue specific ends. It is up to me whether I decide to
campaign. However, we might attempt to finesse that condition by stipulating that
with respect to agential conditions, the evidentiary requirement can be met on the
basis of knowing which ends one has set. That leaves the environmental conditions.
faith in kant 257
It is natural to hold that decision-independent evidence is required for the obtaining

of environmental conditions. It is natural to hold, for example, that evidence is
required for believing that people are likely to vote for me.
On that basis, our principles dictate that with respect to ends for which there are
non-agential environmental conditions, one can rationally set an end, E, only if one
possesses evidence that E’s environmental conditions obtain. Thus, in cases in which
the environmental conditions include the activities of other agents, one must possess
evidence that the agents will be willing and competent to undertake those activities.
The conclusion of Kant’s argument for moral faith was that one can rationally set ends
with environmental conditions for the obtaining of which one doesn’t possess—
indeed, for which one cannot possess—any evidence at all. So, Kant’s claims for moral
faith conflict with the argument we have just developed, and it is incumbent on the
defender of Kant to say something in response. Minimally, they must provide grounds
for rejecting either (P3) or (P4). In doing so, they should say something about the con-
stitutive aims of holding true for practical purposes that supports the rejection of those
principles. That is a large project. I shall make a start on pursuing it in the following
two sections, by developing an alternative conception of holding true for practical pur-
poses. According to the alternative conception, the operative form of holding true
is not a paradigmatic form of belief (so that (P3) is unsupported) and, so, is not subject
to the same evidentiary demands as paradigmatic forms of belief (so that (P4) is
unsupported).
4. Defending Kant’s Principles

Kant’s argument depends upon the principles (P1) and (P2) described in section 3,
purportedly governing the connection between one’s rationally setting an end and
one’s holding true that the end is in principle attainable.
Setting an end for oneself is freely adopting the aim of doing what one can in order
to bring about, or preserve, the end. Setting an end has as upshot its passive counter-
part, having an end, itself a mode of intention. The question whether (P1) and (P2) are
correct is, therefore, a near relative of the question whether (rationally) intending to φ
entails (rationally) believing that one will φ. I’ll approach the question about ends
indirectly, via its connections with the question about intentions.
In attempting to address the question whether intending to φ entails believing that
one will φ, it is important to avoid conflating distinct issues. It is especially important
to avoid two tempting conflations.
First, we should avoid conflating the question whether a rational intention to φ
entails holding true that one will φ with the distinct question whether a rational inten-
tion to φ entails believing that one will φ. Although it may be that all cases of holding
true are cases of believing, it shouldn’t be assumed from the outset that they are.
Reasons to think that intending doesn’t entail believing may be precisely reasons to
think that holding true doesn’t entail believing.
258 Guy Longworth
Second, we should hold distinct two forms that intention can take: on the one hand,
having (or being) decided to φ and, on the other hand, merely having (set) the end of
φ-ing. If I have (or am) decided to φ, then I have the end of φ-ing. However, the converse
entailment fails. For I can have, and act upon, ends that I can’t bring about merely by so
acting—for example, I can have the end of winning a fair lottery. By contrast, I cannot
decide to win a fair lottery; at best, I can decide to try to win it. It would be natural, there-
fore, to expect the cognitive requirements on deciding to φ to differ from those on
deciding on the end of φ-ing. Suppose, for example, that deciding to φ entails believing
that one will φ. And suppose, furthermore, that deciding on the end of φ-ing entails
deciding to try to φ. It would follow that deciding on the end of φ-ing entails believing
that one will try to φ. But it would be implausible to hold, on those grounds, that believ-
ing that one will try to φ entails believing that one will φ. Moreover, given that thought
and talk about what someone intends can target either what they have decided to do or
their wider ends—for example, what they have decided to try to do—we would natur-
ally expect our judgements about the cognitive requirements on intending to be pulled
in opposing directions. On the one hand, insofar as our attention is focused on the
requirements on deciding to φ, we will find it more intuitive that they include holding
that one will φ. On the other hand, insofar as our attention is directed onto the require-
ments on merely deciding on the end of φ-ing, without deciding to φ, we will find it less
intuitive that they include holding that one will φ, and will expect to discern, instead,
somewhat weaker requirements on the rational setting of ends.
Although it is possible to decide on an end that depends upon one’s φ-ing without
thereby deciding to φ, we can nonetheless achieve insight into deciding on an end via
reflection on deciding to φ. Let’s begin, then, by considering principle (P5), governing
rationally deciding to φ:
(P5) One can rationally decide to φ only if one rationally holds true that one will φ.
Two questions about (P5) are central. First, what reason is there to accept (P5) as a
requirement on rational decision, rather than accepting some weaker principle?
Second, how should we understand the nature of the attitude of holding true that is
embedded in (P5)? How, in particular, does that attitude relate to evidence that one will
φ, and in particular evidence that obtains independently of one’s decision to φ? And
how, if at all, does one’s holding that one will φ figure in bringing it about that one will?
One reason for endorsing (P5) is that it seems obviously correct. It seems obvious,
for example, that I can rationally decide to campaign in the election only if I hold
true that I will campaign. Going beyond that, (P5) can figure in explaining some of
the other rational norms that govern deciding to φ. For example, it seems plausible
to accept that any reasonable account of holding true will impose a requirement like
the following:
(P6) One can rationally hold that p, and hold that its being true that p entails its
not being true that q, only if one doesn’t hold that q.
faith in kant 259
(P5) and (P6) can be used to explain why one can’t rationally both decide to φ and
decide to ψ while one holds that its being true that one will φ entails its not being true
that one will ψ. For example, they can explain why I can’t rationally decide to campaign
and decide to take a long vacation, given that I know that campaigning precludes vaca-
tioning. For according to (P5), one’s rationally deciding to φ entails that one rationally
holds that one will φ, and one’s rationally deciding to ψ entails that one rationally holds
that one will ψ. But according to (P6), one can’t rationally hold that its being true that
one will φ entails that one will not ψ and at the same time hold that one will φ and that
one will ψ. So, in conjunction with principles governing rationally holding true, (P5)
can figure in explaining why one can’t decide to do things that one takes to be mutually
incompatible.
Similarly, (P5) can figure, in conjunction with other principles governing rationally
holding true, in deriving an analogue of (P2) that applies to deciding to φ. That is, it can
support an analogue of Kant’s principle, according to which the rational setting of ends
requires holding that those ends are in principle attainable. Plausibly, (P7) governs
rationally holding things true:
(P7) One can rationally hold that p, and hold that its being true that p entails its
being true that q, only if one holds that q.
According to (P7), if one holds that its being true that q is a necessary condition on
one’s φ-ing, then one can rationally hold that one will φ only if one holds that q. But we
have from (P5) that one can rationally decide to φ only if one rationally holds that one
will φ. So, in the same circumstances, one can rationally decide to φ only if one ration-
ally holds that q. Thus, we can derive the following analogue of (P2) for deciding to φ:
(P8) One can rationally decide to φ only if, with respect to all agential and envir-
onmental conditions that one holds to be necessary for its being true that one will
φ, one rationally holds true that those conditions obtain.
Suppose, for example, that one holds that in order to canvass the electorate, one
must leave home by 11 a.m. And suppose that one knows that it is now well past 11 a.m.
In that case, according to (P8), what one holds true precludes one from now rationally
deciding to canvass the electorate. That is, reason would permit one to decide to can-
vass the electorate only if there were a change in what one knew or held true about one’s
circumstances.
To this point, we’ve focused on deciding to φ and haven’t spoken directly to (P1).
However, analogous considerations apply with respect to (P1), on the plausible
assumption that setting an end is at least approximately equivalent to deciding to try.
In both cases, there is a need to explain why certain patterns of decision or end-setting
would not be reasonable, and explaining that requires connecting decision and end-
setting with positive commitments about what one will do or try to do, and so with
what it is possible for one to do or try to do. A mere absence of commitment to impos-
sibility would not support the required explanations. However, although analogous
260 Guy Longworth
considerations apply to both deciding and setting an end, there are also important dif-
ferences between the two cases. In particular, (P1) is weaker than (P5) in a way that
means that it does not rule out the possibility of rationally setting incompatible ends.
For according to (P1), one can rationally set an end that one holds may not be attained.
Since it is possible rationally to hold that it is possible that p and possible that q, even
though it would not be rational to hold that it is possible that (p and q), the principle
therefore fails to rule out the rational setting of incompatible ends. Thus, for example,
it leaves open that I might set vacationing as an end, and set campaigning as an end; it
precludes only my setting as an end vacationing and campaigning (see e.g. Bratman
2009). That result is not itself dramatically implausible. However, if we wished to
defend the claim that perfect rationality is not consistent with the setting of incompat-
ible ends, we could impose it as a requirement, without violence to (P1), by the add-
ition of the following plausible seeming principle of unification:
(P9) One can rationally set oneself a totality of ends, Σ, only if one could rationally
set oneself as an end the conjunction of all ends in Σ.
On this view, it would be imperfectly rational to retain both the end of vacationing and
the end of campaigning. In order to be perfectly rational, I must decide on one end or
the other.
The explanatory power of (P1) and (P5) in underwriting the various rational
requirements that apply to deciding to φ and to setting oneself an end gives us pro
tanto reasons to endorse those premises. However, those reasons might be defeated
by reasons to hold that ordinary thought allows for the possibility of rationally decid-
ing to φ whilst failing to hold true that one will φ. And they might also be defeated by
our failure to make sense of rationally holding true that one will φ on the basis of a
decision to φ. I won’t here consider the first potential source of defeaters. (Holton 2009
discusses defeaters of the first sort, but from a perspective on which holding true is
identified with believing.) However, it is important to say something about how we
are to make sense of holding things true on the basis of decisions. To that end, I pro-
vide the outline of an account of the nature of deciding to φ that has the resources to
make transparent the required connection between deciding to φ and holding true
that one will φ. The account has been developed and defended by Matthew Soteriou
(2013: 257–307). As we’ll see, it has important affinities with Kant’s thoughts about
this topic.
Soteriou’s account of deciding to φ builds on earlier discussions of ways in which
one’s deciding now to φ can figure in future deliberation and action, especially discus-
sions by Michael Bratman (1987; 1999; 2006) and David Velleman (1989; 2000).
Crucially, when one has decided to φ, one’s further practical deliberation will take
place under the constraint that one will φ. Thus, deciding to φ is way of answering, for
practical purposes, the question, what will one do? For example, in deliberating about
what else to do, one will not, without revising or forgetting one’s decision to φ, consider
deciding to do things incompatible with one’s φ-ing.
faith in kant 261
As we’ve discussed, there are good reasons to think that one’s deliberating about
future activities on the basis of having decided to φ is to be explained by appeal to one’s
holding true, on the basis of one’s decision, that one will φ. That would be consistent
with the view, defended by Velleman, that deciding to φ is a matter of forming the
belief that one will φ. However, as Soteriou points out, there are good reasons to think
that, unlike paradigmatic cases of belief, where a subject has decided to φ,
. . . the constraint of treating as true the proposition that she will φ is a constraint on the sub-
ject’s planning that the subject regards as self-imposed. (Soteriou 2013: 287)
Soteriou, following Velleman (2000: 32–55), explains the way in which a subject will
regard their decision as a matter of their imposing constraints on their own deliber-
ation by appeal to the seeming absence of any external source for those constraints:
Acceptance of the truth of the relevant proposition is not one that the subject takes to be
grounded in evidence that she possesses, in so far as she takes herself to be epistemically
entitled, given her evidence, to make alternative decisions and hence to make any one of a
number of other inconsistent assumptions about what she is going to do. So when a subject
decides to φ, and then subsequently plans on the assumption that she is going to φ, the subject
assumes something about her own future on the basis of evidence that, from the subject’s own
point of view, simultaneously licenses her to assume something about her future that contra-
dicts it. Note that even after she decides to φ the subject still takes herself to be epistemically
entitled, given her evidence, to make an alternative decision, and thereby assume something
else about her future. (Soteriou 2013: 287)
To the extent that the constraints imposed on one’s practical deliberation by deciding
to φ are self-imposed, they are similar to the constraints one imposes on one’s theoret-
ical reasoning by supposing something for the sake of argument. Soteriou exploits the
comparison in order to illuminate his proposal about the nature of deciding to φ:
When you assume that p for the sake of argument, you treat p as true, you regard and treat this
constraint on your reasoning as self-imposed, and part of what is involved in treating the con-
straint as self-imposed is your treating the assumption as one that is to be discharged—e.g. with
an outright conditional judgement that is outside the scope of the supposition. Likewise, when,
having decided to φ, you plan on the assumption that you will φ, you regard and treat this con-
straint on your planning as self-imposed. And likewise, I want to suggest, part of what is
involved in treating the constraint as self-imposed is your treating the assumption as one that
is to be discharged. However, in the case of your planning assumption . . . you treat the assump-
tion as one that is to be discharged by the performance of an action that makes the assumption
true. (Soteriou 2013: 288)
Deciding to φ is a matter of imposing on one’s own future practical deliberation and

activity a specific constraint. It is a matter of imposing on one’s deliberation and
activity the constraint of holding that one will make it true that one φs and, so, that
one will φ. In treating the constraint as self-imposed, we view our holding true
as constraining us only via our sustained willingness to remain so constrained.
262 Guy Longworth
In deciding to campaign, I hold true that I will campaign, and treat that as a con-
straint on further practical deliberation. But I treat it as a self-imposed constraint
and, so, as figuring in my practical deliberation only insofar as I remain committed
to campaigning. Thus, in deciding to φ one is constrained by one’s own willingness,
rather than by one’s evidence. Soteriou’s account of the power to decide thereby fits
Kant’s conception of an autonomous will, according to which the autonomy of a per-
son’s will amounts to its being
a free will which, in accordance with its universal laws, must necessarily be able at the same
time to agree to that to which it is to subject itself. (Kant 1788: 5: 132)
Deciding to φ is a matter of imposing on one’s own practical deliberation the con-

straint that one will φ. Similarly, but more generally, setting an end as a matter of
imposing on one’s own practical deliberation the constraint that one will strive to
attain that end. Thus, we have presented an intelligible account of why, and how, prin-
ciple (P1) can be true.
5. Faith and Evidence

Soteriou’s account of deciding to φ makes transparent the connection characterized in
(P5) between deciding to φ and holding true that one will φ. The picture is one in
which, prior to deciding what to do, one is faced with an array of evidence about one’s
capabilities in the circumstances. One is faced with evidence to the effect that there are
some things one cannot do, and so some things that one will not do. If cognizant of that
evidence, one cannot rationally decide to do those things. Thus, I have evidence that
I can’t reach the hall in time to speak to the electorate, and so I can’t rationally decide to
reach the hall in time. One is also faced with evidence to the effect that there are some
things one can do. Typically, one is so presented with a range of things that one can do,
and one’s evidence fails to dictate which amongst them are things that one will do.
To that extent, it is up to oneself which, if any, of those things one decides to do. Thus,
I have evidence that I can canvass any of the houses on this street and it is up to me
which house I decide to canvass. One therefore takes oneself to be free, with respect to
any φ in that range, to decide whether or not to φ. And although one treats a decision to
φ as constraining one’s future practical deliberation, one views that constraint as one
that is self-imposed, and so as a constraint that one will respect only to the extent that
one remains willing to adhere to it. So one does not treat a decision to φ as affecting
one’s evidential situation, by placing one’s not-φ-ing outside the range of actions that
one is then in a position to undertake. Thus, one’s decision to φ is not treated as under-
writing an ordinary theoretical belief that one will φ.
There are things that one has evidence that one cannot do. One cannot decide to do
those things. There are things that one has evidence that one can do. One can decide to
do those. To a good first approximation, reasons for belief suffice for reasons to hold
things true. Given our epistemic limitations, there are liable also to be things about
faith in kant 263
which one lacks evidence either way: potential courses of action for which one lacks
evidence that one cannot complete them and also lacks evidence that one can complete
them. For example, I lack evidence that I can speak loudly enough to be heard at the
back, and also evidence that I can’t. Where one’s φ-ing would fall within that penum-
bral region, is it possible for one rationally to decide to φ, where that would be a matter
of one’s holding true, by way of a planning assumption, that one will φ? The question is
whether reasons for belief are not only sufficient for holding true, but also necessary.
The way we earlier approached this type of question was via the following line of
thought. What determines what one will do is a combination of what one has the power
to do, given one’s circumstances, and how one will make use of that power. What one
has the power to do, given one’s circumstances, is not up to one. But it is up to one how
one will make use of that power. Forming a reasonable view of what one will do, there-
fore, requires forming a reasonable view both of one’s power to act, given one’s circum-
stances and of how one will make use of that power. In making a decision to φ, and so in
imposing a constraint on one’s practical deliberations, one aims to form a reasonable
view about what one will do. One aims to form a view based on a combination of two
factors: first, information about one’s power to act, given one’s circumstances; and sec-
ond, one’s ongoing commitment to act in a particular way in order to discharge the
assumptions that constitutes one’s view about what one will do. One’s view now about
what one will do in the future can therefore turn out to be wrong on either of two
grounds: one might have been wrong about what one had the power to do, given one’s
circumstances; and one might lose one’s commitment to act, by ceasing to impose on
oneself the constraint of seeking to discharge one’s initial assumption about how one
will act, either through changing one’s minds, or through failing to remember what
one had decided to do. That is, one might turn out to be wrong about either environ-
mental or agential conditions.
Now if the extent of the space within which one is entitled freely to impose con-
straints on one’s own practical deliberation is determined by what one is in a position
freely to make true, then that space is determined, in turn, by facts about one’s power,
given one’s circumstances. Since one is not in a position freely to determine facts about
one’s power to act, one is not in a position freely to impose constraints on one’s own
practical deliberation that amount to holding true specific propositions about that
power. Rather, insofar as one’s decision to φ involves one’s holding things true about
one’s power to act, one’s holding those things true will have the same status as other
beliefs about things outside one’s active control, and will therefore be subject to what-
ever evidential demands such beliefs are subject. (That was the function of (P3) in the
argument that was sketched in section 3.) But it is plausible that ordinary beliefs about
things outside one’s active control are reasonable only if one possesses positive evi-
dence. That is, it is plausible that one cannot reasonably believe that p when one merely
lacks evidence that it’s not the case that p. Rather, it is plausible in that case that one
should instead withhold belief. (That was the function of (P4) in the argument sketched
in section 3.) Thus, in cases in which one’s decision to φ involves one’s holding things
264 Guy Longworth
true about the prospective activities of other agents, one will require evidence that
those activities will be forthcoming—evidence concerning the willingness and com-
petence of those other agents. In cases in which one lacks such evidence, one should
withhold belief, and so should not hold true, that the required activities of others will
be forthcoming.
We are now in a position to see what is wrong with that line of reflection. It presup-
poses that holding things true for practical purposes is to be modelled on holding
things true for theoretical purposes. That is, it presupposes that demands on theoret-
ical belief, arising from its specific aims and functions, apply equally to holding things
true for practical purposes. However, our epistemological perspective on what we hold
true for practical purposes is quite different from our perspective on what we hold true
for theoretical purposes. Thus, we are entitled to hold things true for practical pur-
poses on the basis of what we regard as being a free decision, so a decision we regard as
one that, consistently with our evidence, we might have failed to make. Even having
made a decision, and so having come to hold something true for practical purposes, we
do not regard ourselves as having thereby provided grounds for a theoretical belief
about what we will do. For our theoretical beliefs provide constraints on what we can
then decide to do, so that if our theoretical beliefs are inconsistent with its being
true that we will φ, then, unless our theoretical beliefs change, we cannot decide to φ.
If deciding to φ required forming a theoretical belief to the effect that one will φ, then,
having decided to φ, we would be incapable of revoking that decision, without giving
up a belief that is supported by evidence. So, insofar as one has a conception of belief as
governed by evidence, one has reason not to conflate holding things true for practical
purposes with holding things true for theoretical purposes. Thus, at least one of prem-
ises (P3) and (P4) in the argument sketched in section 3 should be rejected.
Kant gestures towards a distinction between holding things true for practical
purposes and holding things true for theoretical—or speculative—purposes in the
following passage:
Thus, in the union of pure speculative with pure practical reason in one cognition, the latter has
primacy. . . . For, without this subordination a conflict of reason with itself would arise, since if
they were merely juxtaposed (coordinate), the first would of itself close its boundaries strictly
and admit nothing from the latter into its domain, while the latter would extend its boundaries
over everything and, when its need required, would try to include the former within it.
(Kant 1788: 5: 121)
Kant’s thought here is the following. Suppose that holding things true for practical
purposes and holding things true for theoretical purposes constituted a single cogni-
tive kind. In that case, we would be liable to be presented with candidates for being
held true with respect to which practical and theoretical reason conflict. These would
be candidates for being held true such that theoretical reason requires us not to hold
them true—since we lack appropriate evidence for holding them true—whilst prac-
tical reason requires us to hold them true—since its aims require setting an end that
faith in kant 265
depends upon holding them true. Kant’s proposal for avoiding otherwise irresolvable
conflicts of that sort is to distinguish holding things true for practical purposes from
holding things true for theoretical purposes. In the context of defending specifically
moral faith, Kant draws the distinction in the following way:
What belongs to duty here is only the striving to produce and promote the highest good in the
world, the possibility of which can therefore be postulated, while our reason finds this think-
able only on the presupposition of a supreme intelligence; to assume the existence of this
supreme intelligence is thus connected with the consciousness of our duty, although this
assumption itself belongs to theoretical reason; with respect to theoretical reason alone, as a
ground of explanation, it can be called a hypothesis; but in relation to the intelligibility of an
object given us by the moral law (the highest good), and consequently of a need for practical
purposes, it can be called faith and, indeed, a pure rational faith since pure reason alone (in its
theoretical as well as in its practical use) is the source from which it springs. (Kant 1788: 5: 126)
Kant proposes that what we hold true for practical purposes is to be viewed, from a
purely theoretical perspective, as a mere hypothesis. That allows that such a holding
true can be detached from the requirement for positive evidential support whilst, at
the same time, leaving open that it is not possible rationally to hold something true in
the face of evidence that it isn’t true. For one can be entitled rationally to accept hypoth-
eses in the absence of positive evidence that they are true; but one must be prepared to
reject them in the face of evidence that they are not true. And we’ve seen that deciding
to φ, or setting E as one’s end, and so holding true for practical purposes that one will φ,
or seek to bring about E, are subject to an analogous pattern of requirements. (Kant
discusses the government of rational hypotheses in 1781/1787: A769/B797–A782/
B810. For useful discussions of Kant’s views about the nature of holding true in
general, see Chignell 2007 and Stevenson 2003.)
We ordinarily think that it can be perfectly rational to decide to φ in the absence of
evidence that we will φ, and thus that we can set an end in the absence of evidence that
the end is in principle attainable. For example, we hold that I can rationally decide to
speak loudly enough to be heard at the back in the absence of evidence either that I will,
or that I won’t, be heard. Thus, we reject the combination of premises (P3) and (P4) in
the argument sketched in section 3. I’ve drawn on the work of Soteriou and Kant in
order to present a theoretical perspective on decision—and more generally, on holdings
things true for practical purposes—that supports our ordinary view. When one decides
to φ, one thereby holds true that one will φ. If one realizes that one will φ only if certain
environmental conditions obtain, then rationally holding that one will φ requires hold-
ing that those environmental conditions do obtain. Similarly, when one decides on an
end, one thereby holds true that one will strive to attain that end and, so, that the end is
in principle attainable. If one realizes that an end is attainable only if certain environ-
mental conditions obtain, then one’s rationally holding that the end is attainable
requires that one hold that those environmental conditions do obtain. Thus, in particu-
lar, in cases in which one knows that the attainability of one’s end is dependent on the
266 Guy Longworth
activities of other agents, one must hold true that those activities will be forthcoming.
However, holding things true for practical purposes differs in nature from holding
things true for theoretical purposes, and so is subject to different requirements. The
differences make space for rationally holding things true for practical purposes which
one couldn’t rationally hold true for theoretical purposes. So, the differences make
space for the possibility of rationally holding true that the activities of other agents on
which one’s successful attainment of an end depend will be forthcoming in the absence
of evidence that those activities will be forthcoming. For example, I can have practical
reasons for holding true that others will campaign for me in the absence of evidence that
they will, as long as I lack evidence that they won’t.
At the end of section 3, I suggested that a full defence of our ordinary view would
require a response to the claim that holding true is governed by the constitutive aim of
holding true only what is true. We are now in a position to see that such a response
must embed an alternative account of the constitutive aims of the setting of ends and,
so, of holding things true for practical purposes. However, our discussion has, from
the outset, been shaped by just such an account. For Kant’s account of practical reason
is one according to which its constitutive aim is a moral aim: the attainment of the
highest good.
Our ultimate moral end, the highest good, imposes upon us the need to set our-
selves mediate ends, what we take to be means to our ultimate end. Our practical and
theoretical limitations mean that the mediate ends that we set are liable to instability.
For instance, our conception of our ultimate end is limited so that it leaves open vari-
ous questions concerning the space between what is morally obligatory for us and
what is morally impermissible for us. It leaves open some questions about the precise
boundaries of the space. And, in addition, it leaves open questions about our optimal
route through it. Thus, with respect to any mediate end that we set, it may turn out that
between setting the end and attaining it, we will cease to constrain ourselves by hold-
ing true that we will strive for that end. For example, I might have decided to stand for
election because I held that winning the election would be an optimal means of fur-
thering moral ends. On realizing that my not standing would support a more efficient
route to those ends, I might shed the self-imposed constraint of holding true that I will
stand. Moreover, limited as we are, we may simply forget some amongst the ends that
we have set. So, from a broadly practical perspective, our ends are liable to a certain
amount of instability.
Our epistemic limitations mean that our mediate ends are liable to an additional
layer of instability. For we often need to set mediate ends that we do not know to be
attainable. With respect to any of those mediate ends, it may turn out that between
setting the end and attaining it, we acquire evidence that the end is in principle
unattainable. If we were to acquire such evidence, then we would be required to stop
holding true that we will strive for the end. I might discover, for example, that it is now
impossible for me to win the election. If we set only those ends that we knew to be in
principle attainable, our striving towards those ends would be to that extent more
faith in kant 267
stable, and we would be more likely to attain those ends. But the gain in stability and
likelihood of success would come at a cost: we would be restricted to deciding on ends
that we knew to be attainable. If our most fundamental aim were success in attaining
whatever ends we set, then the gain in likelihood of success would be privileged. That
might be so, for example, if our fundamental aim in setting ends were the attainment
of knowledge about what we will do. However, insofar as our ultimate practical aims
go beyond success in whatever we decide to attempt, our commitment to those aims
is liable to outweigh our concern to obtain prior guarantees of stability and success.
Thus, my reasons for trying to win the election might trump my desire to avoid uncer-
tainty about my future.
If one knew now that a particular outcome will be forthcoming in the future, one
would be in a position confidently to rely on that outcome. In particular, one would be
in a position to be confident that one will not later come across decisive evidence that
the outcome will not occur. More generally, one would be in a position to be confident
that one’s current view about the future need not be dislodged by genuine reasons.
I might be in that position if I knew that the election was rigged in my favour. One might
lose one’s view in other ways—for example, by losing confidence in the face of merely
apparent counter-evidence, or simply by forgetting. But one will not be forced to shed it
by reason alone. Although we typically lack such a guarantee with respect to what we
decide to do or to strive for, Kant holds that a guarantee is available with respect to the
highest good. For according to Kant, neither of the two potential sources of instability in
our ends applies there. There is no practical source of instability, since there cannot be
genuine practical reasons for giving up the end of achieving the highest good. And there
is no theoretical source, since there cannot be genuine evidence that the highest good is
unattainable. Thus, one can be at least as confident in holding that the highest good is
attainable as one could be if one knew that it were attainable. Holding true for practical
purposes that the highest good is attainable is, therefore, distinctively stable and secure.
However, that does not indicate that holding things true for practical purposes in more
ordinary cases is unreasonable. Rather, it indicates only that the reasonableness of our
holding things true for practical purposes in more ordinary cases is dependent on our
being appropriately sensitive to the potential instability and insecurity of so doing.
Let’s return to the three questions raised in section 2. The answer to (Q1) is that
rationally setting an end requires holding true that the end is in principle attainable.
That requirement trumps weaker requirements due to its superior capacity to explain
constraints on the rational combination of the setting of ends. The answer to (Q2) is
that the required attitude of holding true is not a paradigmatic form of belief. And the
answer to (Q3) is that holding true can be grounded practically rather than eviden-
tially. Typically, holding true is potentially subject to practical or evidential defeat, and
so lacks the stability of knowledge. Kant’s more exigent demands on faith—according
to which it is sustained by absolute moral demands and is insusceptible to evidential
disconfirmation—are designed to align faith with knowledge; there is no reason to
expect that alignment to obtain with respect to quotidian forms of trust.
268 Guy Longworth
6. Conclusion: Faith and Trust

According to Kant, there can be practical grounds for deciding to act, and thus for
imposing on ourselves the constraint of holding true that we will so act. And there can be
practical reasons for setting ends, and thus for imposing on ourselves the constraint of
holding true that those ends will be attained. My aim has been to make space for Kant’s
otherwise plausible view by divorcing that view from some dubious near relatives. Thus,
I’ve explained that such a view needn’t allow that one can decide rationally to do things,
or strive for things, that one knows are unattainable. And I’ve explained that such a view
needn’t be committed to the idea that practical reasons can provide grounds for belief. In
doing so, I appealed to Kant’s specific account of the constitutive aim of holding things
true for practical purposes, according to which that aim is the attainment of the highest
good. However, the same outcome would be achieved if it were shown, more minimally,
that the constitutive aim of holding things true for practical purposes is distinct from the
constitutive aim of belief. (Independently, there need be no commitment to the claim
that one can decide to believe things, for there need be no commitment to thinking that,
simply because deciding to φ entails holding true that one will φ, it follows that deciding
to φ entails deciding to hold true that one will φ. So, even if deciding to φ on practical
grounds entailed believing that one will φ on practical grounds, it need not entail in
addition deciding to believe that one will φ on practical grounds.)
With those resources in hand, let’s return to the question about trust with which we
began. Can it ever be reasonable to trust other people, or to rely upon them, without
evidence of their trustworthiness or reliability, given that trusting them requires
holding true that they will do what they are trusted to do? The answer we have to this
point is conditional: yes, provided, first, that there is not too much evidence that they
are untrustworthy or unreliable and, second, that there are genuine practical reasons
for trusting them or relying upon them. However, it seems obvious that we are con-
stantly presented with genuinely practical reasons for trusting others. That is, it seems
obvious that we are constantly presented with valuable ends the attainability of which
is dependent on others. And typically, at least, our situation is one of lacking evidence
for others’ trustworthiness or reliability, rather than possessing evidence for their
untrustworthiness or unreliability. Plausibly, then, it can sometimes be reasonable to
trust other people, or to rely upon them, in the absence of evidence of their trust-
worthiness or reliability.
That answer leaves open the extent to which trust is sometimes made reasonable in
other ways—for example, by the obtaining of evidential grounds for holding true (in
that case, plausibly, for believing) that someone is reliable or trustworthy. That is, it
leaves open that a full account of the bases of trust might need to include a response of
the third type considered in section 1, an attempt to argue that our epistemic limita-
tions are less stark than the puzzle about trust makes them out to be. And it leaves
open that there may be cases in which the potential cognitive consequences of
trusting someone are dependent on the obtaining of such evidential grounds—for
faith in kant 269
e xample, cases of testimony in which those consequences include knowledge or belief

and, so, cannot straightforwardly be sustained by practical grounds for holding things
true (for relevant discussion, see Audi 2004). However, my aims here have been
limited to sketching out some space for the fourth response to the puzzle, by indicat-
ing ways in which trust might sometimes be reasonable in the absence of evidential
grounds for belief despite the fact that trust involves holding true that another is
trustworthy. From the deflationary perspective that I’ve recommended, the question
whether trust is reasonable in particular cases can then be seen as a matter for ordin-
ary practical deliberation. In general, no appeal need be made to special sources of
evidence or trust-specific ends. However, some general questions about the reason-
ableness of trust remain, and I shall conclude by raising two of them. The questions
concern the extent to which evidential grounds for distrusting others or practical
grounds for trusting them can ever be decisive.
It seems obvious that we must often try to balance practical and theoretical grounds
for trust. Kant himself expressed some pessimism about the results of the trade-off:
We must so conduct ourselves to a friend, that it does us no harm if he were to become our
enemy; we must give him nothing to use against us. We are not, indeed, to suppose that he
may become our enemy, for then there would be no trust between us. But if we give ourselves
entirely to a friend, and entrust him with all the secrets which might detract from our hap-
piness, and might well be divulged if he did become an enemy, then it is very unwise to tell
him these things, since he could either give them away through inadvertence, or use them to
our hurt if he became our foe. (Kant 1784–5: 27: 429–30)
The two questions that I wish to table concern trust in other people’s good disposition
of will, rather than in other competences of theirs on which we might rely. The first
question is this. Could there be cases in which trusting someone is made entirely
unreasonable on evidential grounds? That is, could we know enough about someone’s
poor character that no countermanding practical grounds, including grounds deriv-
ing from morality, would outweigh what we know and thus make it reasonable for us to
trust them? The answer to this question will depend ultimately on answers to further
questions, concerning the extent to which others may be regarded as free, not only
from their own practical perspective, but also from our practical perspective. Can we
have sufficient evidence about the dispositions of others’ wills to rule out their striving
to do what they are trusted to do? Could such evidence ever foreclose on their absolute
freedom to go against all past evidence concerning their character and so redeem
themselves by deciding to warrant our trust?
Second, could there be cases in which practical grounds would make it obligatory to
trust someone, at least with respect to the disposition of their will? Despite Kant’s
animadversions about apportioning complete trust in a friend, we might reasonably
wonder whether there might nonetheless be moral grounds for striving to trust
others, insofar as doing so is made possible by our evidence. And we might wonder, in
particular, whether the highest good could comprise a realm in which every person is
270 Guy Longworth
in fact fully trustworthy with respect to the disposition of their wills, since their wills
are disposed in accord with respect for the moral law, and yet in which not every per-
son is fully trusting. Thus, we might wonder whether fully respecting the demands of
morality, and so setting the highest good as one’s end, in turn imposes on one a require-
ment to strive to have faith in other people, insofar as one isn’t precluded from doing so
by evidence of their liability to malevolence.1
References
Audi, R. 2004. The A Priori Authority of Testimony. Philosophical Issues 14: 18–34.
Bratman, M. 1987. Intentions, Plans, and Practical Reasoning. Cambridge, MA: Harvard University
Press.
Bratman, M. 1999. Faces of Intention: Selected Essays on Intention and Agency. Cambridge:
Bratman, M. 2006. Structures of Agency: Essays. Oxford: Oxford University Press.
Bratman, M. 2009. Intention, Belief, Practical, Theoretical. In Spheres of Reason: New Essays in
the Philosophy of Normativity, ed. S. Robertson. Oxford: Oxford University Press.
Chignell, A. 2007. Belief in Kant. Philosophical Review 116, 3: 323–60.
Gardner, S. 2006. The Primacy of Practical Reason. In A Companion to Kant, ed. G. Bird.
Oxford: Blackwell.
Hieronymi, P. 2008. The Reasons of Trust. Australasian Journal of Philosophy 86, 2: 213–36.
Holton, R. 2009. Willing, Wanting, Waiting. Oxford: Oxford University Press.
Kant, I. 1781/1787. Critique of Pure Reason. Ed. and trans. P. Guyer and A. W. Wood. Cambridge:
Cambridge University Press, 1998.
Kant, I. 1784–5. Moral Philosophy: Collins’s Lecture Notes. In Immanuel Kant: Lectures on
Ethics. Ed. and trans. P. Heath and ed. J. B. Schneewind. Cambridge: Cambridge University
Press, 1997.
Kant, I. 1786. What Does it Mean to Orient Oneself in Thinking? In Religion within the
Boundaries of Mere Reason. Ed. and trans. A. Wood and G. di Giovanni. Cambridge:
Cambridge University Press, 1998.
Kant, I. 1788. Critique of Practical Reason. Ed. and trans. M. Gregor. Cambridge: Cambridge
University Press, 1997.
Kant, I. 1790. Critique of the Power of Judgment. Ed. and trans. Paul Guyer and trans.
E. Matthews. Cambridge: Cambridge University Press, 2000.
Kant, I. 1793. Religion within the Boundaries of Mere Reason. Ed. and trans. A. Wood and G. di
Giovanni. Cambridge: Cambridge University Press, 1998.
Kleingeld, P. 1998. Kant on the Unity of Theoretical and Practical Reason. The Review of
Metaphysics 52, 2: 311–39.
Moore, A. W. 2003. Noble in Reason, Infinite in Faculty: Themes and Variations in Kant’s Moral
and Religious Philosophy. London: Routledge.
1
I’m grateful for discussion and comments to Tom Crowther, Naomi Eilan, Paul Faulkner, Giulia Felappi,
Keith Hossack, Nicola Jamieson, Nils Kürbis, Hemdat Lerman, Daniel Morgan, Johannes Roessler, Thomas
Simpson, Barry C. Smith, Matthew Soteriou, Mark Textor, and an anonymous reviewer. I’m also grateful to
audiences at King’s College London, the London Institute of Philosophy, and Liverpool University.
faith in kant 271
Neiman, S. 1994. The Unity of Reason: Rereading Kant. Oxford: Oxford University Press.
O’Neill, O. 1996. Kant on Reason and Religion. The Tanner Lectures on Human Values 18:
269–308.
Soteriou, M. 2013. The Mind’s Construction: The Ontology of Mind & Mental Action. Oxford:
Stevenson, L. 2003. Opinion, Belief or Faith, and Knowledge. Kantian Review 7: 72–101.
Timmermann, J. 2009. The Unity of Reason: Kantian Perspectives. In Spheres of Reason: New
Essays in the Philosophy of Normativity, ed. S. Robertson. Oxford: Oxford University Press.
Velleman, D. 1989. Practical Reflection. Princeton, NJ: Princeton University Press.
Velleman, D. 2000. The Possibility of Practical Reason. Oxford: Oxford University Press.
Watkins, E. 2010. The Antinomy of Practical Reason: Reason, the Unconditioned and the
Highest Good. In Kant’s ‘Critique of Practical Reason’: A Critical Guide, ed. A. Reath and Jens
Timmermann. Cambridge: Cambridge University Press.
Willaschek, M. 2010. The Primacy of Practical Reason and the Idea of a Practical Postulate. In
Kant’s ‘Critique of Practical Reason’: A Critical Guide, ed. A. Reath and Jens Timmermann.
Wood, A. W. 1970. Kant’s Moral Religion. Ithaca, NY: Cornell University Press.
16
‘Trust is Basic’
Løgstrup on the Priority of Trust
Robert Stern
In the course of his deeply interesting analysis of the place of trust in our lives, the
Danish philosopher and theologian Knud Ejler Løgstrup makes the claim that ‘trust is
basic’, in a sense that somehow puts it prior to mistrust.1 I take it that there is something
intuitively plausible about this remark; but at the same time it is not entirely clear what
it amounts to or involves. What kind of basicness or priority are we talking about here?
My aim in this chapter is to consider different responses to this question that might be
found in Løgstrup’s work, and thus to illuminate what I take to be distinctive in his
account of trust. However, before doing so I will say a little about the role that his dis-
cussion of trust plays in Løgstrup’s writings. These are not well known in Anglophone
philosophy, so some preliminary introductory comments are required.
1. Trust in Løgstrup
Løgstrup was born in 1905 and died in 1981. He lived through the Nazi occupation of
Denmark, and in the middle of this period became professor of theology at the
University of Aarhus in 1943, where he spent the rest of his academic life. He published
1
Løgstrup is not the only philosopher to have this view in some form or other. In particular, some of the
same issues examined below have come up in discussions of Wittgenstein’s remarks about trust in On
Certainty. Cf. Hertzberg (1988) and Lagerspetz (1998, especially chapter 8). While Lagerspetz mainly
focuses on Wittgenstein, he also brings Løgstrup into his final account: see chapter 9. Lagerspetz (2015)
also discusses Løgstrup in chapters 6 and 7. Bob Plant has briefly but illuminatingly connected Wittgenstein’s
views on trust with Derrida’s views, for example the latter’s claim that ‘elementary trust . . . is involved . . . in
every address of the other. From the very first instant it is co-extensive with this other and thus conditions
every “social bond”, every questioning, all knowledge, performativity’ (Derrida 1998: 63); see Plant (2005:
188–90). There is also an interesting parallel between Løgstrup’s claim about the priority of trust over mis-
trust, and Levinas’s claim that while we can of course encounter others with ‘violence, hate and disdain’,
nonetheless what is ‘primary’ because it is ‘presupposed in all human relationships’ is instead ‘the mastery
of the Other [over oneself] and his poverty, with my submission [to the Other] and his wealth’ (Levinas 1985:
89). One could ask questions concerning exactly what ‘primacy’ Levinas has in mind that parallels the
questions we will ask about Løgstrup, where I suspect a similar answer could be given.
‘Trust is Basic’: Løgstrup on the Priority of Trust 273
what is widely seen as his main text, The Ethical Demand, in 1956. This is the work in
which his principal account of trust is developed and which I will focus on in this chapter.
I will also discuss some later works where Løgstrup responds to criticisms (‘Rejoinder’ in
Art and Ethics (1961)), and also somewhat develops his earlier views (Controverting
Kierkegaard (2013a, first published in 1968) and Norm and Spontaneity (1972)).2
It is therefore helpful to say something very briefly about where Løgstrup’s account of
trust fits into the structure of The Ethical Demand. The book begins by focusing on ‘the
religious proclamation of Jesus of Nazareth’, namely the proclamation to ‘love your
neighbour as yourself ’, which Løgstrup says he will try to define in ‘strictly human
terms’ (Løgstrup 1997: 1; 2010: 9). Løgstrup asks how we should understand this proc-
lamation and what it ‘answers to in our existence’3 (Løgstrup 1997: 1; 2010: 9), arguing
in response that it rests on our interdependence, on the fact that ‘the other person must
to such a degree be dependent upon me that what I do and say in the relationship
between us—I alone and nobody else, here and now and not at some other time or in
some other manner—is of decisive importance’ (Løgstrup 1997: 5; 2010: 13). He then
argues that to further understand the nature of this interdependence we need to focus
on trust, as the logic of trust can give us insight into the ethical demand that the proc-
lamation embodies, where he distinguishes between the ethical demand as such and
what might be asked of us by the ordinary conventional requirements of moral life.
Løgstrup then argues that this demand only makes sense if we see life as a gift. He finally
considers objections to his account, including objections to his account of trust. Clearly
this complete picture raises a number of crucial questions which cannot be considered
here, particularly Løgstrup’s characterization of the ethical demand as what he calls rad-
ical, silent, one-sided, isolating, and unfulfillable; how it is to be distinguished from the
requirements of social morality and how they relate to each other, if at all; and what he
means by calling life a gift. But this sketch of the basic structure of The Ethical Demand
should help to see how the discussion of trust in the first chapter fits in, where that struc-
ture is helpfully outlined by Løgstrup in another work as follows:
First I analyze how the life of one person is interwoven with the life of another, and from this
I deduce the content of the ethical demand, which has to do with taking care of the life of the
other person that has been surrendered to us. Some way into the book I make it clear that the
one-sidedness of the demand . . . presupposes that life has been given to the individual person.
(Løgstrup 2007: 10; 1961: 239)
Løgstrup’s account of trust relates centrally to this theme of surrendering one’s life to
another person, as he thinks this is fundamental to the nature of trust. In general, for
2
The Ethical Demand is available in English as Løgstrup (1997). The ‘Rejoinder’ and extracts from the
other works mentioned are available in translation in Løgstrup (2007). In references to Løgstrup, a trans-
lation is given first where available, followed by the most recent Danish edition.
3
The Danish term Løgstrup uses here is ‘tilværelse’, which could also be translated as ‘life’, and indeed
both terms are used by the translators of the English edition of The Ethical Demand. As we shall see, this
connection to life and its proper development is important in what follows, so this should be born in mind.
Løgstrup also uses ‘tilværelse’ as his translation of Heidegger’s term ‘Dasein’.
274 Robert Stern
Løgstrup ‘[t]o trust . . . is to lay oneself open’ (Løgstrup 1997: 9; 2010: 18), which is why
he believes that understanding trust is a good way to understand our interdependence,
without which the proclamation would not make much sense:
If human beings were so independent of one another that the words and deeds of one were
only a dispensable luxury in the life of another and my failure in the life of the neighbour
could easily be made up later, then God’s relation to me would not be as intimately tied up
with my relation to the neighbour as the proclamation of Jesus declares it to be.4 In short, the
intimate connection in which Jesus places our relation to God and our relation to the neigh-
bour presupposes that we are, as Luther expressed it, ‘daily bread’ in the life of one another.
And this presupposition for the intimate connection in the proclamation of Jesus between
the two great commandments in the law can indeed be described in strictly human terms.
(Løgstrup 1997: 5; 2010: 13–14)
Trust is therefore important to Løgstrup because it reveals how that interdependence

works: in trusting another person, I am placed in their hands and make myself vulner-
able to them, while also expecting that ‘surrender’ of myself to play a role in their
response to me; if they do not respond accordingly, I will feel resentment and hurt in a
way that can quickly become moralized, sometimes in exaggerated ways.
This therefore gives us a way to locate Løgstrup in relation to contemporary debates
on trust, which have largely gone on without reference to his work. Most of those
debates trace themselves back to Annette Baier’s key article of 1986 on ‘Trust and
Antitrust’, where she begins by saying that ‘there has been a strange silence on the topic
[of trust] in the tradition of moral philosophy with which I am familiar’ (Baier 1986: 232),5
and then proceeds to introduce certain key distinctions and issues that have now
become central, such as the way the ubiquitousness of trust can make it invisible; the
difference between relying on something or someone and trusting someone; the way in
which trust involves dependence on the goodwill of a person, and thus a vulnerability to
harm, leading to her account of trust as ‘accepted vulnerability to another’s possible but
not expected ill will (or lack of good will) toward one’ (Baier 1986: 235);6 and the conse-
quent role of power within trust relations. Baier’s paper, and related work by her and
others, has opened up a debate between two broad views of trust which have been called
‘predictive’ and ‘affective’.7 On the predictive approach, trust involves dependence where
4
Cf. Matthew 22: 36–40, which has Jesus saying that what he identifies as the second most important
law, namely to love your neighbour as yourself, is like the ‘first and greatest commandment’ to ‘love the lord
your God with all your heart and with all soul and with all your mind’, thus linking love of God with love
of the neighbour.
5
Hertzberg, writing around the same time as Baier, makes a similar point: ‘There does not seem to have
been a great deal of discussion about the concept of trust in recent philosophy’ (Hertzberg 1988: 308).
6
To give Baier’s discussion in slightly more detail: ‘When I trust another, I depend on her good will
toward me . . . Where one depends on another’s good will, one is necessarily vulnerable to the limits of that
good will. One leaves others opportunities to harm one when one trusts, and also shows confidence that
they will not take it . . . Trust then, on this first approximation, is accepted vulnerability to another’s possible
but not expected ill will (or lack of good will) towards one’ (Baier 1986: 235).
7
I am here following Faulkner’s terminology, which is itself partly drawing on Hollis. See e.g.
Faulkner (2014: 1977–8), where he draws the contrast as follows: ‘To say that A trusts S to f on this [pre-
dictive] understanding is just to say that A depends on S f-ing and expects S to f. . . . [But on the affective]
the expectation that this dependence will be satisfied is grounded on evidence that the
trusted person can be relied upon in various ways (e.g. to speak the truth), where this
evidence can come from various sources, such as past experience of her behaviour, and
what one knows about her interests and character. In this sense, I can trust the used car
dealer to sell me a good car, perhaps because she has done so in the past, or I know her
fear of reputational damage will lead her to do so now. By contrast, on the affective
approach, trust still involves dependence, but here that dependence is itself assumed to
play a role in motivating the trusted party to act accordingly. Thus, in such cases you will
feel ‘let down’ by the other, and therefore blame them, whereas if your attitude of trust
was based on evidence of reliability, this would not be appropriate: if our attitude in
trusting was evidential there would be no space for blame of this sort. Thus, if the car
dealer sells me a duff car where I was trusting her in the first sense, I might feel annoyed
at myself for having miscalculated where her interests lie and so having misplaced my
judgement of how she would respond to me by basing that on the wrong evidence, but I
would not feel resentment towards her; it is only if I trusted her in the second sense, and
expected my hopelessness with cars to play some role in her response to me that I would
feel in any way betrayed. And if I expected her to be trustworthy in the second sense, but
found her to only be trustworthy in the first, I might also feel let down, in again realizing
that my vulnerability was playing no real role in her relation to me. As a result, it is com-
mon in the literature to draw a distinction between trust as reliability and hence
dependability in that sense, and trust proper: you might be dependable and I might
depend on you, because your interests make it the case that you will do the best you can
for me, and you are good at this; but this is not the same as being trustworthy, as my
dependence on you plays no role in how or why you serve my interests (just as I might
depend on a rope to hold me, but where this is not really a case of trust in a full sense).8
From what has been said above about Løgstrup’s position, it should be clear that he
is centrally concerned with trust in the second sense, as crucially involving our vulner-
ability to others and the role this plays and is expected to play in their response to us.
Had Løgstrup’s work on trust been more widely known and appreciated at the time, it
might then have had the sort of impact on the debate that was made later by the contri-
butions of Baier and others who have followed her lead.
2. ‘Trust is Basic’
We have now outlined the place that the consideration of trust has in Løgstrup’s
work, and what view he takes of it. I now want to focus on a central part of Løgstrup’s
understanding to say that A trusts S to f is to say that A depends on S f-ing and expects this to motivate
S to f ’. Cf. also Faulkner (2007; 2011) and Hollis (1998).
8
Cf. the contemporary debate, where those who defend the affective view include Annette Baier, Paul
Faulkner, Karen Jones, and Richard Holton, while those on the predictive side include Russell Hardin,
Alvin Goldman, Michael Bacharach, and Pamela Hieronymi. Central texts include: Baier (1986); Faulkner
(2011); Gambetta (1988); Goldman (2011); Hardin (1996); Holton (1994); Hieronymi (2008); Jones (1996).
276 Robert Stern
discussion of trust, namely his claim regarding the basicness of trust, and that is it
somehow prior to mistrust: ‘Trust and distrust are not two parallel ways of life. Trust
is basic; distrust is the absence of trust’ (Løgstrup 1997: 18, note 5; 2010: 28, note 1).9
How is this idea to be understood?
I will consider four main options:
(1) Psychological: trust is the attitude we start out with, not distrust [developmen-
tal priority].
(2) Transcendental: trust is warranted as the default attitude, grounded in the
necessary conditions of our fundamental practices, distrust is not [rational
priority].
(3) Value: trust is a prima facie good, so distrust can only be a privation or defi-
cient form of trust [axiological priority].
(4) Ontological: that trust is possible is not a result of our social arrangements, but
is essential to the proper functioning of human life itself, whereas distrust is
not essential in this way [priority in being].
I will now consider each of the four options above in turn, arguing in the end that it
is the third and fourth that seem to best capture Løgstrup’s position while building on
elements of the other two, so following this development will show us how he came to
understand the nature of trust.
2.1 Psychological priority

In his writing on this issue, it sometimes sounds as if Løgstrup has psychological prior-
ity in mind: namely, as a matter of human psychology and its development, we all first
encounter people with trust, and then come to learn to distrust when things go wrong
(or we are taught they might). The opening chapter of The Ethical Demand where he
first talks about trust could certainly be read this way:
It is characteristic of human life that we normally encounter one another with natural trust.
This is true not only in the case of persons who are well acquainted with one another but also
in the case of complete strangers. Only because of some special circumstances do we ever
distrust a stranger in advance . . . Initially we believe one another’s word; initially we trust one
another. (Løgstrup 1997: 8; 2010: 17)10,11
9
While adequate, this translation is not very precise. A more literal translation of the Danish would be:
‘Distrust is therefore certainly not—as a way of being—equal to trust. Trust is what is fundamental—and
distrust arises out of a lack of trust.’ For the sake of simplicity in this chapter I will continue to stick to the
phrase used in the English translation, that ‘trust is basic’, but the emphasis on fundamentality should be
borne in mind.
10
Again, a more literal and accurate translation of the beginning of this quotation would be: ‘It belongs
to our human life, that we normally encounter one another with natural trust. This is not just the case when
we meet a person we know well, but also holds when we meet a complete stranger.’ The Danish original
implies more clearly that trust is a deep part of that life, and Løgstrup is talking to the reader more directly,
asking them to recognize that this is what we do.
11
Cf. Baier, who speaks about an ‘innate but fragile trust’ (1986: 242), arguing that we must suppose that
‘infants emerge from the womb already equipped with some ur-confidence in what supports them, so that
And slightly later Løgstrup also brings in psychological studies, especially those con-
cerning human development:
[T]he child, in contradistinction to the adult, is never able to trust only partially. To trust with
reservation is possible only for one who has learned to hold back something of herself. But this
the child has not learned consciously and deliberately to do. For her reservation takes place as
a matter of psychic automatism. This is why the disappointed trust, restlessness, and insecurity
which go with it create in the child far-reaching and fateful consequences.
(Løgstrup 1997: 15; 2010: 25)12
Løgstrup seems to be arguing here that when a child distrusts, it takes a very different
form from the adult case, as for the child it is total rather than being selective, which is
why a child who learns to distrust is so psychologically damaged by the experience; but
this in turn suggests that the attitude with which a child begins is not one in which trust
and distrust are both in play (as is typical for adults), but one in which trust is complete,
and hence basic in this sense—it is the attitude from which we start out, but which can
be flipped round into its opposite in exceptional circumstances, with catastrophic
psychological effects when this occurs. Løgstrup also goes on to offer an explanation of
why the child will begin with trust, which is that she is outside the various conventional
norms, where it is those norms that make us ‘hold ourselves in reserve and do not allow
ourselves completely to trust one another’ (Løgstrup 1997: 19; 2010: 28), thereby using
convention ‘as a means for keeping aloof from one another and for insulating ourselves’
(Løgstrup 1997: 20; 2010: 30); but the child is ‘outside convention’, so ‘he or she is able to
trust only without reservation’ (Løgstrup 1997: 20; 2010: 30).
However, while there is reason to think from these passages that Løgstrup might
support this psychological priority claim, there is also reason to think it is not the fun-
damental issue for him. Firstly, to place too much weight on it would be to run counter
to the methodology that he wants to adopt, which he insists is phenomenological
rather than psychological. Thus, while he says that the psychology ‘supports’ the
position he holds, he does not claim that it is the basis for his view, which he identifies
as primarily ‘an analysis of a phenomenological character’ (Løgstrup 1997: 15; 2010: 24).
And elsewhere, he makes plain that he sees a significant difference between scientific
claims and phenomenological ones.13
no choice is needed to continue with that attitude, until something happens to shake or destroy that
attitude’ (Baier 1986: 244).
12
Cf. also: ‘Trust is an original phenomenon. The child comes trustingly into the world, the child
psychiatrist ascertains’ (Løgstrup 1995, II: 355; 2013b: 230).
13
‘[T]here are facts that everyday language is better at establishing than the sciences are. There are
phenomena we can only describe and distinctions we can only express using natural language. That is why
much philosophy remains within the interpretation of the world, of things, and of human existence that is
given in our everyday language. Conversely, everyday language’s “sorting” of things is useless in attempts
to track down the laws in which the exact sciences are interested. In order to track these down, we must
carry out a new and different classification of things, and that is what takes place in the so-called scientific
languages. But as I said earlier, I have not sought to establish scientific laws. This I leave to psychology,
sociology, and similar sciences. In my description of the phenomena, I have only worked with comparisons
278 Robert Stern
Secondly, but relatedly, when Løgstrup deals with subsequent challenges to these
more psychological passages from The Ethical Demand, he does so by downplaying
the ‘scientific’ nature of his position, suggesting that his priority thesis should not
really be understood in these psychological terms. This can be seen clearly in one of
his ‘Rejoinders’, written in response to Henrik Stangerup, who argues against
Løgstrup that we normally meet one another with distrust, where trust only follows
after as a ‘result of the fulfillment of love or friendship’ (Løgstrup 2007: 2; 1961: 229).14
Løgstrup’s reply is as follows:
To this I would say that the disagreement between Stangerup and myself is not merely, as he
presumes, a question of what comes first and what comes last, trust or distrust. He and I take
the words ‘first’ and ‘last’ to mean different things. The difference can be pinned down as
follows: Stangerup is inquiring into which of the two comes first in time, and which comes last,
whereas I see the difference between first and last more as a difference in rank, having my sights
set on the foundational relation. When Stangerup says that distrust comes first, he means—and
this is also how he expresses himself—that in a person’s historically progressing existence, trust
‘follows after’ distrust. When I, on the other hand, say that trust is primary, I mean that distrust
is the negation of trust, and is, as such, founded in trust. (Løgstrup 2007: 2; 1961: 229)
Løgstrup goes on:

Whether Stangerup and I still disagree depends on the position he takes on the distinction
I have made here, and which one could, perhaps, call a distinction between a psychological
status report and an investigation of the foundational relation. For if I am correct in perceiv-
ing Stangerup’s reflections on trust and distrust as such a status report, then I can follow his
reasoning. The question is whether Stangerup, on his part, will concede that it is possible to
practice a philosophical psychology that includes an explanation of the foundational rela-
tions, and whether he can follow my particular reasoning in favor of the foundational relation
that I believe exists between trust and distrust. (Løgstrup 2007: 4; 1961: 231–2)
In these passages, Løgstrup seems to concede that Stangerup could be right that dis-
trust comes first developmentally—or at least, it wouldn’t matter to Løgstrup’s central
point if it did, as that is not the kind of priority he really has in mind. But then, if the
priority of trust over mistrust is not a matter of ‘scientific psychology’ (Løgstrup 2007:
2–3; 1961: 229–30), but some other kind of ‘foundational relation’, what is it?
2.2 Transcendental priority

A second alternative is to argue that the priority is not merely empirical and develop-
mental, but rather takes a transcendental form which in turn makes it ceteris paribus
more rational to opt for an attitude of trust rather than distrust, regardless of how
people may actually behave. That is, while as a matter of psychology particular indi-
and distinctions within the natural language’s interpretation of life. In short, I have stuck to phenomeno-
logical analyses and steered clear of scientific investigations’ (Løgstrup 2007: 9–10; 1961: 238). For a more
extended discussion, see Løgstrup (1987).
14
The article Løgstrup is responding to is Stangerup (1960).
viduals may start out with an attitude of distrust and not trust, this is not really rele-
vant to Løgstrup’s position, as this cannot be said to be the right attitude for people to
have, as our forms of life must in general warrant the attitude of trust over that of
distrust, as otherwise that kind of life would be impossible for us in various ways.
This approach seems to find support in what Løgstrup has to say about the crucial
case of language and speech, where he notes that ‘trust is essential to every conversa-
tion’ (Løgstrup 1997: 14; 2010: 24). In later work, Løgstrup argues that ‘[t]o speak is to
speak openly’ (Løgstrup 2007: 55; 2013a: 100); and he gives an example based on an
actual interview of his wife by a Nazi officer, who was trying to locate Løgstrup’s where-
abouts, to show how difficult it is not to speak openly, even when faced by a ‘destroyer’
to whom telling the truth would lead to disaster, where this difficulty is taken to reflect
the fundamental nature of speech itself:
Let me offer an illustration. Let us imagine that we stand facing a destroyer who is trying to
win us for his cause, but we know that he will shun no means in doing so and that he is not to
be trusted. Face to face with the destroyer, we discover how much effort it takes to remain on
our guard. The thought that, by talking things out, we would be able to dissuade the destroyer
from his destructive enterprise keeps presenting itself; there is no eradicating it once and for
all. We must keep telling ourselves that it is an illusion to think that we could talk things out,
and must continually bear in mind that anything we say will be used to put a third vulnerable
party out of the way. But why is that thought so persistent? Why do we need to make such an
effort to restrain ourselves, and why do we experience doing so as nothing less than contrary
to nature? It is because we are opposing the requirement inherent in speech that speech be
open. To speak is to speak openly.
(Løgstrup 2007: 54–5; 2013a, 100; cf. also Løgstrup 2007: 83–5; 1972: 17–18)
‘The requirement inherent in speech that speech be open’ could be understood as a

transcendental claim: unless speech is open, in the sense that most people speak to
each other truly and honestly, in a way that means they can be trusted in what they say,
speech would be impossible as a form of life. For example, one could not learn to speak
in the first place, and could not learn from testimony so that speech could not serve
this fundamental epistemic role. And even more fundamentally, speech could not
work at all unless people were mostly open and sincere, as otherwise we would have no
way to assess the content of what people were saying, if we could not generally take it
on face value.15
15
Cf. Wittgenstein, where Hertzberg (1988: 308) draws attention to the following passages from On
Certainty (Wittgenstein 1969):
As children we learn facts; e.g., that every human being has a brain, and we take them on
trust. I believe that there is an island, Australia, of such-and-such a shape, and so on and so
on; I believe that I had great-grand-parents, that the people who gave themselves out as my
parents really were my parents, etc. . . .
The child learns by believing the adult. Doubt comes after belief. (§§159–60)
and
I really want to say that a language-game is only possible if one trusts something (I did
not say ‘can trust something’). (§509)
280 Robert Stern
How strong is this transcendental claim? At its strongest, it would be the claim that
all possible forms of speech would become impossible in conditions of dishonesty. But
that may seem to make it implausible, as perhaps we can imagine conditions in which
speech and testimony could still work even under these extreme circumstances. But
even if this were so, all Løgstrup would need is a more modest claim: that for our form
of life, in the way that cooperation and communication work for us, this would be
impossible if people in general could not be trusted to speak the truth: for this would
still warrant us in treating trust as the default attitude, as we could be assured that in
general it is reliable for the conditions in which we find ourselves.16
Thus, it could be argued, ‘trust is basic’ when it comes to speech, as for us dishonesty
and deception using speech must be parasitic on more fundamental honesty and
truth-telling, in which case the default right attitude of the listener should be that of
trust, not mistrust. Thus, just as Kant famously took his universalizability test of false
promising to hinge on the impossibility of this becoming the norm,17 so Løgstrup
could be read in a similar manner, where he puts the point in comparable terms:
. . . we can only be insincere by means of the openness of speech. We can only be untruthful by
dissimulation and by deceiving the other person. By what means? By the openness of speech.
By this means we get the other to swallow the bait of our lies. We can only disregard the open-
ness of speech by making it a feigned openness. Openness can never be eliminated, not even in
the deepest deception. Mere speech involves it. It is just as much of a condition for lying as for
telling the truth. (Løgstrup 2007: 137; 1982: 115)
So, it seems we have a transcendental argument for why we should treat trust as prior to
distrust, even if as a matter of empirical psychology people might start by distrusting
others and then learn to trust: namely, because we know that human life could not
function at all unless people can generally be trusted, we therefore have good grounds
for taking trust as our default attitude to one another.
But then, if we take this line, couldn’t we also have a transcendental claim that refutes
the psychological suggestion that people start by distrusting others? For, how could a
community of distrustful people of this sort learn language or anything about the
world through testimony, if the transcendental claims made above about speech are
right? So isn’t Stangerup also wrong about the developmental empirical claim, and
In this last passage, however, Wittgenstein is arguably talking more about reliance than the sort of trust that
concerns Løgstrup; the translation is somewhat misleading in this respect, as Wittgenstein is using ‘sich
verlassen auf ’, rather than ‘vertrauen’ or ‘glauben’, though cognates of ‘glauben’ are used in §159.
16
Cf. also the quotation from Derrida, cited previously: ‘elementary trust . . . is involved . . . in every
address of the other. From the very first instant it is co-extensive with this other and thus conditions every
“social bond”, every questioning, all knowledge, performativity’.
17
Kant (2011: 73) [Akademie edition 4: 422]: ‘Now, I then see at once that [false promising] could never
hold as a universal law of nature and harmonize with itself, but must necessarily contradict itself. For the
universality of a law that everyone, once he believes himself to be in need, could promise whatever he
fancies with the intention not to keep it, would make the promise and the end one may pursue with it itself
impossible, as no one would believe he was being promised anything, but would laugh about any such
utterance, as a vain pretence.’
can’t Løgstrup now demonstrate that too, so we can also claim priority in the first
sense? Just this combination of views may seem to be what Løgstrup is suggesting in
the following passage:
Initially we believe one another’s word; initially we trust one another. This may indeed seem
strange, but it is part of what it means to be human. Human life could hardly exist if it were
otherwise. We would simply not be able to live; our life would be impaired and wither away if
we were in advance to distrust one another, if we were to suspect the other of thievery and
falsehood from the very outset. (Løgstrup 1997: 8–9; 2010: 17)18
We might thus read Løgstrup as advancing a transcendental claim designed to also

refute the kind of empirical psychological point that someone like Stangerup thinks he
can make.
However, as an interpretation of Løgstrup’s view this stands awkwardly with his
apparent later willingness to distance himself from the psychological claim. But more
importantly, perhaps, it also leaves him vulnerable to a more philosophical objection,
namely that when it comes to the developmental claim in psychology, neither side is
right. For, while the transcendental claim might show that we can’t generally begin
with the attitude of distrust, it still might be said that we could begin with an attitude
that is not really trust either, as it is less substantive than that: for, it is just the absence of
distrust, but not properly trust as such. To see the space for this possibility, consider
the child who asks me for the first time on a long car journey: ‘Are we nearly there yet?’
If young enough, and if this really is the first time it has happened, and if generally our
relations have given her no cause to question me up to this point, it could be argued
that it just doesn’t even occur to her that I might say anything other than the truth, so
she is not taking it that my dependence on her figures in my thinking about her in a
way that we have said is fundamental to the trusting attitude for Løgstrup: she just
takes it for granted that I will answer her correctly. It could thus be said that while she
clearly doesn’t distrust me, she actually doesn’t trust me either, but is in some state
prior to both, where the distinction has not even yet arisen.19 This then would be the
objection to Løgstrup when he writes that ‘distrust is founded on trust’ on the grounds
that ‘distrust arises when the individual has the experience that things do not go as he
has trusted that they would go’, as distrust of this sort may arise not out of ‘disappointed
trust’ but out of the disappointment of one’s expectations, where having such expect-
ations does not amount to trust proper.20 So even if we accept the transcendental
approach to Løgstrup’s priority claim, it would still seem unwise to use this to also
18
The third sentence (‘Human life could hardly exist if it were otherwise’) might be translated more
literally as: ‘It would be hostile to life [livsfjendsk] to behave otherwise’.
19
Cf. Hertzberg (1988: 316): ‘The upshot of this is that it would be misleading to say, “The child behaves
in this way because he trusts the adults”, rather he simply behaves in this way, and out of this relation
(perhaps it could be less misleadingly described as an absence of distrust) there gradually evolve attitudes
which may be called trustful.’
20
Cf. Løgstrup (1980): 223 (my translation).
282 Robert Stern
connect Løgstrup’s claim to the developmental question, where what we start out with
may best be characterized in a more neutral manner altogether.
Nonetheless, even if this is right, we can still use the transcendental approach to
make the first move we considered and to give content to a distinctive kind of priority
thesis: namely, based on a transcendental claim about how speech works, if the ques-
tion of trust vs. distrust arises, it is right to default to trust, all things being equal. So far,
then, we have argued that trust is prior to mistrust because we have transcendental
grounds to think it will be the more warranted attitude than distrust; for example,
given that speech requires openness to be possible at all, if I distrust what people are
telling me, I am more likely to end up with false beliefs than true ones, or at least to lose
out on getting true beliefs. So this seems to show that trust can be given a rational justi-
fication that distrust cannot, based on transcendental claims about what makes human
life (or our kind of human life) possible in the first place.
However, the problem with taking this to be the core of Løgstrup’s priority thesis is
that it would seem to commit him to a predictive rather than affective conception of
trust, whereas we argued previously that the opposite is the case. For the transcenden-
tal claim would appear to ground trust in something other than the role our depend-
ence has in motivating the other person, and instead to ground it in what we take to be
the way in which human life operates, for example that speech must involve openness
where it is this that is said to warrant our attitude of trust. However, while this can
certainly be enough for trust in the predictive sense, we nonetheless argued earlier that
Løgstrup’s view of trust does not take this predictive form, so to treat the priority thesis
in this transcendental manner would seem to run counter to a conception of trust
as he understands it.
This issue relates to the complex question of Løgstrup’s distinction between the
ethical demand itself and our conventional moral norms (although this distinction
cannot be fully explored here).21 But one way to understand it would be in terms of the
predictive/affective contrast drawn above: whereas in trusting people within the con-
ventional norms it is mainly the predictive notion that is in play, and which may there-
fore be given a transcendental grounding, by contrast when we think of trust in relation
to the ethical demand it is the affective notion that is central, and so some other sense
of priority is appropriate. Thus, within the terms of conventional morality, I might
trust you to speak openly and if necessary justify this with the claim that speech could
not operate unless this was generally the case, just as I might trust my car dealer to sell
me a decent car and justify this with the thought that if too many car dealers were
dishonest they would all be out of business; but this is not to justify trust in you in a
deeper sense, as someone responsive to the ethical demand, because this is not to see
my dependence on you as in any way responsible for the truths I expect you to utter.
It may be, however, that there is another way to adopt a transcendental approach
that would avoid this objection. Thus far, we have used a transcendental claim as
21
For further discussion, see Fink 2017.
grounds for the reliability of speech and other social practices, and thus as giving us
rational grounds for trusting those involved in them that outweigh grounds for dis-
trust, so making trust rationally prior. But we might offer a transcendental claim at
another level, namely that unless people were generally trusting in their attitudes to
one another, they could not function at all and so even be capable of distrust, so in this
sense trust is basic: some degree of trust in others is a necessary condition for the pos-
sibility of distrust, which is what makes the former more fundamental than the latter in
this transcendental sense. One writer on trust has put the trust/distrust relation in a
way that might suggest this approach:
Trust, in the broadest sense of confidence in one’s expectations, is a basic fact of social life. In
many situations, of course, man can choose in certain respects whether or not to bestow trust.
But a complete absence of trust would prevent him even from getting up in the morning. He
would be prey to a vague sense of dread, to paralyzing fears. He would not even be capable of
formulating distrust and making that a basis for precautionary measures, since this would
presuppose trust in other directions. Anything and everything would be possible. Such
abrupt confrontation with the complexity of the world at its most extreme is beyond human
endurance. (Luhmann 1979: 4, cited in Lagerspetz 1998: 141)
This seems to be more than just a thesis about a ‘psychological impossibility’, resem-
bling instead a transcendental claim: we need trust in order to even be capable of
distrust, and to this extent it is basic.
However, while avoiding the problems of the first transcendental approach, I would
suggest that Løgstrup would still not subscribe to this position. First, as before, the con-
ception of trust involved is not the same as Løgstrup’s, and it is not clear the transcen-
dental claim would be plausible on his conception. For, while ‘confidence in one’s
expectations’ might be necessary to function at all and so to be capable of distrust, this is
not trust as Løgstrup envisages it, as it is merely predictive trust; but conversely, it seems
less plausible to claim that trust as Løgstrup does envisage it is necessary for us to func-
tion sufficiently to be capable of distrust, as predictive trust would seem to be adequate
for this. But secondly, because this is just a claim of transcendental priority, namely that
trust is a necessary condition for distrust, this position tells us nothing regarding the
normative relation of trust and distrust; that is, it just tells us we can’t have the latter
without the former, but not what makes the former better than the latter. Of course, we
could answer this by going back to the first transcendental approach, as that argues that
conditions for speech and communication mean that trust will be more reliable than
distrust, and so should be preferred on rational grounds; but we have seen reasons to
reject that approach as well, from Løgstrup’s perspective. As we shall now see, however,
the normative question is important to Løgstrup’s view, so is there another option?
2.3 Axiological priority

To appreciate how this might be possible, it is useful to return to the passage with
which we began, and Løgstrup’s apparently paradoxical claim that trust is basic insofar
284 Robert Stern
as it doesn’t really require justification in relation to distrust, because we can only view
the latter as the ‘deficient form’ of the former, and so as inferior to it in first place:
Trust and distrust are not two parallel ways of life. Trust is basic; distrust is the absence of trust.
This is why we do not normally advance arguments and justifications for trust as we do for dis-
trust. To use a modern philosophical expression, distrust is the ‘deficient form’ of trust.
(Løgstrup 1997: 18, note 5; 2010: 28, note 1)
The ‘modern philosophical expression’ Løgstrup is referring to is found in Heidegger,

where in Being and Time Heidegger speaks about deficient modes of being-in-the-
world and of solicitude, for example.22 How does putting things in these terms help us
with our puzzle?
To see how it might, it is worth returning to Løgstrup’s reply to Stangerup. Here,
Løgstrup makes clear that his reason for drawing the contrast between children and
adults was not to make the claim about developmental priority (that trust comes first
in time), but to make a claim about how we see the life of the trusting child to be ‘more
true, more genuine’ than that of the distrustful adult:
The point is, the difference between child and adult lies not only in a person being a child first,
and then becoming an adult; we all find that the two are essentially different kinds of existence.
And do we not, indeed, regard the child’s life as being, in certain respects, more true, more
genuine than the adult’s life—precisely because, among other things, trust plays such an enor-
mous, decisive role? (Løgstrup 2007: 3; 1961: 230)
This passage, I suggest, should be understood in axiological terms, as claiming that the
life of the child is better than the life of the adult, and the role of trust in it is crucial to
making it so; and precisely because it is a good of this sort, we do not need any kind of
argument to justify adopting it over distrust wherever this is possible.23
One way of getting at what Løgstrup has in mind here, I think, is to compare the case
of trust and distrust to that of health and illness. Here it seems plausible to argue that
health is the primary notion, as illness can only be conceived of as the absence of
health, of which illness is the privation or deficient form. And this means we must
accord a prior value to health over illness, so while it always makes sense to ask (e.g. of
a smoker): ‘why choose to be ill rather than healthy?’, it doesn’t generally make sense to
ask (e.g. of someone exercising) ‘why choose to be healthy rather than ill?’; a question
of this sort can only make sense in special circumstances, that are themselves less than
normal or ideal (e.g. of someone starving themselves in a hunger strike, where a fellow
prisoner might ask why the hunger striker has decided to eat the food that has been
given to him). To ask the question is to seek for an explanation for something which
22
Cf. Heidegger (1962: §§12, 83, and §26, 158). Løgstrup makes the connection with Heidegger explicit
in Løgstrup (1950: 30), and in Løgstrup (1995, II: 355; 1982: 220): ‘in Heidegger’s words, mistrust is trust’s
deficient mode’.
23
This approach is also discussed in Rabjerg (2007).
(if you understand the relative significance of health and illness) shouldn’t need to be
explained or even arise.
We might then hold something similar in the case of trust and distrust: that is, we
could hold that trust is the primary notion, as mistrust can only be defined or under-
stood as the absence of trust, of which mistrust is the privation or deficient form. And
this means we must accord a prior value to trust over mistrust, so while it always makes
sense to ask: ‘why decide to distrust rather than trust?’, it doesn’t generally make sense
to ask ‘why decide to trust rather than distrust?’; this question can only make sense in
special circumstances, that are themselves less than normal or ideal (e.g. in conditions
where one has been fooled before). As Løgstrup himself puts it:
In order to establish the foundational relation between trust and distrust, I have mentioned
that one normally does not ask anyone to account for the trust they might have in someone
else, but rather for their distrust of someone else. (Løgstrup 2007: 4; 1961: 231)
To ask the question is to seek for an explanation for something which (if you under-
stand the relative significance of trust over distrust) shouldn’t need to be explained or
even arise as a question.
Løgstrup thus argues that at a conceptual level, it is part of the concept of trust that it
is positively assessed, as an ‘ethically descriptive phenomenon’ (or what we might now-
adays call a ‘thick ethical concept’):
Take trust and distrust, for instance: the positivity of trust and the negativity of distrust are not
some evaluative accretions of which trust and distrust are the subjects, but inhere in the
phenomena themselves. Positivity and negativity, respectively, reside in the very meanings of
these two words. It runs counter to the intrinsic nature of trust, and is contrary to the very
meaning of the term, to evaluate trust as a negative phenomenon. Strictly speaking, we are pre-
cluded from conceiving trust as something negative. We may, of course, appraise trust as a
negative thing, but this can only come about through our applying a perspective to trust in
which we discount what trust itself imparts to us, namely, that it is positive. This is not merely
a theoretical possibility—it does happen that, despite its nature, we appraise trust negatively
because in a particular situation it is dangerous. Trust can be exploited; and so in bringing up
a child we have to caution him or her against a trusting attitude in certain sorts of circum-
stances. But that does not make trust a neutral phenomenon which we are free to conceive of
positively or negatively. It is only possible to evaluate it negatively by flouting its positive nature.
The same applies, of course, to a positive evaluation of distrust: it is possible only in spite of the
negativity of distrust. (Løgstrup 2007: 115; 1972: 48)24
So, Løgstrup argues, trust is like the other ‘sovereign expressions of life’, such as mercy,
love, and hope, that by default can only be evaluated positively, and thus have a priority
over their ‘deficient forms’ as a result.
24
Cf. also Løgstrup (1997: 249; 2010: 279), where Løgstrup identifies trust along with love as one of the
‘realities which we have summarized as the goodness of our human life’, which then means that ‘[t]rust is
there in advance, as surely as reserve and mistrust are a lack of trust and not the reverse’, thereby clearly
suggesting an axiological priority of trust over distrust.
286 Robert Stern
Nonetheless, the question can arise why trust is of positive value; what is it about
trust that makes this the case? Løgstrup makes clear that this is tied up with a proper
understanding of the fundamental operations that govern life, much like those that
relate to health, and being cut off from those operations cannot but cause us damage.
Thus, the attitude of trust is an important part of the human good, without which we
would be greatly impoverished, so that a world in which trust is possible is a better
world than one in which it is not, thus making trust axiologically fundamental.25
But how, more precisely, is this so?
One central reason can be found in the way Løgstrup identifies trust as crucial to a
certain openness we can have to one another, which is lost in conditions of distrust:
Not to let the other person emerge through words, deeds, and conduct, but to hinder this
instead by our suspicion and by the picture we have formed of him or her as a result of our
antipathy is a denial of life. (Løgstrup 1997: 14; 2010: 23)
For Løgstrup, this openness is characteristic of trust and also characteristic of love and
sympathy, because in these relations we take people at face value as they present them-
selves to us and connect to them directly, rather than forming a certain image or pic-
ture of their character, a theory about what makes them tick, and using that to define
them for us. We can form such theories for various reasons, but Løgstrup thinks that
what is important about the trusting relation is that it breaks them down as we go back
to seeing the person again, rather than defining them in terms of our picture of them:
The basic character of trust is revealed in yet another way. In love and sympathy there is no
impulse to investigate the other person’s character. We do not construct an image of who he or
she is . . . If, on the other hand, we are not in sympathy with the other person . . . then we begin
to form a picture of the other’s character . . . However, when we are in the direct association with
that person this picture usually breaks down; the personal presence erases it . . . Only where the
proof of her unreliability has in the most positive sense become an ingrown distrust, or where
the irritation and antipathy have shut me off completely, does the picture continue to stand.
(Løgstrup 1997: 13; 2010: 22–3)26
Distrust for Løgstrup thus amounts to a denial of life because it is to put a picture of the
person in place of the person themselves,27 and to define them in terms of what they
25
Cf. Løgstrup (1995, II: 355; 1982: 220): ‘[E]xternal conditions either cause trust and provide good
conditions for growth, or they harm it at its very source and provide poor conditions for growth’.
26
A more accurate translation of the last sentence would be: ‘Only where it is established that the other
cannot be trusted, and where this has literally become an ingrown distrust, or where irritation and antip-
athy have shut one off completely, does the picture continue to stand.’ There is some discussion of this
aspect of Løgstrup’s position in Lagerspetz (2015: 123).
27
While Løgstrup never gives up the view that this is to be avoided, he later recognizes that one can also
sometimes become so focused on the person that one becomes hypnotized by their presence, as it were,
which can also create problems in its own way, by leading us to just take account of their current concerns:
‘As the meeting [with the other] is absorbing me, it clips my imagination. My imagination cannot move
freely in the other’s history and world, for my attitude, thought, and feeling are narrowed to being an
answer to what is currently occupying the other person and what he requires and expects from me’
(Løgstrup 1983: 51, cited and translated in Bugge 2017: 224–5).
have been rather than allowing them to be capable of making themselves new, just as
(Løgstrup thinks) life can renew itself. Thus, he writes, ‘We might call this a trust in life
itself, in the ongoing renewal of life’ (Løgstrup 1997: 14; 2010: 23).
There are therefore two fundamentally damaging effects in a world where trust is
not possible, which shows how crucial it is to our good—one that involves damage to
the person who is distrusted, and one that involves damage to the person who d istrusts.
The argument for damage to the distrusted person might be put as follows:
(1) As living creatures, we have the capacity for renewal, which must be realized if
our lives are to go well.
(2) This capacity cannot be realized if we are confined by the picture or theory
imposed on us by others.
(3) Trust involves relating to another without a picture or theory.
(4) So being trusted by others enables us to function in the right way.
(5) To distrust is to impose a picture or theory on someone.
(6) So to distrust someone is to risk blocking their capacity for renewal, and thus
to prevent their life going well.
And the argument for damage to the distrusting person might be put as follows:
(1) The capacity of persons for renewal is central to the value of life; it is how lives
develop for the good.
(2) If we fail to recognize (1), we lose our sense of how life can be bettered, and if
we lose this, we lose the ‘zest for life’ or the ‘courage to be’ in our own lives.
(3) To lose one’s ‘zest for life’ or ‘courage to be’ is to be damaged.
(4) To recognize (1) one must not see others as confined by a picture or theory one
has about them.
(5) Trust involves relating to another without a picture or theory.
(6) So trust enables us to recognize (1) and is therefore sufficient to avoid the dam-
age that is (3).
And one might then construct a more general argument that brings in damage to both
parties, through the distrusting relation:
(1) Relating to another directly, without a picture or a theory, is central to having

a proper relation to them as living beings, without interposing anything into
that relation.
(2) Trust relates us to another without a picture or theory.
(3) So trust involves a proper relation to others.
(4) To distrust is to have a picture or theory.
(5) So distrust does not allow for a proper relation to others.
Trust is thus essential to us as a constitutive basic human good, alongside the other
‘sovereign expressions of life’, in the way that distrust is not; distrust is rather a
288 Robert Stern
eprivation of those goods, and so less fundamental, just in the way that illness is a
d
deprivation of health.
To understand what Løgstrup might have in mind here, it is instructive to consider
the example of Charles Myriel, the Bishop of Digne in Victor Hugo’s classic novel Les
Misérables. The Bishop, whose character and history are presented in the first chapter
of the book, then goes on to show not just compassion and pity towards the ex-convict
Jean Valjean, but also trust in allowing him into his home at all, and particularly letting
him sleep with access to the silverware owned by the household. It is clear that while
the Bishop’s sister Mademoiselle Baptistine is prepared to go along with him out of
respect for his judgement and goodness, the Bishop’s housekeeper Madame Magloire
thinks that he has gone too far this time and is profoundly shocked by his actions in
trusting Valjean. What seems to be emphasized by Hugo’s narrative, at least from a
Løgstrupian perspective, is that while everyone else sees Valjean as what he has done
and thus become—a criminal, a vagrant, an outcast—the Bishop (and thus to a lesser
extent his sister) see him as an individual human being standing before the Bishop as
such. Even Valjean seems shocked by the openness the Bishop shows to him and seeks
to remind him of how he should be categorized:
‘Mme Magloire,’ said the bishop, ‘will you please lay another place?’
The man [Valjean] moved nearer to the light of the table-lamp, seeming not to understand.
‘It’s not like that,’ he said. ‘Weren’t you listening? I’m a convict, a felon, I’ve served in the galleys.’
He pulled a sheet of yellow paper out of his pocket and unfolded it. ‘This is my ticket-of-leave—
yellow, as you see. This is why everybody turns me away. Do you want to read it? I can read.
There were classes in prison for anyone who wanted to learn. You can see what it says—“Jean
Valjean, released convict, born in—” not that that matters “—served nineteen years, five years
for robbery with violence, fourteen years for four attempts to escape—a very dangerous man.”
So there you are.’ (Hugo 1982: 85)
The Bishop, however, ignores all this and insists he is not interested. Then, in a letter
from his sister that is subsequently quoted in the text, she ponders on his behaviour,
emphasizing in particular that ‘my brother did not so much as ask the man where he
was born. He did not ask his story. For the story would have included some account of
his crimes and my brother clearly wished to avoid all reference to these.’ She gives the
following explanation of why the Bishop behaved in this manner: ‘He must have
reflected that the man, this Jean Valjean, was sufficiently oppressed already with the
burden of his wretchedness, and that it was better to distract his thoughts and make
him feel, if only for a little while, that he was a man like any other’ (Hugo 1982: 90).
Again, from Løgstrup’s perspective, we might understand Hugo’s point to be that to
trust Valjean just is to see him as no longer defined by his past, whereas conversely to
see Valjean through the eyes of the Bishop’s distrustful housekeeper is not really to see
the person as such, but all the things he stands for—a convict, a felon, a criminal with a
yellow ticket-of-leave. Like Løgstrup’s child, we might also think this gives the Bishop a
‘joy in living, a courage to be’ which the housekeeper, for all that we understand her
sensible caution and reasonable doubts, can never possess, partly because she cannot
see life as capable of the kind of renewal and reform in the same way as the Bishop can,
while also being deprived of the kind of direct interaction with others that his attitude
of trust also makes possible.
Løgstrup is of course not claiming that such distrust is never warranted, or deny-
ing that it could in some sense become ingrained in a person for good reason; but
nonetheless trust is prior to distrust as this could not but cut us off from a better way
of relating to others and to life itself. As someone who lived through the Nazi experi-
ence both in Germany and in Denmark, and the consequent erosion in relations of
trust that this entailed, this must be seen as the fundamental lesson Løgstrup learned:
not that society could not function in such conditions, because in some sense it did,
but that it is still a pathological form of human life, in which important goods were
lost as different and ‘deficient’ kinds of inter-relations took hold that required people
to be committed to a limited way of understanding one another as living beings.
Such is the force of our sense that it is limited in this way, that even when faced with
‘a destroyer’ we may find it difficult not to be open and trusting, as even in such an
encounter it can be hard not to hope that this goodness can be realized, even while
we know it cannot.
It might be objected to this account, however, that if the priority of trust is viewed
axiologically in this manner, and we then explain the value of trust in the way I have
claimed that Løgstrup does, it leaves him open to a fundamental difficulty: namely,
won’t we now have a reason to trust others based on the goodness that such trust brings
to our lives, but isn’t that precisely the wrong reason to trust others, so that if we give it
this axiological basis it cannot then function properly? Indeed, it could be said, we
might think of the Bishop in precisely these terms: because his positive view of the
world is so important to him, this is why he trusts Valjean, thus leading him to trust for
the wrong reason and in a way that blinds him to all the reasons he has to distrust; but
the housekeeper has no such positive view of life, so she is more clear-eyed about the
grounds she has to distrust Valjean, so that for her the attitude of trust is operating as it
should.
Now, Løgstrup himself arguably has an interesting response to this difficulty.28 For,
he suggests, along with all the sovereign expressions of life, it is not possible to trust for
instrumental reasons—for then one no longer is trusting, so that while trusting others
may bring us important goods, there cannot be reasons why we trust. So the ‘wrong
reasons’ problem cannot arise:
This unconditionality [of sovereign expressions of life, like trust] manifests itself in the fact
that as soon as an expression of life is called upon to serve another purpose than its own, it
disappears or is transformed into its opposite. . . . Mercy consists in an impulse to free another
person from suffering. If it serves another purpose, such as stabilizing society, it is replaced by
28
For helpful, more general discussions of this issue, see Williams (2002: 90–3) and Faulkner (2011:
174–5).
290 Robert Stern
indifference towards the other person’s suffering. The ulterior motive transforms mercy into
its own opposite.
This is why the spontaneous expressions of life defy all justification. The very moment we
seek to give a reason for them, we make them contingent upon that which we present as our
reason, and they become corrupted right then and there. We have made them a means to
obtain a goal other than their own: a means for the goal that is present in the justification.
(Løgstrup 2007: 128; 1982: 107)
And this argument seems very plausible in the case of trust, at least as Løgstrup con-
ceives it: for, as we have seen, to trust is to be open to the other person, to see them for
themselves rather than through a picture or a theory. But, if you trust the other for the
good such trust brings you, as a way of helping retain a positive view of life, this is just
another way to cut oneself off from the person concerned, by focusing in on yourself
instead. This is why, if we did find out that the Bishop trusted Valjean because he wishes
to retain the goods that come with trust, we would no longer say he trusted Valjean at
all. Thus, while the axiological view can explain why ‘trust is basic’ in the manner we
have explained, it does not threaten to undermine the grounds for trust in a way that
would be problematic, for by the logic of Løgstrup’s account of trust, the value of trust
cannot serve as the reason for trust and so distort trust in this way.
2.4 Ontological priority

We have seen, then, that we can take Løgstrup’s intriguing claim that ‘trust is basic’ in
an axiological manner: that is, a world in which trust is possible is better than a world
in which it is absent, and not just because lack of trust will damage or make impossible
other human relations (though it doubtless will), but because the world will be
deprived of goods intrinsic to trust itself, which are taken away or threatened once we
live in a world of distrust. In this way, then, trust is not parallel or equal to distrust, but
rather what is basic: not because we start out by trusting and then learn to distrust; or
because we have better reasons for trusting than distrusting; or because we cannot
distrust unless we trust; but because trust is of prior value, and thus distrust can only be
its deprivation or deficient form.
However, this is not quite the end of the story. For, in Løgstrup’s way of thinking
(which cannot be fully set out here), this kind of axiological priority is closely related
to another form of priority that can be attributed to trust: namely, what I have called
ontological priority, which is captured in Løgstrup’s remark (to which his claim that
‘trust is basic’ is added in a footnote) that ‘[t]rust is not of our own making; it is given’
(Løgstrup 1997: 18; 2010: 27).29 For Løgstrup, essentially what this means is that we
do not create or bring about trust as a practice or norm, in the way that we bring
about practices or norms like driving on the left, marriage, or even property, which
govern our various social institutions in ways that we hope are for the best. These
29
A more accurate translation would be: ‘Trust is not down to us. It is given.’
practices or norms are brought into being by us in a contractual or quasi-contractual

manner, and are thus goods that we introduce into the world and over which we have
control. But there are other structures which are fundamental to life itself, of which
we are part, that we could not bring about in this way as without them we could not
come to be at all, and trust (along with the other sovereign expressions of life) are
structures of this sort. Without trust we could not function as the creatures we are in
the first place, given our vulnerability and interdependence. In this sense, then, ‘trust
is not of our own making’, but is something given with the nature of human life as
such, and thus a ‘good’ for which we are not ourselves responsible, and for which we
can therefore claim no credit.30 By contrast, distrust is not ontologically basic in this
way, as human life could function perfectly well without distrust—it only becomes
required because we distort life through our own selfishness, which is why it is
not essential to the proper functioning of human life itself, and why the fact of its
existence is to our discredit.
It should be clear, therefore, that on Løgstrup’s account, the claims of axiological and
ontological priority need to be thought together: the latter claim is not a value neutral
one, for example like the claim that individuals are prior to social agents as the latter
are not possible without the former. Rather, the ontological priority of trust stems from
the fact that trust is a requirement or condition for the proper functioning of human
life, which is what makes it something we do not create ourselves (ontological prior-
ity), but also makes it a fundamental good and thus prior in this sense too, as having a
value that is also not attributed to it by us (axiological priority).31
This also shows why the two kinds of priority we rejected—psychological and
transcendental—may nonetheless come to have some place in Løgstrup’s account in
a suitably modified way. For while we argued that Løgstrup’s view does not operate at
just a psychological level, and we can now see why, nonetheless we can also see why
he appeals to psychological evidence of the damaging effects of distrust on children
and adults. And likewise, while we also contended that Løgstrup’s view does not
argue primarily for the rational priority of trust over distrust on the grounds that the
latter makes the former possible, nonetheless his claims about the ontological prior-
ity of trust do contain what might be thought of as a world-directed transcendental
claim, which is that trust is a necessary condition to the proper functioning of
human life on which distrust is parasitic, and thus is a normative structure in which
we are grounded as a ‘given’, rather than something we create for ourselves and for
30
Cf. Løgstrup (1997: 141; 2010: 161, translation modified), where Løgstrup rejects the thought that
trust and love can be subtracted from our evil—‘as though trust and natural love were not given to man,
but were his own achievements and could be credited to the account of the self ’.
31
Cf. Løgstrup (2007: 115; 1972: 48), where both claims come together when Løgstrup writes of a
phenomenon like trust: ‘Whether such phenomena are positive or negative, good or bad, is not first deter-
mined in our evaluation of them; it is not first decided in our engaging with them. They make me their own
before I make them my own. They have intimated to me what is good and bad before I consider the matter
myself and evaluate it. This is the reason for calling the positive expressions of life sovereign.’
292 Robert Stern
which we can claim any credit.32 The hope is, therefore, that having pulled apart
these strands in Løgstrup’s complex conception of trust and also shown how they
relate to one another, we can now see more clearly what it means for him to claim
that trust is basic, and what makes that claim of significant interest.33
References
Baier, Annette. 1986. Trust and Anti-Trust. Ethics 96: pp. 231–60.
Bugge, David. 2017. The Out-Side In-Sight: Løgstrup and Fictional Writing. In What is Ethically
Demanded? K. E. Løgstrup’s Philosophy of Moral Life, ed. Hans Fink and Robert Stern. Notre
Dame, IN: University of Notre Dame Press, pp. 216–33.
Derrida, Jacques. 1998. Faith and Knowledge: The Two Sources of ‘Religion’ at the Limits of
Reason Alone. Trans. Samuel Weber, in Religion, ed. Jacques Derrida and Gianni Vattimo.
Oxford: Polity Press, pp. 1–78.
Faulkner, Paul. 2007. On Telling and Trusting. Mind 116: pp. 875–902.
Faulkner, Paul. 2014. The Practical Rationality of Trust. Synthese 191: pp. 1975–89.
Fink, Hans. 2017. Løgstrup’s Point: The Complementarity Between the Ethical Demand and
All Other Moral Demands. In What is Ethically Demanded? K. E. Løgstrup’s Philosophy of
Moral Life, ed. Hans Fink and Robert Stern. Notre Dame, IN: University of Notre Dame
Press, pp. 54–79.
Gambetta, Diego (ed). 1988. Trust: Making and Breaking Cooperative Relations. Oxford: Basil
Blackwell.
Goldman, Alvin. 2011. Experts: Which Ones Should You Trust? In Social Epistemology:
Essential Readings, ed. A. Goldman and Dennis Whitcomb. Oxford: Oxford University
Press, pp. 109–33. (Original edition 2001.)
Hardin, Russell. 1996. Trustworthiness. Ethics 107: pp. 26–42.
Heidegger, Martin. 1962. Being and Time. Trans. John Macquarrie and Edward Robinson. Oxford:
Blackwell.
Hertzberg, Lars. 1988. On the Attitude of Trust. Inquiry 31: pp. 307–22.
Hieronymi, Pamela. 2008. The Reasons of Trust. Australasian Journal of Philosophy 86:
pp. 213–36.
Hollis, Martin. 1998. Trust Within Reason. Cambridge: Cambridge University Press.
72: pp. 63–76.
Hugo, Victor. 1982. Les Misérables, trans. Norman Denny. Harmondsworth: Penguin Books.
Jones, Karen. 1996. Trust as an Affective Attitude. Ethics 107: pp. 4–25.
Kant, Immanuel. 2011. Groundwork of the Metaphysics of Morals. Ed. and trans. Mary Gregor
and Jens Timmerman. Cambridge: Cambridge University Press.
32
Cf. Løgstrup (1962: 532): ‘Suspicion lives at the expense of trust, the evil will to overcome the other is
parasitic on the created possibility of life.’
33
I am extremely grateful to generous comments from David Bugge, Paul Faulkner, Hans Fink, Bjørn
Rabjerg, Tom Simpson, and to the audience at the conference on trust in Oxford at which this chapter was
originally presented.
Lagerspetz, Olli. 1998. Trust: The Tacit Demand. Dordrecht: Kluwer.

Lagerspetz, Olli. 2015. Trust, Ethics and Human Reason. London: Bloomsbury.
Levinas, Emmanuel. 1985. Ethics and Infinity: Conversations with Philippe Nemo. Transl. R. A.
Cohen. Pittsburgh, PA: Duquesne University Press.
Løgstrup, K. E. 1950. Kierkegaards und Heideggers Existenzanalyse und ihr Verhältnis zur
Verkündigung. Berlin: Erich Blaschker Verlag.
Løgstrup, K. E. 1961. Kunst og Etik. Copenhagen: Gyldendal.
Løgstrup, K. E. 1962. Kristendom uden Skabelsestro, in Vindrosen 9: pp. 523–35; trans. Hans Fink
and Robert Stern, available at: https://ethicaldemand.wordpress.com/resources-and-link/.
Løgstrup, K. E. 1972. Norm og Spontaneitet. Copenhagen: Gyldendal.
Løgstrup, K. E. 1980. Trampolinens Genmæle, Fonix 4: pp. 222–34; trans. Kees van Kooten Niekerk
and Robert Stern, available at: https://ethicaldemand.wordpress.com/resources-and-link/.
Løgstrup, K. E. 1982. System og Symbol: Essays. Copenhagen: Gyldendal.
Løgstrup, K. E. 1983. Kunst og Erkendelse. Copenhagen: Gyldendal.
Løgstrup, K. E. 1987. Fænomenologi og Psykologi. In his Solidaritet og Kærlighed og Andre
Essays. Copenhagen: Gyldendal, pp. 116–40; trans. Hans Fink and Robert Stern, available at:
https://ethicaldemand.wordpress.com/resources-and-link/.
Løgstrup, K. E. 1995. Metaphysics. Trans. Russell L. Dees, 2 vols. Milwaukee, WI: Marquette
University Press.
Løgstrup, K. E. 1997. The Ethical Demand. Trans. Theodor I Jensen, rev. and ed. with an intro-
duction by Hans Fink and Alasdair MacIntyre. Notre Dame, IN: University of Notre
Dame Press.
Løgstrup, K. E. 2007. Beyond the Ethical Demand. Trans. Susan Dew and Heidi Flegal, ed. Kees
van Kooten Niekerk. Notre Dame, IN: University of Notre Dame Press.
Løgstrup, K. E. 2010. Den Etiske Fordring. Aarhus: Klim.
Løgstrup, K. E. 2013a.Opgør med Kierkegaard. Aarhus: Klim.
Løgstrup, K. E. 2013b. Ophav og Omgivelse: Betragtninger over historie og natur, Metafysik III.
Aarhus: Klim.
Luhmann, Niklas. 1979. Trust and Power. Chichester: Wiley.
Plant, Bob. 2005. Wittgenstein and Levinas: Ethical and Religious Thought. London: Routledge.
Rabjerg, Bjørn. 2007. Løgstrups Kritik af Kierkegaard: Den Uendelige Kvalitative Forskel på
Fortabelse og Kærlighed. Res Cogitans 1: pp. 20–58.
Stangerup, Henrik. 1960. En Replik i Diskussionen om den Etiske Fordring, Perspektiv 7:
pp. 40–4; trans. Colin Roth and Eskil Irminger, available at: https://ethicaldemand.wordpress.
com/resources-and-link/.
Williams, Bernard. 2002. Truth and Truthfulness: An Essay in Genealogy. Princeton, NJ:
Princeton University Press.
Wittgenstein, Ludwig. 1969. On Certainty. Trans. Denis Paul and G. E. M. Anscombe. Oxford:
Blackwell.
Index
acceptanceâ•‡ 4, 7, 10, 47, 77, 87–8, 138, 172n.15, 189–90, 195, 197–8, 203, 206, 208–11, 220,
181, 219–20, 221n.23, 225–8, 261 223–4, 226, 228, 234, 236–7, 239–40, 242,
accountabilityâ•‡ 5, 7, 39–43, 46–8, 91, 93, 231, 246–7, 275, 288
237, 243, 245 beliefâ•‡ 3–5, 7–11, 15, 19, 22–4, 25, 29, 33, 36,
actionâ•‡ 8–10, 15–16, 23, 26, 41–2, 51, 91n.2, 56–7, 93, 109, 113–15, 119, 123–4, 126,
93, 96, 100, 104, 109, 116, 119, 124–6, 130–2, 136, 138–40, 144, 146, 149,
129–31, 133, 135–40, 142, 144–6, 150–1, 154–5, 161–75, 179–82, 184–8, 191–3,
164, 165n.6, 169, 171n.15, 173n.16, 178, 196, 197n.3, 201–2, 209, 214–16,
180–7, 192, 198, 204, 218n.14, 235, 218n.14, 220, 223–4, 228, 231–2, 236,
245–6, 260–1, 263 238–40, 242, 247, 252, 255–8, 260–4,
adviceâ•‡ 55–60, 63, 156 267–9, 279n.15, 282
affectionâ•‡ 73, 75, 79n.13, 152, 221 Bennett, J.â•‡ 165, 167
affectiveâ•‡ 36, 154, 168–9, 174, 181, 183–4, 196, betrayalâ•‡ 8, 35–8, 40, 46, 51–5, 58–9, 63, 66,
201, see also trust, affective; emotion 68–71, 72–4, 77–80, 82–3, 85, 87, 90–1,
agency 104, 123, 153–4, 175, 181, 187–8, 205–6,
individualâ•‡ 8, 22–3, 51, 62, 90–2, 94, 97–9, 221, 227–8, 231, 236, 241, 243, 246, 275
101–3, 137, 140, 145, 172–5, 178, 182, bioethicsâ•‡5
244–5 bipolar obligationâ•‡ 39, 41, 46
collectiveâ•‡ 7, 51, 129–31, 136, 139–44, 146–7, blameâ•‡ 7, 40–3, 45–8, 104, 206n.11, 243, 245, 275
231, 235 Brandom, R.â•‡ 237
conditions ofâ•‡ 256, 259, 263 Bratman, M.â•‡ 172n.15, 260
akrasiaâ•‡ 52, 100, 182 Buchak, L.â•‡ 190, 198n.4
Alston, W.â•‡ 165n.7 Burge, T.â•‡ 209–10, 211n.14, 242
altruismâ•‡159
amoralistsâ•‡ 102, 105 childrenâ•‡ 2, 61–2, 84, 92, 123, 126n.20, 146, 151,
Anderson, E. S.â•‡ 245 203–4, 210, 276–7, 279n.15, 281, 284–5,
angerâ•‡ 168–9, 245 288, see also infants
Anscombe, E.â•‡ 4, 174, 192 Coady, C. A. J.â•‡ 4
Antarctic Resupplyâ•‡ 177, 184–6, 188–90 cognitiveâ•‡ 32, 130–2, 146, 180–4, 187, 202, 209,
anti-reductionismâ•‡ 3–4, 242 255, 258, 264, 268
appreciationâ•‡ 31–2, 46–7, 85, 174, 216–17, 230, Cohen, J.â•‡ 172n.15
236, 244 Coleman, J.â•‡ 189, 198, 207
Aristotle/Aristotelianâ•‡ 32, 199, 210 collective
assertionâ•‡ 4–5, 10, 56–8, 217–18, 228–9, 236–9, actionâ•‡ 92, 129–31, 135–40, 142–3, 145–7,
244, 246 231, 235, 240, 246–7
assuranceâ•‡ 4–5, 45, 51–60, 63, 130, 134–5, agentsâ•‡ 7, 129, 134–5, 137, 141, 143, 145–6,
137–8, 140n.13, 141, 192, 201, 204, 206, 230–4, 237–8
223, 239–44, 280 attitudesâ•‡ 36, 92, 130, 135, 137, 139, 140–3,
theory of trustâ•‡ 51, 55, 63 232, 242
view of testimonyâ•‡ 4, 241 responsibilityâ•‡ 145, 238, 240, 243, 246
Austin, J. L.â•‡ 4, 219n.18 commitmentâ•‡ 1, 4–5, 8, 61–2, 66–7, 78n.12, 90,
authorityâ•‡ 4, 7–8, 39–40, 43, 46–7, 62–3, 143, 101, 121, 130n.1, 144, 167, 172–4, 179,
185, 188, 211, 215 181–2, 190, 221n.22, 234, 237, 240–1, 244,
awarenessâ•‡ 22, 24, 37, 72, 93, 95, 106, 123 246, 259, 263, 267–8
compensationâ•‡ 39, 43, 45, 187
Bacharach, M.â•‡ 130, 135–6, 139 competenceâ•‡ 36, 41, 53, 92, 94–5, 99, 105–6,
Baier, A.â•‡ 2, 36, 39–40, 51–2, 123, 149n.2, 183, 155, 168–9, 188, 196, 201–4, 232, 248,
243, 274 257, 264, 269
behaviourâ•‡ 14, 28, 80, 92–4, 96, 100, 104, 110, complianceâ•‡ 46n.15, 47, 101, 208
125–6, 129, 131, 137–8, 141, 146, 159n.16, conative attitudesâ•‡ 183, 184, 215, 220
296â•… index
confidenceâ•‡ 15n.2, 18, 22–3, 33, 48, 100, 123, Elster, J.â•‡ 110
135, 154, 196, 204, 206, 219, 224, 252, 267, emotionâ•‡ 46n.15, 78, 85, 130, 132,
274n.6, 276n.11, 283 168–9, 180, 202, 215–6,
contemptâ•‡ 41–2, 53–4 see also affective
contractarianâ•‡ 2, 93–4, 106 encouragementâ•‡ 22, 25, 28–9, 32, 48
Cook, K.â•‡ 3 enemyâ•‡ 195, 269
cooperationâ•‡ 1–3, 6, 8–10, 110–13, 116–7, enkratic coherenceâ•‡ 52
129–30, 132, 145–6, 159, 180, 195–6, environmentâ•‡ 208–10, 211n.14, 235, 256–7, 259,
197n.3, 198–201, 203, 207–8, 263, 265
214–5, 280 epistemicâ•‡ 4, 185, 188–90, 238–43, 245, 247,
coordinationâ•‡ 9–10, 130, 133–5, 137–8, 140–3, see also reasons
146, 218, 220–4, 228 constraintsâ•‡178
counting onâ•‡ 2, 8, 36–7, 45, 90–4, 96–7, 99–107, limitationsâ•‡ 251, 262, 266, 268
150, 152, 168–9 needsâ•‡60
courageâ•‡ 22, 33, 165n.6, 217, 229, 287–8 responsibilityâ•‡ 240, 246
Craig, E.â•‡ 210n.13 epistemologyâ•‡ 4, 10, 33, 195, 235, 264
cunningâ•‡ 25, 195, 207 essentialismâ•‡208–9
esteemâ•‡ 22, 24–5, 28
Darwall, S.â•‡ 39, 182 ethicsâ•‡ 5–6, 10, 33, 41, 62, 85, 97n.4, 102, 178,
Dasgupta, P.â•‡ 180 215, 242, 285
Davidson, D.â•‡ 66 evidenceâ•‡ 4, 9–11, 22, 24, 32–3, 55, 57–8,
D’Cruz, J.â•‡ 84 119–20, 122–3, 126, 137–8, 152, 153n.6,
deceptionâ•‡ 4, 113–14, 141, 280 154n.9, 155, 158, 159n.15, 163, 165–6,
decision theoryâ•‡ 134–5, 136, 180, 198n.4 168–9, 174, 177–9, 181–93, 201, 204–6,
decisionâ•‡ 81, 93, 99, 129, 134–5, 137–40, 142–3, 215, 217, 218n.13, 223–4, 226, 228,
162–3, 169–70, 173, 179, 186, 189–90, 199, 240, 242, 251–4, 256–8, 261–70,
204, 210, 257–65 275, 291
demandâ•‡ 10, 24n.10, 122, 251, 267, 273, 276, evidentialismâ•‡ 179, 180, 184
278, 282 evidentialist constraintâ•‡ 9, 177–81, 183–4, 187,
dutyâ•‡ 33, 41, 74, 159n.16, 181, 226, 252, 265 190, 192
democracyâ•‡ 211, 233 expectationâ•‡ 7–8, 33, 36–8, 40, 45, 48, 54, 64–5,
denialâ•‡ 104, 113, 187, 236, 286 70–1, 73, 75, 78n.11, 80–7, 90–6, 102,
deonticâ•‡ 7, 40–3, 45–8, 186, 226 105–6, 111, 115, 122, 126n.19, 129–32, 134,
dependabilityâ•‡ 16–18, 20n.7, 21–2, 23n.9, 24–7, 137, 141, 154, 168, 180–1, 191, 196, 199,
30–2, 210–11, 275 201, 204, 206–8, 210, 215, 219, 227–8, 275,
dependenceâ•‡ 5, 7–8, 32, 38, 92, 94–6, 100–1, 281, 283
103–7, 109, 119, 122–3, 172n.15, 274–5, explanationâ•‡ 1–2, 4, 10, 17, 70–1, 79, 92, 117–8,
281–2, 291 126, 137, 140, 172, 174, 183–4, 191–2,
dependence-responsivenessâ•‡ 205–8, 210, 212 197–202, 204–8, 211–2
Derrida, J.â•‡ 272n.1, 280n.16
desireâ•‡ 6, 75, 85, 114–15, 118, 124–5, 131, faithâ•‡ 7, 23–4, 27, 33, 45, 120–1, 123, 218n.14,
165, 179–80, 184–9, 196, 198, 215, 227, 251–5, 257, 262, 265, 267–8, 270
231, 241, 267 Family Manâ•‡ 186, 189
dilemmaâ•‡ 98, 117, 136, 200, 248 Faulkner, P.â•‡ 5, 40n.5, 75n.8, 162, 202, 222n.24,
disappointmentâ•‡ 8, 27, 35–7, 44–6, 48, 51–4, 224, 239, 247, 274n.5
58–60, 63–4, 66, 78, 83, 91, 154, 181, fiduciaryâ•‡ 6, 79
222, 281 Fricker, M.â•‡ 240–2, 247
discretionâ•‡ 39, 104, 106, 151, 155, 157, 159 Friedrich, D.â•‡ 43, 73, 75n.7, 76–7, 218n.14
dishonestyâ•‡ 280, 282 friends/friendshipâ•‡ 9, 17, 19, 21n.8, 33, 40,
disillusionmentâ•‡ 83, 85 46–8, 61, 70, 72, 75, 79–80, 84, 91, 97–9,
dispositionâ•‡ 14, 17, 20, 196 102–3, 107, 110, 133, 150, 156, 158,
distrustâ•‡ 6, 11, 79n.14, 120–1, 125n.17, 126, 150, 161, 171, 178, 181–2, 185, 187, 190,
154–5, 169, 181, 214, 230–1, 233–4, 241, 200, 226–7, 228n.33, 230, 234,
247–8, 269, 276–91 269, 278
doxasticâ•‡ 67, 162 Frost-Arnold, K. L.â•‡ 79, 170n.12
Durkheim, E.â•‡ 3 Fukuyama, F.â•‡ 3
indexâ•… 297
Galton, F.â•‡ 237 imaginingâ•‡ 100, 123, 172n.15, 173, 222–3

Gambetta, D. inactionâ•‡ 51, 246
Gauthier, D.â•‡ 112–13 infantsâ•‡ 2, 123, 146, 203, 276n.11, see also children
Gilbert, M.â•‡ 130n.1, 134, 240 institutionsâ•‡ 5, 59, 132n.6, 144, 157–9, 184,
Godâ•‡ 7, 47, 158, 252–5, 274 197–204, 206–8, 210–11, 214, 232–3, 243,
Gold, N.â•‡ 135n.9, 137 245, 290
goodnessâ•‡ 127, 252, 285n.24, 288–9 insultsâ•‡ 4, 86, 192–3, 242
goodwillâ•‡ 1, 36, 46, 73–6, 79n.13, 101, 119, 121, integrityâ•‡ 90, 131, 142, 144, 196, 202
168–9, 196, 199, 201–4, 217nn.10–11, 222, intentionâ•‡ 8, 10, 15, 33, 53, 55–6, 58–60, 62–8,
232, 274 92, 100, 139–40, 145, 150, 164n.5, 167–8,
gratitudeâ•‡ 81–3, 86, 153–4, 181, 243 172n.15, 173–4, 179, 205, 210, 214–16,
Grice, P.â•‡ 55 218–20, 232, 241, 257–8, 280n.17
group(s)â•‡ 10, 94, 129, 137, 140, 145, 159, 198, 202, interactionâ•‡ 6, 8, 31, 36, 44, 86, 99, 103,
230, 232–5, 237–43, 246–8, see also collective 119–12, 115, 117–9, 122, 124, 130–2,
ethics ofâ•‡ 239, 243–5 135, 138, 168, 178, 200–2, 204, 207, 210,
membersâ•‡ 129–30, 136–7, 139, 141, 143, 146, 230–1, 239, 289
235, 240, 244, 246–8 interactive relianceâ•‡ 222, 225
reliabilityâ•‡ 232–5, 238, 240, 243, 245, 247 interdependenceâ•‡ 136n.11, 140n.13, 273–4, 291
testimonyâ•‡ 10, 235, 237–40, 242–3, 245–7 international relationsâ•‡ 6, 230
trustworthinessâ•‡ 102, 230, 232–4, 238–41, intimacyâ•‡ 40, 43, 46, 61–2, 86, 98, 100, 157, 181,
243, 245, 247 202, 243, 274
valuesâ•‡ 141, 245, 248 intuitionsâ•‡ 23, 61–3, 99, 154, 163n.3, 195,
Guide casesâ•‡ 56–7, 60 208–11, 253
guiltâ•‡ 40–2, 48, 154 invitationâ•‡98
Hampshire, S.â•‡ 161n.1 Jesusâ•‡ 273, 274n.4

happinessâ•‡ 253–4, 269 Jones, K.â•‡ 35–9, 74, 75n.7, 154n.7, 168, 174, 181,
Hardin, R.â•‡ 109, 111, 149n.1, 180, 196, 197n.3 184, 204
Harding, M.â•‡ 73
Haslanger, S.â•‡ 98n.7 Kant, I.â•‡ 11, 35, 37, 41, 44–5, 109, 251–2, 257,
Hawley, K.â•‡ 3, 78, 120, 122–3, 150, 162, 189 264–5, 280
health/illnessâ•‡ 11, 159, 284–6, 288 Kavka’s toxin puzzleâ•‡ 167n.11
Heidegger, M.â•‡ 284 Keren, A.â•‡ 182n.2, 191n.7
Hertzberg, L.â•‡ 274n.5, 279n.15, 281n.19 knowledgeâ•‡ 1, 3–4, 37, 45, 57, 72, 82, 86, 94,
Hieronymi, P.â•‡ 164nn.4–5, 165n.6, 167, 110–11, 117–19, 123–4, 135–7, 149, 151,
170nn.13–14, 171–2, 174, 182, 196n.2, 180, 195, 209–10, 220, 228, 231, 237, 239,
220n.20 244, 256, 267, 269, 272n.1, 280n.16
Hi-Lo gameâ•‡ 9, 133, 136, 141–2 Kohn, M.â•‡ 3
Hinchman, E. S.â•‡ 239, 242 Kolodny, N.â•‡ 68n.11
Hindriks, F.â•‡ 136n.10 Kornblith, H.â•‡ 209
Hobbes, T.â•‡ 202 Korsgaard, C.â•‡ 126
Hodgson, G.â•‡ 134n.7, 233 Kosfeld, M.â•‡ 153n.5
Hollis, M.â•‡ 3, 247
Holton, R.â•‡ 35, 38–9, 91n.2, 132, 153n.5, La Salle, R. deâ•‡ 200–2, 206–7
154n.7, 162, 163n.3, 171, 180–1, 184, Lackey, J.â•‡ 238, 240–1
220n.20, 224–5 Lagerspetz, O.â•‡ 272n.1
honourâ•‡ 3, 8, 71, 82–8, 110, 122, 183 Lahno, B.â•‡ 132n.6, 134n.7
hopeâ•‡ 32–3, 46–8, 55, 81, 111, 114–15, 117, 126, Lahroodi, R.â•‡ 245
168, 174, 241, 255, 285, 289 Leaf casesâ•‡ 53–4, 59–60, 63
Horsburgh, H. J. N.â•‡ 82, 149n.1, 181, 184 Levinas, E.â•‡ 272n.1
Huijts, N. M. A.â•‡ 204 List, C.â•‡ 233
Hume, D.â•‡ 4, 37, 202, 214, 220–1, 223 Locke, J.â•‡ 2, 6
Hurka, T.â•‡ 182 Løgstrup, K. E.â•‡ 11, 41, 122, 125n.17, 272, 281,
286–90, 292
ignoranceâ•‡ 104, 111, 113, 115, 151 loveâ•‡ 8–9, 11, 40, 46–9, 122, 149–50, 152, 158,
illness, see health/illness 168, 273, 274n.4, 278, 285–6, 291n.30
298â•… index
loyaltyâ•‡ 8, 70, 77, 79, 80, 114, 159, 185, 226, prideâ•‡168–9
243, 248 promiseâ•‡ 1–2, 5, 7, 10, 24–5, 27–8, 31–2, 39,
Luhmann, N.â•‡ 3 43–6, 51–3, 56–8, 60–8, 72, 76–9, 81–4, 87,
90, 121, 177, 181, 214–15, 217–29, 231,
Maggs, C.â•‡ 159n.15 280n.17
Markovits, D.â•‡ 46 prudenceâ•‡ 97, 114, 166–7
Mathieson, K.â•‡ 232 psychologyâ•‡ 3, 14, 18n.6, 19, 28, 136, 140, 214,
McGeer, V.â•‡ 81, 97n.5, 177 218n.14, 248, 276–8, 280–1
McKenna, M.â•‡ 41n.8 of trustâ•‡ 70, 215, 220
McLeod, C.â•‡ 202 Pufendorf ’s Pointâ•‡ 48n.18
McMyler, B.â•‡ 182, 220n.20, 239 Putnam, R.â•‡ 3
mentalâ•‡ 2, 99, 172–3, 178–81, 183–4, 198
Midden, C. J. H.â•‡ 204 Rapport, F.â•‡ 159n.15
Miller, K.â•‡ 130n.1 rationalityâ•‡ 3, 8, 11, 52, 54, 64–5, 69, 100–1,
mistrustâ•‡ 66–9, 150, 272, 276, 278, 280, 282, 110–13, 118, 125–6, 131, 134–8, 178–9,
284n.22, 285, see also distrust 181, 183, 189–90, 198, 205, 207, 247, 252–3,
self-mistrustâ•‡ 52, 65, 67–8 255–8, 260, 263, 266–8, 278, 283
Möllering, G.â•‡ 126n.19 of cooperationâ•‡ 110, 113
Moore-paradoxical sentencesâ•‡ 191–2 of relianceâ•‡ 110, 112–14, 124–6
moralityâ•‡ 2–3, 52, 91, 97–9, 102, 107, 126, 252–3, Raz, J.â•‡ 165n.6, 184–5, 188, 191, 216n.6
269–70, 273, 282 reactive attitudesâ•‡ 9, 38–41, 44–5, 47–8, 78,
Moran, R.â•‡ 4, 192, 239, 242 153–5, 158–9, 231, 243–6
motivationâ•‡ 8, 15, 24, 75, 109–11, 113–14, reasons
117–18, 123–4, 127, 138, 155, 159, 179, epistemic,â•‡ 5, 55–6, 170–1, 177–80, 182–4,
181–2, 195–6, 197n.3, 199, 205, 207, 210, 186–8, 191, 201, 205, 239, 241
215, 221, 225, 245–6 exclusionaryâ•‡ 185–8, 191n.7
first orderâ•‡ 185, 187, 191
Nickel, P. J.â•‡ 223n.27 interpersonalâ•‡ 55, 63
non-cognitiveâ•‡ 162–3, 168–9, 181–2 moralâ•‡ 170, 181–2, 184, 199
non-evidentialismâ•‡ 52, 181–4, 187–8, 191, non-epistemicâ•‡ 178, 180, 182, 186, 188,
223, 253 191n.7
normsâ•‡ 2, 30, 32–3, 90, 97–8, 100, 102, 105–6, second-orderâ•‡ 178, 185, 187, 191
198, 210, 215, 258, 282 reassuranceâ•‡ 109–10, 122, 125
reductionismâ•‡ 4, 240
obligationâ•‡ 8, 24, 39–41, 43–4, 46–7, 52, 55, Reid, T.â•‡ 3
58–61, 63–5, 70–88, 90, 106, 181, 186, reinforcementâ•‡ 22, 27–8
218–19, 221–2, 223n.27, 225–8, 245, reliabilismâ•‡ 57n.4, 235
see also duty reliabilityâ•‡ 8, 10, 14, 15n.2, 52, 55–8, 63, 65,
O’Neill, O.â•‡ 5, 71n.1, 223n.27 94, 109–10, 124, 130n.1, 190, 192, 196,
optimismâ•‡ 36, 54, 99, 157, 168, 196, 199, 201, 199, 201–2, 204–6, 209n.12, 210, 222,
203–5, 218n.14 224n.28, 225, 232–40, 242–8, 251, 268,
organizationsâ•‡ 230, 232–3, 237, 241, 243–7, 254, 275, 283
see also collective agents, groups relianceâ•‡ 1, 7–10, 14–16, 18–25, 32, 35–7, 46,
Output Conditionâ•‡ 197, 199 51–8, 65, 69, 73–5, 78, 80, 91n.2, 92,
Owens, D.â•‡ 84n.21, 218n.13, 226, 228 109–16, 119–21, 124–6, 130–2, 141, 150–5,
157, 172n.15, 181, 187, 196, 198–9, 201,
Pagden, A.â•‡ 3 203–8, 210, 212, 217n.11, 218n.14, 219–20,
Paretoâ•‡ 133, 136n.11 221n.23, 222–3, 225–8, 231–6, 239, 241,
Parkman, F.â•‡ 201 243, 246–8, 251, 280n.15
participant stanceâ•‡ 38–40, 44–46, 99, 103, 132, religionâ•‡ 6–7, 273
138–9, 141–3, 146, 198, 228 resentmentâ•‡ 7, 38, 40–3, 45–7, 78–9, 83, 85,
patient–physician relationsâ•‡ 5 217n.11, 222n.24, 231, 241, 243, 245,
Peirce, C. S.â•‡ 236 274–5
Pettit, P.â•‡ 18n.6, 83n.19, 114–15, 205n.10, ring of Gygesâ•‡ 18, 28, 32
222n.24, 233 Role Reversalâ•‡ 61
phenomenologyâ•‡ 4, 180, 208, 211, 277, 278n.13 Romney, M.â•‡ 231
Pink, T.â•‡ 218n.14, 221n.23 Ross, A.â•‡ 205
indexâ•… 299
Scanlon, T. M.â•‡ 67, 90, 93, 95–6, 105–6, 228 infantâ•‡ 123, 203–4, 210
Schmid, B.â•‡ 130 interpersonalâ•‡ 51–3, 55, 58–9, 69, 132, 161,
Schmitt, F.â•‡ 232 163–4, 171n.15, 173–5, 215–17, 218n.14,
Schroeder, M.â•‡ 165n.6 232, 247
Searle, J.â•‡ 130n.1 intrapersonalâ•‡ 51–3, 55, 59, 66, 69
second-personalâ•‡ 24n.10, 35, 39–47, 123, 240–1 moralizedâ•‡ 52, 73, 78–80, 84
self-trustâ•‡ 49, 51, 53, 65–6, 68, 164 predictive/non-predictiveâ•‡ 91, 96, 105, 204,
Sellars, W.â•‡ 130n.1 247, 274, 282–3
Shiffrin, S.â•‡ 46 reciprocalâ•‡ 7, 38–42, 46–8, 121, 126
signalling/signalsâ•‡ 14, 94–5, 99, 104–5, 106, 110, therapeuticâ•‡ 80–3, 85, 87, 170, 233
113–14, 143, 145, 157, 181, 205, 207, 210 Trust Gameâ•‡ 9, 110–11, 114, 118–19, 124–6, 195,
Simmel, G.â•‡ 3 198, 203–4, 207
Simpson, T.â•‡ 72n.3, 73n.5, 195n.1, 199n.5 trust–reliance distinctionâ•‡ 231–2, 234, 239, 246–7
Sober, E.â•‡ 114 trust-responsivenessâ•‡ 14, 16–33, 101, 104, 205n.10
sociology/sociologistsâ•‡ 3, 157, 198, 204, 277n.13 trustworthyâ•‡ 5, 10–11, 20–1, 25–6, 28, 30–2, 48,
Soteriou, M.â•‡ 261–2 58–9, 63–6, 70–1, 80–3, 85–7, 90–2, 94,
Southwood, N.â•‡ 43, 73, 75n.7, 76, 77, 218n.14 96–7, 99–107, 109–10, 111n.4, 113–9,
Stangerup, H.â•‡ 280, 284 122–4, 126, 157, 161–2, 168, 177, 179–82,
Strawson, P. F.â•‡ 38, 40–1, 47, 132, 153n.5, 181 184–8, 190, 196, 201–3, 204n.9, 207,
Stroud, S.â•‡ 182 209n.12, 216, 222–3, 230–41, 243–8,
Sugden, R.â•‡ 130–1, 134–6, 139 251–2, 268–70, 275
Surowiecki, J.â•‡ 237 richly trustworthyâ•‡ 94, 99, 104–5
sympathyâ•‡ 3, 11, 214, 286 Two Promisesâ•‡ 61–3
Sztompka, Piotrâ•‡ 233 two-placeâ•‡ 9, 118–20, 149–53, 155–8
team reasoningâ•‡ 9–10, 130, 134–41 untrustworthyâ•‡ 65, 81, 97n.4, 107, 117, 122–3,
testimonyâ•‡ 3–5, 7, 10, 56–9, 60, 63, 164–6, 187, 169, 230, 233–6, 236–7, 238, 241, 243–6,
191–2, 233, 235–43, 245–7, 269, 279–80 248, 268
Thomas, D. O.â•‡ 181, 184 Uslaner, E.â•‡ 157–8
Thompson, C.â•‡ 243
Thompson, E. P.â•‡ 158n.12 Velleman, D.â•‡ 260–1
three-placeâ•‡ 9, 99, 102, 118–9, 121–3, 149–53, viceâ•‡245
156–7 virtueâ•‡ 3, 18, 21, 32, 80, 82–3, 85, 97n.4, 98, 114,
Titmuss, R.â•‡ 159 132n.6, 182, 245, 248
Tollefsen, D.â•‡ 240, 242 vulnerabilityâ•‡ 4, 19, 41, 43, 46–7, 53–4, 74–6,
transcendentalâ•‡ 4, 11, 276, 278–83, 291 78–80, 85–6, 90, 100, 102, 106, 118, 126n.19,
trust 129, 130n.1, 132, 141–2, 144, 151, 183,
absence ofâ•‡ 120, 197n.3, 216, 276, 283–5 187–8, 227–8, 232, 274–5, 279, 281, 291
affectiveâ•‡ 5, 162, 168, 183, 207, 247, 274, 282
axiology ofâ•‡ 11, 283–5, 289–91 Walker, A. D. M.â•‡ 81–2
cognitiveâ•‡ 10, 137, 180–2, 187–8, 191, 210, Walker, M.â•‡ 71n.2
215, 220 Watson, G.â•‡ 41n.8
conjunctive accountâ•‡ 153, 180, 188n.5 Williams, B.â•‡ 4, 109–10, 115–17, 119, 121,
contractualâ•‡ 118–9, 121–2 167n.9
disappointedâ•‡ 51–3, 59, 277, 281 Williamson, T.â•‡ 237
empowering theory ofâ•‡ 7, 14, 18–19, 21–2, 27, wisdom of crowdsâ•‡ 237–8, 244
29–30, 33 Wittgenstein, L.â•‡ 272n.1, 279n.15
evidence ofâ•‡ 20, 26, 29 wrongingâ•‡ 8, 71–2, 77–82, 85–7, 225
fulfilment ofâ•‡ 6, 177–8, 184, 186–7, 189
genuineâ•‡ 78, 80, 82, 85, 130–2, 140–1, 146, Young, I. M.â•‡ 245
155, 157, 195–6, 203–4, 209
of institutionsâ•‡ 158–60, 164 Zagzebski, L.â•‡ 188

Paul Faulkner, Thomas Simpson - The Philosophy of Trust-Oxford University Press (2017)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Paul Faulkner, Thomas Simpson - The Philosophy of Trust-Oxford University Press (2017)

Uploaded by

Copyright:

Available Formats

OUP CORRECTED PROOF – FINAL, 01/11/2017, SPi

The Philosophy of Trust

List of Illustrations vii

15. Faith in Kant 251

2.1 Responsiveness to evidence of trust 26

Stephen Darwall, Yale University

Robert Stern, University of Sheffield

1. Trust and Cooperation

2 Paul Faulkner and Thomas Simpson

In juxtaposition to the equality presumed by members of a club, Baier proposes infant–

2. Trust and Knowledge

4 Paul Faulkner and Thomas Simpson

it may be an injury not to be believed’ (1979: 150). In believing someone, I show my

3. Trust and Social Philosophy

6 Paul Faulkner and Thomas Simpson

intelligence; international relations; government and corporate surveillance; and

8 Paul Faulkner and Thomas Simpson

canvassing a ‘trust-based’ solution. According to this solution the problem is dissolved

10 Paul Faulkner and Thomas Simpson

instance, on an expedition to the Antarctic whose success required a supply drop, it

12 Paul Faulkner and Thomas Simpson

1. Trust and Reliance

the empowering theory of trust 15

16 Victoria McGeer and Philip Pettit

the empowering theory of trust 17

To illustrate the distinction between dependability and durability in our usage,

18 Victoria McGeer and Philip Pettit

the empowering theory of trust 19

2. The Case for the Empowering Theory

20 Victoria McGeer and Philip Pettit

the empowering theory of trust 21

22 Victoria McGeer and Philip Pettit

2.2 The situational enhancement of dependability

2.2.1 tHE ENCOURAGEMENT ARGUMENT

the empowering theory of trust 23

2.2.2 the request-based argument

24 Victoria McGeer and Philip Pettit

2.2.3 the esteem-based argument

the empowering theory of trust 25

26 Victoria McGeer and Philip Pettit

2.3 Modelling situational enhancement

Sensitivity to reasons of + Sensitivity to trustor’s

Figure 2.1 Responsiveness to evidence of trust

the empowering theory of trust 27

Figure 2.2 Trust-responsive dependability

2.4 The situational reinforcement of durability

28 Victoria McGeer and Philip Pettit

the empowering theory of trust 29

3. Trust-Responsiveness on the Empowering Theory

30 Victoria McGeer and Philip Pettit

3.2 Trust-responsiveness is context dependent

the empowering theory of trust 31

3.3 Trust-responsiveness is a work in progress

32 Victoria McGeer and Philip Pettit

There may be an attraction, of course, in the prospect of achieving an increasing level

3.4 Ascribing trust-responsiveness is subject to moral norms

the empowering theory of trust 33

34 Victoria McGeer and Philip Pettit

1. Baier, Jones, and Holton on Trust: The Participant

trust as a second-personal attitude (of the heart) 37

person. If, consequently, the expectational element in Jones’ account is to capture

Strawson calls these ‘participant attitudes’ because ‘they belong to involvement or

trust as a second-personal attitude (of the heart) 39

2. Participant, Reciprocating Attitudes: Deontic

2.1 Responsiveness to evidence of trust 26

f urther here.10 There is a sense in which shame is a reciprocal attitude of contempt as