A. Coliva - Extended Rationality_A Hinge Epistemology (2015)

Extended Rationality
Palgrave Innovations in Philosophy
Series Editors: Vincent F. Hendricks, University of Copenhagen and

Duncan Pritchard, University of Edinburgh.
Titles include:
Mikkel Gerken
EPISTEMIC REASONING AND THE MENTAL
Kevin Meeker
HUME’S RADICAL SCEPTICISM AND THE FATE OF NATURALIZED
EPISTEMOLOGY
Ted Poston
REASON AND EXPLANATION: A DEFENSE OF EXPLANATORY COHERENTISM
Aidan McGlynn
KNOWLEDGE FIRST?
E.J. Coffman
LUCK: ITS NATURE AND SIGNIFICANCE FOR HUMAN KNOWLEDGE AND
AGENCY
Jonathan Matheson
THE EPISTEMIC SIGNIFICANCE OF DISAGREEMENT
Forthcoming titles:
J. Adam Carter
THE PROSPECTS FOR RELATIVISM IN EPISTEMOLOGY
Annalisa Coliva
THE VARIETIES OF SELF-KNOWLEDGE
Julian Kiverstein
THE SIGNIFICANCE OF PHENOMENOLOGY
David Pedersen
POLITICAL EPISTEMOLOGY: Epistemic Theories and Knowledge Institutions
Christopher Pincock and Sandra Lapointe (editors)
INNOVATIONS IN THE HISTORY OF ANALYTICAL PHILOSOPHY
Palgrave Innovations in Philosophy

Series Standing Order ISBN 978–0–230–36085–3 (hardback)
(outside North America only)
You can receive future titles in this series as they are published by placing a standing order.
Please contact your bookseller or, in case of difficulty, write to us at the address below with
your name and address, the title of the series and the ISBN quoted above.
Customer Services Department, Macmillan Distribution Ltd, Houndmills, Basingstoke,
Hampshire RG21 6XS, England.
Extended Rationality
A Hinge Epistemology
Annalisa Coliva
University of Modena and Reggio Emilia, Italy
© Annalisa Coliva 2015
Softcover reprint of the hardcover 1st edition 2015 978-1-137-50188-2
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No portion of this publication may be reproduced, copied or transmitted
save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency,
Saffron House, 6–10 Kirby Street, London EC1N 8TS.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The author has asserted her right to be identified as the author of this work
in accordance with the Copyright, Designs and Patents Act 1988.
First published 2015 by
PALGRAVE MACMILLAN
Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited,
registered in England, company number 785998, of Houndmills, Basingstoke,
Hampshire RG21 6XS.
Palgrave Macmillan in the US is a division of St Martin’s Press LLC,
175 Fifth Avenue, New York, NY 10010.
Palgrave Macmillan is the global academic imprint of the above companies
and has companies and representatives throughout the world.
Palgrave® and Macmillan® are registered trademarks in the United States,
the United Kingdom, Europe and other countries.
ISBN 978-1-349-50563-0 ISBN 978-1-137-50189-9 (eBook)
DOI 10.1057/9781137501899
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources. Logging, pulping and manufacturing
processes are expected to conform to the environmental regulations of the
country of origin.
A catalogue record for this book is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Coliva, Annalisa, 1973–
Extended rationality : a hinge epistemology / Annalisa Coliva, University
of Modena and Reggio Emilia, Italy.
pages cm.—(Palgrave innovations in philosophy)
Includes bibliographical references and index.
1. Justification (Theory of knowledge) 2. Knowledge, Theory of. I. Title.

BD212.C65 2015
1219.3—dc23 2014049722
For Leonardo and Maria Elisa
This page intentionally left blank
Contents
Acknowledgments ix
Credits x
List of Abbreviations xi
Introduction 1
1 Moderatism about Perceptual Warrants 18
1 The architecture of perceptual warrants 1:
the liberal view 20
the conservative view 28
the moderate view 33
4 Moderates are legion (while liberals and
conservatives are not) 39
5 Moderatism and the content of perception 43
6 Summary 54
2 Further Motivation for Moderatism 57
1 Against liberal Mooreanism 58
2 Against entitlements 64
3 Against a priori warrants 71
3.1 Wedgwood and a priori bootstrapping 72
3.2 Peacocke’s rationalism 78
3.3 Sosa and rational intuitions 82
4 Summary 84
3 The Bearing of the Moderate View: Transmission
Failures, Closure, Easy Knowledge, and Bootstrapping 86
1 Wright’s and Davies’s original failure 88
2 Davies’s new failure and its failure 90
3 Another kind of failure 92
4 Moderates, TF2, and Moore 94
5 Transmission failures 1&2: a compatibilist approach 96
6 Transmission failures 1&2 and closure 101
7 Transmission failures, easy knowledge, and
bootstrapping 108
vii
viii Contents
8 A possible counterexample: Pryor’s (?) failure to

see this failure 114
9 Summary 118
4 The Extended Rationality View 119

1 Naturalism, pragmatism, and Wittgenstein vis-à-vis
Humean skepticism 121
2 The extended rationality view and Humean skepticism 127
3 The extended rationality view, Cartesian skepticism,
closure, and “abominable conjunctions” 136
4 The extended rationality view and epistemic relativism 139
5 The extended rationality view and the Oblomovian
challenge 145
6 The extended rationality view and truth 147
7 Summary 150
5 The Extended Rationality View Extended 153

1 The extended rationality view and the principle of
the uniformity of nature 154
2 The extended rationality view and the past 158
3 The extended rationality view, testimony, and
other minds 161
4 The extended rationality view and basic logical laws 166
5 Summary 180
Notes 181
Bibliography 205
Name Index 211
Subject Index 213

Acknowledgments
This book has developed over a considerable span of time. Some parts,
extensively reworked to make them fit within the structure of the book,
have been published previously. I will therefore avoid repeating my
thanks to those people who were either in attendance at presentations of
the relevant material, or who gave me comments on previously written
versions and whom I thanked when those papers appeared in print. Here
I wish to express my gratitude to those who took the trouble of reading
and commenting on the entire manuscript. In particular, I would like to
thank Yuval Avnur for extensive feedback and for unfailing support and
encouragement throughout the years, especially during a fantastic stay at
Scripps College, thanks to an O’Brien Distinguished Professorship in the
winter of 2014. My deepest gratitude also goes to Martin Kusch for his
comments and criticisms on an earlier draft of the first four chapters of
the book and for the suggestion of connecting this work more clearly to
my earlier book Moore and Wittgenstein: Scepticism, Certainty and Common
Sense. Many thanks also to Danièle Moyal-Sharrock for the suggestion of
calling the view at the heart of this book “extended rationality” and for
her constant support in promoting what we call “hinge epistemology”.
We share the view that it is only this way that Wittgenstein’s contribu-
tion to epistemology will not remain a topic merely for scholars inter-
ested in the history of analytic philosophy. I would also like to thank
Duncan Pritchard for proposing to have a seminar on the final draft of
the book at Eidyn Research Centre (Edinburgh) and Adam Carter for
organizing it. Special thanks are due to all commentators, whose sugges-
tions and criticisms I have tried to take into account while revising
the copy-edited version of the manuscript. Namely, Alan Millar, Aidan
McGlynn, Modesto Alonso-Gomez, Silvan Wittwer, Kevin Wallbridge,
Natalie Ashton and Jie Gao. Thanks are also due to Mikkel Gerken and
Jesper Kallestrup, for long and stimulating discussions over the years and
during the seminar at Eidyn. Finally, my deepest thanks to Eva Picardi,
Paolo Leonardi, Carla Bagnoli, Giorgio Volpe, Sebastiano Moruzzi,
Michele Palmira, Delia Belleri, Alessia Pasquali, Eugenio Orlandelli,
Filippo Ferrari, Raban Reichmann, Luca Zanetti and everyone at the
COGITO Research Centre, with whom I discussed an earlier draft of the
book and without whom none of what I have been able to write over the
last five years would have been possible.
ix
Credits
Chapter 1, §5 contains a translation of my paper “Percepire le ragioni?”,

Iride, XXV/65, 2012, pp. 117–130.
Chapter 3, §3–5 and 8 contain revised and enlarged material stemming

from my paper “Varieties of failure (of warrant transmission – what
else?!)”, Synthese, 189/2, 2012, pp. 235–254.
Chapter 4 is an extensively revised and enlarged version of my papers

“Moderatism, transmission failures, closure and Humean scepticism”,
in E. Zardini, D. Dodd (eds) Scepticism and Perceptual Justification, Oxford
University Press, Oxford, 2014, pp. 248–272; and “Moore’s proof, liberals
and conservatives. Is there a (Wittgensteinian) third way?”, in A. Coliva
(ed.) Mind, Meaning and Knowledge: Themes from the Philosophy of Crispin
Wright, Oxford University Press, Oxford, 2012, pp. 323–351.
Chapter 5, §4: an earlier version appeared as “Lessons from Pascal Engel.

Achilles, the tortoise and hinge epistemology for basic logical laws”,
in A. Meylan, J. Dutant, A. Logins and D. Fassio (eds) Liber Amicorum.
Hommage à Pascal Engel, 2014, pp. 696–715, http://www.unige.ch/lettres/
philo/publications/engel/liberamicorum.
x
List of Abbreviations
OC Wittgenstein L. (1969) On Certainty (Oxford: Basil Blackwell)

BIV Brain in a vat
xi
Introduction
It is natural to think that justifications stem from a system of assump-

tions. Take, for instance, the mathematical case. Theorems – that is,
justified mathematical propositions – are obtained through proofs, given
certain axioms. Indeed, different sets of axioms constitute different theo-
ries, which give rise to different sets of justified mathematical proposi-
tions. However, no matter how many sets of axioms are possible, and no
matter whether the propositions eventually justified are different, the
former are clearly necessary in order to generate proofs and therefore
warrants for certain less basic mathematical propositions.
Take a more humdrum case. We see a ball roll between two poles.1
We form the belief that a goal has just been scored. Yet, clearly, that
belief is justified only insofar as it is assumed that a football match is
being played and this, in turn, justifies us in inferring that supporters of
the team whose player sent the ball between the poles will be cheering.
If it were a different game, whose point is actually that of letting the
ball roll between the poles in one’s mid-field, while players of the other
team should prevent that from happening, we would not be justified
either in believing that a goal has just been scored, or in inferring that
supporters of the team whose player sent the ball between the poles will
be rejoicing. Similar examples could be multiplied ad libitum.
Many further interesting examples can be found in Wittgenstein’s On
Certainty (OC). For instance, he noticed that geological beliefs about the
specific age of the Earth could only be justified by taking for granted
that the Earth had existed for a very long time. Only that way could
fossils and other evidence be brought to bear on the issue of the specific
age of the Earth.2 To see why, consider the hypothesis that the Earth,
with all its fossils, had just been created five minutes ago. We would still
have those fossils yet they could not be taken to prove anything about
1
2 Extended Rationality
the specific age of the Earth. If a specific belief about the age of the Earth
can be justified at all, it is only thanks to those fossils and the collateral
general assumption regarding the long existence of the Earth.
Wittgenstein made a similar point concerning what we regard as
evidence in favor of a specific historical event.3 For instance, we think
that our belief that Napoleon was defeated at Waterloo in 1815 is justi-
fied thanks to documents and testimonies to that effect. However, those
very documents and pieces of testimony would not justify that belief at
all, if the Earth had been created just a few minutes ago with all those
documents and putative testimonies. For the fact that an event took
place in 1815, two centuries ago, presupposes that the Earth has existed
for a very long time. Therefore, in order to possess a justification for
specific historical and geological propositions it appears that we must
take it for granted that the Earth has existed for a very long time, so that
everything we regard as evidence in favor of specific historical or geolog-
ical propositions can actually play that role. If that assumption were
not in place, those very fossils, documents, and putative testimonies
would be no evidence whatsoever in favor of those specific historical or
geological propositions. He called these presuppositions, which make
it possible for us to have justifications for ordinary empirical proposi-
tions, “hinges” – hence the subtitle of this book. Here are the relevant
passages:
All testing, all confirmation and disconfirmation of a hypothesis

takes place already within a system [of assumptions]. And this system
is not a more or less arbitrary and doubtful point of departure for
all our arguments; no, it belongs to the essence of what we call an
argument.
That is to say, the questions that we raise and our doubts depend on
the fact that some propositions are exempt from doubt, are as it were
like hinges on which those turn.
That is to say, it belongs to the logic of our scientific investigations
that certain things are in deed not doubted.
But it isn’t that the situation is like this: We just can’t investigate
everything, and for that reason we are forced to rest content with
assumption. If I want the door to turn, the hinges must stay put.
(OC 105, 341–343)
Yet, the main bulk of this book will not deal with the case of beliefs about
the distant past and their characteristic presuppositions regarding the
Introduction 3
Earth’s long existence. Rather, it concerns the structure of perceptual justi-

fication. That is, the kind of justification we have, based on current sense
experience, for propositions about mid-size objects in our environment,
which contribute to the content of our beliefs. Examples of this kind of
proposition are: “Here is my hand”, based on seeing one in front of one’s
nose; “Here is a PC”, based on watching the screen while writing this very
Introduction; and so on. The claim at the heart of the present work is that,
when specific empirical beliefs are at stake, perceptual justification is only
possible based on a system of assumptions. That is to say, it is not enough
merely to have a certain type of experience – hand-like or PC-like – in
order to justify the corresponding empirical proposition, “Here is a hand”
or “Here is a PC”. Rather, we need one or more general assumptions that
allow us – no matter how defeasibly that might be – to bring those expe-
riences to bear on a world populated by physical objects. Why so? The
reason will become apparent in the first chapter, which criticizes a contem-
porary version of the thesis that experience by itself (absent defeaters) is
enough to give one a defeasible justification for ordinary empirical propo-
sitions. For, it will be claimed, experience by itself underdetermines the
propositions that could legitimately be taken to be justified by it.
Compare this with the case of “A goal has just been scored”. We take
our experience of seeing a ball roll between two poles to justify that
proposition only thanks to already taking for granted that a football
match is being played. However, as we saw, that very experience could
be just the same and, if a different game were being played, a different
proposition (or set thereof) would be justified. Going back to the case
of a hand-like experience: just by itself it could equally justify “Here is
a hand”, “I am hallucinating having a hand”; “I am a BIV (a brain in
a vat) who is having a hand-like experience”, and so on. Hence, taking
that experience to partly justify “Here’s a hand”, rather than any propo-
sition compatible with that very experience, depends on already taking
for granted that we are interacting with a world populated by phys-
ical objects, that our sense organs mostly work correctly (and, possibly,
some other propositions, for example “I am cognitively lucid and not
a victim of massive perceptual and cognitive deception”). Alternatively,
consider the case of a certain visual experience and the belief “Here is
a PC”, which we take to be justified partly on the basis of that experi-
ence. Again, it is only by taking for granted that we are interacting with
physical objects, and that our sense organs are mostly reliable, that we
can take that experience as a justification for “Here is a PC”, rather than
for “I am dreaming of seeing a PC” or “I am a BIV having a PC-like expe-
rience”, and so on.
Notice, moreover, that the general propositions I claim must be

assumed in order for our experiences to bear legitimately onto other
propositions about mid-size objects in our environment, so that the
latter are justified, are not needed to give us an indefeasible justification
for these more specific empirical propositions. Ceteris paribus – that is,
given those very assumptions and experiences – we could still be facing
papier-mâché hands and PCs, for instance. What we need those assump-
tions for is to be able to overcome what one might call “our cognitive
locality”. Namely, we need them in order justifiably to go beyond our
experiences and to bring them to bear on a universe populated by phys-
ical objects, whose precise identity and properties can, of course, still
escape us in certain circumstances, rather than take them to be caused
by appropriate neurological stimulations. To be more precise: if a certain
kind of evidence e, like a perceptual experience, is compatible with mutu-
ally incompatible kinds of propositions, namely propositions about
mid-size physical objects (P) or about BIVs being stimulated so as to
have those experiences, absent any causal interaction with the relevant
physical objects (Q), in order for e to accrue to a justification for proposi-
tions of kind P rather than Q, some extra condition has to be met. It is
only in this way that we will have a justification for propositions of kind
P and will be within our rights in taking a given experience, which is a
mind-dependent kind of evidence, to bear on propositions about mind-
independent objects.
Hence, to repeat, the claim at the heart of this work is that percep-
tual justification can take place only thanks to a system of very general
assumptions, such as “There is an external world”, “My sense organs
work mostly reliably”, “I am not a victim of massive perceptual and
cognitive deception”, and so on. A problem as old as the very history
of epistemology – epitomized by “Agrippa’s trilemma” – concerns the
epistemic status of these assumptions. In the quest for justification
each horn of this trilemma is thought to be problematical: either we
end up providing circular justifications; or we embark on an infinite
regress; or stop with unjustifiable and therefore a-rational and arbitrary
assumptions.
Suppose we hold that each assumption, in its turn, needs to be
warranted, in order for it to generate perceptual justification, together
with the appropriate kind of experience. For, one may think, it is only
if these assumptions are justified that our ordinary empirical beliefs
based on them will rest on a secure base and will therefore be justified.
Consider the football case: it is only if I am independently justified in
believing that a football match is being played that my experience of
Introduction 5
seeing a ball roll between two poles provides a justification for “A goal
has just been scored”. I think that in this case there is no dispute. Why
so? Because it is indeed very easy to see how that assumption can be
independently justified, for instance: I know that I paid for a ticket to
the football match between teams A and B in the stadium where I am
now sitting, watching the game; or, I know that every Sunday a football
match is played in the stadium where I am, roughly at this time, and
that today is Sunday; or else, if I am watching the match on television, I
know that it has been advertised as the football match between the two
teams; or that commentators keep repeating that this is a crucial football
match, or saying that the team that prevails will win the World Cup, and
I know that the World Cup is a football tournament; and so on.
Yet, as soon as we move away from the football example, things
become much more complicated. Consider the historical case and the
very general proposition that the Earth has existed for a very long time
before our birth. One might think that that proposition is justified by
a lot of our specific historical beliefs based, in their turn, on testimo-
nies, both personal and documentary, often recorded in academic texts.
However, as we saw before, those testimonies and documents could be
just the same and yet have appeared and been recorded in academic
books only a few minutes back. Therefore, clearly, it is not to be expected
that a justification for such a general proposition could be obtained by
inferring to it starting with premises that are justified just as long as that
very proposition is taken for granted. That kind of justification would
ultimately be circular and it would be no justification at all.
Nor is it to be expected that justification for it could ensue from
coherence between it and our further beliefs. Justifications are epis-
temic goods – to put it in most general terms – that should speak to the
truth of what they are supposed to justify. Yet, starting with the same
evidence – apparent testimonies, documents and academic records – we
could just as well produce a different and yet entirely coherent system
of propositions. In that system the general assumption is that the Earth
has just been created replete with everything we find in it and the corre-
sponding specific empirical propositions are like “It looks as if Napoleon
was defeated at Waterloo about three centuries ago”. Nothing makes the
first system of beliefs more likely to be true than the second one. If we
think otherwise it is either because we are more used to it and therefore
think that it is epistemically kosher; or else it is because we consider its
specific beliefs justified and think that this, in turn, gives us a justifica-
tion for its basic presuppositions. However, in the former case, we would
conflate our willingness to endorse a given system of beliefs with proof
of its truth and, in the latter, we would try to provide a circular justifica-
tion for its basic assumptions, starting from beliefs that are justified only
insofar as those very assumptions are taken for granted. We discuss this
issue at length in Chapter 3.
Another possibility is to think that we have a priori justification for
“The Earth has existed for a very long time”. Where would it come from,
though? Intuition is an appealing answer, but only momentarily, because
one then faces the problem of explaining its nature and workings. This
remains one of the philosophically most arduous tasks.4 Perhaps we have
some kind of a priori yet inferential justification, coming from reflection
on the very meaning of the terms involved. Notice, however, that this
would immediately be hostage to the particular theory of meaning we
are prepared to subscribe to. For it is only on the basis of some inferen-
tial-role semantics, which takes either a holistic or a molecularist form,
that we can sensibly claim that, for instance, it is constitutive of the
meaning of “Earth” that it has existed for a very long time. Yet, a direct
referentialist could simply say that “Earth” refers to the planet we are all
living on now, whether it has existed for a very long time or only for five
minutes, and that this is the meaning of “Earth”.
Faced with this kind of difficulty – to repeat, distrust in justifications
for general assumptions stemming from specific beliefs that would be
justified only by already taking them for granted; as well as in coher-
ence theories of justification, and mistrust in intuition and in inferential
a priori justifications stemming from meaning-constitutive considera-
tions – recent years have seen the emergence of yet another proposal,
which belongs to the a priori camp broadly construed. This proposal
provides for non-evidential warrants, called entitlements, for very
general background presuppositions, such as, “The Earth has existed for
a very long time”. Entitlements however, at least in the main way in
which they are currently thought of,5 are not meant to speak to the
truth of these propositions. Yet, if this is the case, it is very hard to see
how entitlements could be genuine epistemic warrants for them, since
they are neither evidential nor guides to the truth of the relevant propo-
sitions, capable of providing a viable solution to the original problem
they were meant to address. Namely, the problem of how these general
assumptions could actually be epistemically justified.
Similar considerations to the ones just rehearsed for “The Earth has
existed for a very long time” could be made for “There is an external
world”, “My sense organs work mostly reliably” and “I am not a victim
of massive perceptual and cognitive deception”, which, arguably, are the
presuppositions thanks to which our sensory experiences can be taken
Introduction 7
(defeasibly) to justify our beliefs about mid-size objects in our environ-

ment. If this were the situation, since we can neither provide immediate
justification for these propositions, nor mediate ones, it would seem
that the skeptical outcome would ensue. That is to say, it would seem
that the only plausible alternative would be to hold that these are just
a-rational assumptions and that, even if we think we are justified in
believing ordinary empirical propositions, we are not.
I think that in broad outline this is the path that (save for consid-
erations regarding coherence and entitlements) led Hume to his skepti-
cism. However, it is again Hume who, to my mind, offered the first seeds
to try to escape it, as paradoxical as that might seem. Seeds that were
developed much later on, in a different direction, by Wittgenstein in
On Certainty, as I think Peter Strawson was the first to recognize in his
Scepticism and Naturalism. Some Varieties (1985).
According to Hume, we cannot help believing that there is an external
world, so that our sensory experiences are constantly brought to bear on
a world populated by mid-size objects that are taken to exist independ-
ently of our minds, even when they are not directly perceived by us. For
Hume, it is part of our psychological constitution that we cannot but
form beliefs and devise actions accordingly. That is the way we live. That
is the human condition; but notice that, for him, the human condition
is the Humean condition of being forced by nature to follow certain
forms of psychological and practical conduct that fall outside rational
sanction. Rationally, however, we have to recognize that our most basic
beliefs are not justified and neither are our more specific empirical beliefs
based on perceptual evidence.
Wittgenstein, in contrast, put forward the view that even though we
cannot justify these very general assumptions (or indeed, in his view,
even more specific ones which are equally necessary for certain sorts
of empirical practices and inquiries), we cannot help but make them
thanks to our upbringing within a community that shares language and
certain epistemic practices or, more generally, a form of life. However, his
idea was that the human condition is not the Humean one at bottom.
That is to say, one in which there is an unbridgeable gulf between what
reflection imposes on us – that is, the recognition that all justification
for ordinary empirical propositions rests on unwarrantable assump-
tions – and what we cannot help doing, given our psychological and
more culturally determined nature – like going on living as if, thanks
to those assumptions, our ordinary beliefs were justified. Rather, the
human condition, in his view, is one in which we simply have to recog-
nize that whatever degree of justification we possess for our ordinary
empirical beliefs, and that we do in fact possess, it takes place within

a system of assumptions, which are neither justified nor justifiable.6
Therefore, according to Wittgenstein, the human condition is impor-
tantly different from the Humean one, primarily because justifications
are indeed possible, at least for ordinary empirical propositions, but only
thanks to a system of unwarrantable assumptions.
This is the kind of picture about the structure of perceptual justifica-
tion that I wish to present and defend in some detail in this book. It
can been seen, among other things, as the attempt to make good one of
the horns of Agrippa’s alleged trilemma. According to that trilemma, no
justification is ever possible because there are no immediately justified
propositions, which can serve as the basis for all others,7 and so the quest
for justification ultimately leads to an infinite regress; nor can justifica-
tion be produced in a circular way8 or by resting on unjustified assump-
tions. The view I will present and defend here agrees that, when it comes
to very general propositions, such as “There is an external world”, we
cannot immediately justify them (whatever that might mean as we have
briefly explored above), nor can we justify them in a circular way by
advocating beliefs which are justified only as long as these assumptions
are already taken for granted. However, it aims to vindicate the idea that
even if these assumptions are neither warranted nor warrantable, they
can serve to produce a justification for ordinary empirical propositions,
once we enjoy the appropriate kinds of experience.
For reasons that will become apparent in the first chapter, I call this
view the “moderate” conception of perceptual warrant, as it can be seen
as lying in between the so-called “liberal” view, proposed in recent years
by Jim Pryor, and the “conservative” view defended mostly by Crispin
Wright. We need not enter into the details of these positions here. In
outline, the first one corresponds to the intuition that perceptual justifi-
cation is not theory-laden. As long as there are no defeaters, our percep-
tual experiences give us an immediate justification for ordinary empirical
propositions such as “Here is a hand”. In contrast, the conservative view
has it that a warrant for ordinary empirical propositions can be had only
if certain general assumptions are independently justified.
The idea I want to defend is that, contrary to the liberal position, we
need assumptions to overcome our cognitive locality, that is if we want
to form defeasibly justified beliefs about physical objects in our environ-
ment based on our experiences. Yet, contrary to the conservative view,
these assumptions need not be warranted, for, in fact, they cannot. The
reasons why they cannot are explored in Chapters 2 and 3, following the
kind of reasoning we have quickly rehearsed here for the case of “The
Introduction 9
Earth has existed for a very long time”, and by analyzing the details
of several contemporary attempts to provide independent justifications
for “There is an external world”, and the like. For present purposes, let
me stress that the moderate architecture of the structure of perceptual
warrant just says that a specific empirical proposition P, for instance
“Here is a hand”, is perceptually justified iff one has the relevant kind
of experience, such as a hand-like one, and the background assump-
tion that there is an external world is in place (possibly together with
other ones such as, “My sense organs are mostly working reliably”, “I
am not the victim of massive perceptual and cognitive deception”, and
so on). Since this definition is compatible with various ways of thinking
of the status of such an assumption, which range from an externalist
positing that the world is just like that, to making it the content of a
doxastic attitude of a specific subject, moderatism will be introduced
in Chapter 1 as a family of possible views and not as just one single
position. Yet, they would all be different species of the same genus –
the genus I call, following the Wittgensteinian metaphor, “hinge episte-
mology” – because they all hold that perceptual justifications take place
“within a system” (OC 105) of assumptions, that is of propositions, that
lie beside the route of inquiry and that make justifications within the
inquiry possible in the first place.
Furthermore, these species of the same genus are compatible with
different accounts of how we should think of the content of perceptual
experience in order for the latter partially to constitute a justification for
ordinary empirical beliefs. Indeed, it is my conviction that the moderate
architecture of the structure of perceptual warrants has been endorsed, in
one version or another, by many different philosophers, like naturalists
of a Humean persuasion (provided they were prepared to forsake Hume’s
skeptical attitude at the reflective level), Wittgenstein in On Certainty,
and naturalists inspired by him, like Strawson. Also, pragmatists would
turn out to be moderates, in my view, as would those externalists about
the nature of perceptual justification who are prepared to recognize a
role for general assumptions, like Ernest Sosa in recent writings.
One could then be tempted to think that moderatism inspired by
some of Wittgenstein’s considerations in On Certainty would offer
only a momentary relief from skeptical worries for – the train of
thought would go – it would remain that if those assumptions are not
justifiable, then they may well turn out to be false. Hence, nothing
guarantees that our epistemic practices rest on a secure basis. Yet this,
according to Wittgenstein, would be right only if it made sense to
call those assumptions into question. That is to say, it would be right
only if those assumptions were in the business of epistemic appraisal

at all; if it made sense to apply to them the very categories of truth
and falsity and, more importantly and less contentiously, the very
categories of being justified/unjustified, or even known or unknown.
But the main thrust of On Certainty, at least according to the kind
of, so-called, “framework reading” I myself have put forward,9 is that
those very general assumptions are not like ordinary empirical prop-
ositions, contra what G. E. Moore held. Rather, they are similar to
rules; that is to say, they play a normative role and, like rules, are not
subject to truth or falsity, nor to assessment in terms of justification
or lack thereof.10 Compare with “Stop at traffic lights when red”. It is
intuitive to think that it does not correspond to a pre-ordinate fact,
and so that it does not make sense to think of it as either true (or false)
in any robust sense of that word. Nor, for the same reason, would it
make sense to think of it as either justified – that is as supported by
further facts or experiences – or as unjustified – as disconfirmed by
further facts and experiences. If “There is an external world” is rele-
vantly similar to “Stop at traffic lights when red” then the skeptical
worry that, being unjustified, it might turn out to be false would be
off target and due to a mistaken conception of the very nature of that
“hinge”.
I myself embrace the Wittgensteinian view that justifications for ordi-
nary empirical propositions are possible thanks to a system of assump-
tions, that is owing to a system of more general propositions, which, as
such, cannot be justified. However, I do not wish to endorse the view that
these assumptions are rules, devoid of empirical content, if that is indeed
Wittgenstein’s considered view on the issue.11 Yet, if this seems to be a
sensible avenue to explore as far as the status of “There is an external world”
is concerned, it actually seems to be in danger of re-opening the door to
the skeptical challenge. For now, how would one block the conclusion that
this is merely an assumption we make which, however, is actually unjusti-
fied and therefore not rational, exactly as a skeptic would hold? This is the
challenge I take up in the fourth chapter of the book, where I present and
defend the extended rationality view. Accordingly, if either empirical, or
coherentist, or a priori kinds of warrant for “There is an external world”
are unattainable and entitlements are only putative epistemic warrants,
then we may try and defend the epistemic legitimacy of that assumption
by claiming that, even though unwarranted, it is in fact constitutive of
epistemic rationality itself. Just as both rules and moves are part of the
game so, I argue, both constitutive assumptions and perceptual justifica-
tions, which are possible thanks to them, are part of epistemic rationality.
Introduction 11
To ban constitutive assumptions from epistemic rationality simply because

they are not warranted (as they cannot be), like skeptics do, is due to too
narrow and unmotivated a conception of the extent of epistemic ration-
ality. Namely, one that confines it to perceptually justified beliefs only.
In contrast, epistemic rationality extends beyond the latter to those very
assumptions that make it possible to produce ordinary perceptual justifi-
cations and that therefore make it possible to have the kind of practice of
forming, assessing, and withdrawing from empirical beliefs on the basis of
perceptual evidence, which is itself constitutive of our very notion of epis-
temic rationality. If so, it turns out that we are actually mandated by epis-
temic rationality itself to assume “There is an external world”. However, a
rational mandate is not an epistemic warrant – namely, an epistemic good
that speaks to the truth of what it is meant to warrant. Skeptics are right
to think that we have no such warrant for “There is an external world”.
However, they are wrong to think that, for that very reason, that proposi-
tion falls outside the scope of epistemic rationality and that, for that very
reason, we cannot have perceptual warrants for our ordinary empirical
beliefs.
Hence, the final and specific version of hinge epistemology I endorse
has it that thanks to epistemically rationally mandated assumptions
such as “There is an external world”, together with appropriate courses
of experience, we can and do have perceptual justifications for ordinary
empirical beliefs such as “Here is a hand”. However, to repeat, this is
the species of the hinge epistemology genus I endorse. It is not the only
possible one; even though I am convinced it is the one that has the best
prospects of success, because it speaks to the skeptical challenge, albeit
by developing an indirect response to it. Namely, not a response that
contradicts the skeptic by providing what he is looking for and main-
taining that it is impossible to attain, that is warrants for “There is an
external world”. Rather, the extended rationality view is a response that
shows that the skeptical quest is somehow illegitimate when it comes
to very general propositions like “There is an external world”, as it asks
for warrants that cannot be obtained and it is based on too narrow and
unmotivated a conception of epistemic rationality.
The extended rationality view also speaks to the peculiar relativist
challenge that, unlike other forms of epistemic relativism originated
by reflections in philosophy of mathematics and science, were to target
the assumptions of the basic practice of forming, assessing, and with-
drawing from empirical beliefs on the basis of perceptual evidence.12
The thought would be this: we are familiar with the view that different
mathematical or logical theories are possible and hence that there can
be different sets of axioms that would generate proofs for different

theorems. Why not think, then, that even though our conception of
epistemic rationality has its own background presuppositions, such
as “There is an external world”, other ones would be possible too? In
particular, those which forsook that assumption and didn’t take our
experience to bear on beliefs about mind-independent objects, yet
could account for that very experience.13 In response to this challenge,
I claim that this alternative conception would not even have the means
to account for the very content of our perceptual experience, which
is objective, that is to say such as to be as of objects and/or proper-
ties out there, and not as of mere subjective variations, like sensations
of cold and heat. Nor would it be able to explain perceptual constan-
cies, namely the phenomenon whereby no matter how far or near an
object we move, and so no matter how many changes in perceptual
input there are, we keep having a representation as of the same object
or property.14 Arguably, all this happens without the intervention of
concepts, through the workings of partially encapsulated perceptual
faculties. Hence, it would be difficult to explain the very content of
our perceptual experiences if it were not assumed that our perceptual
faculties must operate, or at least have evolved, in an environment
populated by physical objects. If, in contrast, one wished to say that
those perceptual representations are indeed the result of the operation
of concepts, rather than of the subpersonal and automatic workings of
our perceptual faculties, then it would actually turn out that the best
way of explaining them would involve an appeal to the very assump-
tion this alternative model should have dispensed with, namely that
there are mind-independent objects. For, if one insisted that perceptual
constancies and the like were, in fact, the product of the intervention
onto perceptions of our concepts, our overall cognitive system should
be such as to embed a general rule that unifies subjective and scat-
tered representations in such a way as to produce a representation of an
object or a property out there.
Finally, I consider what I dub “the Oblomovian challenge”, after
Goncharov’s novel Oblomov. I take that challenge to consist in the idea
that we may opt out of playing the game of epistemic rationality alto-
gether, as it were. Not in order to play any other game, with different
characteristic assumptions, as a relativist would say, but simply in order
not to play that game, or alternative ones in the same ballpark, at all.
I hold that only faced with the Oblomovian challenge can practical
considerations be appealed to in order to motivate our acting the way
we do. Yet this challenge is not epistemological in nature, but rather
Introduction 13
practical or even psychological. Therefore, it is only to be expected that

it will be responded to by appealing to that kind of consideration.
Hence, the overall picture I will be presenting and defending is that
perceptual justification is possible only thanks to some general assump-
tions, which are mandated by epistemic rationality itself and that are
universal as long as we are dealing with creatures like us that have to rely
on their senses to form beliefs about physical objects in their environ-
ment and that have a certain kind of perceptual experience. Whether it
is conceivable to imagine different creatures who can either rationally
intuit truths about physical objects, rather than arrive at them through
the operation of their senses, or that there might be creatures with alto-
gether different perceptual experiences, it is not an issue that should
really preoccupy us. For what we are trying to understand is the human
condition and the kind of epistemic security for some of our actual
beliefs that can be attained.
To repeat, at the heart of the moderate conception lies the claim that
the assumptions that make the acquisition of perceptual warrants possible
are not warranted. For we do not possess, nor can possess, any proof of
their truth. Yet, given the extended rationality view, those assumptions
are rationally mandated, and, thanks to them and to appropriate courses
of experience, we do possess a perceptual justification for many ordinary
empirical beliefs.
As will become apparent, a number of important consequences follow
from such a general picture. For example, it follows that the principle of
closure for justification under known entailment is not unconditionally
valid. For “Here is my hand” entails “There is an external world”. Yet,
while we can justifiably believe the former, we cannot justifiably believe
the latter. Furthermore, we have to recognize that beside the kind of
warrant transmission-failure principle originally presented by Wright,15
according to which an argument cannot generate (or enhance one’s
previous) warrant for a conclusion if, and only if, the warrantedness of
its premises depends on already possessing a warrant for its conclusion,
there is another kind of warrant transmission-failure principle, which is
indeed at issue in the kinds of case that are of most interest to philoso-
phers. Namely, the one according to which an argument cannot generate
(or enhance one’s previous) warrant for a conclusion if, and only if, the
warrantedness of its premises depends on the very assumption of its
conclusion. It is for this reason that also on the moderate architecture
of perceptual warrant, and not only on its conservative counterpart,
Moore’s argument (“Here is a hand. If there is a hand here, there is an
external world. Therefore, there is an external world”) is not cogent.
Furthermore, it is because of this kind of transmission-failure that boot-

strapping arguments designed to produce warrants for very general
beliefs, such as “My sense organs are mostly working correctly”, out of
specific perceptual beliefs justified by means of occurrent perceptions
would not be cogent either. In Chapter 3, the details of these important
consequences of the moderate view will be explored at length.
Finally, in Chapter 5, I explore the possibility of bringing the extended
rationality view to bear on all other aspects of epistemic rationality.
Namely, on the case of those hinges, like the principle of the uniformity
of nature, that allow us to form inductive justifications for generalized
empirical propositions. The same goes for the case of “There is a past”,
which is a hinge of the diachronic aspect of epistemic justification, that
is the one in which we form (or retain) justifications for propositions
about the past based on memory. Moreover, the extended rationality
view will be extended to some presuppositions of the social aspect of
epistemic rationality, where we form justified beliefs on the basis of testi-
mony. Finally, it will be applied to the case of modus ponens, considered
a basic rule of inference, constitutive of the deductive aspects of epis-
temic rationality.
I think it is a merit of the present proposal that it can be extended to
various domains, beside the perceptual one. If, in a Carnapian spirit, the
interest and even the plausibility of a philosophical view can be measured
by its fruitfulness, among other factors, I think the moderate conception,
developed along the lines of the extended rationality view, does well in
this respect. For, as already mentioned, it allows us to shed new light
on issues such as the validity of the principle of closure, the nature of
transmission-failure, the status of arguments such as Moore’s. Moreover,
it can help illuminate the nature of inductive justification, of memory-
based and testimonial warrants, and the status of basic logical laws.
However, let me add some caveats before ending this Introduction.
First, I would like to warn the reader that the bulk of the book will
contain detailed discussions of other theorists’ views, especially Pryor’s
and Wright’s. Anyone familiar with the recent epistemological litera-
ture will know how central and influential their debate has been. The
moderate view comes out nicely as an alternative to both, just as the
extended rationality view is best appreciated if one sees how it distances
itself from Wright’s notion of entitlements. However, as this Introduction
should have made apparent, the moderate view can be seen as a devel-
opment of one of the horns of Agrippa’s trilemma, and can actually
be considered as a development of some Humean and, especially, of
some Wittgensteinian ideas. Its force resides, I think, in the fact that it is
Introduction 15
intrinsically plausible, as it gives us the means to vindicate the intuitive

view that we do have perceptual justifications for our ordinary empirical
beliefs, as well as – when developed along the lines of the extended
rationality view – the means to defuse skeptical and relativist challenges.
Yet, by placing it in the context of a crucial contemporary discussion in
epistemology, its plausibility and interest will actually be augmented by
showing how it does better than several of its best present-day competi-
tors on a number of fronts. Other views, beside Pryor’s and Wright’s will
be discussed, some of them with a fair degree of detail. Again, this should
help us appreciate how the moderate view does better than its competi-
tors on a number of fronts. However, not all possible epistemological
positions will be discussed. This would clearly be impossible. Notice,
moreover, that since the specific version of the moderate position I will
be developing is of an internalist brand (more on this shortly), it makes
sense, I think, to take issue mostly with its contenders in the internalist
camp. Surely, some readers will be disappointed because they will feel
that positions that are closer to their hearts haven’t been discussed, at
least not with the degree of detail they would have liked; or, conversely,
they will complain that positions that are not that close to their hearts
have been discussed too extensively. It is always difficult to strike the
right balance, one that will make all possible readers happy, so I will be
content to make some of them happy. To repeat, I will discuss mostly
those that are the best developed views within the internalist camp and
that can be taken to be paradigmatic examples of relevant alternatives
to my own theses. Consequently, a certain familiarity with these posi-
tions will be presupposed. However, I do not think the discussion will
ever proceed in such a way that the points at issue will be totally opaque
to a reader who is not familiar with them, for these views will always be
introduced and explained before any critical discussion. This, however,
is not an introductory book. Hence, those readers who find themselves
unhappy with my characterizations, lack of detail, or clarification of the
relevant views would presumably benefit from reverting to the relevant
literature. It will be a success if what I say makes apparent to them the
interest and relevance of the literature taken into account in this book.
Second, I would like to make some terminological remarks. I will
be talking indifferently of justifications and warrants, meaning, if not
otherwise indicated, evidential, perceptually-based warrants/justifica-
tions. I take them to be epistemic goods that ought to speak to the truth
of what they are meant to warrant or justify. Hence, having a hand-
like experience, together with the assumption that there is an external
world, absent defeaters, will epistemically support the proposition “Here
is a hand”. To put it contentiously, it will make it (more) true that there

is a hand there, rather than nothing at all, or something other than a
hand. Indeed, I will deal mostly with propositional justifications, that
is with whatever speaks or can be considered to speak to the truth of
a given proposition “in the abstract space of reasons”. That is to say,
independently of whether that proposition is the content of a subject’s
actual belief, and independently of which further beliefs a given subject
might actually have. It is then a further question how those justifica-
tions turn into doxastic ones, that is into warrants for actually held
beliefs that have the relevant propositions as content. Moreover, it is a
different issue how specific collateral beliefs make those propositional
justifications rationally available to the subjects who entertain those
collateral beliefs.
Another terminological point concerns the use in this book of expres-
sions like “internalism” and “externalism”, in epistemology and not in
the theory of mental content. I wish to give minimal characterizations
of them, for it is well-known that any number of more specific readings
are proposed in the literature. Minimally, an internalist conception of
justification or warrant has it that the factor(s) that constitute(s) a justi-
fication in the relevant domain must be accessible, at least in principle,
to the subject. Obviously, this requires some element of idealization,
for we will have to think of subjects who have the relevant conceptual
repertoires and who are endowed with the relevant information. I think
that these idealizations, however, are legitimate. After all, epistemology
is a normative and not a descriptive discipline, so it is free to think of
its subjects as suitably idealized. The important point is that they are
never idealized beyond reason, that is by ascribing to them abilities and
knowledge that exceed human capabilities and knowledge. Externalists
will therefore be those who, in contrast, think that those justifying
factors need not be available to the subject, not even in principle.
Furthermore, in order to increase the intelligibility of the book, I
have added a summary, at the end of each chapter. To the same end, I
have written a brief introduction to each chapter that presents the main
claims defended directly in that part of that work, with no detailed refer-
ence to the existing literature against which some of the positive claims
are advanced in the course of the chapter. To the same purpose, I have
added definitions, when needed and possible, of various positions, in
the hope that this effort will forestall possible misunderstandings.
Finally, let me make a meta-philosophical consideration. It is my view
of philosophy that several coherent positions can be maintained vis-
à-vis a given philosophical issue, or set thereof. And that it is possible
Introduction 17
to adjudicate between them, at least pro tempore, by assessing how each

of them fares with respect to a set of questions and challenges that are
recognized as characteristic of the debate at issue. Of course I do think,
and will offer motivations to concur with my judgment, that the views
I propound fare at least equally well or even better, but certainly no
worse, than their competitors regarding the distinguishing questions at
the heart of the debate on the structure of perceptual justifications and
the problem of external world skepticism. Yet, it would be a sign of intel-
lectual arrogance to think that one’s competitors have been conclusively
proved wrong. It is in this spirit that I present the views contained in
this book. My hope is that a picture will emerge which, given its merits
and its fruitfulness, insofar as it can be shown to shed new light on
a number of important issues and to be applicable in wider domains
beside the one at the heart of this book, will be considered worth taking
seriously as one important contender in this complex and exciting phil-
osophical battlefield.
1
Moderatism about Perceptual
Warrants
We do think of having perceptual justifications for our ordinary empirical

beliefs. Hence, we think that, when we look at our hand while holding
it up in front of us, we do have a justification based on that perceptual
experience for a belief such as “Here is a hand”, taken to be about a
mind-independent object. When we consider how such a warrant can
be had, however, things become incredibly complex and it is difficult to
see exactly what is needed to achieve that end.
The aim of this chapter is to present an account of the epistemic
structure of our empirical judgments based on the deliverance of our
perceptual systems, which I dub moderatism. Accordingly, perceptual
experiences can constitute a defeasible warrant for corresponding beliefs
about mid-size objects only thanks to some very general assumptions.
Chief among them is “There is an external world”. Such an assumption
is needed in order to surpass what we may call our cognitive locality.
The intuitive idea is that experiences can be subjectively just the same
irrespective of whether or not they are brought about by a causal interac-
tion with physical objects. In order to take them to bear onto a universe
populated by material objects, it ought to be assumed that there is such a
universe with which we are, at least mostly, in contact. Such an assump-
tion, however, does not prevent one from occasionally going wrong, due
to possibly unfavorable conditions. Therefore, the perceptual warrant
one obtains through perception is defeasible. Yet, the relevant assump-
tion allows subjects to bring their experiences to bear on beliefs about
material objects.
As I characterize it in the first instance, moderatism is in fact a family
of views, for the details of what it means to assume “There is an external
world” can be spelled out in several ways, which would give rise to
different types of moderatism – that is, to different species of the same
18
Moderatism about Perceptual Warrants 19
genus. Thus, even though I end up favoring one possible variant, I think
the moderate camp extends beyond that. I present some examples of
views propounded in the history of philosophy, which belong to it.
Equally, I defend one particular conception of the nature and content of
perceptual experience and the conditions in which it can play a justifica-
tory role with respect to the corresponding empirical beliefs. Yet again,
this will be but one possible conception of it. Hence, although this is
the view of perceptual content I favor, I think moderatism is compatible
with those different conceptions.
The discussion shows how moderatism fares better than some of
its main competitors in the internalist camp.1 A first, quite intuitive,
view, called “liberal” and put forward mainly by Jim Pryor, holds that
just by having a certain experience, with a given conscious representa-
tional content, that P, absent reasons to doubt that it is being formed
in normal conditions, one would thereby acquire an empirical warrant
for the corresponding belief. This position is presented and criticized
in §1. Its main fault is that it is very difficult to see how it can help us
overcome our cognitive locality. For our experiences could subjectively
be just the same had they been produced in unfavorable conditions. Yet,
absent reasons to doubt that such uncongenial conditions obtain, we
would be ipso facto and implausibly justified in believing “Here is my
hand” and in disbelieving “I am dreaming of having a hand here” or “I
am hallucinating having a hand”.
The liberal view can be seen as occupying one extreme in the range
of possible (internalist) theories regarding the structure of empirical
warrants. Namely, the one that is the least demanding regarding the
conditions that need to be satisfied in order to possess such a warrant.
At the other extreme, and still in the internalist camp, lies the, so-called,
conservative view of the structure of empirical warrant, largely due to
Crispin Wright.2 This is a much more demanding theory for it requires
that, besides having a certain experience, certain collateral factors are in
play, such as further information, which, in order to be epistemically in
good standing, must be justified in its turn. The intuitive motivation for
the conservative view is that these collateral factors seem to be neces-
sary in order for one’s experience to be brought to bear on the intended
class of beliefs – i.e. beliefs about mind-independent, material objects
in our surroundings. Arguably, however, the requirements imposed by
the conservative view for the attainment of perceptual warrant are too
demanding (§2).
Such a discussion paves the way to an alternative, intermediate posi-
tion which, quite naturally, can be dubbed “moderate”, since it avoids
both previous extremes. Accordingly, an empirical perceptual warrant is

seen as depending on more than a mere experience, absent reasons to
doubt that conditions are normal, but on less than having, besides such
an experience, a warrant for some general propositions. In particular, in
the moderate view the mere assumption of these general propositions
is seen as providing the needed informational setting within which
one’s experience can be brought to bear on propositions about material
objects. The moderate position is presented in §3.
Although the moderate label is new, the theory is not, or so I will
argue. In fact, it can be seen at work, at least implicitly, in a number of
theories regarding the structure of empirical warrants that have been
proposed throughout the history of philosophy. The reasons behind
such widespread acceptance are presented and discussed in §4. To borrow
the Wittgensteinian metaphor, the idea is that in order for justifications
for ordinary empirical propositions to obtain, some background presup-
positions must stay fixed, like hinges that have to stay put for the door
to be able to turn.
Finally, this chapter addresses the issue of how we should think of
the content of perceptual experience for it to play a justificatory role,
together with some collateral assumptions, with respect to our ordinary
beliefs about specific empirical objects. When dealing with the content
of perception, two issues must be kept apart. The first is establishing the
metaphysical nature of perception and hence the conditions in which
a creature can be said to perceptually represent its environment. The
second, instead, focuses on the epistemic problem of determining the
requirements that need to be satisfied for a subject not only to have a
given perception, but also to be in a position to avail himself of that
perception in the course of obtaining a warrant for his empirical beliefs.
There is no obvious reason to think that these two issues should be
solved by imposing exactly the same set of conditions on perception.
In this vein, it is argued (in §5) that while non- or a-conceptual crea-
tures can have quite coarse-grained or basic perceptions, only creatures
endowed with a conceptual repertoire can use their perceptions as part
of their justification for specific empirical beliefs.

the liberal view
Nothing seems more intuitive than the idea that just by having an
experience with the representational content of a red table one would
thereby acquire a warrant to believe that there is a red table where one
seems to see it. Of course, we all know that sometimes things can be a
little tricky: one may be in a room with mirrors and the table may actu-
ally be behind one’s back; or the lighting conditions may not be optimal
and the table may look red while in fact it is white and bathed in red
light. Yet, if there is no reason to think that these vagaries obtain, just by
having the relevant experience one would have a warrant to believe that
there is a red table in front of one. This is the gist of the, so-called, liberal
view regarding the architecture of empirical warrant, put forward and
defended in depth by Jim Pryor in several papers in the early 2000s.3
Liberal account of perceptual justification: a belief about specific

material objects that P is perceptually justified iff, absent defeaters,
one has the appropriate course of experience (typically an experience
with representational content that P).4
Now, it has to be pointed out that the notion of justification (or warrant)
under discussion here is that of an evidential, internalist kind. That is
to say, the role of warrant is played by the internal state of a subject –
namely his perception with a certain available content – that is to say,
a conscious representational content that P, of which the subject is
conscious. It is also a state that, for conceptually sophisticated enough
creatures, can be articulated: conceptually endowed subjects could actu-
ally respond to the question “How do you know that there is a red table
in front of you?” by saying “Because I see it”.5
Furthermore, here we are concerned with, so-called, propositional
warrants. Namely, those warrants there are, in the abstract space of
reasons, for a given proposition. Propositional warrants, therefore, exist
independently of whether a subject forms the corresponding belief.
Suppose a subject is absent-minded. He is in front of a red table and
sees it. Yet, he has no time to, or interest in, forming such a belief.
Nevertheless, we can actually say that, since he satisfies the condition
imposed by the liberal view to have a perceptual warrant, the proposi-
tion “There is a red table” is justified, as would be the belief that there is
such a table were he to form it on that basis, precisely by virtue of there
being such a propositional justification. Indeed, were he to entertain the
relevant belief on the basis of his perception, he would be doxastically
justified in holding it. For, so-called, doxastic warrants are those propo-
sitional warrants which attach (or fail to attach) to one’s actually held
beliefs, once the latter are entertained on the basis of the former.
Recent dissatisfaction has been expressed with the ordinary account
of doxastic warrants, which sees them as a subspecies of propositional
ones.6 Let me therefore speak to this general concern first. In my view,

the idea of propositional justification is captured nicely by what happens
in mathematics, where there are warrants for propositions, namely
proofs of the relevant theorems. Furthermore, mathematical proofs are
accessible, at least to some human beings who exercise their knowledge
and cognitive faculties at their best. This is enough to make sense of
the idea that those mathematical propositions can be justified for all
of us and, in particular, for those who do not have the means either
to grasp those justifications or to entertain the relevant beliefs, namely
those containing germane theorems.7 Alternatively, we can think of
the case presented in the previous paragraph, where there is a proposi-
tional warrant for the proposition “There is a red table” if a subject has
an appropriate course of experience (taking for granted, for the sake of
argument, that the liberal account is correct), even if he actually fails to
form the corresponding belief.
Notice that there is a third notion that might be in play but it needs
to be carefully distinguished from those of propositional and doxastic
warrant. Namely, what might be called a rationally available warrant.
Now, suppose our subject is convinced that sometimes he is given a pill
that makes him hallucinate red tables; then suppose he actually sees a
red table and forms the corresponding belief. His belief would be justi-
fied, yet he could not avail himself of that warrant, given his collat-
eral beliefs. We could therefore say that that propositional, and even
doxastic, warrant is rationally unavailable to him or, equivalently, that
it is “rationally obstructed” from him. It has to be stressed that these
collateral beliefs do not actually destroy one’s warrant, because there are
no real defeaters in this case. Rather, we may see them as “hypotheti-
cally undermining” it.8 In sum, warrants are always propositional ones,
yet sometimes they attach to propositions that are also the content of
subjects’ actual beliefs (formed on the relevant bases), and may or may
not be rationally available to them, depending on their further beliefs
and cognitive situations at large.
Given these specifications, it should now be clear why the liberal view is
so appealing and, indeed, so “liberal”. For, it easily makes sense of the fact
that even creatures who are not conceptually sophisticated and are thus
unable to articulate their warrants, could nevertheless be justified, propo-
sitionally, and even doxastically, were they to form the relevant beliefs.
For, to that effect, it would be enough for them, in the normal run of
cases, to have the appropriate kind of conscious perceptual experience.
A word of caution is needed though. Several factors make the liberal
view different from the well known position put forward by Tyler Burge
regarding, what he labels, “perceptual entitlements”.9 According to

Burge, entitlements are warrants a subject may have even if he is not
in a position to articulate them. For instance, by having the relevant
kind of perception, absent defeaters, a child who formed the belief that
there is a red table in front of him, would have an entitlement for it,
even if, ex hypothesi, he could not articulate it (say because he does not
have the concept of seeing yet). So much is common between entitle-
ments and Pryor’s liberal conception of perceptual warrant. However,
there are several important differences. First, the liberal view is not
backed by an externalist story as to why the relevant experiences would
provide subjects with a warrant for their beliefs, in the way Burge’s view
of entitlements does. In particular, it is not part of the explanation of
the nature of these warrants that they satisfy the natural norm that is
characteristic of perceptual representations – roughly, to represent the
environment around a perceiver correctly, at least with respect to the
kind of surroundings within which the perceptual faculty evolved to
produce correct representations of it. Furthermore, the role played by
conscious experience is much more relevant in Pryor’s account than in
Burge’s. While for the former it is a necessary condition for experience
to play a justificatory role, for the latter it is not – or, at the very least, it
is not clear that it is.10 Indeed, Pryor goes so far as to maintain that the
mere phenomenal content of an experience – the specific way in which
things feel to us when we have certain experiences – could be enough
for it play a justificatory role with respect to at least some classes of
belief, such as “I have a headache”.11 Nothing like this can be found in
Burge, who is very clear about the fact that the phenomenal does not
amount to the representational and that only representational percep-
tual content, which (probably) need not even be conscious, could play
an entitling role.12
The reader may wonder which notion of perceptual warrant is to be
preferred. Now, I am skeptical of the existence of natural norms charac-
teristic of perceptual representation. It seems dubious that the notion
of representation carries with it a priori, as Burge has it, a commitment
to its veridicality. In particular, the notion of representation seems
simply to involve the idea of providing a potential layout of the world
around a perceiver, which can be correct or incorrect, depending on
whether it corresponds, or fails to correspond, to that particular envi-
ronment. Yet, were it incorrect, and had it always been incorrect, the
representation would be a representation nonetheless. Nor does the
very notion of representation seem to me to entail that one must be,
or have been, at least in the past, in causal relations with the object it
allegedly represents. There can be representations of unicorns even if

they never existed. Nor does it seem to me to be a natural norm that
perceptual representations are factive and therefore entail the correct-
ness of their content. To hold that much seems a conventional aspect,
perhaps fostered by the use of English, which does not hold in other
languages – such as Italian – or at least not as clearly as in English (but of
course, such a use could be stipulated in either language). Now, without
recourse to natural norms characteristic of perception, Burge’s position
does not seem to be substantially different from forms of reliabilism that
have been extensively criticized, at least as theories of justification, and
from which he wishes to steer away. Furthermore, I am suspicious of a
notion of warrant that is compatible with the possibility that subjects
affected by blind-sight – and more generally by perceptual representa-
tions they are not conscious of – could have an entitlement for their
perceptually based beliefs. For, in that case, what would differentiate a
warranted belief from a mere shot in the dark, at least from a subjective
point of view?
However, in a more conciliatory spirit, one may hold that there are
two notions of justification or warrant along the lines of Burge’s enti-
tlements, one more externalist in kind, which takes care of the truth-
tracking aspect of warrants, and one of an internalist fashion, along
the lines of Pryor’s perceptual warrants (augmented by some further
conditions we discuss in the following), that takes care of the aspects
of warrant related to the responsible formation, assessment, and revi-
sion of one’s beliefs. Such a pluralist view seems enough to motivate
an inquiry, like the present one, within the internalist camp.13 Notice,
lastly, that this pluralist position is not the one put forward by Burge
himself. For, according to him, the divide lies merely in the articulated/
non-articulated nature of perceptual warrant, while the pluralist view
has it that, whether or not perceptual warrants are articulable, they
come in two species. Namely, those which depend on correctly tracking
the truth by means of whatever mechanism allows us to do so, which
need not depend on any conscious element; and those which depend
on the subject’s having a conscious experience which, in favorable
conditions, tracks the truth, independently of whether a subject is able
to articulate it.
In light of these considerations, let me now return to the issue of
the propositional nature of warrants. Despite the fact that psychological
plausibility and cognitive parsimony are a plus of any theory of warrant,
the previous discussion of propositional, doxastic, and rationally avail-
able warrants should help dispel the impression, which still holds captive
many theorists,14 that the issue of the architecture of empirical warrants

could be decided by reference to such considerations. As we have seen,
propositional warrants may (or may not) exist independently of what
subjects think of them (vs. rationally available ones), and also independ-
ently of whether they are actually able to articulate them – provided,
however, such warrants are articulable, at least in principle, by concep-
tually endowed creatures – and even independently of whether subjects
form the corresponding beliefs (vs. doxastic ones). Hence, considerations
of psychological plausibility and cognitive parsimony cut no ice with
respect to the issue of the proper architecture of empirical warrants.15 In
particular, they neither rule out alternative, “heavier” accounts, which
impose further constraints on the attainment of perceptual justification,
such as the ones we will consider in the next sections; nor, pari passu, do
they lend immediate support to the liberal view.16
Let us now consider if there are reasons that directly speak against the
liberal position. Here the discussion will have to be partial, because only
in the following chapter do we consider arguments against Mooreanism –
i.e. the view according to which Moore-like arguments are epistemically
felicitous – which is often coupled with the liberal position.
The main reason for dissatisfaction with the liberal view is that it allows
that experiences could be subjectively the same, irrespective of whether
or not they have been produced by causal interaction with appropriate
physical objects. Hence, in order to have a defeasible warrant for a belief
about specific material objects, something over and above a mere experi-
ence should be in place. One could put the point in terms of “cogni-
tive locality”.17 It seems that as long as we are merely concerned with
perceptual experiences and do not avail ourselves of some externalist
story to provide an account of why they should, at least mostly, put us
in touch with material objects out there, we remain confined within the
realm of experience which, as such, is not sufficient to get us outside of
it, that is, to warrant beliefs about material objects. This does not mean
we should think of experience in terms of sense data, conceived of as
mental entities or as a “veil”, which wraps us up. Rather, the point is
that even if, metaphysically, experiences were individuated by reference
to material objects they allegedly represented, and even if it were the
case that we were lucky enough to have mostly veridical experiences,
there would be no subjectively available reason to hold that they are in
fact, at least mostly, caused by causal interaction with physical objects.
If everything looks internally just the same to one, why would one be
warranted in believing, upon having a hand-like experience, “Here is
my hand” rather than “I am hallucinating a hand”, or “I am dreaming
of having a hand”, and so on? In such a predicament, it would seem

natural to think that if we can surpass our cognitive locality, it would
be because of some extra information that – as implicit as it may be –
allows us to consider our perceptual experiences as bearing on physical
objects. Of course, even once this collateral information is in place, it
is not guaranteed that specific perceptual episodes are veridical. Yet, we
would be within our rights in taking experiences generally to represent –
no matter how defeasibly that might be on specific occasions – entities
which go beyond experience itself, such as mind-independent material
objects.
Hence, in the internalist framework, which is the background of
the liberal position, there is a genuine issue of having no means to
transcend our cognitive locality. Furthermore, any solution we might
propose would depend on smuggling in an assumption of veridicality
either in the general background conditions – in the metaphysics, as
it were – or in the domain of the information potentially accessible to
a subject – that is in his informational setting. The first option would
be externalist in spirit; the second would be more internalist-friendly,
but it would represent a substantial departure from the liberal view
anyway. Indeed, as we shall see, it characterizes both conservative and
moderate reactions to it, notwithstanding their respective differences.
Another way of putting the same point would be to say that, within
the present internalist framework, perceptual justification cannot be
basic, that is to say, unaided by collateral factors and merely a deliver-
ance of the perceptual system.18 Again, the reason why this is the case is
that perceptions by themselves do not allow us, from an epistemically
internalist standpoint, warrantedly to transcend them, as they would
always be subjectively indistinguishable no matter how they have been
brought about.
One further source of potential difficulty for the liberal view stems
from the idea that perception immediately justifies the corresponding
beliefs by courtesy of the absence of defeaters. Now, how should we
think of the latter? As we saw, they cannot be just internal factors,
such as countervailing beliefs, which may well be very misguided. For,
according to Pryor, those beliefs would place perceptual warrants ration-
ally out of reach, but would not make the beliefs based on one’s percep-
tions any less justified. If, in contrast, defeaters are external factors, they
may well be beyond a subject’s ken. Yet, the “defeasible-warrants-only”
reply we encountered before would then flounder. Reflect: if the defeaters
are states of affairs beyond a subject’s ken, affecting either his environ-
ment or his cognitive powers (or both), in the envisaged circumstances a
subject would not have any perceptual warrant for the belief that there is
a red table in front of him, no matter how defeasible that warrant might
be. In contrast, if defeaters were external factors, which are somehow
given to a subject, the psychological plausibility and the cognitive parsi-
mony of the liberal view would be greatly diminished. For a subject
should now be able, in order to have a warrant for his perceptually
based belief, to conceptualize – at least to some extent – the absence of
defeaters. Certainly, the theory could not be ruled out, for these consid-
erations, as we saw, do not directly bear on the epistemic aspects of the
proposal and therefore on its prospects. Yet, they would make it less
palatable than it first seemed.
Finally, a consequence of working within an internalist framework
is that warrants must be articulable, at least in principle, provided
subjects have the necessary conceptual repertoire. This is indeed an
important aspect of the theory and one that makes it superior to
its externalist counterparts at least when two, different, “cognitive
projects” are at stake. One of them, as we saw before, is the project of
policing one’s own beliefs and therefore being able to check whether
they are epistemically supported in such a way as to maintain them,
or else withhold from them or even revise them, in an epistemically
responsible way. That is to say, by being oneself in control of these
operations so as to be accountable for them. Another project, which
we will consider at much greater length in the next chapters, is that
of being in the same ballpark as a skeptic vis-à-vis the issue of being
in a position to know, or at least warrantedly believe, that things are
as they actually seem to us. More specifically, we need a shared notion
of warrant (or justification) whereby it makes sense to ask whether
one knows that one has it (or is at least warranted in believing that
one does).
Now, such a notion of warrant is, I think, legitimate, whether or not
it is the only right one. It is legitimate from a descriptive point of view,
because it is regularly at work in our practice of giving and asking for
justifications for our beliefs. Moreover, it is the kind of notion Pryor
makes use of within his liberal account. Hence, let us see how it fares
with respect to the kind of second-order issues just rehearsed. Now,
suppose that, whether or not one has any skeptical preoccupation in
mind, the issue were raised as to whether one’s experience of a red table
is veridical. The kind of reply that the liberal view would condone would
be “Yes, that is how my sense experience represents things as being”. Yet,
the whole point of the initial question is precisely to give expression to
the worry that one’s experience would be identical whether or not it is
appropriately caused by the interaction with the red table. Just repeating
that one’s experience looks that way would be neither here nor there.19
If, in contrast, a liberal suspended judgment over these second-order
issues, this would show that his internalist notion of warrant is inca-
pable of fulfilling one of its alleged purposes. Namely, that of providing
a subject with a reason to claim that things are as they look to him.
Hence, a liberal could either refuse to answer these questions, or else
come up with a further explanation. The former option would not be
intellectually satisfying, while the latter is of course viable. But, as we
shall see, it is not in keeping with the story that Pryor actually tells in
defense of Moore-like arguments against external world skepticism, nor
is it consistent with the explanation he gives of, so-called, easy knowl-
edge (we will consider them in the next chapter).
Nevertheless, I think these considerations are enough to give us pause
and a reason to explore possible alternatives to the liberal view. Let me
stress that I am not aiming to provide knockdown arguments against
the liberal view, or against any other position for that matter. The real
project is to explore the positive suggestion that the architecture of
empirical warrants be explained along moderate lines. To such an end,
it is sufficient to show that the main alternatives are problematical and
that the moderate position can avoid the problems that beset its compet-
itors. Thus, as already pointed out in the Introduction, the overall meth-
odological standpoint endorsed in this work is that different coherent
philosophical accounts of the structure of perceptual justification are
possible, with their various consequences for a large number of issues,
such as the analysis of Moore-style arguments, the validity of princi-
ples such as closure of warrant under known entailment, the problem
of external world skepticism, and the nature of epistemic rationality.
They have to be assessed, though, on the basis of a careful examination
of their pros and cons, in order to reach a form of reflective equilibrium.
My hope is to show, throughout this work, how moderatism can do
better, and certainly no worse, than its rivals on each of these issues and
thereby provide reasons to endorse it.
2 The architecture of perceptual warrants 2: the

conservative view
The take-home message of the previous discussion of the liberal position

is not that an experience cannot ever justify one in believing its content,
for example, that there is a red a table in front of one. Rather, the bottom
line is that, if it does, it does so only within a larger informational setting,
comprising, in particular, the proposition that there is an external world.

This way, perceptual justifications – consisting of a given experience and
of these collateral assumptions – allow us to overcome our cognitive
locality.
This is indeed part of the gist of Crispin Wright’s, so-called, conserva-
tive position with respect to the structure of empirical warrants.20 What
Wright adds to this picture (or considers to be entailed by it), is that for
these assumptions to be epistemically in good standing, they need to be
independently warranted. Wright can be seen as the most resourceful
contemporary supporter of such a view and, for this reason, the discus-
sion to follow will focus on him.21
As is well known, Wright presents his template in a very general fashion,
as holding of arguments such as ZEBRA and MOORE. Here they are:
ZEBRA
(I) Here is a zebra

(II) If there is a zebra here, then it is not a cleverly disguised mule
(III) This is not a cleverly disguised mule
MOORE
(I) Here is my hand

(II) If there is a hand here, there is an external world
(III) There is an external world
Take ZEBRA. The basic point is that just by having a zebra-like experi-
ence one is not immediately in a position to justify the belief “Here
is a zebra”. Something more is needed, in particular the information
contained in the conclusion – that that animal is not a cleverly disguised
mule. For, to repeat, that experience by itself would be compatible with
its being produced by a cleverly disguised mule. Hence, such collateral
information must be in place, at least implicitly, in order for us to be
allowed to form, in an epistemically responsible way, the belief that that
is a zebra. However, such an assumption, in order to be epistemically
kosher in its turn, needs to be independently justified. Hence, here is
the definition of the
Conservative account of perceptual justification: a belief about

specific material objects that P is perceptually justified iff, absent
defeaters, one has the appropriate course of experience (typically an
experience with content that P) and it is warrantedly assumed that
there is an external world (and possibly other general propositions,

such as “My sense organs work mostly reliably”, “I am not the victim
of massive cognitive deception”, etc).
Three things are worth mentioning at this stage. Firstly, Wright, like
Pryor, is concerned with propositional warrants. Hence, once again,
considerations regarding psychological plausibility and cognitive parsi-
mony are not relevant in order to assess the view under discussion. In
particular, the subject himself is not required to entertain the collateral
assumptions that allow his experience to constitute a warrant for (I);
even less to be able to provide a warrant for them. The important point
is that such a warrant be available in the abstract space of reasons and be
accessible, at least in principle, by us (perhaps only by epistemologists).
Secondly, the warranted collateral information on the basis of which a
subject’s experience can constitute a perceptual warrant for (I) is needed,
yet not in order for one to have an indefeasible warrant for (I). Suppose
a DNA test has been made on the animal in the pen, which has revealed
that it is not a mule. Still, it has not yet confirmed whether it is a zebra
or any other animal, perhaps an unknown species living in some remote
part of Africa. The perceptual warrant one would have for (I), thanks to
one’s experience and this warranted collateral assumption, would still
be defeasible. The point is that that warranted assumption, together
with one’s on-going experience, makes it more likely that the animal in
the pen is a zebra, than just the mere zebra-like experience. However, it
is still possible that that animal is not a zebra.
Finally, and connectedly, in ZEBRA it is clear that an independent
warrant for the conclusion can be obtained. A warrant, that is, which
is not itself dependent on the kind of procedure – i.e. outer observa-
tion – by means of which, in ZEBRA, one would try and warrant the
conclusion that that is not a cleverly disguised mule. That the warrant
for the conclusion cannot be provided by ZEBRA itself is obvious once it
is acknowledged that, in order for its first premise to be warranted at all,
warrant for its conclusion would be needed already. Hence, ZEBRA cannot
provide one with a first warrant for its conclusion, precisely because
such a warrant would already be needed in order to have a warrant for
premise (I) in the first place.
Let us now consider MOORE. Again, the basic point is that for one’s
hand-like experience to constitute a warrant for “Here’s my hand” it
is not enough merely to have that experience. A collateral assumption
is needed to the effect that one is entitled to bring that experience to
bear on a realm of mind-independent objects. Hence, once again, the
information contained in the conclusion is needed in order to have

warrant for the first premise. As before, however, such an assumption is
made in an epistemically proper way just in case it can be independently
warranted in its turn.
Again, being concerned with propositional warrants, the view does
not require a subject to be able actually to entertain the content of (III),
that there is an external world. Nor does it require him to be able to
provide a warrant for it. Rather, it suffices that there be such a warrant
in the abstract space of reasons and that it be accessible, at least in prin-
ciple, by us (if only by epistemologists).
Once more, such a warranted assumption is not needed in order to
have an indefeasible warrant for (I), but only to be allowed to tran-
scend one’s cognitive locality and be epistemically within one’s rights
in bringing one’s hand-like experience to bear on a realm of mind-in-
dependent objects. However, it could still be the case that that is not
a human hand, after all, but a papier-mâché one. Still, it would be a
physical object.
Now, the problem arises with the requirement that such an assump-
tion be independently justified. That is to say, that there be a warrant
for it which does not stem from the kind of procedure by means of
which, in MOORE, one would try to warrant the proposition that there
is an external world, i.e. outer observation. Again, the reason why one
could not get such a warrant from MOORE is simply that, as we have just
seen, in order to have warrant for (I) in the first place, one already needs
to have warrant for (III), according to Wright. Hence, MOORE cannot
provide one with a first warrant to believe its conclusion. More gener-
ally, arguments such as ZEBRA and MOORE, present a form of epistemic
circularity, in Wright’s view. For they purport to produce a warrant for
their conclusion when in fact such a warrant is already needed in order
to have a warrant for their premises. As Wright puts it, these arguments
fail to transmit warrant from their premises to their conclusion. For,
even if the former are warranted after all, it is only because the latter
are already warranted in their turn. Perhaps a less contentious (or less
confusing) way of putting the point is simply to say that, due to their
epistemically circular nature, these arguments cannot produce a first
warrant to believe their conclusion. For such a warrant would already
be needed in order for their premises to be perceptually warranted in
their turn.22
Here, however, lies the disanalogy between ZEBRA and MOORE. They are
both epistemically circular, but while for ZEBRA it is obvious that there can
be independent ways of warranting its conclusion, it is not at all clear that
the same holds for MOORE. For no argument ultimately based on percep-
tual warrants would do, since it would display the kind of circularity that
besets MOORE. Yet, it is difficult to see how a priori arguments based on
reflection on the very notion of an external world could provide a warrant
to believe that it exists. In very general terms, they would resemble those
attempts at proving God’s existence by reflecting on the very concept of
God.23 Surely, there can be more refined ways of trying to provide an a
priori justification for the existence of an external world,24 but it is dubious
that they would be successful. One genre of such arguments holds that it
is a necessary condition in order to be able to represent specific objects,
either in perception (as we saw in passing in our discussion of Burge), or
in thought (consider Putnam’s “Brains in a vat”), that they have existed
and that we have – perhaps only phylogenetically – been in causal contact
with them. Now, whether or not it may be a correct condition to impose
in order for our minds to be able to represent a given content, either
perceptually or in thought, it is dubious that it would provide a warrant for
holding that physical objects exist now, as we form the allegedly percep-
tually warranted belief that there is a hand in front of us.25 Yet, we seem
to need such a warrant for our present beliefs to be actually perceptually
justified. The idea that we might be within our epistemic rights to believe,
now, that there is a hand here, because our best philosophical theories tell
us that we must have encountered physical objects in the past, is of very
little comfort. Firstly, because the warrant we would allegedly have would
be hostage to the success of these very theories. Secondly, because along
these lines the present attempt to transcend our cognitive locality could
very well be thwarted.26
Are there any other ways in which we may try to obtain a warrant for
“There is an external world”, beside MOORE-like arguments and a priori
ones? As connoisseurs of the literature will know, Wright has argued that
a (Humean) skeptic thinks there are not and for this very reason ends
up being a skeptic about the existence of an external world. Yet, Wright
thinks this skeptical outcome is due to culpable ignorance of a third
possibility. Namely, that there could be an “entitlement” for “There is an
external world”. An entitlement, in his view,27 is a non-evidential kind
of warrant. This means that it is neither perceptual, nor testimonial,
nor a priori. It is a propositional warrant that it has always been there
in the abstract space of reasons and that Wright has redeemed for all of
us. That is to say, it is a kind of internalist warrant that Wright has been
able to access and to place at our disposal by means of his philosophical
inquiry. Yet, and this is the real crux of the matter, it is not a warrant that
speaks to the likely truth of the proposition it is supposed to warrant.
We consider Wright’s entitlements more extensively in the next

chapter. For now, suffice it to say that a warrant which does not speak to
the likely truth of “There is an external world”, simply is not in the skep-
tic’s ballpark. Precisely because a skeptic asks us to provide a reason to
think it is more likely true than not that there are physical objects, so as
to allow us to be within our epistemic rights in taking our current sense
experience to bear on a world of mind-independent entities. Whatever
the merits of Wright’s entitlements, they are not supposed to give us
that much. Therefore, I conclude that a skeptic is right to maintain that,
sadly, there are no warrants for those collateral assumptions that are
needed for us to have perceptual warrants for our ordinary empirical
beliefs, according to the conservative account of the architecture of
perceptual warrants.
Now, does the preceding entail the skeptical conclusion that, after
all, we have no perceptual warrants for our specific empirical beliefs
either? What is to be noticed is that such a conclusion only follows on
a conservative account of the structure of empirical warrants. For it is
only in such a view that warrant for “There is an external world” would
be needed to have a perceptual warrant for “Here is my hand”, given
one’s current sense experience. Yet, notice that between the Scylla of
the liberal architecture of empirical warrants, which didn’t allow us to
transcend our cognitive locality, and the Charybdis of the conservative
structure of empirical warrants, with its skeptical consequences, lies the
alternative of the moderate account of the nature of empirical warrants,
to which we will now turn.

the moderate view
The take-home message of the previous section is that it is very hard to

see how “There is an external world” could be warranted and yet, given
the discussion in §1, it seems that without that collateral information,
our sense experience could not, just by itself, allow us to transcend our
cognitive locality. The key idea of the moderate position is that such
an assumption needs to be in place for us to be within our epistemic
rights in taking our current sense experience to bear on a world of mind-
independent entities; yet, to forsake the requirement that it need be
warranted. Hence, the moderate version of the architecture of empirical
warrants has it that for one to have a perceptual warrant for, say, “Here is
a red table”, one needs to have a red table experience and the assumption
must be in place that there is an external world (and possibly that our
sense organs work mostly reliably and that we are appropriately cogni-
tively alert and related to the world so as to exclude skeptical scenarios,
such as dreaming hypotheses, or evil demons, and BIVs). Given these
collateral assumptions, we can legitimately transcend our cognitive
locality and take experiences at face value to justify the corresponding
empirical beliefs. More precisely:
Moderate account of perceptual justification: a belief about specific

one has the appropriate course of experience (typically an experi-
ence with content that P) and it is assumed that there is an external
world (and possibly other general propositions, such as “My sense
organs work mostly reliably”, “I am not the victim of massive cogni-
tive deception”).
A few points of clarification are in order. Moderates, like liberals and

conservatives, are concerned with propositional warrants. Hence, it is not
required that subjects are actually able to entertain the proposition that
there is an external world, even less that, were they able to entertain it,
they had to do it explicitly any time they go about obtaining a perceptual
justification for their ordinary empirical beliefs. The basic idea, therefore,
is that the information that there is an external world figures as one of
the constitutive ingredients of perceptual warrants in the abstract space
of reasons, as it were. Yet, it is important, in order to remain within the
internalist camp, that such a proposition be graspable and articulable,
at least in principle, by subjects endowed with the relevant conceptual
repertoire. But, of course, there is also a sense in which those subjects
who do have the relevant conceptual repertoire would be able actually
to entertain it and take it as a datum, which they can, on occasion, make
explicit so as to form or claim perceptual warrants, thanks to concomi-
tant appropriate sense experiences.
Therefore, there are in fact two, equally legitimate, senses in which we
can say that the assumption28 that there is an external world is one of
the constitutive ingredients of our perceptual warrants. The first sense
may be called propositional and means simply that the proposition that
there is an external world does figure as an ingredient of perceptual
warrants in the abstract space of reasons. Add to that that it must be
graspable and articulable, at least in principle, for the ensuing warrant
to be of an internalist fashion. The second, call it the doxastic, sense has
it, instead, that such a proposition should actually be entertained by
subjects endowed with the relevant conceptual repertoire and be part of
what they would offer were they requested to make explicit their own
warrants for holding that there is, say, a red table in front of them.29
This seems to preclude the possibility that, at least in this second sense,
young children and the unsophisticated could have a perceptual justifi-
cation for their beliefs about specific material objects.
Now, as for the previous positions, I do not think this would actu-
ally speak against the moderate conception of perceptual warrant. For
the merits and faults of this position, like those of the liberal and the
conservative ones, have to be assessed with respect to propositional
warrants. Yet, it would certainly be a plus if moderatism could speak
to the problem of the justification of perceptual beliefs held by young
children and conceptually unsophisticated creatures, which, to repeat,
must not be confused with the claim that the theory entails they would
lack propositional warrants for their beliefs.
Now, it seems to me that even if a child were unable to entertain the
proposition that there is an external world, and would thus be unable to
offer it as part of his grounds for his perceptually based beliefs, one could
grant him that assumption, provided he were able to take part in a prac-
tice which has that very assumption as its rational precondition. Hence,
suppose the child said things, such as, “The red table is in the kitchen”,
while he is not there seeing it, or “Someone moved the red table from the
kitchen” while he is in the kitchen and realizes that the red table is no
longer there. We could then say that, at least implicitly, he considers the
table as a mind-independent object and brings his experiences to bear
on to such a mind-independent entity. Hence, even if he does not have
the concept of an external world, as such, and is in no position to make
explicit his assumption to that effect, he should be granted a conception
of physical objects as mind-independent entities, if only implicitly.
Hence, the assumption that there is an external world, or at least that
there are mind-independent objects, can also be granted, in the doxastic
sense, to children and unsophisticated creatures, as long as they meet
the previously mentioned requirements. Furthermore, it can certainly
be granted to those subjects who, while having the necessary concep-
tual repertoire, do not explicitly consider it each time they form a belief
based on the deliverance of their perceptual experiences, so long as they
themselves meet the same requirements that are imposed on children
and the unsophisticated.30 Here is a summary of the distinction intro-
duced thus far.
To assume that Q propositionally: in the abstract space of warrants, a

propositional warrant for P depends on having an appropriate course
of experience (typically an experience with content that P) and on

the proposition Q (that there is an external world, and so on);
To assume that Q doxastically: in order to have a rationally avail-
able warrant for one’s belief that P a subject S ought to have an appro-
priate course of experience (typically an experience with content
that P) and to be able to grasp and articulate the assumption that Q
(that there is an external world, and so on) or else act and judge in
conformity with it.
A second issue worth considering at this stage is what it means to assume

that there is an external world in this doxastic sense. In particular, we
have to distinguish three possible types of doxastic “assuming”. Firstly,
there is a kind of assuming that merely entails acting as if P were true.
This would be the sense in which one may entertain an assumption in
thought and see what follows from it, without any special commitment
to its truth,31 or even while thinking (or knowing) that it is in fact false.32
Secondly, there is a kind of assuming that involves a commitment to the
truth of what figures as its content, but it is consistent with the fact that
it might actually be false.33 Finally, there is a species of assuming that
simply posits the truth of P and draws out the consequences. I think
externalist theories ultimately do that, since they actually say, “given
that it is in fact the case that there is an external world, with which we
causally interact, thus-and-thus follows”. Hence, they assume the exist-
ence of the external world and start building their respective theories
from that.
To assume1 that P: to act or judge as if P were true, even if one knows

that P is false.
To assume2 that P: to be committed in one’s acts and judgments to
the truth of P, even if P itself, unbeknownst to one, might be false.
To assume3 that P: to posit that P is true, even beyond the possibility
of recognition.
Now, I think the second and the third options are both consistent with
the moderate position, while the first is not, because it is merely suppo-
sitional. However, my own preferences go to the second one. Hence,
the mode of assumption characteristic of the moderate position as I am
characterizing it, has it that the truth of “There is an external world” is
not dogmatically posited, even if it is a tenet of the theory that in our
actions and thoughts we are committed to it.
Then there is a further issue, which has to do with the kind of truth
we predicate of “There is an external world”, when it is propositionally
or doxastically assumed. My preferences go to an anti-realist conception
of truth. That is to say, “There is an external world” isn’t considered as
tracking a mind-independent reality, which would be such even if we
had no warrant to believe it. Nor is it taken to be true because all the
evidence we have speaks in its favor. For all that evidence depends on
already taking for granted that there is an external world. As we shall see
in Chapter 4 (§6), “There is an external world” is true in a different, more
minimal sense, which is consistent with the idea that that proposition
is true because it is held fast by us, given our Weltbild, and with the view
that, for such a reason, no evidence can in the end speak against it.
One might think that this way of characterizing the moderate position
makes it immediately unsuitable for meeting any skeptical challenge. For it
seems that a skeptic is precisely inviting us to provide a warrant to believe
that “There is an external world” tracks a mind-independent fact. I think
this conclusion would be too hasty, but I have to beg the reader to bear
with me, until we actually address the skeptical challenge in Chapter 4.
Let me just table one idea, though. Skepticism comes in different varieties,
but, first and foremost, it is an epistemic challenge that has to do with
how we can be warranted in believing certain things, or, more generally,
with how we can be rational in doing so.34 Even someone with anti-realist
metaphysical proclivities can confront that skeptical challenge. For they
would see it as the challenge of redeeming the rationality of making the
assumption that there is an external world, whether or not “There is an
external world” is taken to track a mind-independent fact, or its truth is
thought of in a more minimal sense. Notice, moreover, that the kind of
realist metaphysics that underpins externalist positions would not really
be better off vis-à-vis the skeptical challenge. For, as we just saw, that meta-
physical thesis is, after all, a dogmatic assumption. Now, a skeptic could
either challenge his opponent to demonstrate its truth, without conceding
it to him, not even for the sake of argument; or else, he could concede that
as a matter of fact there is an external world, but then ask the opponent to
rationally redeem such a claim. In the former case, the anti-realist and the
realist would have to face exactly the same challenge; whereas in the latter
case the externalist seems better off, but, as our discussion of bootstrap-
ping arguments will reveal,35 this is more a prima facie impression rather
than a solid starting point. Furthermore, it is important to keep in mind
that a realist notion of truth is a skeptic’s most powerful ally. For it will
give him the resources to make play with the idea that, despite everything
looking to us to be a certain way, things might be otherwise.
Let us recap the main features of the moderate position as I have

been characterizing it so far. According to such a view, our perceptual
warrants depend on a certain course of experience, absent defeaters,
together with some very general assumptions (that there is an external
world, that we are not victims of lucid and sustained dreams, that our
sense organs are mostly reliable). These assumptions have to be under-
stood, in the first instance, as propositions that figure as constitutive
ingredients of our perceptual warrants in the abstract space of reasons,
together with appropriate kinds of experiences and absent certain
defeaters. This way, contrary to the liberal view, we can actually surpass
our cognitive locality – that is, we can bring our experiences to bear on a
realm of mind-independent entities. This is, of course, compatible with
the fact that our specific warrants are defeasible and that we may be
mistaken about the identity, the properties, and even the existence on a
given occasion of a particular object. Since no warrants for these general
assumptions are required, the fact that it is difficult to see how there
could be any does not make the warrants based on them impossible to
obtain, contrary to what would happen in the conservative view.
Furthermore, the moderate position, as I have been characterizing it,
has a story to tell about what it means doxastically to assume that there
is an external world. It consists in being committed in our thoughts and
actions to the existence of an external world. Moreover, to assume doxas-
tically that Q does not entail that subjects have to explicitly entertain its
content, or even actually be able to do it. It is enough that they comply
with a practice that has as its rational precondition the commitment to
the existence of an external world. Of course, this is entirely compatible
with the fact that those subjects who do have the necessary conceptual
apparatus ought, on occasion and if requested, to offer their grounds
for their perceptually based specific empirical beliefs and to be able to
mention such an assumption.
In sum, the moderate position holds that we do have perceptual
warrants for our ordinary empirical beliefs, but these arise only within
a system of assumptions, or hinges, which, while not being in turn
warranted, or even warrantable, make it possible for us to transcend
our cognitive locality and therefore to bring our experiences to bear
on a world populated by mind-independent entities. Obviously, the
most serious challenge the moderate position has to face is the one
posed by a kind of skepticism of Humean descent that challenges the
claim that such an assumption is after all rational and non-arbitrary,
as no warrant for it can be provided. We will take up that challenge in
Chapter 4 (§2).
Lastly, let me address one final worry. Namely, the so-called leaching
problem.36 The trouble is this: if all we need is the assumption “There is an
external world”, as opposed to a corresponding warranted belief, in order
to have a perceptual warrant for “Here is a hand”, upon having a hand-
like experience, we will merely have a warrant for that belief conditional
upon that assumption, but we will not have any real epistemic warrant for
it. In response, it is important to stress that the moderate view is, prima-
rily, a constitutive thesis about the very nature of perceptual warrants. As
we saw, moderatism tells you what it takes to have a perceptual warrant;
namely a certain course of experience together with an assumption about
the existence of an external world. Once those ontologically constitutive
ingredients are in place, a genuine epistemic warrant for propositions,
such as “Here is a hand”, ensues. Hence, moderatism does not merely
tell you what the necessary (and perhaps jointly sufficient) conditions
for having perceptual warrant are; nor does it amount to the view that
perceptual warrant is simply conditional on an assumption. Rather, it tells
you what ingredients constitute a genuine, non-conditional, perceptual
warrant. Thus, it is all to be expected that the final product – a perceptual
warrant – be something over and above its constitutive components – an
experience and a general assumption – just like the color pink is some-
thing over and above its constitutive elements – red and white. It is then
a separate issue, which we shall examine in Chapter 3, whether such a
warrant can transmit to propositions entailed by the ones for which one
has such a justification. However, for now, let us consider another issue,
namely, the width of the moderate position.
4 Moderates are legion (while liberals and

conservatives are not)
The use of the term moderate to indicate a certain view about the struc-
ture of perceptual warrant is new, as well as the appearance of this theory
within current debate on the structure of perceptual warrants. However,
if we look at the history of philosophy moderates are legion, while
liberals and conservatives are not.37 For instance, moderatism is the
outcome of certain forms of naturalism, which, in a Humean fashion,
hold that our ordinary empirical beliefs are justified within a system of
assumptions that it is natural for us to make, which, however, are not in
turn justified. Following Strawson (1985), we could think of this form
of naturalism either as due to our psychological constitution, again in a
Humean fashion, or else, in a more Wittgensteinian spirit, as due to our
upbringing within a certain community. In the former option, it would
be part of our cognitive endowment that we take it for granted that

there is an external world and, on that basis, use our sense experience
to justify ordinary beliefs about material objects in our surroundings.
In the latter option, in contrast, it is because we have been brought up
within a community that behaves in conformity with that assumption
that we cannot help making it.
A point in common to both forms of naturalism is the fact that neither
holds that these assumptions are warrantable or need be. Usually, their
response to the skeptical challenge consists in pointing out that a skeptic
asks for reasons and grounds where there are none or there cannot be
any. This reply, however, is not entirely satisfactory, unless it is shown
that we are, nevertheless, epistemically within our rights in assuming
them and that we are not just making arbitrary assumptions;38 or, at
least, that we are not making assumptions which other forms of life
could not make while still being epistemically rational. For this reason, in
Chapter 4 (§2), I will not develop the moderate position in a naturalist
fashion. Yet, the moderate architecture of perceptual warrant can also be
seen as operative within such a naturalist framework.
Similarly, moderatism is a tenet of certain pragmatist positions,
according to which we have pragmatic warrants for the assumption that
there is an external world. In particular, such an assumption is neces-
sary for us to maintain our valuable and useful practice of acquiring,
assessing, and withdrawing from ordinary empirical beliefs in light of
our experiences. Yet, on these pragmatist readings, we would have no
epistemic warrant for such an assumption. That is to say, we would have
no warrant that speaks to its likely truth. Hence, from this pragmatist
viewpoint, it must be the case that our perceptual warrants are due not
only to appropriate courses of experience but also to this, and possibly
other, collateral assumptions which, as such, neither are nor can be epis-
temically warranted.39
Like the forms of naturalism we mentioned at the outset of this
section, the kind of pragmatism here considered is unsatisfactory when
taken in connection with skepticism, for it actually surrenders to it by
acknowledging that we do not have epistemic warrants for these basic
assumptions and that we actually fail to be epistemically rational in
holding them, even though we may be perfectly practically rational in
doing so. For this reason, I will not develop moderatism in a pragmatic
fashion. Yet, the moderate architecture of perceptual warrant can also be
seen as operative in that kind of framework.
Arguably, Wittgenstein in On Certainty, even on a reading that does
not follow Strawson in making him a sui generis kind of naturalist, also
turns out to be a moderate.40 For, according to him, our hinges – namely,

our most deep-seated certainties – are not grounded in reasons, yet they
must stay fixed for us to be able to acquire our ordinary warrants. More
specifically, they cannot be epistemically justified for any alleged warrant
for them would actually presuppose them. Yet, on such a reading of On
Certainty, hinges play a normative role. In particular, they play the role
of norms of evidential significance, for they determine what can count
as evidence for what. Assuming that “There is an external world” was a
hinge for Wittgenstein, it would then turn out that its assumption, with
no warrants in its favor, would be necessary to have ordinary perceptual
warrants, based on sensory experience, for ordinary empirical beliefs
about physical objects. Hence, Wittgenstein too would be a moderate.41
For, even if the relevant assumptions were considered as norms and not
as empirical propositions, which are semantically assessable, they would
still have to be in place to generate ordinary empirical warrants for one’s
beliefs about material objects.
This view has, I think, a better chance than the ones seen so far in
providing some elements that could be useful in order to confront the
skeptical challenge. For, as we have seen, by being necessary to the produc-
tion of ordinary warrants, there could not be any evidence against them.
Furthermore, being norms, it would actually be a categorical mistake to
call them into doubt. For sensible doubts can be raised only with respect
to empirical propositions and on the basis of reasons, which would neces-
sarily be lacking in the case of hinges. Yet, the idea that hinges are norms
is problematical. For norms are not accountable to a pre-existing reality,
while Wittgenstein’s hinges seem, on the face of it, to be susceptible of
semantic evaluation. “Nobody has ever been on the Moon” is a hinge for
Wittgenstein and many would be inclined to say that although it was true
when On Certainty was written, it is false now. I think that a Wittgensteinian
would have ways of responding to this problem. For instance, he could
exploit the idea that there are doppelgangers, that is, sentences that look the
same while they play entirely different roles. Therefore, “Nobody has ever
been on the Moon” at the time when Wittgenstein was writing expressed
a hinge and hence a rule, which could not be semantically assessed. Yet its
doppelganger, namely “Nobody has ever been on the Moon” as we may
utter or write it now, looks exactly like the other sentence, yet it expresses
an empirical proposition, which is actually false.42 Or else, one may
retain a more nuanced idea according to which it is one thing to consider
the content of a sentence and another to determine what role it plays
in context. From this viewpoint, “Nobody has ever been on the Moon”
would have a descriptive content, and would thus describe a possible
state of affairs like any ordinary proposition. Yet, in context, it could be

removed from doubt and actually play a rule-like role. By varying contex-
tual factors, it could cease to play such a role and be re-immersed in the
flux of ordinary empirical propositions that are semantically assessed, and
could actually turn out to be false.43 Yet, it is clear that these strategies are
interesting ways to make sense of Wittgenstein’s position, but it is much
less clear whether they are viable accounts of hinges as such. In particular,
there seems to be no obvious need to maintain that hinges and, specifi-
cally, “There is an external world”, are norms. More precisely, given the
difficulty of actually making sense of this idea, and the possibility of doing
without it, I think it is best to avoid any commitment to their normative
nature. Hence, up to the present point and in the following, I have devel-
oped and will explore further a moderate position that is not committed
to that view. Still, the Wittgensteinian strategy in On Certainty would be
compatible with the basic tenet of moderatism. Namely, that empirical
justifications take place within a system of assumptions which are not, in
turn, justifiable. It is indeed for this reason that Wittgenstein’s remark in
OC 105 – “All testing, all confirmation and disconfirmation of a hypoth-
esis takes place already within a system [of assumptions]. And this system
(...) belongs to the essence of what we call an argument (...)” – would be a
suitable ex ergo of this study, and that the position developed in this work
is called a hinge epistemology.
Finally, as I have already alluded to in the previous section, external-
ists too might also be moderate and hold that, given that it is in fact the
case that there is an external world, our experiences can be brought to
bear on specific mind-independent entities. Now, this version of moder-
atism is not satisfactory to my mind for various reasons, including that
it looks dogmatic vis-à-vis the skeptical challenge. A skeptic could even
grant that there is an external world, yet ask the very simple question
“How do we know there is?”.44 In such a predicament an externalist
seems obliged either to repeat himself and say “Because there is”, or else
engage in bootstrapping arguments which do not look very convincing
(see Chapter 3, §7 for a discussion). For these reasons, the kind of
moderate position I will be concerned with in this book will not be
developed along externalist lines. Yet, it remains that externalists may
join the moderate camp, if they so wish.
This survey has no aspirations to being comprehensive or historically
accurate, for there can be different readings of those philosophers I have
mentioned and because other authors in the history of philosophy,
whom I have ignored here, may have views in keeping with the moderate
architecture of empirical warrants. Yet, let me point out why it is only to
be expected that moderates are legion while liberals and conservatives
are not. The reason is simple: the moderate view concerning the struc-
ture of empirical warrants removes the uncommittal attitude towards
“There is an external world” held by liberals, which precludes the possi-
bility of overcoming our cognitive locality. Yet, it is not as demanding as
the conservative view, which requires warrants for “There is an external
world” that seem unattainable, thus leading to the unpleasant result
that we would have no perceptual warrant for our ordinary empirical
beliefs either. Hence, small surprise that, after all, moderates are and
have been thicker on the philosophical ground than the present-stage
debate between liberals and conservatives makes it seem.
5 Moderatism and the content of perception
So far, we have been talking about the role of collateral assumptions

within the structure of perceptually based warrants. However, we have
not said anything about perception itself and what conditions it must
fulfil in order to play a justificatory role. I think that an account is
owed on that score too, within the project of explaining how percep-
tual warrants are constituted. This section will deal with that issue. Let
me point out from the start, however, that the moderate architecture
of empirical warrant could be retained even if one disagreed with the
way in which I account for the constraints perceptions have to meet
in order to play a justificatory role, once the relevant assumptions
are propositionally in play. The reason why I am addressing the issue
of perceptual content is that, in my opinion, it is important to try
and offer a comprehensive account of perceptual justification which
also deals with the issue of the nature of the content of perception,
usually considered to pertain only to philosophy of mind. As we shall
see, that issue does have important connections with epistemology,
even though it will not affect the problem of the role of background
presuppositions for the obtaining of perceptual justifications.
A good starting point for tackling the issue is to briefly consider John
McDowell’s rendition, in Mind and World,45 of Wilfrid Sellars’ attack
against the “Myth of the Given”.46 Roughly, the idea is that if we think
of experience as devoid of any structure and therefore of a representa-
tional content with correctness conditions, like sense data theorists do,
or like those who think that experiences, as such, have only a phenom-
enal content, then our perceptual experiences could not justify those
empirical beliefs which are based on them. For how could an experi-
ence devoid of any representational structure support the truth of a
specific empirical proposition, such as “Here is a red table”, which is
a structured entity that represents the world correctly or incorrectly?
Moreover, such an experience would be compatible with a multiplicity

of possible conceptually structured propositions, not all of which would
be relevant to the specific belief which we would like to justify by means
of it. For instance, it would be compatible not only with “There is a red
rectangular-shaped object here”, but also with “There is a white rectan-
gular-shaped object bathed in red light”, or with “There is a hologram
of a red rectangular-shaped object”, or even with “There are two juxta-
posed triangular red objects here”, and so on.
If “the Given” is an unstructured datum then it cannot play a justi-
ficatory role with respect to our beliefs. However, if one endorses the
idea that only beliefs can justify other beliefs, in a coherentist fashion,47
because they are the only entities that possess the right kind of structure
to play such a role, then we incur other problems. First, if beliefs are never
justified by experience, although they are caused by it, then none of our
beliefs will ever be justified by how things are in the world; contrary to
our deep-seated intuition that at least when perceptually based beliefs
are at stake this would be the case. Rather, beliefs will always be corrobo-
rated on the basis of other beliefs, as long as they comply with some
coherentist principles. Yet, maximally coherent systems can be very
different one from the other. How should we choose between them?
Which one would give us the (largely) correct account of reality?
The second problem raised by a coherentist response to the myth of
the given is the one that perhaps is dearest to McDowell. Namely, the
Kantian issue of how our concepts can have empirical content if they are
merely caused by experience. For, if the experience has merely a causal
role in the formation and application of our concepts, then nothing
precludes the subsequent inverted spectrum possibility from obtaining.
There could be people that, in front of red objects, have a green experi-
ence, which they conceptualize as red, while others, in front of the very
same objects, would have a red experience, which they also conceptu-
alize as red. If we grant that both have the concept RED,48 it follows that
that concept is not individuated, at least in part, by how things appear
to one. For the different experience in the two cases would not lead to
two different concepts. Hence, concepts would not have an empirical
content after all. Indeed, they would turn out to be just abstract rules
that a cognitive system could implement even if it didn’t have any color
experience, as long as it were able to categorize all and only red objects
on the basis of whatever kind of information the system possessed,
which may well have nothing to do with how that color looked to us.
In order to solve both these problems, as is well known, McDowell
puts forward the view that the content of perception is conceptual. This
means that in experiencing the world by means of our senses we passively

deploy our conceptual abilities. Hence, on the one hand, our concepts
are endowed – one would like to say replete – with empirical content.
On the other hand, perceptual content is structured; indeed it has prop-
ositional structure, and can thus serve as a justification for the beliefs
formed by taking it at face value. Moreover, according to McDowell,
in perception we take in facts – the world itself – in an already concep-
tualized way. Accordingly, we are in touch directly with the world, yet
we take it in in such a way that it is immediately suitable to justify our
empirical beliefs, because it is already structured and conceptualized just
like those propositions we go on to believe if there are no reasons to
think that our sense organs might have somehow deceived us.49
Here I will not take issue with the idea that in perception we take
in facts and with McDowell’s ensuing disjunctivism, to the effect that
whenever our experience is not veridical, it is not an instance of the
psychological type “perception”. I think this view is extremely problem-
atic. Yet a discussion of it would take us too far afield.50 Let me just say
that, for experiences to be appropriately structured so as to constitute a
warrant for specific empirical beliefs (by courtesy of some very general
assumptions, according to moderatism, whether or not a subject is able
to entertain them), there is no need to think of them as ways of taking
in facts. It is enough that they be structured so as to match the proposi-
tions which are believed on their basis. The epistemological question
to be addressed is: can they have such a structure only if they involve
the passive exercise of concepts? Before considering it in more detail,
however, let me say a few words about another issue raised by McDowell’s
position, namely, the metaphysical problem of whether only creatures
endowed with the relevant conceptual repertoire can have a perceptual
representation of the world around them.
As is well known, McDowell has been fiercely criticized for holding a
view which, on the face of it, seems to imply that only human adults
could perceive their surroundings. Indeed, this seems to run contrary
to both common sense and scientific inquiry, which concur in granting
perceptions to infants and even animals who, arguably, lack the concep-
tual repertoire that would be needed, on McDowell’s account of the
content of perception, to have those very perceptual representations.51
Furthermore, he has been criticized on the score that his theory would
provide circular possession conditions for observational concepts,
because one would need to have the concept RED already in order to
perceive something as red. This would pose insurmountable problems
for those who, like McDowell, think that our concepts are acquired
and are not innate. For how could one offer a psychologically plausible
account of how humans can acquire the concept RED, say, if subjects
must already possess it in order to perceive red objects as red and thus
have the kind of perceptual experience on which their acquisition of
that concept should, allegedly, be based?52
Another canonical objection is the one relative to the finesse of the
grain of perception that exceeds our color vocabulary, for instance. As I
have maintained elsewhere,53 although it is true that our color percep-
tion is more fine-grained than our color lexicon, we can form demon-
strative concepts, such as THAT SHADE OF RED, which, as short-lived as they
might be, since we may be unable to recognize that shade on future
occasions, may well be operative while perceiving the shade in ques-
tion and while discriminating it from others in one’s perceptual scene.
Yet, in my view, this is no comfort to the metaphysical thesis regarding
perception that McDowell is committed to, namely, that only creatures
with the relevant conceptual repertoire can have the corresponding
kinds of perceptions. For what the argument regarding the finesse of
grain shows, when supported by the considerations just advanced, is
that we can have fine-grained concepts that match the content of our
color experience. However, it does not establish that only creatures with
those concepts can have those experiences. Hence, I think it is useful to
distinguish two theses:
Dependence thesis: a subject can have a perception with the repre-

sentational content p if and only if he has the concepts that are
necessary to characterize it in a canonical way.
Immediate conceptualization thesis: for those who possess the rele-
vant concepts, an experience with content p is immediately concep-
tualized – that is, independently of judgment – by means of the
concepts that are necessary to characterize its content in a canonical
way.
The argument from the fine-grained content of experience is silent on

the first thesis. Hence, it is powerless to adjudicate the issue between
conceptual and non-conceptual theorists, which hinges precisely on
that point. Similarly, the argument that conceptualists have often
invoked in defense of their position – namely the one from the justi-
fication of empirical beliefs, according to which justificatory relations
hold only between conceptual contents, and therefore perceptual expe-
riences can justify empirical beliefs, as such, only if they have a concep-
tual content – is equally silent on the first thesis. For the second thesis
would be enough to account for the conceptualist intuition that only

conceptualized experiences could justify the corresponding beliefs.
We can therefore conclude that there are no arguments in favor of the
first thesis, while there are actually two against it. Namely, the argument
from the perception of infants and animals and the one from circularity,
especially in its “acquisition” version. Thus, I think the answer to the
metaphysical question whether only creatures endowed with a concep-
tual repertoire can perceive the world around them must be answered
in the negative.
We can now return to the epistemological problem, encapsulated in
the argument from the justification of empirical beliefs, of whether only
conceptualized experiences can play a justificatory role with respect to
the corresponding beliefs. Again, as is well known, two theorists espe-
cially – Christopher Peacocke and Tyler Burge – have put forward the
view that although perceptions are not conceptual, they have repre-
sentational contents and therefore correctness conditions. Moreover,
they have both maintained that perceptions can, unaided by judgment,
provide a warrant for the corresponding empirical beliefs.
Their characterization of the very content of experience is partially
different, though. They both think it is a kind of map of the environ-
ment around the perceiver. However, Peacocke introduces the notion
of “scenario” – a sort of relatively coarse-grained way of filling out the
space around the perceiver, which can be correct or incorrect – and that
of “protoproposition” to account for those kinds of cases where the
scenario by itself does not discriminate between different ways of filling
in the perceptual scene.54 Burge, in contrast, does not make use of these
notions. However, he also thinks that perceptual content is a kind of
map that fills in the space around the subject and has singular and
general elements. That is to say, it singles out specific objects (or shapes,
or colors, etc.) and attributes properties to them (being to the right of, to
the left of, above or below something else). Now, the number of percep-
tual attributives, for Burge, is limited and consists of spatial, temporal,
numerical ones, of attributives that indicate food and, possibly, conspe-
cifics and mates.
However, the problem remains of how a perception that is not concep-
tualized and has no propositional structure could justify an empirical
belief. Obviously, it cannot do it if by “warrant” one means a justifica-
tion, which must be constitutively articulable by a subject. For articu-
lability requires conceptualization in judgment. Hence, the proposal
would end up embracing a sophisticated form of coherentism, where
experiences have non-conceptual and non-propositional structures, yet
are given that kind of structure in judgment, which, in its turn, enters
the justification of further beliefs and judgments.55
It is perhaps for this reason that Peacocke and Burge countenance a
different notion of warrant, called “entitlement” (not to be confused
with Wright’s notion of entitlement),56 which would allow experiences
as such – that is, independently of being conceptualized and of having
propositional structure – to play a genuinely justificatory role (once
certain further conditions are met).
According to Burge, as we have already remarked, entitlements do
not require subjects to have access to them, so they need not even be
conscious experiences, nor do they require subjects to be able to justify
why they are warrants. Rather, it is enough that the perceptual system
works properly and therefore satisfies the natural norm that character-
izes it. Roughly, that norm requires that the system produces correct
representations of the environment, at least in the kind of surround-
ings in which the system has evolved, in such a way as to be capable of
representing them correctly. Hence, when placed in a different kind of
environment, the system would still produce an entitlement, although
it would not yield a correct representation. Burge’s entitlements are
therefore clearly defeasible. Yet, since they are not enough to produce a
warrant for general beliefs such as “There is an external world”, they do
not give rise to the unpleasant consequence of providing an entitlement
for such a belief when the system is placed in a kind of skeptical envi-
ronment. Finally, in order to have an entitlement for a given belief, the
transition from the perception to the belief based on it must preserve
the former’s content, while there are no defeaters.
This last requirement is actually a little problematical. For if defeaters
are conceived of internistically, as reasons to think that one’s percep-
tion and further cognitive operations may be unreliable, then subjects
should be able to conceptualize them and even to conduct inquiries to
make sure whether they obtain. This, however, would prevent young
children and the unsophisticated from having perceptual entitlements
for their beliefs. Yet, this would spoil the very notion of entitlement
that, in this context, is meant precisely to overcome the unpalatable
result that ordinary notions of justification would have with respect to
the epistemic status of young children’s beliefs. Burge is aware of this
problem and suggests thinking of defeaters “externistically”, as unfavo-
rable conditions which need not lie within a subject’s ken.57 However, it
now seems that by being placed in uncongenial environments subjects
would, after all, lose their entitlements for their perceptual beliefs,
because these unfavorable conditions would indeed obtain. Burge’s
notion of entitlement, therefore, risks being very close to traditional

reliabilist accounts of justification. Furthermore, as we saw before (§1),
the very idea is dubious that there is a natural norm characteristic of
perception which should hold a priori, and, if one does without it, it
becomes even clearer that entitlements do not represent a real improve-
ment over reliabilism.
Let us now turn to Peacocke’s position. Some general tenets are similar
to Burge’s. In particular, that a subject need not know that he has a
perceptual entitlement in order to have it and even less how it manages
to give him a warrant for his beliefs. The differences concern both some
details and some general background assumptions. Let us start with the
former. Firstly, for Peacocke the transitions from perceptual contents to
the corresponding beliefs are “relatively a priori”, in particular when
beliefs do not go beyond the concepts whose possession conditions are
exhausted by the capacity to exercise them when one has the corre-
sponding experiences. An example of such a transition would be the
one from a red perceptual content to the belief “This is red”. Secondly,
Peacocke offers a sort of evolutionistic explanation as to why such tran-
sitions would lead to mostly true beliefs. For his idea is that perceptual
systems have evolved so as to yield correct representations of subjects’
surroundings. Therefore, if we take our experiences at face value we
will form mostly true beliefs, as long as the latter require the exercise of
merely observational concepts.58
Turning now to general background assumptions, it has to be noticed
that Peacocke’s position is crucially different from Burge’s insofar as, for
him, rational transitions are those that are such from a subject’s own
point of view. Hence, in making the transition from a red perceptual
content to the corresponding belief, a subject must know what it means
to be red, be aware that he has sufficient reasons to move from the expe-
rience to the belief, and do so because he is aware of those reasons.59 A
further element, which helps shed light on the differences between the
two proposals, is that for Peacocke an experience must be conscious in
order to give a subject an entitlement for the corresponding belief. As
we saw before (§1), in contrast, Burge’s position seems to be compatible
with the claim that even a subject affected by blind-sight could have an
entitlement for the belief based on it. Hence, Peacocke is much more
of an epistemic internalist than Burge. Indeed, one might even wonder
whether he is actually making use of a notion of warrant that is crucially
different from the usual, internalist one. Certainly, there is no require-
ment that the entitlement be articulable, yet a subject should know, at
least implicitly, a great number of things with respect to the conditions
in which warrants are supposed to obtain. Now, are we really sure young
children and the unsophisticated do? The standard move is to make
all this knowledge implicit. However, this strategy is problematical,60 at
least when the conditions a subject should meet in order to have it are
not clearly specified.
Nevertheless, let us now consider an objection that I think can be
raised against both accounts and that will allow us, hopefully, to put
forward a proposal about the content of perceptual experience in order
for it to play a justificatory role. We could put the point by asking: how
many genuinely perceptual entitlements would we have? For, according
to Burge, perceptual attributives are very limited and, for Peacocke,
purely observational concepts are few. Hence, perceptual entitlements
are either too many – many more than those we would, intuitively, wish
to allow for – or too few.61 Let us consider a subject who has a perception
with the content of a small spherical red shape. This content is compat-
ible with a great number of beliefs, such as “Here is a red sphere”, “Here
is a red apple”, “Here is a red ball”, “Here is a red spherical candle”, and
so on. So, if that perception warrants each of those beliefs, we would
end up having perceptual entitlements not only for the belief we have
actually formed on that basis, that there is a red apple, but also for many
more, some of which would be incompatible with the one we have
formed. If, in contrast, and as Burge’s and Peacocke’s writings seem to
suggest, that perception gives one an entitlement to believe only “Here
is a red spherical shape”, then perception as such would provide us with
entitlements for very few beliefs. Even the humdrum belief that there
is a red apple in front of us would have to be justified by the interplay
between our perceptual and conceptual systems. As before, there are only
two ways in which this can happen: either by mobilizing beliefs, or else
by letting concepts shape our very perceptions, as McDowell has been
maintaining. The former option would actually incur a sophisticated
form of coherentism, as we saw. The latter would not, but it would repre-
sent an important concession to McDowell’s position. I would actually
recommend opting for this second horn of the dilemma. Interestingly,
however, this would show that the debate between conceptual and non-
conceptual theorists is much less sharp and clear-cut than is usually
thought of. For the two positions need to interact in fruitful ways if
we wish to answer both the metaphysical problem of perception – and
so allow that many more creatures beside human adults could actually
enjoy fully representational perceptual states – and the epistemological
problem of how experiences, as such, can justify our beliefs based on
them. The kind of solution I am proposing is a kind of divide et impera. For
non-conceptualists are right with respect to the metaphysical problem,

while conceptualists are right regarding the epistemological one.
Before analyzing this position in more detail let me review yet another
proposal, recently made by Susanna Siegel.62 According to Siegel, in
perception itself, without the aid of concepts, we are presented with
“rich” contents. She writes: “When you see a bowl of fruit, you can
(usually) recognize the kinds of fruit in the bowl” (p. 78) and, according
to her, such a recognitional ability is purely based on the content of
one’s perceptions unaided by concepts. Notice that for Siegel such recog-
nitional abilities may well be based on the representation in visual expe-
rience of natural kind properties (cf. p. 114). Now, I think she is right
to hold that we can learn to recognize kinds of fruit, of tree, and so
on. I also agree with Siegel that these recognitional abilities may alter
the phenomenology of one’s visual experiences. To that effect, Siegel
proposes a nice discussion of how knowledge of Cyrillic can change the
phenomenological aspects of one’s visual experience while looking at
that kind of alphabetical character. Finally, I agree with her that some
recognitional abilities regarding, say, moms, dads, types of food and
one’s immediate and familiar surroundings,63 may take place without
possession of the relevant concepts. What, however, I find difficult to
accept is that recognitional abilities about natural kind properties do
not involve the exercise, however passive, of a conceptual repertoire.
That is to say, it may be that without the aid of the concept APPLE a
child can recognize the fruit in the bowl as the same one he saw the day
before and it may obviously be the case that he is able to distinguish
it visually from pears and oranges. However, this ability has nothing
to do with the ability to visually recognize the property of being an
apple taken as a natural kind one. To be able to do that seems clearly to
require the concept of natural kind, such that one could be sensitive
to the fact that two identical looking objects may not be instances of
the same natural kind (e.g. water and twater, jade), and to the fact that
two different looking objects can be instances of the same natural kind
(e.g. water and ice). Precisely because natural kind properties are inde-
pendent of the way things look to us, I take it that it is not possible to
perceive natural kind properties as such. Therefore, I agree that percep-
tions have a representational content that is not exhausted merely by
shapes and colors and that represents objects (and people) as having
specific, or at least characteristic contours, colors, odors, textures, tastes,
sounds and voices. I also agree that, based on these experiences, we can
learn to recognize and distinguish trees and apples from other kinds of
object. However, I disagree that in experience, unaided by concepts, we
represent more than that and, in particular, I do not agree that we repre-
sent natural kind properties.
One may then wonder whether the kind of recognitional abilities
here under discussion would not suffice for possession of concepts,
as Ruth Millikan (1998) claims. In that case, it would be clear that
concepts would already figure in perception. Yet, even granting
that these recognitional abilities are not based on the exercise of
concepts – or do not themselves constitute the possession of the
relevant concepts – it is quite clear that the proposal would not take
us too far with respect to the solution of the underdetermination
problem. Take an environment in which there are apples of the deli-
cious variety, candles with the same shape and color, and decorations
in the house resembling apples. The visual experiences would all be
relevantly similar. Thus, once more, experiences unaided by concepts
would not be sufficiently fine-grained to justify “Here is an apple”,
any more than “Here is an apple-looking candle”, contrary to what
the entitlement theory maintains. Hence, it seems to me that if we
want experiences to justify – no matter how defeasibly that might
be – the relevant belief, concepts have to be in play somehow. That
way, the experience could be taken at face value to form the corre-
sponding belief, absent defeaters. By contrast, without the exercise of
concepts already in experience, what belief would that very experi-
ence be a guide to? They would all be on a par. To repeat, the point
of requiring concepts to enter experiences is not to make them inde-
feasible warrants for the corresponding beliefs for, after all, it may
be the case that the apple-looking candle has been put in the bowl,
unbeknownst to the subject who is accustomed to his environment
and knows that usually apples are in the kitchen and apple-looking
candles are not. Rather, the point is to allow experiences to be guides
to relevant beliefs, which may then be true or false. Without the
exercise of concepts, they are not sufficiently fine-grained to guide
a subject, raised in that kind of environment, to any of the possible
beliefs licensed by those very experiences.
Thus, I submit that we have to grant that creatures without concepts
can also have genuine perceptual representations of their surround-
ings (and their bodies). Yet, in order for these perceptual experiences
to enter the structure of empirical warrants, they must involve the
passive exercise of concepts. Furthermore, the underdetermination
problem reviewed before, teaches us that these concepts had better be
as fine-grained as the ones that enter the corresponding beliefs and
so exceed merely observational ones.64 Otherwise, we would end up
having either too many or too few perceptual justifications. Hence,

basically, the perceptual content of creatures endowed with concepts is
partially different from that of those who do not possess them. While
the latter simply have a kind of map of their environment with some
specific perceptual attributives, the former have a perceptual content
that is refined and restructured in such a way that its elements fill in
a conceptual, propositional structure. That is to say, while non-con-
ceptual creatures merely enjoy a sort of “pictorial” representation of
a red sphere in front of them, creatures endowed with the relevant
concepts see that there is a red apple in front of them, where this propo-
sitional and conceptual structure is immediately presented to them in
the very perception itself, and has all the empirical content perception
provides and which mere entertainment of that proposition in thought
lacks. Finally, the passive exercise of concepts in conscious experience
complies with the internalist requirement that perceptual warrants
be, at least partly, internal states of a subject, which are given to him,
although this does not require the articulability on his part of his own
warrants. For there is no need for a subject to be able to conceptualize
his perceptual experience as a seeing that things are thus. It is enough
that he actually sees that there is a red sphere in front of him. Yet, such a
warrant is indeed articulable, at least in principle, by those who happen
to have the relevant conceptual repertoire. Hence, it has all the charac-
teristic features of a respectable internalist justification.
Finally, let me briefly go back to the issue of the epistemic status of
the beliefs of young children and the unsophisticated. It is part of the
picture I have been advocating that non-conceptual creatures cannot
have perceptual justifications, while they can have perceptions. Yet, it
has to be noticed that they could not have beliefs either. For beliefs
require concepts, which, ex hypothesi, they lack. Thus, if we are dealing
with early-age infants or non-conceptual animals, this would not be a
problem. However, on reflection, the view is also not problematic when
taken in connection with older children. For they will have as many
perceptual warrants as they are capable of forming, given their concep-
tual resources, which, crucially, according to the present proposal, match
the number of beliefs they are actually able to entertain.65 To sum up
and conclude: perceptions as such, that is unaided by judgments, yet by
courtesy of some general background assumptions, which need not be
conceptualized by a subject, can justify empirical beliefs based on them.
Yet, this can be the case only for those creatures who have a concep-
tual apparatus and are therefore in a position to exercise it passively in
perception.
6 Summary
To summarize and conclude, in this chapter we have presented the

moderate position regarding the structure of perceptual justification.
Accordingly,
Moderate account of perceptual justification: a belief about specific

with content that P) and it is assumed that there is an external world
(and possibly other general propositions, e.g. “My sense organs work
mostly reliably”, “I am not the victim of massive cognitive decep-
tion”, and so on).
The moderate account has the means to surpass our cognitive locality.
A problem that, I have argued, besets liberal positions. According to
them,
Liberal account of perceptual justification: a belief about specific

with content that P).
But, given the internalist camp to which the liberal view belongs, it is
unclear how merely having a certain course of experience, absent reasons
to doubt that there is an external world, may suffice to entitle subjects
to form justified beliefs about specific material objects. For experiences
would be subjectively identical even if they were normally produced in
unfavorable circumstances. Hence, unless it is implicitly assumed that
things are otherwise, it is unclear why one would have any more justi-
fication for “Here is a hand” over “I am hallucinating having a hand”,
or “I am dreaming of having a hand”, and so on. To stress, the problem
of cognitive locality ought not to be conflated with the issue of the
indefeasibility of perceptual justification. Background assumptions are
needed to bring experiences to bear on beliefs whose content concerns
physical objects in one’s environment. This is entirely compatible with
the fact that there may be new evidence to defeat one’s justification for
a specific belief of this kind.
The moderate view solves the problem of our cognitive locality
without requiring that the general assumptions that need to be in
place, beside the appropriate course of experience, to have the relevant
warrants, should be warranted in their turn. This is hardly a surmount-

able problem when it comes to propositions like “There is an external
world”. For this reason, the moderate view is superior to the conserva-
tive one, according to which,
Conservative account of perceptual justification: a belief about

specific material objects that P is perceptually justified iff, absent
defeaters, one has the appropriate course of experience (typically an
experience with content that P) and it is warrantedly assumed that
there is an external world (and possibly other general propositions,
e.g. “My sense organs work mostly reliably”, “I am not the victim of
massive cognitive deception”, and so on).
I have clarified the specific meaning I attach to the idea that certain
propositions should be assumed, in order to have perceptual warrants,
while insisting that moderatism is a family of views, of which I am
endorsing one possible variant. On my understanding of the notion
of “assuming”, there is a propositional and a doxastic sense of that
term. For the former, we should think of the proposition assumed as
one ingredient that, together with appropriate experiences, gives rise to
perceptual justification in the abstract space of reasons. For the latter, we
can grant an assumption, that is a commitment to the truth of “There
is an external world”, to all those creatures who, while perhaps unable
to conceptualize that proposition, manifest commitment to it in their
behavior. For they act and speak in ways which only make sense by
taking them to be shaped by a commitment to the existence of mind-
independent objects. Still, conceptually endowed creatures ought to be
able to articulate such an assumption, in order to be credited with the
ability to assume it doxastically. The specific version of the moderate
view I endorse, can then be defined as follows:
Moderate account of perceptual justificationAC: a belief about specific

material objects that P is perceptually justified iff, absent defeaters, one
has the appropriate course of experience (typically an experience with
content that P) and one is committed in one’s acts and judgements to
the existence of an external world (and possibly other general propo-
sitions, e.g. “My sense organs work mostly reliably”, “I am not the
victim of massive cognitive deception”, and so on).
I have also shown why the moderate view can be seen at work in the writ-
ings of several important philosophers, both modern and contemporary.
I have claimed that this should not be very surprising given the benefits
attached to the moderate position, that is the possibility of overcoming
our cognitive locality while avoiding the too demanding conditions
posed by the conservative view, which, not by chance, is often called
skeptic, as it is likely to set the standards for the acquisition of percep-
tual warrants so high that they cannot be met, thus opening the way to
skepticism.
I have then proposed one particular conception of the content of
perceptual experiences, which I think allows us to explain how we may
actually have perceptual warrants. I have argued that while it is possible
for creatures without the relevant concepts to enjoy certain perceptual
experiences, the latter have to be conceptualized in order to give rise to
justifications for specific empirical beliefs. Let me stress in closing that
this is not part of the moderate position. For that view is non-committal
with respect to the kind of content perceptual experiences must have in
order to play a justificatory role vis-à-vis our specific empirical beliefs.
However, it is a useful and important complement of the moderate view
that fills in its details in one possible and plausible way. Yet, the fate of
the moderate position does not depend on actually endorsing precisely
this view of the content of perceptual experiences.
2
Further Motivation for Moderatism
In this chapter, I offer further motivation to embrace the moderate posi-

tion by considering in more detail some criticisms that can be leveled
against both the liberal and the conservative view. I start by examining
some additional worries against the liberal position when taken in
connection with arguments such as Moore’s proof of an external world
(§1). Matching the liberal position with Mooreanism – that is, the view
that Moore’s argument is cogent and that it can produce a first justifi-
cation to believe its conclusion – would seem to be a promising anti-
skeptical strategy for a liberal to pursue, and indeed a very natural one
given his views about the structure of perceptual justification. However,
as we shall see, it runs into trouble. I then turn to Wright’s proposed way
out of the skeptical outcome of the conservative position by appealing
to the notion of entitlement, which I find wanting (§2) for several inter-
connected reasons. Finally, I critically consider some recent proposals,
aimed at buttressing the idea that there are a priori independent warrants
for the conclusion of arguments, such as Moore’s proof of an external
world (§3), while, in some cases, attempting a combination with the
liberal view. In particular, I consider the positions put forward by Ralph
Wedgwood (§3.1), Christopher Peacocke (§3.2) and, very briefly, one
recently presented by Ernie Sosa (§3.3).
The present discussion, together with the one in the previous chapter,
collectively make – I think – a good case in favor of the moderate posi-
tion. For, on the one hand, the moderate view avoids the problems that
beset the liberal position and, on the other hand, it forsakes the require-
ment that there be an independent warrant, whether genuinely a priori
or non-evidential such as an entitlement, which appears very difficult
to come by. Hence, it offers the best prospect for vindicating the idea
that we do have, after all, the perceptual warrants for ordinary empirical
57
propositions we think we have. The discussion in these first two chapters

will, therefore, offer sufficient motivation for the moderate conception
of the structure of empirical warrants, whose consequences, for Moore-
style arguments, external world skepticism, and epistemic relativism,
will be considered in the following two chapters.
1 Against liberal Mooreanism
Since its first appearance, the liberal view has been subject to scrutiny
and many theorists working in this area have found fault with it. In
this section, I will not go over all the criticisms which have been raised
against it. Rather, I will concentrate on those few that, however, seem to
me to go to the heart of the proposal. They are all concerned with the
consequences of the liberal position once taken in connection with argu-
ments, such as Moore’s, for the existence of an external world and other
similar ones. For this very reason, theorists who do not want to embrace
these consequences, and yet are rather persuaded by the liberal architec-
ture of empirical warrant, have proposed to divorce the liberal view from
its Moorean consequences.1 I will have something to say against this
move later on in this chapter (§3). For the time being, let me consider
the objections to what we may call liberal Mooreanism, which is in fact
Pryor’s position. Coupling liberalism and Mooreanism, as we shall see,
is indeed a natural move to make, given a liberal’s understanding of the
structure of perceptual justification. To repeat, on that view, one’s hand-
like experience, absent reasons for doubt, would be enough to give one a
defeasible warrant for “Here is one hand”, which would then transmit to
the conclusion of Moore’s argument that “There is an external world”,
which would then turn out to be a perfectly cogent argument. Of course,
this seems an odd result, but, according to Pryor, it is due to the fact
that we usually take Moore’s argument to be directed at a skeptic who
already doubts that there is an external world. That is to say, a skeptic
thinks that it is likely to be false that there is one. Given such a collateral
doubt, he will not consider (I) to be warranted in the first place. For such
a reason the argument will fail to persuade Moore’s opponent. Indeed,
Moore’s argument would be dialectically ineffective, yet perfectly all right
from an epistemic point of view.2
The first argument against liberal Mooreanism we will consider should
be familiar from the writings of Roger White, but it has been echoed in
works by Wright, Nicholas Silins, and Ralph Wedgwood.3 In my very
early writings on the topic, I had intuitions which went in the very same
direction.4 As we have repeatedly seen, liberalism is the view according
Further Motivation for Moderatism 59
to which it is enough to have a perceptual warrant for an ordinary

empirical proposition that p – for example, “Here is my hand” – merely
to have a hand-like experience, while lacking reasons to doubt that there
is an external world. Now, one basic worry about liberalism is that Pryor
himself concedes that that very experience would be compatible with
its being produced in non-standard conditions. However, its occurrence,
from the liberal point of view, should warrant “Here is a hand” and
thereby, equally and at once, disprove a proposition incompatible with
p, which, however, entails the same evidence, such as p* “I am dreaming
of there being a hand here”. Intuitively, however, in such a case one’s
experience, just by itself, would be neither here nor there. That is to say,
it would justify neither the former nor the latter proposition. Indeed, if
one were allowed to model present intuitions along Bayesian lines,5 the
occurrence of that experience would actually increase the probability of
p*.6 If it appears otherwise – that is, if it seems that a hand-like experi-
ence does not justify p* – it is only because there is already an at least
implicit presupposition that one’s experience is produced in favorable
circumstances, namely, through the interaction with a material world,
by means of the reliable operation of one’s sense organs, while awake.
Hence, even if the probability of p* goes up, it should be noticed that, as
far as we are concerned, it does not become greater than that of p.
One could put the same point by considering the situation of an
open-minded subject.7 Namely, of someone who had never considered
before whether he has hands or may be just dreaming of them, and
who opened his eyes for the very first time and had what in all respects
looks like a veridical hand-like experience. Yet, according to the theory
under scrutiny, which is based on an internalist notion of warrant, he
could have that very same experience even if he were indeed dreaming
of a hand in front of him. Intuitively, in such a predicament, it would
seem odd that that experience, just by itself, could give him a warrant
to discard the uncongenial scenario and, consequently, a warrant in
support of “Here is my hand”. Another way of putting the same point
could be this: if one’s hand-like experience increases the likelihood both
of p and p*, how could it by itself justify p over p*?
Now, theorists interested in raising the Bayesian version of this objec-
tion have pointed out that it would create problems only for a variant
of Moore’s argument that had a specific anti-skeptical conclusion, whose
falsity, however, would entail one’s present experience. That is to say,
for an argument like (I) “Here’s my hand”, based on one’s current sense
experience; (II*) “If there is a hand here, I am not a BIV hallucinating
having a hand”; therefore (III*) “I am not a BIV hallucinating having a
hand”. This clearly is not Moore’s original argument which went from
(I) to (III) “There is an external world”, via the relevant entailment. The
reason why the Bayesian version of the argument is considered to be
neutral is that the falsity of (III), in what we might call “real MOORE”,
would not predict that one could have a hand experience, but, rather,
that one didn’t have any experience, since one would not exist.
However, this reasoning is not compelling. Indeed, the non-existence
of an external world would not impair the possibility that there are selves,
thought of as something like Cartesian egos capable of enjoying hand-
like experiences. Nevertheless, confronted with the previous qualifica-
tion one may react in two ways. One could say – as, for instance, Nico
Silins does – that the Bayesian argument does not show what is wrong
with “real MOORE”. An explanation of what is wrong with it could turn
out to be favorable either to Pryor’s or Wright’s position. That would
depend on whether one explained the problem with “real MOORE” either
by appealing to the notion of dialectical ineffectiveness or to the notion
of transmission failure understood along Wrightian lines, which entails
the correctness of the conservative position. If that were the reaction, it
would weaken the present objection, which would in fact turn out to be
neutral in the actual debate between Pryor and Wright.
Alternatively, one might try to generalize the argument by noticing
that “real MOORE” entails the truth and warrantedness of “There is an
external world with a (or my) hand in it (which I am seeing)”, since that
would just be due to the conjunction of the premise and the conclu-
sion which, ex hypothesi, are both true and warranted. Now, the simple
thought would be that no warrant for the general presupposition that
there is an external world could be provided by an argument such as
Moore’s. For, given the previous considerations about one’s hand-like
experience, it seems arbitrary to suppose that it would justify “There is
an external world with a hand in it (that I am seeing)”, which is what
Moore’s proof would deliver, rather than, say, “There is an evil demon
deceiving me about having a hand, which in fact I am just dreaming”
or “I am a BIV presently hallucinating a hand”. Generalized in this way,
the objection would apply to “real MOORE” and would actually engage
with the Pryor/Wright debate by showing why Mooreanism – that is,
the view that Moore’s argument could produce a warrant to believe its
conclusion – is defective.
Nevertheless, it seems to me that an explanation of why it seems
that a hand-like experience would increase the probability of “Here is
a hand”, rather than “I am dreaming of a hand”, or even “I am a BIV
hallucinating a hand”, is still owed and is not provided by someone

like Silins (and Wedgwood, as we shall see in §3.1) who wants to hold
onto the liberal architecture of empirical warrants. This is the problem
of surpassing our cognitive locality we dealt with in the last chapter. To
repeat, if one’s experiences could be just the same no matter how they
are produced, why should they justify beliefs about material objects
rather than their skeptical counterparts? It seems entirely arbitrary to
take them to favor the former rather than the latter. This, to stress,
is not to be conflated with the problem of the defeasibility of one’s
perceptual warrants for specific empirical beliefs. We are not looking
for justifications that would resist any increment in one’s information.
We are, rather, looking for conditions whose satisfaction would allow
us to take a mind-dependent kind of evidence to bear on beliefs whose
content is eminently mind-independent. Otherwise, we would never
have a justification for the latter kind of belief, neither defeasible nor
indefeasible.
By contrast, both moderates and conservatives would have the resources
to explain our bias in favor of “Here is a hand”, rather than “I am a BIV
with a hand-like experience”. They could say that if it seems to us that a
hand-like experience favors “Here’s a hand” over its skeptical counterpart,
it is because we are, most probably implicitly, assuming we are in the safe
scenario where we are causally interacting with material objects and are
not just merely dreaming of them. Notice, however, that to acknowledge
this much does not, ipso facto, commit one to holding that these presup-
positions need or should (already) be warranted in their turn. It may
just be that the degree of confidence we have in these assumptions, for
whatever reason – even inculturation – be such as to guarantee that no
matter how much the subjective probability of p* went up, it would not
become greater than that of p. Hence, the previous considerations do not
immediately speak in favor of (some version of) the conservative view.
Therefore, if it appears to us that a hand-like experience would be enough
to give us a defeasible warrant for the corresponding belief only because
we are already assuming that there is an external world with which we
are causally interacting, the view that better accounts for the structure
of empirical warrants is one which removes the non-committal attitude
towards “There is an external world” held by liberalism. Moderatism is
indeed such a view, as we saw at length in the previous chapter (§3). Of
course conservatism has exactly the same effect, but it then faces other
problems, which we have already partly seen in Chapter 1 (§2) and will
further explore in the following (§2).
Another variation on this very theme is as follows. Consider the

argument
RED WALL
(I) Here is a red wall (based on one’s current sense experience)

(II) If this is a red wall, it is not a white wall bathed in red light
(III) This is not a white wall bathed in red light
Now, according to the liberal conception of the structure of perceptual

warrant, RED WALL is cogent, for no warrant for (III) is needed in order to
have warrant for (I), as long as one has no reason to doubt (III). Hence,
on that account, RED WALL would be perfectly fine from an epistemic
point of view. This, however, sounds very weird.
As is familiar, Pryor agrees that RED WALL sounds odd, but only because –
perhaps implicitly – we place that argument within a dialectical setting
whereby we imagine that there is an opponent who doubts that (III) is
the case.8 That is to say, the opponent would think it likely that after
all one may well be seeing a white wall bathed in red light. Given that
collateral doubt, such an opponent will not take (I) as perceptually justi-
fied on the basis of his current sense experience. Therefore, the argu-
ment will obviously be dialectically ineffective against that opponent,
but this is not enough to turn it into an epistemically circular argument.
To stress, the argument would be epistemically all right, but dialecti-
cally impotent when propounded against an opponent who was already
doubting its conclusion. Therefore, in general, the dialectical setting
within which the argument is presented is important inasmuch as it can
affect the possibility of a subject finding it cogent. Still, it does not affect
its epistemic structure and so its objective cogency.
The objection to this reply is similar to the previous one: suppose one
were just open-minded with respect to (III) and did not have any posi-
tive reason to think it false. The liberal view predicts that the argument
should be effective for such a subject, but it does not seem to be so. The
reason, as before, is that if it is a tenet of the theory that the very same
experience could be had if one were just seeing a white wall bathed in
red light, it is totally unclear how it could, ipso facto, justify (I) over
its uncongenial counterpart. If it seems otherwise it is only because we
are, perhaps implicitly, assuming (III), or, even more generally, that we
are interacting with a physical world in environmentally normal condi-
tions. However, if that is right, we cannot acquire a warrant to believe
(III) – or some more general proposition needed to have warrant for (I)
in the first place – by running RED WALL. For, intuitively, warrant for (I)
would presuppose (III) (or some other more general proposition). RED
WALL would thus be epistemically circular and not just merely dialecti-
cally defective.9
One last observation concerns Pryor’s understanding of the skeptical
position, in the case of Moore’s proof, and his construal of a doubter
of (III) in the case of RED WALL. Of course, in the latter case, Pryor may
well want to stipulate the epistemic position of his imagined opponent
along his chosen lines. Still, it is inaccurate to depict a skeptic about
the existence of an external world as someone who thinks it is likely
to be false that it exists.10 For a skeptic is someone who, on the basis of
specific arguments, thinks neither “There is an external world” nor its
negation can be warranted. He therefore suspends judgment on both.
Obviously, what sounds damaging to us, given our usual Weltanschauung,
is the suspension of judgment regarding “There is an external world”.
However, this by no means commits a skeptic to favor its negation. As
said, the skeptic position, thought of at its best, is an agnostic one. If so,
it is not clear that a skeptic would find Moore’s proof wanting for the
reasons suggested by Pryor. For it is not obvious that agnosticism about
(III) should lead one to denying that (I) is perceptually warranted. In
particular, it would not in the liberal understanding of the structure of
empirical warrants, while it would in the moderate or the conservative
view. For it is only in the latter views that a commitment to (III), or even
its warrantedness, would be seen as necessary for a perceptual warrant
for (I). However, liberalism predicts that a skeptic who was agnostic
about (III) should consider (I) warranted, unless he obstinately refused
to accept the liberal view. Hence, on scrutiny, Pryor does not even have
a prima facie explanation for why a skeptic should find Moore’s proof
wanting, in light of the liberal view, once confronted with it, unless he
simply stubbornly declined to take that view seriously. Ironically, then,
the explanation Pryor has offered of a skeptic’s reaction vis-à-vis Moore’s
proof, in terms of dialectical ineffectiveness, could work only given a
different account of the structure of empirical warrant.11
Therefore, as we have seen, the liberal position seems quite naturally
to give rise to both Mooreanism – that is, the view that Moore’s proof is
a perfectly cogent argument – and to “easy knowledge”. Furthermore, it
has a serious difficulty in explaining why we do not find either MOORE
or RED WALL cogent. Alternatively, if it somehow manages to explain this
impression, it is by placing those arguments in very specific and argu-
ably contentious contexts, while granting their epistemic soundness in
all others, where it is dubious that one would really have the inclination
to condone them. Hence, the implausible consequences the liberal posi-

tion gives rise to seem to add to our pre-existing motivations to discard
that conception of the architecture of empirical warrant and to look for
an alternative.12
2 Against entitlements
As we saw in the first chapter, if one endorses the conservative view

on the structure of empirical justification and takes warrant for very
general assumptions, such as “There is an external world” and “I am not
the victim of a lucid and sustained dream”, to be necessary in order to
have a warrant for “Here is a hand”, when based on one’s current sense
experience, a skeptical outcome is likely to ensue. For if, in addition,
one thinks that these general assumptions cannot be warranted either
evidentially or a priori, it seems quite obvious that no warrant for the
premises of Moore-like arguments could ever be obtained. Absent such
warrants, we would end up forming the beliefs that we have hands and
that there is an external world with no warrant for either, just like a
Cartesian and a Humean skeptic would maintain.
Wright has attempted to avoid this result by pointing out a common
lacuna in both Cartesian and Humean forms of skepticism.13 Namely,
the idea that warrants for these general assumptions should somehow be
earned or acquired by means either of empirical evidence or of a priori
reasoning. This conception grounds Cartesian forms of skepticism, for
we could never know if a test used to verify whether we are dreaming
right now has actually been executed, or has merely been dreamt of.
Lacking a warrant for “I am not the victim of a lucid and sustained
dream”, and holding the principle of closure for warrant under known
entailment,14 by contraposition, we would lack warrant for any specific
empirical belief we might form on the basis of our current sense experi-
ence. Similarly, in the case of Humean skepticism, the lack of warrant
for the general assumptions needed for a warrant for the premises of
Moore-like arguments – that is, “There is an external world”, or “My
sense organs are generally reliable” – would result in a lack of warrant
for them as well.
This observation paves the way to Wright’s “unified” solution to both
forms of skeptical argument.15 Wright sees himself developing some of
Wittgenstein’s ideas in On Certainty (§§341-343, 151 in particular, but
see also §§56, 82, 318, 628), according to which it would belong to the
“logic” and “method” of our inquiry that some propositions be exempt
from doubt. Accordingly, he proposes a strategy to claim possession of
an unearned warrant – an entitlement in his terminology – for those

very assumptions. Entitlements are warrants that are not earned either
through the collection of empirical evidence, or by means of a priori
reflection on the concepts involved in these general assumptions, or
in other beliefs, such as the ones that figure as premises in Moore-style
arguments. According to Wright, they have always been there independ-
ently of us, in the abstract space of reasons, by virtue of the very epis-
temic structure of our usual inquiries. In this sense, they are unearned
or for free. Yet, Wright thinks he can provide arguments to (re)claim
them, that is, to convince our skeptical opponents and ourselves that
we do have them after all. More precisely, Wright thinks we have an
entitlement to accept, rather than believe, “I am not now dreaming”
and “There is an external world”. An acceptance differs from a belief, in
his account, because while the latter has evidence as input and leads to
behavior and/or belief as output, the former has the same output but has
no evidence as input.
Furthermore, Wright thinks that entitlements are a peculiar kind of
warrant, for they do not speak to the (likely) truth of certain corner-
stone propositions, as Wright calls the general assumptions we have
been talking about all along. As he writes:
In general, it has to be recognised that the unified strategy can at most

deliver a sceptical solution ... Sceptical solutions concede the thrust of
the sceptical arguments they respond to ... The unified strategy like-
wise concedes the basic point of the sceptical arguments to which it
reacts, namely that we do indeed have no claim to know, in any sense
involving possession of evidence for their likely truth, that certain
cornerstones of what we take to be procedures yielding knowledge
and justified belief hold good. (Wright 2004a, p. 206)
I will not address the exegetical issue of whether Wright’s reading of

Wittgenstein’s position in On Certainty is plausible. I will just note that
there are alternative ones: we have already encountered Strawson’s in
the previous chapter and we have already mentioned the fact that,
according to several readings, Wittgenstein regarded hinges either as
norms or as propositions with a normative role, for which no warrant –
evidential or otherwise – would be possible, let alone needed.16
According to Wright though, to claim these entitlements we need
two different arguments, which we consider in the order in which he
presents them. That is to say, we will consider the argument he offers to
vindicate our entitlement for “I am not now dreaming” first, and then
move on to the argument he proposes in order to claim our entitlement

to “There is an external world”.
Let us consider a cognitive project, the failure of which would not be
worse than the costs implied by not undertaking it, and whose success
would be better. Let P be a presupposition of such a cognitive project if
doubting P would imply doubting about the importance, or the possi-
bility of competently carrying out that particular project. Let us addi-
tionally suppose that: (i) there is no sufficient reason to believe that P is
false; (ii) every attempt to justify P would call for other presuppositions
that are no more certain than P, in such a way that if one wished to
justify P, one would implicitly commit to an infinite regress of cognitive
projects.
Let us assume that P is identical to “I am not dreaming now”. Clearly,
if P were called into doubt, we would not be able to think that any
cognitive project (in particular, an empirical one) could be competently
accomplished. Yet, precisely because if we were dreaming we could
not know that, we do not even have a reason to believe that P is false.
Moreover, any attempt to justify P would imply further presuppositions:
for example, that our senses are functioning appropriately and that there
is an external world, which is correctly represented in our experience. If,
though, we wished to justify these presuppositions, we would have to
be in a position to justify P, thus finding ourselves stuck in a regress of
cognitive projects.
According to Wright, we have to conclude that “I am not dreaming
now” is a presupposition of all our cognitive projects (both of an empir-
ical and of a rational kind), for which we have an entitlement. Therefore,
it is rational to believe that one is not dreaming now, even though there
is no way to acquire an evidential warrant for this belief.
Although the Cartesian skeptical paradox has been blocked, this does
not entail that the Humean one has been too, according to Wright.
For, in his view, the previous argument only allows one to vindicate
an entitlement for “I am not dreaming now” and, therefore, for our
being rationally entitled to use our perceptual and cognitive faculties.
However, it does not provide an entitlement for our trust that there is a
world inhabited by physical objects, which these faculties would allow
us to know. A different argument is needed to vindicate what he calls
“entitlement of substance”.
As Wright himself admits, the following is a much more sketchy argu-
ment than the previous one, and it could be summarized thus. In order
to have an objective conception of experience, one has to think of it
as arising from an interaction with objects that exist independently
of it. Since we have an objective conception of experience, we have to

conceive of objects that exist independently of our experience. Wright is
perfectly aware that this argument does not establish our ontology, but
merely makes it admissible. Nevertheless, he maintains that this is the
best that can be done in order to vindicate our entitlement for “There is
an external world”.
It should be noted that this second argument is grounded in a
conception of experience, whose objectivity requirement dictates that
it is thought of as concerning objects that exist independently of it.17
However, on the face of it, an idealist could maintain that, for experience
to be objective, it is sufficient that it displays uniformities that do not
necessarily depend on thinking about objects as existing independently
of our experiencing them. For example, instead of regarding my experi-
ence of the computer screen in front of me as objective only if it can
be considered as brought about by the interaction with an object that
exists independently of its being experienced, one could maintain that
it is objective insofar as it displays spatio-temporal uniformities, which
may ground certain generalizations. For instance, if I look in this direc-
tion, I will have this particular experience of a square-shaped patch, with
certain colors and a certain brightness. If I look from a different angle, I
will have an experience of a more or less rectangular patch, with certain
shades of color and a certain brightness. If tomorrow I look in the same
direction as the one in which I am looking right now, I will have the same
(or very similar and coherent with past ones) experiences as those of
today. I can therefore decide to call this bundle of experiences “computer
screen” and not that object whose existence should be independent of
them, and which is supposed to cause them. Moreover, an idealist could
maintain that presupposing the uniformity and coherence across time of
experiences might depend on the fact that some material object is there
(a computer), which exists independently of its being perceived, would
imply presupposing something which our own experience does not
warrant: for all we know, this uniformity could depend on the fact that
an evil scientist, or a demon, is triggering those representations in our
mind. Let me stress that these considerations are not meant to support
idealism, but only to draw attention to the fact that, if we follow Wright,
we run the risk of warranting the rational legitimacy of our assuming
that there are physical objects by means of a transcendental kind of argu-
ment, which rests on controversial premises. In Chapter 4 we scrutinize
whether anything better can be done about this.18
Let us now consider the more general issue of the nature of Wright’s
entitlements. Some authors (Pritchard 2005b, Jenkins 2007, Williams
2012) have argued that they are not epistemic warrants – that is, warrants
that corroborate the truth of the propositions of interest – but pragmatic
ones. That is to say, they do not give us reasons for believing that it is
true that we are not dreaming, or that there is an external world, but
merely pragmatic reasons for acting as if things were so, since this would
allow us to preserve our “cognitive projects” that are chiefly useful and
important to us.
I believe there is some truth in this objection, but also that its strength
can only be appreciated by relating it to the following problem, which
I believe is the main problem with Wright’s strategy. As I have already
explained, Wright aims to establish that the skeptic wrongly main-
tains that our basic assumptions are not warranted. Certainly, these
assumptions cannot be evidentially (that is, by experience or by a priori
reasoning) warranted, but, nonetheless, they are warranted “for free”,
in his view. However, entitlements do not, according to Wright, provide
us with reasons, albeit fallible, to believe that it is true that we are not
dreaming or that there is an external world. Rather, they just concede the
conclusion of the paradox, to the effect that there are basic assumptions,
which ground many of our “cognitive projects”, that are – as such – not
warrantable. Hence, at best, entitlements give us a reason to think that it
is not rationally mistaken or unmotivated to rely on assumptions that –
as such – are unwarranted and unwarrantable. Thus, it is not the skeptic
who has a partial notion of warrant, which may be amended by means
of the notion of entitlement, to make these assumptions ultimately
warranted. Instead, Wright does not realize that entitlements, which
are ultimately just a priori arguments, are not warrants to believe that
certain (first-order) propositions are true, but merely ways of reassuring
ourselves that assuming those propositions without a warrant is ration-
ally admissible. That is to say, Wright has, at best, provided warrants (of
an a priori kind) for “It is rational to assume that there is an external
world (or that I am not dreaming)”, but not for “There is an external
world” or for “I am not dreaming right now”. Hence, he has provided a
warrant for holding true that trusting P is rational, but he has discovered
no warrant – that is, an epistemic good that corroborates the truth of
P – for P itself.19
At this point, the question arises whether Wright’s entitlements at
least succeed in the task of telling us why trusting in certain – as such –
unwarranted (and unwarrantable) propositions is, after all, epistemically
rational. Even in this case, however, difficulties are just around the corner.
As I have already briefly remarked, the first of Wright’s arguments seems
to show that, since forming empirical beliefs based on our experiences
is an important cognitive project for us – I would say an indispensable

one – we must accept that we are not dreaming. This, however, is a
clearly pragmatic warrant, whose obtaining owes to the fact that the
epistemic method, which has “I am not dreaming now” as its basis, is
useful or even necessary for us. The skeptic may – I take it – have no
reason to object that it is useful for us, if not indispensable, to assume
those propositions. He could therefore insist that, clearly, this does not
prove that those propositions are epistemically warranted – that is, that
they are likely true. Thus, it seems that Wright’s anti-skeptic strategy is
unsuccessful, even with respect to the task of explaining why assuming
certain unwarranted (and unwarrantable) propositions is rational in the
light of epistemic rationality, as understood by the skeptic.
The second argument, in contrast, tells us that, in order to have the
conception of experience that we actually have, we cannot but assume
that there are physical objects. I already pointed out that this is not
obvious. To repeat, an idealist could argue that an objective conception
of experience may depend on regularities and uniformities among our
perceptions, with no need for an assumption regarding the existence
of an external world. Nevertheless, one may accept the idea that we
actually have a conceptual scheme in which the category of physical
object is not empty. One could then argue that it is useful for us to have
such a conceptual scheme, and are therefore warranted in assuming that
there are physical objects. However, at this point, the skeptic could obvi-
ously admit that, in order to have the conceptual scheme we actually
have, it is necessary for us to assume that there is an external world,
though he could hold that this does not rule out that it is metaphysi-
cally possible that the descriptions for which we deploy it are systemati-
cally incorrect.
At this point, one may appeal to semantic externalism and claim
that, for our thoughts (and their linguistic counterparts) to have their
contents, the physical objects they are about must exist. This argument
was put forward in several versions by authors such as Putnam (1981),
and surprisingly, Wittgenstein (1969, §369), and has been revived, in
different guises, by authors, such as Peacocke (2004), and Wedgwood
(2013), as we shall presently see.
In response to these considerations, two things are worth noticing.
Firstly, that semantic externalism is a controversial position; hence,
this argument could be contested were internalism in the philosophy
of mind and language to be established. Secondly, that even if exter-
nalism about mental (and linguistic) content were correct, it would only
entail that one must have been, or at least that one still finds oneself at
times, in a causal relationship with physical objects, since this would

suffice to guarantee that our thoughts (and assertions) have a certain
content. This, however, does not establish that we are in a position to
know that we still are, or that we are right now, interacting with those
objects. In other words, the semantic thesis – provided it is correct –
implies an ontological–metaphysical thesis. However, it in no way
affects the epistemic problem raised by skepticism. This should come
as no surprise since, after all, semantic externalism has always been
open to epistemic objections, as it posits individuation conditions for
contents which outrun a subject’s discerning capacities. In the case of
Oscar and his twin, famously discussed by Putnam (1975) (or in cases of
singular thoughts à la Evans (1982) and McDowell (1984)), two subjects
may entertain thoughts whose contents are phenomenologically iden-
tical, such as “This is water” and “This is my hand”, yet only one may
have a thought about actual water, or the relevant singular thought. The
opacity of semantic content, which is a consequence of semantic exter-
nalism, has therefore always favored its compatibility – under certain
conditions – with skeptical scenarios.
To conclude, it seems to me that Wright’s strategy in terms of entitle-
ments does not ultimately solve the skeptical paradox. It does not offer
the kinds of warrant that corroborate the truth of propositions like “I
am not dreaming” and “There is an external world”, but it only offers
warrants for maintaining that it is rational to assume certain proposi-
tions are – as such – unwarranted. Moreover, on closer inspection, the
justifications he provides for the rationality of assuming those proposi-
tions are, at most, of a pragmatic kind, and therefore altogether compat-
ible with a skeptical position.
To stress, whether or not he has succeeded in providing epistemic
warrants to believe “To assume that there is an external world/that one
is not the victim of a lucid and sustained dream is rational”, and I think
there are reasons to doubt that he has, I submit that Wright’s own enti-
tlements had better not be understood as epistemic goods produced (or
discovered) by philosophical reflection that turn a-rational assumptions
into rational ones; nor as philosophical arguments which provide one
with a priori warrant for “There is an external world” or “I am not the
victim of a lucid and sustained dream”.20 Rather, they should be under-
stood as philosophical arguments that, if successful, would provide one
with a priori warrant to believe that the assumptions that there is an
external world, or that one is not the victim of a lucid and sustained
dream, which are unwarrantable, are rational.21 Notice, in fact, that if he
tried to say that entitlements are ultimately first-order a priori warrants
obtained through philosophical reflection for “There is an external

world” of “I am not the victim of a lucid and sustained dream”, this would
create a tension. For, in that case, they would have to speak to the likely
truth of those assumptions, while, officially, Wright has been concerned
to deny that entitlements could achieve that much.22 Furthermore, if
that were what entitlements were supposed to do, then we would have
a priori reasons for the relevant contents and so we should believe them
and not merely assume them. For, according to Wright’s distinction,
assumptions, contrary to beliefs, are a species of acceptance that does
not have evidence – a priori or otherwise – as input. However, if entitle-
ments were a priori warrants for “There is an external world” and the
like, then we would have a priori evidence to believe them.
The surprising outcome of this discussion is therefore that, ironically,
the most charitable interpretation of Wright’s strategy consists in saying
that, contrary to his official pronouncements, he is in fact proposing
a moderate conception of the architecture of perceptual warrant.
Accordingly, assumptions that are not epistemically warranted, together
with a certain experiences, provide warrant for ordinary empirical
propositions. Consequently, entitlements are in fact a priori warrants
that speak in favor of believing that such unwarranted assumptions are,
after all, rational.23 Yet, as we have seen, it is dubious that his way of
redeeming them would show why these unwarrantable assumptions
are rational by the very lights of epistemic rationality itself. Hence, I
conclude that also on this – to my mind more charitable – interpreta-
tion, Wright’s anti-skeptical strategy is not successful.
3 Against a priori warrants
In the previous sections we provided further reasons for discomfort

with liberal and conservative views, beside the ones in Chapter 1, and
therefore have offered a motivation to explore the alternative to both,
namely moderatism. Insofar as these two first chapters are concerned
with supporting the moderate position, vis-à-vis its main competitors in
the internalist camp, our job should be over and we could now explore
the consequences of this view, both with respect to Moore-style argu-
ments and Humean skepticism (issues taken up in Chapters 3 and 4
respectively).
Before closing this chapter, however, I think it is useful to critically
examine some prominent attempts at providing first-order a priori
warrants for very general assumptions, such as “There is an external
world”, that have recently been advanced in the literature. One might
see them, at least partly, as potential defenses of the conservative view,

inasmuch as they would bear in the business of providing it with what
it would need in order to be feasible at all, once it is acknowledged
that the entitlement strategy fails. For they provide those warrants for
collateral assumptions which, from a conservative view, have to be in
place for one’s sense experience to supply one with a perceptual warrant
to form specific empirical beliefs. Some supporters of these views will
resist being connected with conservativism (Wedgwood and Peacocke),
because they claim to be sympathetic to liberalism, while others will not
(Sosa). Still, at the price of some interpretative injustice, I believe it is
useful to consider them at this point.
Furthermore, quite independently of their specific relations with
the conservative view, it is clear that these theories compete with the
moderate view, at least as long as it is a tenet of the latter that no warrant
for very general assumptions, such as “There is an external world”, could
be provided – either a posteriori, or a priori or through entitlements.24
Hence, I consider a position recently put forward by Ralph Wedgwood,
which is presented as a sort of mid-way course between Pryor’s liber-
alism and Wright’s conservativism. I then turn to Peacocke’s view, which
could also be characterized as a sort of middle ground between Pryor
and Wright, although its author does not present it that way. Finally,
we consider Sosa’s recent attempt to provide a priori warrants – in fact
knowledge, in his view – for general assumptions such as “There is an
external world”. To repeat, the common trait between these proposals is
that they all try to provide a priori justification for these general assump-
tions. As we shall see, however, this entails a substantial and problem-
atic revision of the very notion of a priori.
3.1 Wedgwood and a priori bootstrapping

According to Wedgwood, who follows Silins (2007) on this score, a
semi-formal proof can be set up by employing the resources of prob-
ability calculus (thereby following White 2006), which demonstrates
that an experience with a representational and phenomenal content,
as of a hand, cannot by itself warrant “I am not dreaming that I am
seeing a hand in front of me now”. That experience, by itself, might
even raise the probability of the opposite, specific skeptical hypothesis
(cf. §1). This, according to Wedgwood, shows – against Pryor’s Moorean
liberal position – that the existence of a warrant for ruling out that one
is in the specific skeptical scenario compatible with the experience one
is currently undergoing is a necessary consequence of having a perceptual
warrant for “Here is a hand”. However, according to Wedgwood (still
Moorean Rationalist
Liberal Pryor Wedgwood

(and Silins)
Conservative ? Wright
following Silins) and contrary to what conservatives hold, this justifi-

cation does not explain why one’s experience is a warrant for “Here is
a hand”. Hence, the novelty of Wedgwood’s proposal lies precisely in
maintaining that because, as liberals have it, one is immediately justi-
fied in believing propositions such as “Here is a hand” on the basis of
one’s experience, one can achieve an explanation of why one is justified
in believing that there is an external world or that one is not having
a lucid and persistent dream. If we wanted to represent the range of
possible positions licensed by the divorce from the liberal position and
Mooreanism, and if we called the position of those who think there can
be independent justifications – either a priori, or by means of entitle-
ments – “rationalism” for “I am not now dreaming of there being a
hand”, we could identify Wedgwood’s position as follows.25
The kind of a priori justification that we have for these propositions
depends, according to Wedgwood, on a bootstrapping argument, which
draws on the fact that one is justified in forming empirical beliefs directly
on the basis of one’s occurrent experiences, as long as one has no reason
to believe the opposite. However, this bootstrapping argument is slightly
different from the usual ones, for it is a priori. The argument is rather
complex, and we need not look at it in detail. It is sufficient to notice
that, according to Wedgwood, the practice of taking one’s experiences at
face value in order to form the corresponding beliefs is rational, provided
there are no countervailing reasons. The same holds for the practice of
forming beliefs by deductive means, and by means of inferences to the
best explanation, provided one starts from rationally believed premises.
Thus, it is rational to form the belief that I have a hand, if I have an
experience with that content and I have no countervailing reasons. I can
also believe a host of other propositions based on appropriate percep-
tual experiences. By means of an inference to the best explanation, I
can therefore rationally conclude that my experiences are largely reli-
able, which is incompatible with skeptical hypotheses like “I am having
a lucid and persistent dream” and “There is not an external world”.
This process of rational reasoning, also called “bootstrapping a priori
reasoning”, may lead any subject who has the relevant concepts and
cognitive capacities to believe the negation of the skeptical hypotheses,
and explains why we have an a priori justification to do that. Finally, it

is important to note that, if Wedgwood is right, one would have a priori
justifications for empirical contingent propositions such as “I am not
dreaming at this moment”, which would, as a matter of fact, be a priori
justified even if they were false, were one in a skeptical scenario.
As already mentioned, this argument rests on the assumption that
the practice of taking experiences at face value in the absence of reasons
to the contrary, in order to form the corresponding empirical beliefs,
is rational. Wedgwood explicitly admits that he has not taken any step
towards showing the legitimacy of this assumption. Clearly, this assump-
tion is going to be rejected by a Humean skeptic since, from his point of
view, unless one has a justification to rule out that one is in a skeptical
scenario already (albeit a specific one), one cannot take one’s experience
at face value in order to form the corresponding empirical beliefs.
Elsewhere Wedgwood (2011) does provide an argument to establish
the rationality of forming empirical beliefs by taking experiences at face
value, absent countervailing considerations. This argument hinges on
the fact that to have some concepts, it is necessary to form beliefs that
imply the use of those concepts on the immediate basis of experiences
with a corresponding representational content (an analogous argument
can be found in Peacocke 2004).
It is not relevant for present purposes to go into the details of the
proposal, or expound on the questionable idea that the absence of collat-
eral beliefs makes these belief-forming practices allegedly “a priori”. After
all, they are still epistemically grounded in experience. What has to be
emphasized, rather, is that even if it were the case that those practices
must have led us to true beliefs, at least in the past, this does not prevent
us from relying on them even in skeptical scenarios in which they would
not give rise to any true beliefs, as Wedgwood himself acknowledges.
It is therefore unclear in what sense they can be employed to give a
compelling response to the skeptical paradox, at least when the latter
is raised with respect to our present epistemic situation. We shall come
back to this issue shortly.
There is also a further problem with the stability of Wedgwood’s
argument since it is unclear whether his position does not ultimately
collapse onto either of the two theories from which he intends to
depart, to some extent. For if there is an immediate justification for
“Here is a hand”, given one’s sensory experience and the absence of
countervailing reasons, why could not this also justify the negation of
the skeptical hypotheses? What is wrong with an inference from “Here
is a hand” – which, ex hypothesi, is an immediately justified premise – to
“There is an external world” or “I am not now dreaming”? This remains

unexplained since Wedgwood does not deny the principle of epistemic
closure and he denies that a Moore-style argument is an example of
warrant transmission failure.26 Wedgwood’s intermediate position seems
therefore to collapse onto Pryor’s Moorean liberal thesis, or at least entail
that Moore’s proof would enhance one’s independently acquired a priori
warrant for “There is an external world”. The further idea, tabled in
a recent paper (2012), that our enduring belief in the existence of an
external world may be supported by coherentist considerations, besides
being the warranted result of an inference such as Moore’s, does not
change the fact that Wedgwood, like Pryor, would then be committed to
Mooreanism. To repeat: in the end a Moorean argument would provide,
or enhance, one’s previous warrant for the occurrent belief in the exist-
ence of an external world, even if, as an enduring belief of ours, the latter
would be justified not through a Moore-style argument but by coheren-
tist considerations.27
A first possibility, to avoid the conclusion that a Moore-style argu-
ment provides a justification for the occurrent belief in (III), would
consist in exploiting the generalized Bayesian argument I presented in
§1. Yet, Wedgwood should also explain why it seems natural to us that
the probabilities of “Here is one hand” go up whenever we have a hand-
like experience (absent reasons to doubt). Such an explanation, as we
have seen, seems necessarily to involve the abandonment of the original
liberal position about the structure of empirical justification, and to rely
on (either the moderate or) the conservative architecture of empirical
warrants. Hence, along these lines it would be difficult to rescue liber-
alism, albeit deprived of its Moorean elements. Let me explain: the liberal
view has it that no warrant for (III) is needed to have warrant for (I). The
generalized Bayesian argument, in contrast, would block Mooreanism
only if (III) were already independently warranted (or at least assumed)
and such a warrant were needed in order warrantedly to believe (I) –
“Here is a hand”. Hence, the fact that despite the raise in probability
of uncongenial local skeptical hypotheses the probability of “Here is
one hand” would still go up, would necessarily be due to the fact that,
at least as far as rationally available justifications are concerned,28 the
warranted belief in (III) (or at least its assumption) would play an epis-
temic role with respect to how subjects acquire a warrant for (I) – “Here
is a hand” – through experience.
The second possibility, which might be open to Wedgwood only if he
were prepared to allow for non-trivial cases of failure of transmission of
warrant, would be to hold that Moore’s argument is a case of transmission
failure once this notion is explained along Wright’s favorite lines. That is,
by saying that in order to have a justification for “Here is one hand” one
needs already to have a warrant for “There is an external world”, or “I am
not now dreaming”. Hence, one cannot acquire a justification for these
propositions by running Moore’s proof. Clearly, however, Wedgwood’s
position would then collapse onto Wright’s conservativism.
Finally, in order to avoid falling back onto either Pryor’s or Wright’s
positions, Wedgwood should refrain, once more, from denying that
there are non-trivial cases of warrant transmission failure. Moreover,
he should provide an alternative explanation – different from both
Wright’s and the one I have offered by generalizing the Bayesian
argument – for the reason why Moore’s argument cannot transmit
the warrant one would have for its premises, on a liberal account of
perceptual warrant, to its conclusion.29 The aforementioned distinc-
tion between enduring beliefs, for which warrant is never transmitted,
according to Wedgwood, and occurrent mental states figuring in an
inference as premises and conclusions for which warrant is always
transmitted, according to him, seems unable to impugn the cogency
of Moore-style arguments, at least when occurrent beliefs are stake. I
therefore conclude that Wedgwood does not have a plausible explana-
tion of why “real MOORE” fails and that what he actually says makes
his views collapse either onto Pryor’s or onto Wright’s, or even on
the moderate position. In short, Wedgwood’s position is unstable and
therefore fails to be a viable alternative to either liberalism, or conserv-
ativism (or indeed to the moderate view).
The prima facie appeal of Wedgwood’s position (and of Silins’) I think
depends, at least potentially, on a conflation between justifications actu-
ally possessed by subjects – that is, the rationally available warrants we
encountered in Chapter 1 (§1) – and propositional justifications. Such a
conflation is perhaps caused by Wedgwood’s endorsement of a “robust”
form of internalism, according to which talk about justifications is
invariably talk about justifications as rationally available to subjects. For
it is only from the point of view of these justifications that it might
seem prima facie plausible to maintain that subjects need have none
to believe the negation of skeptical hypotheses in order to form justi-
fied beliefs about material objects in their surroundings, on the basis of
the deliverances of their perceptions, absent reasons for doubt. Yet, to
repeat, the dispute between Wright and Pryor (as well as between them
and moderates) concerns propositional justifications and, if this is the
case, Wedgwood’s intermediate position (and Silins’) loses a lot of its
intuitive appeal.
Finally we saw that, according to Wedgwood, there are empirical

propositions such as “I am not dreaming now” which are a priori justi-
fied and which would continue to be so even if they were false, were
one for instance to find oneself in a skeptical scenario. This is not the
right place for a thorough discussion of the notion of a priori. One
must acknowledge, however, that, if things were so, it would become
apparent that the arguments aimed at providing an a priori justification
for these propositions would not in any way corroborate their truth. The
paradoxical consequence would ensue that being in possession of an a
priori justification for them would be entirely compatible with the truth
of the skeptical hypothesis, which would make the effectiveness of this
strategy with respect to solving the skeptical paradox at least dubious.
Let me stress that the objection here is not that any feasible anti-skep-
tical strategy should give us certainty or indefeasible warrants to believe
anti-skeptical conclusions. Rather, it consists in noticing that nothing
in the present strategy allows for raising the probability, as it were, that
skeptical scenarios do not obtain.
In more detail: as is well known, John Hawthorne (2002) has defended
the view that there is a priori knowledge of deeply contingent truths.
Recently, Yuval Avnur (2011) and John Turri (2011) have, rightly in my
opinion, pointed out that Hawthorne’s example, based on conceiving of
a disembodied being who would have a priori knowledge of the contin-
gently true conditional “If I undergo experiential life history L, then T is
true”, is unconvincing for a number of reasons. Turri has then gone on
to provide another example to the same effect. This time the contingent
truth allegedly known a priori would be “The most unlikely possible
event is not presently occurring”. This is not the place to discuss this
example, though I personally find it dubious. For one thing, it seems
to me that, for all I know, the most unlikely event could be presently
occurring unbeknownst to me. Hence, it is not clear in what sense one
could have knowledge – let alone a priori – of a proposition incompat-
ible with that. Nevertheless, interestingly, Turri also persuasively rejects
Kripke’s example of a priori knowledge of a contingent truth regarding
the length of the standard meter. Equally interestingly, in a footnote he
seems to raise some qualms about what I would personally take as a clear
example of a priori knowledge of a contingent truth, namely, knowledge
of “I exist”. For my existence is not necessary – that is, there are possible
worlds in which I do not exist – yet I can now know a priori that I do, if
indeed I do. That is to say, I can know that I exist now, by mere reflec-
tion on the concepts involved, in particular regarding the first-person
concept. Thus, I do agree with Hawthorne and Turri that there may be
examples of a priori knowledge of contingent truths, although we disa-

gree about what would count as an example of such knowledge.
Yet, this is a long way short of being able to know a priori the contin-
gent truth that I am not the victim of a lucid and sustained dream or
that there is an external world. Therefore, although the motivation, as
Turri explicitly recognizes, for searching for deeply contingent a priori
knowledge is – among other things – to try and provide such a kind of
rebuff of the skeptic, it is by no means clear that, so far, it amounts to
anything more than just a hope.
Finally, to repeat, Wedgwood is concerned with a priori justification,
not with knowledge. Hence, in his view, we could have a priori justifi-
cation for “I am not the victim of a lucid and sustained dream” even if
we were in such a skeptical scenario. However, if that is the case, then
it is unclear what damage it would make to a skeptic. For we could be a
priori justified in believing “I am not the victim of a lucid and sustained
dream” when it is in fact false and, in particular, we would not thereby
have any actual reason to think that it is more probably true than false.
Hence, we would not have any means to exclude that, unbeknownst to
us and irrespective of our justifications, we are in fact victims of a lucid
and sustained dream. That is to say, we could not exclude that, after all,
we are in a skeptical scenario! Yet, such a possibility has always been
enough to run skeptical arguments and it is unclear how successful a
purportedly anti-skeptical strategy would be if it condoned it.
3.2 Peacocke’s rationalism

In this section, we briefly consider Christopher Peacocke’s position, with
respect to the justification of empirical beliefs based on perception and
Moore’s proof, as presented in his The Realm of Reason (2004).
According to him, perceptual experiences – that are conscious mental
events – can immediately justify one’s empirical beliefs, without appeal
to intermediate beliefs, or further mental states about one’s own experi-
ence. This “directness” suffices, for Peacocke, to make the transitions
from the perceptual experience to the corresponding belief “relatively a
priori”.30 Moreover, the transition from the former to the latter is reason
giving, or “entitling” in his terminology, because it is truth-conducive
and rational. In particular, Peacocke thinks that there are instance-indi-
viduated perceptual experiences, that is, perceptual experiences whose
content that p is immediately available to one because it does not require
any other background knowledge or belief, that are such that, if the
world cooperates, are caused by the obtaining of p. Therefore, an experi-
ence of a red object or of a round-looking one are instance-individuated;
while that of a Mac, or of a baseball bat, are not. In general, the criterion
to distinguish between these two kinds of perceptual experience is that,
in the instance-individuated case, if the judgment based on the relevant
experience is false it is so either because the environmental conditions
are abnormal or because a subject’s perceptual system is malfunctioning.
Furthermore, he holds that it is constitutive of at least instance-individ-
uated perceptual experiences, and so a priori knowable, that they have
the representational content that p because such a content is caused by
the obtaining of p, at least relative to normal environmental conditions
and to the proper functioning of a subject’s perceptual system.31 Thus,
according to Peacocke, if one is having an instance-individuated percep-
tual experience, then, absent reasons for doubt (about the environmental
conditions and/or the proper functioning of one’s sense organs), one is
rationally entitled to take it at face value to form the corresponding
judgment.32 That is to say, one would have a non-inferential, internalist
justification for the relevant belief,33 which, however, one may well be
unable to articulate. Moreover, the reason why the entitlement would
hold a priori is that the concepts involved in the relevant beliefs are such
that their possession conditions are exhausted by being able to apply
them based on appropriate instance-individuated experiences. Now, it
is unclear whether an experience of a hand is instance-individuated,
on Peacocke’s view of the matter, but supposing it is, just by having it,
absent reasons for doubt, one would have a justification – an entitle-
ment, in Peacocke’s terminology – to believe “Here is a hand”.
Clearly, Peacocke’s position, so far, has several connections with
Burge’s (Chapter 1, §§1, 5). The main differences have to do, first, with
Peacocke’s requirement – which I find absolutely plausible – that experi-
ences should be conscious events in order to give rise to entitlements,
while, for Burge, they need not be. Second, they have to do with his
account of why the experience would, in the appropriate conditions,
constitute an entitlement for the corresponding belief, which, as we
have just seen, proceeds by reference also to the possession conditions
of the concepts involved in that very belief. Whereas for Burge, as we
saw in the previous chapter, the explanation, though also a priori,
depends merely on reflections on the very notion of representation and,
in particular, of perceptual representation.
However, Peacocke’s view also bears important similarities to Pryor’s
liberal account of the justification of empirical beliefs based on one’s
experience. For it is a tenet of Peacocke’s position that there is no need
for an independent justification of very general beliefs such as “There
is an external world” or “I am not the victim of a lucid and sustained
dream”, in order to possess a perceptual justification for a specific empir-

ical belief, given an appropriate course of experience.
Contrary to Burge (2010) and similarly to Pryor, however, Peacocke
claims that Moore’s proof is epistemically fine as it is (if we could have
an entitlement for its first premise). Hence, it would give us justification
to believe that there is an external world or that we are not envatted
brains or victims of lucid and sustained dreams. In accordance with
Pryor, Peacocke also thinks that the proof is wanting only for dialectical
reasons, when it is propounded in the context of an argument with a
skeptic who, given his collateral doubts about the truth of the conclu-
sion, would not allow that its first premise, in particular, is warranted.
For Peacocke, however, this is not the end of the story. After all, given
what we have just seen, he might simply have maintained that the
warrant for “Here is a hand” would transmit to “There is an external
world”, or “I am not the victim of a lucid and sustained dream” and
thereby rebut the usual skeptic who doubts this kind of conclusion.
Instead, Peacocke provides us with what he considers an a priori argu-
ment in favor of these very conclusions. In his view, this argument
should give us the final explanation of why it is correct, absent reasons
for doubt, to take one’s instance-individuated experiences at face value
to form the corresponding judgments. However, as we shall see, it
equally offers an a priori reason to believe that there is an external
world, and that we are not victims of lucid and sustained dreams or of
evil demons. Hence, it is what we may call the master argument against
a skeptic challenging the view that there may be entitlements for our
perceptually based beliefs.
The argument is roughly this.34 (1) We have experiences with repre-
sentational content that p, like the visual experience of a hand. (2) They
are complex events. (3) As such, they are in need of complexity-reducing
explanations. (4) An explanation of their occurrence by natural selec-
tion meets such a requirement by holding that the visual experience of
a hand I am having is due to the cooperation between an environment
containing hands and a visual system evolved in such an environment
to be able to represent hands, contrary to skeptical hypotheses. (5) A
complexity-reducing explanation is more likely to be true than a non-
complexity-reducing one. (6) It is an a priori truth that it is rational to
hold that things come about in a way in which they are more likely to
have come about. (7) Therefore, it is more rational to assume that expe-
riences are the result of the operation of a suitable cognitive mechanism
that has been selected through natural selection to produce, by and
large, correct representations of the environment around us. (8) Hence,
it is more rational to assume that there is a material world, which is by

and large correctly represented by experiences.
Several worries could be raised against this argument. For instance,
against step (5), it may be suggested that it is not at all clear that
complexity-reducing explanations are more likely to be true than more
complex ones and, insofar as they are, we would know that on empir-
ical, so a posteriori, grounds. Furthermore, the notion of a complexity-
reducing explanation is itself quite vague. I take it that, cognitively for
instance, though not physically, it is less complex to think that, say,
God has created everything just as it is, human beings included, than
to think that everything in the world is the result of very complex
events, whose initial conditions are quite difficult to conceptualize,
let alone to know. Nor is it clear that complexity-reducing explana-
tions should be ontologically reductive and that the same types of
state could not figure in the explanation of other states. Take inten-
tional explanations of intentions in which a subject’s further mental
states are usually appealed to. They would not be reductive explana-
tions, but they would be perfectly respectable ones all the same. This
would have a bearing on step (4) in Peacocke’s argument. For he claims
that skeptical scenarios involving evil demons or mad scientists, as
opposed to scenarios appealing to natural selection, would not provide
complexity-reducing explanations of the representational properties
of experiences, as they would presuppose the existence of intentional
states in general, and of experiences in particular, had either by evil
demons or by mad scientists. Notice, furthermore, not all skeptical
scenarios would have to appeal to the existence of mad scientists or
evil demons.35 One could, for instance, simply revert to the dreaming
hypothesis and raise at least local doubts with respect to the fact that
right now, given the actual course of our experience, we could know, or
at least justifiably believe, that there is a hand in front of us and that
we are not merely dreaming it.36
Hence, first of all, if this argument is not successful – and there are
reasons to think it is not – Peacocke would have failed, by his own lights,
to show that we are entitled to believe “Here is one hand” when we have
a hand-like experience, absent reasons for doubt. Thus, he could not
really join forces with Pryor in maintaining the liberal view of the struc-
ture of empirical warrants. Furthermore, he would not have provided an
a priori, independent warrant for the conclusion of Moore’s proof either.
In addition, given the failure of his defense of his own variety of liber-
alism, with the consequent failure of the attendant Moorean response to
skepticism, he would be left with no response to it at all.
Secondly, and more importantly I think, even if this argument

succeeded the very fact that Peacocke feels the need to propose it should
make us suspicious about the kind of strategy he is actually pursuing.
For as we have just seen, according to him, we are entitled to take certain
experiences at face value to form the corresponding beliefs, where the
transition is unmediated by further beliefs. Moreover, we can explain
why we are so entitled and the explanation need not be available to
ordinary folks. All this seems to me perfectly reasonable, yet silent on
whether an entitlement for either “There is an external world” or “I am
not now dreaming” (or any other general anti-skeptical proposition) is
needed in order to have an entitlement of a perceptual kind for “Here
is a hand”. To repeat, it is obvious that a subject need not be able to
entertain such an explanation. All the same, it seems that Peacocke,
by providing an explanation as to why we are entitled to these general
presuppositions and by seeing it as part of the explanation of why we
have a perceptual entitlement for ordinary empirical beliefs such as
“Here is a hand”, is, after all, assuming that, in the abstract space of
reasons at least, warrant for type-III propositions is needed to actually
have a perceptual warrant for type-I propositions. Put slightly differently,
Peacocke’s position seems an alternative to Wright’s only when ration-
ally available warrants are concerned, but it seems to collapse onto it
when propositional ones are at issue.
3.3 Sosa and rational intuitions

A different and currently much sketchier proposal, which, for this
reason, will occupy us very briefly here, is the one put forward by Ernest
Sosa (2010). He claims that propositions like “There is an external
world” are actually known a priori (and hence true). According to Sosa,
we can explain how we gain knowledge of propositions like “There is an
external world” by appealing to the notion of rational competence, even
though we are not yet in a position to explain in full detail how this
competence may work. The proposed account is as follows.
A subject S has a rationally competent human intuition if and only
if: (i) understanding P prompts him to assent to P, independently of the
fact that S has any other reason that leads him to give his assent to P
(besides his understanding of P); (ii) the mechanism that prompts him
to give his assent to P, on the basis of the comprehension of P, independ-
ently of any other rational basis, is an epistemic competence; (iii) this
mechanism usually operates in human beings during their normal
development, since it provides subjects with the constitutive concepts
of P together with the inclination to assent to P.
What is relevant for our purposes is that such rational competence

would give rise to intuitive knowledge of P, that is, knowledge that is not
based on experience. For Sosa maintains that knowing specific empirical
propositions like “Here is a hand” on the basis on one’s own experience
depends on knowing propositions such as “There is an external world”.
The latter knowledge cannot, however, be gained by means of experi-
ence. Hence, it would be a priori, though stemming from the exercise
of our epistemic competence, whose nature and workings we ignore, at
least at present.
As I said, Sosa’s proposal is still very sketchy and it is hard to assess
it properly. Though the question immediately arises whether it adds
anything to the classical a priori idea that we may gain knowledge
of “There is an external world” simply via reflection on the concepts
involved in the characterization of that proposition. The question also
arises of what the proposal can do against the objection that, no matter
how intuitive it may appear to us to give our assent to “There is an
external world”, it is nevertheless the case that we could find ourselves
in a skeptical scenario. In other words, although giving our assent to
“There is an external world” is natural for us, provided we have the
appropriate concepts, such a spontaneous assent does not seem suffi-
cient to establish that it is likely true there is an external world, and so
we know that there is.
One could obviously urge that, if such an assent is the product of the
exercise of a real epistemic competence, this guarantees that it takes
as its objects only true propositions. However, at this point it becomes
incumbent on a theorist to say something more about the nature of
this competence, in order to guarantee that it keeps track of the truth.
Moreover, one can imagine that this competence has been activated
in conditions such that it would give rise only to true beliefs (indeed,
instances of knowledge) but that, at a subsequent stage, we have been
placed in a skeptical scenario (while unaware of this), in such a way that
the competence remains operative but no longer guarantees knowledge
of those propositions which are believed through its exercise. Hence,
in this case as in Wedgwood’s, we would reach the counterintuitive
conclusion that we can have a priori justified, yet false beliefs. This
once again casts doubts on the anti-skeptical effectiveness of such a
strategy.
At the end of this examination of some contemporary attempts to
provide a priori justification for the belief that there is an external
world, that we are not BIVs, or victims of lucid and sustained dreams,
we can say that none of the important views put forward by Wedgwood,
Peacocke, or Sosa have passed muster. For this reason I think that once it
is further acknowledged that the Moorean liberal position and Wright’s
entitlement strategy in defense of the warrantedness of these proposi-
tions are both wanting, the obvious result is that no warrant, either
empirical, or a priori, or through entitlements for them is forthcoming.
Since there are no other warrants, I believe we must start living with the
consequences of this finding. Namely, we must come to terms with the
possibility that these propositions are assumptions we make and, as we
saw in the previous chapter, are fundamental to perceptual justifications
for our ordinary beliefs, yet are not themselves warranted or warrant-
able. This, in effect, is the gist of the moderate position, whose conse-
quences for the nature of Moore’s proof and similar arguments, and for
external world skepticism are addressed in Chapters 3 and 4.
4 Summary
In this chapter, we have seen the problems that beset the attempt to
meet the skeptical challenge once either liberal or conservative views
are endorsed. With respect to the former, we have seen how, when
coupled with Mooreanism, that is, the view that Moore’s argument is
cogent despite appearances to the contrary, it gets into trouble for it
is difficult to see how a hand-like experience, which would be subjec-
tively indistinguishable no matter how it comes about, could actually
provide one with a justification for “There is an external world with a
hand in it” rather than for a proposition like “I am a BIV hallucinating
having a hand”. Furthermore, we have seen how problematic the claim
is that Moore’s proof would be merely dialectically ineffective when
propounded against an opponent like a skeptic. For a skeptic is best
thought of as an agnostic regarding the existence of the external world.
However, if one were such an agnostic and the liberal view were right,
one should rationally give up one’s agnosticism. Conversely, one could
remain an agnostic only by stubbornly sticking to an erroneous concep-
tion of perceptual justification. Skepticism would be a kind of disease
one should not catch, or should cure oneself of. Yet, this does not seem
to do it justice.
We have also seen how some recent attempts to decouple the liberal
position from Mooreanism, such as Wedgwood’s and less explicitly
Peacocke’s, are actually problematical. It is not clear, when looked at
in detail, that they do succeed in sticking to the liberal position, and
it is even less clear that they succeed in providing an independent a
priori justification for “There is an external world”, capable of meeting
the skeptical challenge. We have also looked at Sosa’s recent attempt to

show that we have independent and a priori knowledge of “There is an
external world”, and have found it equally wanting.
We have analyzed Wright’s effort to provide non-evidential justifica-
tions, that is, entitlements, for propositions such as “There is an external
world”, as part of his anti-skeptical strategy. We have pointed out how
entitlements do not, by definition, speak to the truth of the proposi-
tions they are meant to warrant. They are therefore only superficially
similar to first-order epistemic warrants for given propositions. What
they do, at most, is provide a justification, of an entirely a priori kind
and therefore evidential, for the second-order proposition “To accept
that there is an external world is rational”. However, when looked at in
this light, it turns out that they merely provide a pragmatic, not an epis-
temic, warrant for such a second-order proposition, that is, a justifica-
tion that speaks to the truth of the targeted (second-order) proposition.
Hence, I have argued that Wright does not successfully meet the skep-
tical challenge. For he neither provides an epistemic warrant for “There
is an external world”, nor for “To assume that there is an external world
is (epistemically) rational”.
The main upshot of this chapter, therefore, beside showing the inef-
fectiveness of some versions of both liberal and conservative views vis-
à-vis the skeptical challenge, is the realization that an epistemic warrant
for “There is an external world” is extremely hard to come by. This, I
take it, is a strong motivation to start taking the moderate position seri-
ously, inasmuch as it looks very likely to offer us the best prospect for
vindicating, within an internalist framework, the intuition that we do
have perceptual warrants for ordinary empirical propositions, such as
“Here is a hand” when we happen to have a hand-like visual experience.
For, on the one hand, it avoids the problem of surpassing our cognitive
locality that besets the liberal position, and on the other hand, it avoids
the requirement that there should be an apparently unattainable inde-
pendent warrant for “There is an external world” that leads conservative
views to failure. Therefore, it has all the credentials to qualify as the best
possible explanation of how we can have the perceptual warrants we
think we have.
3
The Bearing of the Moderate View:
Transmission Failures, Closure,
Easy Knowledge, and Bootstrapping
Consider the following arguments:
ZEBRA
(I) Here is a zebra;

(II) If this is a zebra, then it is not a cleverly disguised mule;
(III) This is not a cleverly disguised mule.
MOORE
(I) Here is a hand;

(II) If there is a hand here, then there is an external world;
(III) There is an external world.
OTHER MINDS
(I) Here is a person who is in pain;

(II) If there is a person in pain here then other minds exist;
(III) Other minds exist.
PAST
(I) The Earth has existed for a very long time (based on scientific and
testimonial evidence);
(II) If the Earth has existed for a very long time, there is a long-term
past;
(III) There is a long-term past.
86
The Bearing of the Moderate View 87
TESTIMONY
(I) Socrates died by drinking hemlock (based on all testimonial

evidence about that event at our disposal);
(II) If Socrates died that way, all our testimonial evidence (about that
event) at our disposal is reliable;
(III) All our testimonial evidence (about that event) at our disposal is
reliable.
RED TABLE
(1) The table is red (known by means of visual perception);

(2) My visual perception produced the belief that the table is red (known
through introspection);
(3) My visual perception produced a true belief that the table is red (1, 2);
(4) Repeat;
(5) My visual perception is reliable.
Intuitively there is something dubious in all these cases. But what exactly,
if anything? Diagnoses of problem(s) vary considerably. As is well known,
Fred Dretske would say that ZEBRA and MOORE, for instance, are cases in
which the principle of closure of epistemic operators under known entail-
ment fails. Crispin Wright, in contrast, maintains that these arguments
are affected by transmission failure. In order for the premises to be justi-
fied an independent justification for the conclusion is needed. Therefore,
these arguments cannot provide a first justification (nor enhance one’s
previous justification).1 Jim Pryor and others, in contrast, have claimed
that, at least in cases such as MOORE, the problem is not epistemic, but
dialectical. That is to say, the argument is cogent – it does provide justi-
fication for its conclusion – but it is dialectically ineffective – that is,
unable to convince someone who is doubtful of the conclusion. As for
RED TABLE, theorists like Stewart Cohen and Johnathan Vogel would argue
that it exhibits an illicit form of bootstrapping, while others, like Michael
Bergmann, would say that it is merely dialectically ineffective.
As this brief recognition makes apparent, what is at issue is of utmost
importance from an epistemological point of view. For it makes a huge
difference whether an argument is cogent or not, or merely dialectically
ineffective when propounded in certain contexts. In the former case,
its rationality is impugned; in the latter, in contrast, only its contextual
propriety is.
This chapter addresses these issues, starting with a presentation of
Wright’s and Martin Davies’s early views on arguments such as ZEBRA
and MOORE.2 We then consider Martin Davies’ recent change of view,3

according to which there is more than one kind of transmission
failure, beside the one he and Wright originally exposed. More specifi-
cally, Davies thinks that a second kind of transmission failure has to
be countenanced. He connects both kinds of failure of transmission
of warrant with two different kinds of epistemic project and with the
exploration of whether the current dispute on the nature of perceptual
warrant between conservatives, such as Wright, and liberals, such as
Jim Pryor, has a bearing on them. We will leave the latter issue to one
side,4 since we have already dealt with it in the first chapter. Here I
wish merely to point out why Davies’s second kind of transmission
failure is indeed no such thing (§2). I then move on to canvass another
kind of transmission failure, different from the one studied by both
Wright and Davies (§3) and dependent on the moderate conception of
the structure of empirical warrants. I investigate its bearing on Moore’s
proof (§4), its relationship with Wright’s kind of transmission failure
(§5), and with the closure principle (§6). I then connect it with the
issues of easy knowledge and bootstrapping arguments (§7), and, in
closing, I defend it from criticisms that can be elicited from Pryor’s
recent work (§8).5 By so doing, the case for the moderate view is rein-
forced, for the moderate conception will be shown to have fruitful
applications with respect to several important philosophical issues,
including the cogency of arguments (such as the ones reviewed), the
failure of the principle of closure, and the analysis of easy knowledge
and bootstrapping arguments.
1 Wright’s and Davies’s original failure
As we saw in previous chapters, there is, nowadays, a considerable debate

in epistemology6 concerning whether Moore’s proof of an external
world – (I) Here is a hand; (II) If there is a hand here, then there is an
external world; (III) There is an external world – exhibits what Wright
first called “failure of transmission of warrant”. To repeat, Wright’s idea
is that Moore’s proof is one of a range of arguments which, while valid
and, at least in a normal context, proceeding from warranted premises,
is nevertheless powerless to produce a first warrant (or to enhance one’s
previous warrant) to believe their conclusions. For prior possession of
warrant for the latter is needed for possession of warrant for (at least one
of) their premises in the first place. The phenomenon is often illustrated
by means of the following example:7

(II) If this is a zebra, then it is not a cleverly disguised mule;
The idea is that normal perceptual experience of a zebra in front of

one can provide a warrant for (I) only if one is already warranted in
supposing (III); that one is not, in fact, perceptually interacting with a
cleverly disguised mule or in any other way dealing with an environ-
ment in which visual appearances are grossly misleading. Thus, if one
possesses a perceptual warrant for (I) – as is normal when going to a zoo
and seeing what looks like a zebra in a pen – one will also, as required by
closure of warrant across entailment (which Wright does not question),
possess a warrant for (III).8 Still, in Wright’s view, one cannot acquire a
first warrant to believe (III) (or add to one’s pre-existing reasons to believe
it) by running such an argument. For, to repeat, one’s perceptual warrant
for (I) depends not just on the content of one’s current sensory experi-
ence – which could be indistinguishable, at least in principle, even if one
were indeed interacting with a cleverly disguised mule – but also on the
collateral information that, among other things, one is not suffering any
form of gross deception, in particular not perceiving a cleverly disguised
mule. That information, in turn, is rationally accepted just in case one
has some form of independent warrant. Hence, an antecedent warrant for
(III) is needed, in Wright’s view, for one’s sense experience to provide an
(albeit defeasible) warrant for (I).9 If so, however, the argument is epis-
temically circular (or question-begging) and cannot, therefore, produce a
first (or enhance one’s previous) warrant to believe in its conclusion.10
As we saw in previous chapters, this argument draws merely on the
following conceptual resource: it takes theory to infer from a phenom-
enon, such as a perceptual experience, to its likely11 cause. For its content is
compatible, at least in principle, with its being produced in many different
ways – dreams, holograms, appropriate stimulations of one’s brain, and
so on. That motivates the thought that more is required than just one’s
current sense experience, in order to have (a defeasible) perceptual warrant
for ordinary empirical beliefs such as (I). That it might seem otherwise,
given the phenomenology of our usual ways of forming warranted beliefs
about ordinary physical objects in our surroundings, is a mere byproduct
of the fact that the actual architecture of perceptual warrant remains,
most of the time, implicit. Of course, to say this doesn’t mean to side with
Wright. For it may still be disputed, in various ways, as we saw in the first
chapter, that the structure of warrants is as he thinks it is.
For a long time Davies shared Wright’s understanding of transmission

failure, his diagnosis of the failure of Moore’s proof, and other arguments
which seemingly exhibit the same kind of structure, such as McKinsey’s
argument against the compatibility between self-knowledge and
semantic externalism, and Putnam’s argument against the (epistemic)
possibility that we might be brains in a vat. Recently, however, reacting
to Pryor’s defense of the liberal view and his attendant diagnosis of the
failure of Moore’s proof as due to a mere dialectical shortcoming, Davies
has changed his view on transmission failure as well. More precisely, he
has urged that a second kind of failure of transmission of warrant should
be countenanced.
2 Davies’s new failure and its failure
Davies introduces his second notion of transmission failure in the

context of a quite elaborate discussion of how he thinks the Wright–
Pryor dispute should be understood. Rightly, in my opinion, he points
out that that dispute concerns the structure of propositional warrants for
ordinary empirical beliefs, such as (I) “Here is a hand”. To repeat a point
already stressed in the first chapter, propositional warrants are those
warrants there are for a particular proposition, which are independent
of a subject’s actual beliefs and of how he conceives of those warrants.
Hence, they must be contrasted both with doxastic warrants (those prop-
ositional warrants which attach or fail to attach to a subject’s actual
beliefs), and with those warrants a subject takes himself to have given his
collateral beliefs, which are rationally available to him.12 To illustrate:
if one has a perception of a hand in front of one (and, depending on
one’s theory of perceptual warrant, there is also an independent warrant
for (III), that there is an external world), one would also have a propo-
sitional warrant for the belief “Here is a hand”, even if one does not in
fact form that belief. If, in contrast, one does form that belief, then one
will have a doxastic warrant for it. Finally, a subject could have collateral
beliefs that prevent him from availing himself of the warrant he has for
the belief “Here is my hand”. For instance, he may think he has been
given a pill that causes him to hallucinate objects in his surroundings. In
this case, he cannot rationally avail himself of that warrant he neverthe-
less has for his belief.
As we have already repeatedly seen, in Wright’s view that warrant is
afforded not just by one’s current sense experience with the represen-
tational content as of a hand where one seems to see it, but also by
the collateral assumption (III) that there is an external world, which,
in turn, needs to be warranted to be rationally held. As is by now

familiar, Pryor denies this: he thinks that one’s current sense experi-
ence, with its particular representational (and phenomenal) content,
is enough to give one a – to be sure defeasible – warrant for (I), without
the need of any prior warrant for (III), provided one has no (reason to)
doubt about it. According to Wright, Moore’s proof then fails because
of its patent circularity – because warrant for (III) is already needed
to have the perceptual warrant for (I), which, via the entailment,
would give one a warrant for (III) – while no charge of circularity can
be sustained against it on Pryor’s understanding of the structure of
perceptual warrant. In his view, the proof fails simply because it is
usually advanced against a skeptic, who, according to Pryor, thinks
(III) is (likely to be) false. However, surely, if one thinks that (III) is
(likely) false, one cannot take one’s current sense experience as a
warrant for (I). Hence, when Moore’s proof is presented against such
an opponent, it will fail to convince him because, given his collateral
beliefs about (III), he will be bound to think that the proof starts with
an unwarranted premise. Thus, on Pryor’s view, the proof fails merely
for dialectical reasons.
Now, Davies thinks that Pryor’s notion of dialectical failure is in fact
a second kind of failure of transmission of warrant. For, in his view, it is
important to recognize that Pryor’s allegedly dialectical failure actually
occurs within a specific kind of epistemic project: one in which one is
trying to “settle the question” as to whether there is an external world,
against an opponent who thinks it is (likely) false there is.13 In such a
project what is at issue is not the structure of propositional warrants,
but, rather, which warrants, among propositional ones, one can appro-
priate given one’s own experiences and collateral beliefs.
If so, no matter whether the structure of propositional warrants is as
Wright thinks it is, or as Pryor thinks of it, one’s antecedent disbelief
in (III) will make one’s (putative) perceptual propositional warrant for
(I) rationally unavailable. Hence, in Davies’s view, one’s collateral beliefs
prevent the perceptual propositional warrant for (I) being transmitted
to (III).
Now, is Davies’s alleged second kind of transmission failure really a
case of transmission failure? A moment of reflection suffices to return
a negative answer to this question. For, trivially, there seems to be no
reason to deny that the perceptual propositional warrant does transmit
(in Pryor’s view of the matter), but remains rationally out of reach. The
problem is not one of failure of transmission of a propositional percep-
tual warrant for (I), but rather of the rational unavailability14 to the given
doubting subject of that warrant, in light of his collateral beliefs. To

clarify: on Pryor’s understanding of the structure of empirical warrants,
one’s experience as of a hand would give one a propositional warrant
to believe (I), which would transmit to (III). Still, given one’s collateral
antecedent belief that it is (likely) false that there is an external world,
one could not rationally take oneself to have that warrant. However,
on this latter understanding of the notion of warrant, there would be
no (rationally available) warrant for (I) in the first place. Hence, there
could be no failure of transmission of warrant, for that presupposes the
existence of a warrant for the premises that somehow cannot reach the
conclusion of a logically valid argument that proceeds from it. Therefore,
whatever phenomenon Davies is trying to draw attention to, I suggest
it has to be conceived of as something other than a kind of failure of
transmission of warrant.15
3 Another kind of failure
Despite the failure of Davies’s second kind of failure, I do think that

there are other ways in which an argument can exhibit transmis-
sion failure and be epistemically circular and, for this reason, question-
begging. Here, I wish to canvass one such further kind of transmission
failure, leaving the question of whether there are more to further
investigation.
For this second kind of transmission failure, let us go back to Wright’s
(and early Davies’s) kind of transmission failure. As we saw, the thought
was that,
Thesis 1: a certain experience with a given representational (and

phenomenal) content would not be enough to give one a (defeasible)
warrant for a specific empirical belief such as (I) “Here is a hand”.
Hence, according to Wright (and early Davies),
Thesis 2: in order for that experience to be a warrant for (I), some

extra information16 must be assumed, in particular that there is an
external world (with which one is causally interacting and which is
mostly correctly revealed to one through sense experience).
However, according to Wright (and early Davies),
Thesis 3: that information can rationally be assumed only if one has

warrant for it.
The circularity, then, in Wright’s (and early Davies’s) account of the

matter, depends on Thesis 3. For warrant for (III) – that there is an external
world – would already be needed in order to have warrant for (I) – “Here
is a hand”. Hence, the perceptual warrant one would have for (I) could
never give one a first warrant to believe (III). Therefore,
Transmission failure 1 (TF1): An argument fails to transmit warrant

from its premises to the conclusion (in way 1) if and only if warrant
for the conclusion is already needed in order to have warrant for its
premises in the first place.
I think, however, that Thesis 2 would already be enough to generate a –

of course, different – kind of transmission failure. The simple thought
behind this claim is this: how could one acquire a warrant to believe
a conclusion of an argument if the mere information contained in the
conclusion were already needed – in the sense of having to be assumed –
in order for the warrant for the premises to exist at all, in the first place?
Let me make this idea more vivid by considering proofs in the math-
ematical case, which were what originated Wright’s ideas on transmis-
sion failure at the outset. As will become apparent, this analogy is not
straightforward. Still, I think it may help to get our intuitions going
for it is beyond the scope of this chapter to try to defend it in detail.
Mathematical proofs – think, for convenience, of proofs in Euclidean
geometry – are generated within mathematical theories and these, in
turn, depend for their existence on certain axioms, that is, on certain
propositions that are assumed without proof.17 Now, why can axioms
not be proved, or be proved only circularly? Clearly, because they are the
primitive pieces of information on which any proof within that theory
depends. What this means is that they are needed to generate warrant
for any other mathematical proposition within a given theory. Put in
mathematical terms, axioms are needed in order to prove certain prop-
ositions that, once proved, count as that theory’s theorems. Thus, the
reason why axioms cannot be proved within a mathematical theory is
that they need to be assumed already to give any proof – so to generate
any kind of warrant – within the theory.
This therefore suggests that another way in which a failure of trans-
mission of warrant may occur is as follows:
Transmission failure 2 (TF2): An argument fails to transmit warrant

from its premises to the conclusion (in way 2) if and only if this very
conclusion needs to be assumed already in order to have warrant for
its premises in the first place.
The mathematical analogue of TF1 would, in contrast, be this: a given

proposition cannot be proved within a mathematical theory – and,
therefore, it cannot become a theorem of that theory – if proving it –
that is, generating warrant for it – depended, at some stage, on already
having a warrant for it. That is to say, if it depended on its already being
a theorem of the theory.18
One might then suggest that the case of mathematical axioms is
somehow spurious – in fact, a degenerate case of TF1 – for it is true
that axioms cannot be proved within the theory, still they are rationally
assumed because we have a priori, independent warrant for them. I am
skeptical of this view, I prefer to consider axioms in a mathematical theory
as stipulations, which, as such, cannot be independently warranted.19 True,
once these stipulations have been made, one may think of having an a
priori warrant to believe that, say, points (in Euclidean geometry) are
indivisible or that parallel lines will never intersect. However, even if we
had such a priori knowledge, it would come for free from a stipulation
which, as such, cannot be warranted. Furthermore, if a priori warrants
somehow certified – or at least made more probable – the truth of what
is warranted thereby, the vexed issue of the status of Euclid’s axiom
five – according to which there is only one parallel line to another line r,
which passes from point P – may make us suspicious of the fact that we
could really have such a kind of warrant. Hence, one may just stick to
the view that these axioms are stipulated and accepted without any kind
of warrant and, as such, used to generate warrant for propositions that
follow from them. If so, the attempt to prove them within the theory
they constitute would give rise to TF2, for one would try to prove them
by using propositions that are warranted only as long as those axioms
are assumed without there being a warrant for them.
This the point of the analogy with the mathematical case which I think
is instructive; yet clearly our aim is to extend this sort of consideration
to the case of assumptions such as “There is an external world”, which
are not mere stipulations but contingent truths. Let us therefore leave
the mathematical case aside and proceed with the task of accounting for
the status of Moore-like arguments.
4 Moderates, TF2, and Moore
Beside the mathematical case just reviewed, I think that reflections on

the very nature of perceptual warrants should also lead us to acknowl-
edge TF2. In particular, it is important to realize why moderate posi-
tions about the nature of perceptual warrant ought to return a negative
verdict on Moore’s proof as an attempt to provide warrant for our belief

in the existence of an external world. All moderates would say that the
conclusion of that argument – that there is an external world – must
already be assumed in order to have a perceptual defeasible warrant for
(I) – “Here is a hand” – in the first place. However, if perceptual warrant
depends for its existence on assuming (III), then an argument which,
proceeding from that warrant, aimed at warranting (III) itself, would in
fact presuppose the very piece of information it was supposed to provide
warrant for. Hence, an argument such as Moore’s would exhibit TF2.
Accordingly, it could not produce a warrant for its conclusion.
To help clarify this point, consider that it is the assumption of the
conclusion which allows one to enter the first premise of Moore’s proof,
so that, in effect, the structure of that proof would be akin to:
[Q]
P
P→Q
———
Q
It is worth noticing, moreover, that, as already remarked in Chapter 2,

if one were allowed to model present intuitions in terms of probability
calculus, the occurrence of a hand-like experience just by itself would actu-
ally increase the probability of (specific) skeptical hypotheses (provided
closure held), such as “I am currently dreaming of seeing a hand in front
of me”. If it does not increase the latter’s warrantedness, it is because we
are already, at least implicitly, assuming the negation of such a skeptical
hypothesis, and, in particular, of causally interacting with a world popu-
lated by physical objects. Yet, it is very difficult to see how one could
acquire warrant for these assumptions by going through something like
Moore’s proof. For it would be a very peculiar form of bootstrapping,
whereby warrants acquired by courtesy of certain assumptions would
straightforwardly produce warrants for those very assumptions! Therefore,
it seems that moderates had better allow for TF2 and connectedly return a
negative verdict on the cogency of arguments such as MOORE.
Furthermore, if one then reflects on the fact that warrant is here
conceived of in an internalist fashion, as something to which one
could (at least in principle) appeal to redeem or claim the rationality of
holding a certain belief, the kind of circularity involved would become
even clearer. For, if asked to claim our warrant for (III), we would end up
saying that assuming there is an external world, one is entitled to take
one’s sense experience as of a hand as a (defeasible) warrant for (I), and

that, via the entailment and holding closure, that gives one a warrant
for (III). Nevertheless, at that stage the obvious question to ask would
be what right one has to assume (III). This, however, is not a question
that Moore’s proof itself can answer for, from this viewpoint, that proof
can be run only if that assumption is already made. That is why I think
Moore’s proof would prove, once more, unable to provide a first warrant
to claim that accepting (III) is the rational thing to do. For, to repeat, one
could claim warrant for (I), and hence for (III), only by being already
warranted in claiming that assuming (III) is rationally legitimate.20
Therefore, we can see that, in general, in order for a proof to be
cogent – no matter in which domain it is propounded – warrant for its
premise(s) should neither depend, for its very existence, on the warrant-
edness of the conclusion (TF1), nor on the assumption of its conclusion
as such (TF2).
5 Transmission failures 1&2: a compatibilist approach
Let us now turn to the issue of the relationship between TF1 and TF2. In
my view, transmission failures 1 and 2 are compatible for whenever no
independent warrant for the conclusion of an argument can be provided and
to assume such a conclusion is nevertheless necessary for a certain body
of information to warrant its premises, TF2 occurs.21 In contrast, if inde-
pendent warrant for the conclusion can be acquired (and to assume it
is necessary for a certain body of information to warrant the premises),
TF1 takes place.
Hence, for instance, ZEBRA is an example of TF1:
(I) Here’s a zebra;

(II) If this is a zebra, this is not a cleverly disguised mule;
For, as we saw in Chapter 1, we can get independent warrant for “This

is not a cleverly disguised mule” – independent, that is, of the specific
sensory experience which gives us warrant for (I), which should then
transmit to (III) via ZEBRA. For instance, we could run a DNA test that
would exclude that the animal in front of us was a mule in disguise,
even though it could not tell us whether it was a zebra.22,23 In contrast, I
think we could not get an independent warrant for “There is an external
world”. For such a conclusion is much more general than the one in
ZEBRA. Thus, it would make no difference to the case if, instead of (I)
in MOORE, we had (I*) “Here is a foot” (or any other kind of object).
Hence, any kind of perceptual warrant bearing on a proposition like
(I) would be in the same boat as the one for (I) itself. Furthermore, as
argued in previous chapters, there is no prospect of getting any other
kind of warrant for (III) because that would commit us to implausible
positions, in my view. That is to say, either a traditional conservative
view, according to which there should be a priori warrant for proposi-
tions such as “There is an external world”; or else, to Wright’s position
that countenances non evidential warrants – entitlements – for such a
presupposition.24
In general, therefore, what is to be expected is that TF2 will occur
whenever putative arguments designed to confer warrant on very general
propositions are put forward. That is to say, TF2 affects those arguments
that proceed from a premise warranted on the basis of one’s everyday
experience, which entails a conclusion about, for instance (beside the
existence of an external world), the fact that our sense organs are mostly
working reliably, that we are not victims of lucid and sustained dreams,
that there are other minds, that there is a past, that there are uniformi-
ties of nature, and possibly some more. Characteristically, in all these
cases the conclusion of the argument must already be assumed in order
to have warrant for its premises in the first place. Further investigation
will reveal whether other kinds of argument could exhibit TF2 beside
the ones just mentioned. In other cases, where independent warrant for
a conclusion entailed by a logically valid argument (and needed in order
to have warrant for the premises in the first place) can be attained, the
diagnosis of the problem will have to appeal to TF1.
Consider, for instance, the following argument for the existence of
other minds. I see someone moaning and crying, while bleeding, so I
reason as follows:
OTHER MINDS
(I) Here is a person in pain;

(II) If there is a person in pain here then other minds exist;
(III) Other minds exist.
Now, it seems that in order to have a warrant for the first premise it is not
enough merely to take my experience at face value and that the assump-
tion of the conclusion is also needed. Again, the assumption is needed in
order to bring my current experience to bear on a belief about a person,
and not to make my warrant indefeasible. However, it is an assumption
for which it is difficult to see how one could have independent warrant.
For that warrant cannot be empirical if OTHER MINDS is defective. Nor is

it easy to see how it could be a priori. Arguably, the notion of a person
entails the idea that persons have mental states. Yet, merely by having
the experience just described one cannot have an immediate warrant for
“There is a person there”. Hence, one may have an a priori warrant that
if someone is a person, then they have mental states, but one cannot
have an a priori warrant to believe that what is causing the relevant
kind of experience, in this scenario, is indeed a person. Finally, if one is
suspicious of Wright-style entitlements, one would not have an entitle-
ment for “There are other minds” either.25 Hence, OTHER MINDS seems to
fit the template for TF2. For the assumption of the conclusion is needed
to have warrant for the premises in the first place, though it cannot be
independently warranted. Yet, given TF2, by running that argument I
would not acquire any warrant to believe its conclusion.
Let us now investigate if similar considerations might hold for Moore-
style arguments purportedly designed to confer warrant to propositions
such as “The Earth has existed for a very long time” or “People are gener-
ally reliable”.
Suppose you run a test on a fossil and the test indicates that it is one
billion years old. You now reason as follows:
EARTH
(I) Here is a one billion years old fossil;

(II) If this fossil is one billion years old, the Earth has existed for a very
long time;
(III) The Earth has existed for a very long time.
Here again it seems that one has warrant for (I) just in case the very long
existence of the Earth is assumed. For the evidence you might have for
(I), such as chemical tests, is compatible with the fact that everything
was created a relatively short time before and made to look as if it were
very old. It is not very probable, but the assumption that the Earth has
existed for a very long time seems to be needed in order for our geolog-
ical inquiry to be possible at all. Of course, it might turn out that the
test on the fossil has not been carried out appropriately and that, there-
fore, it is not exactly one billion years old. But, to reiterate, our present
scientific investigation seems to rest on the assumption that the Earth
has existed for a very long time. Now, can we warrant it independently
of EARTH? It is difficult to see how. For arguments based on testimony
would display a structure similar to EARTH and would thus presuppose
(III) in order to have warrant for their respective first premises. A priori
arguments on the very notion of Earth, in contrast, would presumably

draw out what we mean by “Earth” but would not establish that what
we believe about it is (likely to be) true. Finally, entitlements would be
problematic for reasons already largely explored. Hence, I conclude that
it seems quite likely that EARTH is also a case of TF2.
One might wonder whether the assumption underlying our warrant
for (I) in EARTH is (III) or, rather, something like (III*) “We are not being
deceived in such a way that everything looks to us as if very old, while
in fact it is not”. I am actually open to this suggestion, but notice that
either it would be insufficient to turn EARTH into a cogent argument,
or else it would lead to a slightly different example of TF2 from EARTH.
For one might think that the Earth is precisely an instance of things we
ought not to be deceived by in only seeming to be very old. Moreover, it
is only because we are not being deceived about it that something like a
fossil, which is part of its surface, can generally be taken to be very old.
Then our best scientific procedures will help us determine how old it
actually is (at some approximation, of course). Alternatively, one might
think that the following is affected by TF2,
PAST
(I) The Earth has existed for a very long time (based on scientific and
testimonial evidence);
(II) If the Earth has existed for a very long time, there is a long-term past;
(III) There is a long-term past.
For the existence of a long-term past seems to be a necessary assumption

in order to bring our scientific and testimonial evidence to bear on the
long existence of the Earth. It does not seem to be empirically warrant-
able, since arguments to that effect would exhibit the same structure as
PAST. Nor does it seem to be warrantable either a priori or by means of
entitlements. Hence, PAST would be a case of TF2.
Let us now turn to the following argument, based on past evidence,
UNIFORMITY
(I) The sun will rise tomorrow (based on memory and inductive
inference);
(II) If the sun rises tomorrow, then what has constantly happened in
the past will repeat in the future;
(III) What has constantly happened in the past will repeat in the
future.
Here again there is something intuitively wrong with the argument.

For what allows one to bring one’s past evidence to bear on (I) is precisely
the assumption of (III). However, such an assumption does not seem to
be empirically warrantable for arguments designed to that effect would
exhibit the same kind of structure as UNIFORMITY. Nor is it easy to see
how one can provide a priori warrants for (III). The usual worries with
respect to entitlements hold in this case too. Hence, UNIFORMITY seems to
be another case of TF2.
Finally, I would like to consider the following argument,
TESTIMONY
(I) Socrates died by drinking hemlock (based on all the testimonial

evidence about that event at our disposal);
(II) If Socrates died that way, all our testimonial evidence (about that
event) at our disposal is reliable;
(III) All our testimonial evidence (about that event) at our disposal is
reliable.
Again, there seems to be something amiss with this argument. After all,
we got (I) thanks to all our testimonial evidence, but can we ascertain
it independently? Clearly not. For TESTIMONY cannot give us a warrant
to believe (III). Yet, we cannot warrant (III) independently since we
cannot go back and check, as it were. Nor can we warrant it a priori
or through entitlements. Thus, no independent warrant for (III) is
possible and yet assuming (III) seems to be necessary in order to have
warrant for (I) in the first place. Therefore, TESTIMONY also seems to be
an example of TF2.
We can now stop assembling examples and leave it to further inves-
tigation to see if other arguments, structurally similar to the ones just
considered, exhibit TF2. Characteristically, in all these cases the conclu-
sion of the argument must already be assumed in order to have warrant
for its premises in the first place. In contrast, as we saw with ZEBRA, where
independent warrant for a conclusion entailed by a logically valid argu-
ment (and needed in order to have warrant for the premises in the first
place) can be attained, the diagnosis of the problem will have to appeal to
TF1. Yet, clearly, TF2 is a more philosophically interesting kind of trans-
mission failure. For, after all, in philosophy, we are not that interested
in zebras and cleverly disguised mules. Rather, we are mostly concerned
with understanding the relationship we have with our most deep-seated
commitments, encapsulated in the conclusions of MOORE, OTHER MINDS,
EARTH and/or PAST, UNIFORMITY and TESTIMONY. We will consider this issue
at length in Chapter 5.
6 Transmission failures 1&2 and closure
Let us now turn to the relationship between our two kinds of transmis-
sion failure and the principle of closure under known entailment for
epistemic operators such as warrant. As is well known, the details of the
formulation of this principle are tricky.26 Yet the following characteri-
zation will do for present purposes, where “warrant” is taken to mean
propositional warrant.
Closure principle (for warrant): If S has a warrant for P, S knows that

P entails Q, and competently deduces Q from P, then S has warrant
for Q.
Closure seems fundamental in allowing people to extend the scope of

their warranted beliefs by means of competent deductive reasoning.
For such a reason, it is almost universally considered as uncondition-
ally valid. Notable exceptions, although they were concerned with
closure for knowledge rather than warrant, are Robert Nozick and Fred
Dretske.27 Based on their preferred account of knowledge and on the
attempt to resist Cartesian skepticism, they argued that one may know,
for instance, that one has two hands and also that the existence of one’s
hands entails that one is not a brain in a vat, yet lack knowledge of the
consequent. The accounts of knowledge proposed in that connection
were in terms of truth-tracking and relevant alternatives. Roughly, on the
one hand knowledge would require that one’s beliefs tracked the truth,
while, on the other hand knowledge would require excluding only rele-
vant alternatives. Now, while “I have two hands” would track the truth,
“I am not a brain in a vat” would not, for one would believe it even if
one were in fact a brain in a vat. In a similar vein, while knowing “I have
two hands” requires excluding certain relevant alternatives, it would not
require excluding that one be a brain in a vat. Thus, one would know the
former, even if one were in no position to exclude the latter, thus failing
to know that one is not.
Connectedly, the denial of closure allowed Nozick and Dretske to
counter Cartesian skepticism by claiming that although we lack knowl-
edge of “I am not a brain in a vat”, this does not entail that we do not
have knowledge of humdrum truths, such as “I have two hands”. For,
failing closure, one cannot infer that since one has no knowledge of “I
am not a brain in a vat”, one does not have knowledge of “I have two
hands” either.
We will not dwell on the huge discussion their account of knowledge
has given rise to for, after all, we are not so much concerned with knowl-
edge as with warrants. What is important to notice is that closure merely
imposes a consistency requirement to the effect that if one has warrant
for P and knows that P entails Q, this prohibits the possibility that one
lacks warrant for Q. However, closure is silent on whether warrant for
Q is dependent on the reasoning that holding closure would lead to
the conclusion that Q is warranted. Therefore, for instance, one could
already have warrant for Q, say of a testimonial nature, and then have
perceptual warrant for P, know that P entails Q, and then acquire a new
warrant to believe Q. Yet, if one’s reasoning were in fact a case of TF1,
it would not produce any new warrant to believe Q, yet closure would
nonetheless hold by virtue of there being an independent warrant for Q.
To see that this is the case, consider ZEBRA

(II) If this is a zebra, this is not a cleverly disguised mule;
Let us take supporters of TF1 to be saying that that argument cannot

provide a first warrant to believe its conclusion because in order to have
warrant for (I) (III) must already be warranted in the first place. Since (III)
can be independently warranted by, for instance, running a DNA test
and supposing that such a test has excluded that that animal is a mule,
closure would hold.
Take MOORE and those who think it is a case of TF1, because warrant
for its first premise depends on there already being a warrant for its
conclusion. Unless they want to end up embracing skepticism, they
will have to say either that we have an independent a priori warrant
for its conclusion, or an entitlement for it – where an entitlement is a
non-evidential kind of warrant, but a warrant all the same. On these
accounts, MOORE would be an instance of TF1, yet closure would hold
in that case too.28
What about the relationship between TF2 and the closure principle?
On that view, things look worse for closure. For it is denied that assump-
tions such as (III) – that there is an external world – are in any way
independently warrantable. Moreover, it is a tenet of the moderate posi-
tion – spelled out along the lines presented so far – that one could not
acquire a warrant by running a valid argument which proceeds from
premises that, once those assumptions are made (and one has a hand-
like experience as well), are warranted.
I actually believe that the failure of closure for warrant is indeed a
consequence of embracing TF2 and a consequence one can live with
because of the limited number of cases in which one should favor the
moderate conception of warrant (for independent reasons) and thus
allow for TF2.29 Connectedly, we can now see that while ZEBRA exhibits
TF1 but no failure of closure, Moore’s proof exhibits TF2 and therefore
a failure of closure. Thus, it appears that we have reached an explana-
tion of why closure (for warrant) must fail, and does so only in certain
cases. That is, in all and only those cases where the assumption of the
conclusion is necessary in order to have warrant for the premises and
the conclusion cannot be warranted, evidentially or otherwise. Therefore,
it would be more correct to say that closure does not hold uncondition-
ally, rather than say that it fails tout court. However, for ease of exposi-
tion, I will go on talking of closure failure.
This – I think – is an important result for, as connoisseurs of Dretske’s
work will know, it is often unclear why closure for warrant should fail.
Indeed the motivation offered seems often to depend on a conflation
between failure of transmission – TF1, in fact – and failure of closure.
Wright, however, has long been concerned to show that TF1 and failure
of closure for epistemic warrant are two different phenomena and that
the former does not entail the latter. However, we can now see that
there is something to Dretske’s intuition. For failure of closure is indeed
entailed by transmission failure, but by TF2, not TF1.
Actually I think this result is acceptable. For closure would fail only
in a very limited number of cases and for principled reasons. Thus, it
does not have the disastrous consequences regarding the cogency and
epistemic utility of most of our inferences often denounced by those
who have been concerned to defend closure at all costs.30 Consider, for
instance, the following arguments:
FINGERS
(I) John has two hands;

(II) If John has two hands, he has ten fingers;
(III) John has ten fingers.
Suppose you are told (I) and have no reason to think that your informant
has been deceitful, or that John might not have normal hands. You will
have warrant for (I). Then, given your background knowledge of the
fact that each human hand has five fingers and a bit of arithmetic, you
will know (II). Thus, you will have warrant for (III). Hence, I do not
wish to dispute that competent deduction from warranted premises to a
conclusion logically entailed by them can help us augment our knowl-
edge or acquire warrants – as defeasible as that might be – for its conclu-
sion. However, notice that in FINGERS we can get independent warrant
for (III) and this is what secures that closure holds in this case too. For,
whether or not the argument is warrant transmitting – and on the face
of it, it seems so – the fact that, at least in the abstract space of reasons,
as it were, there is a warrant for (III), which is independent of FINGERS
itself, guarantees that if (I) is warranted and (II) is known, (III) is also
warranted.
In a similar vein, consider the following argument, where the first
premise is grounded on your reading the page number on the last page
of this book:
PAGES
(I) This book on the shelf has 120 pages;

(II) If this book on the shelf has 120 pages, it has more than 100
pages;
(III) This book on the shelf has more than 100 pages.
Here again closure holds because we can independently warrant (III).

It would hold whether or not PAGES is an example of TF1. One might
be tempted to think that if you want your perceptual evidence – for
example, reading the number 120 on the last page of the book – to
bear on (I) – “This book has 120 pages” – you may need a warranted
collateral information that this is not a defective copy in which only
the last few pages have been printed. Now, if someone maintained
that PAGES is a case of TF1, by going through it we could not acquire
a warrant to believe (III). Yet, closure would hold because, at least in
the abstract space of warrants, there is an independent warrant for
(III), to the effect that the principle that if P knowingly entails Q,
and one has warrant for P, then one has warrant for Q too, would be
respected.
Therefore, I think there are actually no reasons to be concerned with
the kind of failure of closure I have been embracing as a consequence of
TF2. For, to repeat, it would not preclude the holding of that principle on
most occasions. Namely, when we are engaging in everyday reasoning
either by simply drawing out the consequences of our warranted beliefs
(if TF1 occurs, like in ZEBRA and perhaps PAGES), or by gaining warrant for
the conclusion of arguments for whose premises we already possess a

warrant (if no TF1 takes place, like in FINGERS). Furthermore, I think we
should not be worried about these sporadic failures of closure, since TF2,
which entails them, is indeed a consequence of a view about the archi-
tecture of empirical warrants, once construed at its best, that seems to
me far more plausible than its rivals – that is, the moderate conception
of empirical warrant.31
Recently, however, Duncan Pritchard has provided a different strategy
to defend the universal validity of closure. Firstly, he presents a more
demanding version of the closure principle. According to which,
If S has rationally grounded knowledge that P, and S competently

deduces from P that Q, forming a belief that Q on that basis while
retaining her rationally grounded knowledge that P, then S has a
rationally grounded knowledge that Q. (Pritchard, forthcoming,
Chapter 4)
Secondly, Pritchard claims that the closure principle so understood has

to do with knowledge and knowledge entails belief. Hinges, however,
are not objects of belief and therefore they do not qualify as candidate
propositions to be substituted for Q. Consequently, closure does not fail
with respect to them, instead it is not applicable to them.
Notice, however, that Pritchard and I work with two different notions
of closure in mind and that mine is much less demanding than his. For it
talks of warrant, rather than knowledge, and it mentions no specific atti-
tudes subjects should have with respect to the propositions which can
be plugged into the schema. My own version of that principle imposes
merely a consistency requirement, as we have just seen.
Formulating closure is notoriously a hard task and I will not try to
settle the matter here. All I wish to point out is that my less demanding
formulation is consistent with any attitude one might have towards
the relevant Q (in particular), even one which is different from belief,
such as an assumption, in the committal sense I favor (see Chapter 1,
§3). It is also noncommittal with respect to the notion of warrant it
mentions, that is, it could be either evidential or non-evidential, like an
entitlement. For such a reason, it is fully compatible with the retention
of closure by the lights of a conservative of Wright’s ilk, in a way in
which Pritchard’s version is not. For, on that understanding of closure,
a Wrightian would have to say that closure does not apply since the
relevant Qs are neither the objects of belief nor, a fortiori, of knowledge.
Pritchard’s account of closure therefore “overshoots”. That is to say, it
declares closure inapplicable in cases where different theorists would say

that it is applicable and actually satisfiable and satisfied. This, in fact,
takes us to the general worry with Pritchard’s account. Namely, that by
endorsing such a demanding notion of the principle, it runs the risk
of ruling out, by definition, all cases which would be counterexamples
to its universal validity by banning them as cases in which the prin-
ciple would not apply. Therefore, I think that a weaker account, which
does not prejudge the issue of whether closure holds universally or not,
is preferable. It is then an open and substantial issue whether closure
holds unconditionally or not. Thus, I see no cogent reason to abandon
the weaker formulation of closure and, consequently, to forsake the
claim that a moderate will have to hold that, on such an understanding
of closure, it actually does not hold unconditionally.
Finally, we should briefly mention the challenge posed by Keith
DeRose,32 according to whom the denial of closure would yield what
he provocatively calls “abominable conjunctions”. Although DeRose
made his point with respect to knowledge, I think a similar charge
could be raised against those who deny that closure for warrant holds
(unconditionally). For the following conjunction is indeed a conse-
quence of the theory presented so far, namely, “I have warrant to
believe that I have a hand, but I do not have a warrant to believe
that there is an external world”. Indeed, there is no denying that this
conjunction sounds odd and even embarrassing, at least on the face of
it. Two replies to it are prominent in the recent literature, the first from
Ernest Sosa, while the second one has been put forward by Duncan
Pritchard.33
According to Sosa, it is one thing to clarify the conditions in which
we have knowledge (or warranted beliefs, in our case) and quite another
to explain the legitimacy of certain assertions. That is to say, there is no
reason to think that a correct account of knowledge (or warranted belief)
should also yield an account of appropriate knowledge ascriptions. For
Sosa, epistemology is concerned with the former issue, while the study
of the conditions in which it would be correct to make a knowledge
claim (or a claim to warranted belief) would pertain, at most, to the
philosophy of language.
It seems clear that this strategy should explain why it would sound
strange to make assertions that one’s theory predicts to be true. Moreover,
it would be embarrassing to confine oneself in a position whereby one
would possess a truth, which, however, could not be appropriately
asserted. Moreover, DeRose’s abominable conjunctions do not seem to
be less abominable if transposed at the level of judgment. Therefore, they

raise a deeper problem than the one Sosa seems to acknowledge with his
reply.
Pritchard, in contrast, has maintained that the assertion sounds weird
because it violates the Gricean maxim of quality, which prescribes to
assert only that for which one has justification (or even knowledge).
Now, if it is true that we do not have justification to believe that there is
an external world (or that we are not dreaming, or that our sense organs
are mostly reliable), then it is clear why it would be pragmatically infe-
licitous if we asserted it.
However, in this case too I think that the problem is harder than the
reply makes it seem. For the odd assertion is not “There is an external
world”, but “I have warrant to believe that I have a hand, but not to
believe that there is an external world”. In addition, this would be a true
assertion, by the lights of those who deny that closure holds (uncon-
ditionally), that they therefore think to know, or at least warrantedly
believe. Thus, it is a conjunction they are in a position to assert in
conformity with Gricean maxims. As before, moreover, the conjunc-
tion would be odd if it were merely entertained in thought rather than
asserted. Therefore, it is far from clear that we can explain what is going
wrong by reverting to the Gricean machinery that is supposed to govern
merely verbal communication.
I think we will only be in a position to address DeRose’s challenge
fully by the end of Chapter 4. For now, let me just anticipate that the
view I will eventually put forward will have the means to characterize
the second conjunct in such a way that it will not sound (that) strange
anymore, when taken in connection with the first one.
We can thus summarize as follows the outcome of our discussion so
far with respect to closure.
Liberals Conservatives Moderates
Experience + Experience+ Experience+

No reasons to doubt (III) W* (III) Ass (III)
W (I) P W (I) P W (I) P
W (II) P→Q W (II) P→Q W (II) P→Q
W (III) Q W/W* (III)Q (III) Q
(W= evidential warrant; W* = either a priori or non-evidential warrant, depending on one’s

variety of conservatism and whether or not one retains closure for W).
7 Transmission failures, easy knowledge, and

bootstrapping
As is well known, one notable consequence of the conservative architec-

ture of perceptual warrant and of TF1 is that arguments such as,
RED WALL
(I) Here is a red wall (based on one’s current sense experience);

(II) If this is a red wall, it is not a white wall bathed in red light;
(III) This is not a white wall bathed in red light;
will not be cogent. For they cannot produce a first warrant to believe their
conclusion, since warrant for it is needed in order to have a perceptual
warrant for (I) in the first place. By contrast, as is equally well known, on
the liberal conception of the structure of perceptual warrant, RED WALL
is cogent, for no warrant for (III) is needed in order to have warrant
for (I). On that account, RED WALL would be an example of easy knowl-
edge. Again, as is familiar, Pryor agrees that this sounds odd, but only
because – perhaps implicitly – we place that argument within a dialec-
tical setting whereby we imagine that there is an opponent who doubts
that (III) is the case. Given that collateral doubt, such an opponent will
not take (I) as perceptually justified on the basis of one’s current sense
experience. Therefore, the argument will obviously be dialectically inef-
fective against that opponent, but this is not enough to turn it into an
epistemically circular argument.
Now, given what we have seen regarding the compatibility of TF1
and TF2, it is important to notice that in the moderate conception of
the architecture of perceptual warrant there is no need to offer a third
account of what is amiss with arguments such as RED WALL. The reason
is precisely that (III) does not figure among those background assump-
tions that must be in place, in the moderate conception of perceptual
warrant as I have characterized it, in order to have that kind of warrant
for ordinary empirical beliefs such as (I). As we saw, the background
assumptions involved are only very general ones, like “There is an
external world”, “My sense organs are generally reliable”, “I am not
the victim of a lucid and sustained dream” (and possibly more). To
repeat, the assumptions at issue are merely those that are always needed
in order to bring one’s experience to bear on beliefs about physical,
material objects in one’s surroundings. Furthermore, it is for this kind
of assumption that, in my account, TF2 occurs. For these are the only
assumptions for which – I have argued – it is very difficult to see how
one could have an independent warrant. Connectedly, they are the

only ones that give rise to failures of closure in those arguments that
aim to provide evidential justification for them. However, (III) in RED
WALL is not of the same ilk. For we can get independent warrant for it,
for instance by means of testimony. Hence, I am inclined to consider
RED WALL an example of TF1. To be fully explicit, the reason why I do
not concur with Pryor’s diagnosis is that, even if one had no positive
doubt about (III) but were just open-minded with respect to it, perhaps
because the issue had never occurred to one, then the simple experi-
ence as of a red wall, internistically conceived in the way in which
Pryor recommends, would not be enough to provide one with a warrant
to believe (I). After all, as we saw in Chapter 2 in connection with argu-
ments such as MOORE, it would actually increase the probability of “This
is a white wall bathed in red light”.34 Therefore, to sum up, I submit
that RED WALL would be a case of TF1.
Now, let me turn to, so-called, bootstrapping arguments. These argu-
ments have originally been designed to bring out shortcomings in those
theories that allow for the possibility of basic knowledge.35 More specifi-
cally, although the liberal view is one such theory, for it allows that one
can have immediate justification for a lot of ordinary beliefs about mate-
rial objects in one’s surroundings based on one’s current sense experi-
ence (and this, together with the truth of what is believed and one’s
belief in it, would yield knowledge)36, the objection has usually been
directed against externalist forms of basic knowledge. Namely, those
which we may generally characterize as holding that S’s belief that P
amounts to knowledge if and only if it is reliably produced, where the
notion of reliability in question can be spelled out in various ways but
it never requires the appreciation on a subject’s part of the conditions
which make the process reliable. Nowadays, it is mostly explicated in
terms of safety: S’s belief that P is reliably produced if and only if, in
most nearby worlds, P is true, if S believes that P, as a result of R. Hence,
here is an example of a bootstrapping argument:
RED TABLE
(1) The table is red (known by means of visual perception);

(2) My visual perception produced the belief that the table is red (known
through introspection);
(3) My visual perception produced a true belief that the table is red (1, 2);
(4) Repeat;
(5) My visual perception is reliable.
The idea is that a supporter of basic knowledge is committed to RED

TABLE. However, since RED TABLE is clearly defective, so much the worse for
basic knowledge. In particular, it seems odd that one could gain knowl-
edge of the reliability of one’s sight by means of successive applications
of it.
Jesper Kallestrup recently maintained that the diagnosis of what is
wrong with arguments such as RED TABLE should be divorced from the
attack against basic knowledge.37 In his view, once we recognize that
they exhibit transmission failure, we can, on the one hand, diagnose
what is wrong with them, while leaving it open that basic knowledge,
in particular in the form of safety reliabilism, is correct. I am not sure
that basic knowledge would remain intact after exposing what is wrong
with bootstrapping arguments. Because, after all, the charge is that theo-
ries trafficking in basic knowledge entail that arguments such as RED
TABLE should turn out, implausibly, to be correct. Furthermore, any form
of transmission failure depends on thinking of the architecture of the
relevant kind of warrants (or of knowledge) as information dependent.
Therefore, the diagnosis of what goes wrong in bootstrapping arguments
would actually show that thinking of those warrants (or knowledge) as
basic was a mistake from the very beginning.
Now, supporters of basic knowledge could rejoin that such a diag-
nosis of what is wrong with bootstrapping arguments would beg the
question against them. Accordingly, nowadays they lean towards the
view that bootstrapping arguments are epistemically fine and, in partic-
ular, that whatever epistemic good one has for the premises transfers to
their conclusion. Yet, they are dialectically wanting if presented against
opponents who already doubt the conclusion. For they will not take
the premises to be warranted or known and therefore fail to form a
warranted or known belief in the conclusion.38
However, this reply is not very convincing. In particular, it seems the
essence of arguments that they should give those who are open-minded
with respect to their conclusions a reason to believe them. This is not
a dialectical property, but an epistemic one. That is to say, through an
argument we do not merely want dialectically to win our opponent to
our own point of view, by means of whatever dialectical trick. Rather, we
would like to provide reasons that would rationally compel any rational
subject to embrace the conclusion we are supporting by means of that
very argument. Another way of putting the same point is to consider
that the kind of epistemic arguments we are concerned with here could
be deployed in one’s mental arena, where there is no issue of dialec-
tical effects, in order to provide oneself with a good epistemic reason to
believe their conclusion. Such a project requires that one starts out as
open-minded with respect to the conclusion, at least pro tempore, even if
one were intimately convinced of its truth. Hence, bootstrapping argu-
ments are epistemically defective. A die-hard supporter of their epis-
temic propriety could insist that they are all right as long as they take
us from warranted or known premises to warranted or known conclu-
sions. That is fine, as far as it goes. Notice, however, that this shows
nothing about their capacity rationally to convince someone who was
open-minded about their conclusion. To think otherwise would amount
to conflating arguments which preserve closure of the epistemic oper-
ator under known entailment, with arguments that are transmissive of
that operator – that is, that are such to produce a warrant to believe, or
knowledge of their conclusions. Therefore, to come up with the diag-
nosis that bootstrapping arguments are examples of transmission failure
is not a way of begging the question against those who think they are
epistemically fine, because one would presuppose the falsity of theories
committed to basic knowledge. Rather, it amounts to offering the right
diagnosis of what kind of epistemic fault afflicts them qua arguments,
even if they retain some other epistemic virtue, such as closure preserva-
tion. Hence, the right attitude for a supporter of basic knowledge is to
acknowledge that bootstrapping arguments are epistemically defective
qua arguments. Consequently, he should either give up basic knowledge
or admit that he has nothing to say about why they are bad and yet this
is not a sufficient reason for him to abandon basic knowledge. After all,
all theories in this domain have their pitfalls!
Nevertheless, I will now assume that arguments like RED TABLE exhibit
transmission failure and what I will do is to try to determine what kind
of transmission failure they display.
In his discussion, Kallestrup focuses mainly on,
GAUGE
(6) The gauge reads “full” (formed by reading “full” on the gauge);
(7) The tank is full (6);
(8) The gauge is reading correctly (7, 8);
(9) Repeat;
(10) The gauge is reliable.
Hence, I will follow him, and his numbering, for ease of exposition.
Kallestrup rightly notices that the shorter bootstrapping arguments –
those from (1) to (3) in RED TABLE and from (6) to (8) in GAUGE – are already
problematic. In his view, their failing is epistemic rather than merely

dialectic. In particular, considering GAUGE, he writes (2011, p. 17):
If S either suspends judgment (not stemming from doubt) or ques-

tions (8), she is rationally prevented from believing (7). Such doubt or
indifference thus deprives S of the opportunity of coming to know (7)
on the basis of what the gauge reads. Put differently, unless S already
believes (8), S is rationally precluded from believing and hence
knowing (7) on the basis of the gauge’s deliverances.
He then goes on to notice that this failure is epistemic because “no argu-
ment is cogent unless the premises can be known while withholding
belief regarding its conclusion. Otherwise, the argumentative purpose
of overcoming prior doubt or deciding what to be believe would be
thwarted.” (ibid.)
Notice that this view of what goes wrong in some bootstrapping argu-
ments comports very well with TF2. For lack of a suspension of belief in
the conclusion – that is, the assumption of the conclusion – is taken to
be necessary (and sufficient together with one’s course of experience) to
have warrant for the premises, which cannot, thus, produce a warrant
for the conclusion. However, it turns out that Kallestrup wants to draw
a distinction between various bootstrapping arguments. For, according
to him, even if S believes (8), this does not make GAUGE a cogent argu-
ment. The argument is not cogent, in his view, because there can be
independent justification for its conclusion – for example, one might
measure the level of petrol in the tank directly –39 and such a justifica-
tion would be needed in order to go from (6) to (7) in an epistemically
acceptable way. This suggests that, given the kinds of bootstrapping argu-
ments Kallestrup actually considers, the reason why they fail is slightly
different from the one he offers. It may be phrased as follows:
Unless S already warrantedly believes (8) (or even knows it), S is ration-
ally precluded from believing and hence knowing (7) on the basis
of the gauge’s deliverances. But no argument is cogent unless the
premises can be known while having no warrant for its conclusion.
Thus, as a matter of fact, GAUGE is unsatisfactory because it is affected by

TF1 rather than TF2. Now, does this mean that TF2 has no role to play
in an account of what goes wrong in bootstrapping arguments? I do not
think so. To see why, consider that there are more general bootstrapping
arguments than GAUGE and RED TABLE. Consider, for instance:
P, Q, R, ... N
(1) P, Q, R, ... N (whatever belief based on the deliverances of one’s

overall perceptual system at time t);
(2) My overall perceptual system produced the beliefs P, Q, R, ... N (based
on introspection);
(3) My overall perceptual system produced the true beliefs P, Q, R, ... N
(1, 2);
(4) Repeat;
(5) My overall perceptual system is reliable.
RED TABLE*
(1) The table is red (based on one’s visual perception);
(2) My visual system produced the belief that the table is red (based on
introspection);
(3) My visual system produced the true belief that the table is red (1, 2);
(4) I am not now dreaming (3).
The difference is that P, Q, R, ... N and RED TABLE* are such that, contrary to
RED TABLE and GAUGE, no independent warrant for their respective conclusions
could be obtained.
This suggests that we should endorse a two-pronged diagnosis of what
is wrong with bootstrapping arguments. Kallestrup’s original explana-
tion, which pairs with TF2, holds in the following cases:
Any bootstrapping argument that has as a conclusion something

about the general reliability of our senses tout court, or about the
reliability of one of our sensory systems vis-à-vis skeptical scenarios,
would be epistemically circular and affected by Transmission-failure
2, since no independent warrant for its conclusion could be obtained.
In particular, acceptance of the conclusion would be needed in order
to have warrant for the premises in the first place, yet no argument is
cogent unless the premises can be known while being open-minded
regarding its conclusion.
In contrast, whenever a bootstrapping argument is such that inde-

pendent warrant for its conclusion could be obtained, and no warrant
could be obtained for the premises unless one had not only belief, but
also a warrant for the conclusion, its failure should be diagnosed as due
to TF1, rather than TF2. Accordingly, as we have anticipated, Kallestrup’s
explanation of what is wrong with the bootstrapping arguments he

actually considers, should be as follows:
Unless S already warrantedly believes (8) (or even knows it), S is ration-
ally precluded from believing and hence knowing (7) on the basis
of the gauge’s deliverances. But no argument is cogent unless the
premises can be known while having no warrant for its conclusion.
Hence, the distinction between TF1 and TF2 allows one to offer a more
nuanced and, to my mind, more convincing explanation of what goes
wrong in various kinds of bootstrapping arguments. Furthermore, TF2
allows one to properly diagnose what goes wrong in the most philo-
sophically interesting cases of bootstrapping argument, like P, Q, R, ... N
and RED TABLE*. After all, in philosophy we are not so much interested
in gauges and specific cognitive faculties as in being safeguarded against
the general unreliability of our senses and the obtaining of skeptical
scenarios. Therefore, as this and the previous sections have shown, the
distinction between TF1 and TF2 seems to be crucial for the explanation
of why different kinds of argument, whether or not of a bootstrapping
nature, are not cogent.
8 A possible counterexample: Pryor’s (?) failure to

see this failure
As is well known, Jim Pryor40 accepts TF1. Indeed, he allows that ZEBRA is
an example of it. He also accepts the principle of closure. However, as is
familiar, he maintains that Moore’s proof is not an example of transmis-
sion failure. For, in his view, one need not have any antecedent warrant
for the conclusion of that argument – that there is an external world – to
have a (of course defeasible) perceptual warrant for its premise, given
one’s current sense experience as of a hand, and that lack of doubt
suffices to that end. For, in effect, he actually denies Thesis 1 – that
is, that absent defeaters an appropriate perceptual experience is not
enough to justify the corresponding empirical belief – at least for what
he considers perceptually basic empirical beliefs, that is, beliefs that are
immediately licensed by the very content of one’s perceptions.41
What I wish to inquire into further is Pryor’s recent denial of the reality
of the phenomenon that I have called TF2, which occurs in the context of
a complex discussion of “When warrant transmits”, as his title reads.42 The
purpose of the discussion is to show that there are only two contenders:
liberals and conservatives. Moderates – not called such by Pryor – who
merely think that assuming (III) is needed to have a perceptual warrant for
(I) in the first place, and that this would suffice to bring about a second kind
of transmission failure, have no place in town.43 Pryor actually discusses
two possible positions that claim that Moore’s proof exhibits a kind of trans-
mission failure other than TF1. He calls them the “warrant-making factor
model” and the “anti-underminer model”, respectively. Accordingly, “if B
is among the factors that make W be a warrant for A, then your W-based
warrant for A cannot transmit to the claim that B obtains”; and “if B is an
anti-underminer for W, then your W-based warrant for A cannot transmit
to B”. According to Pryor, an anti-underminer hypothesis is one which is
not an underminer for one’s warrant for A and that entails the falsity of
some undermining hypothesis for one’s warrant for A. Hence, for instance,
“There is an external world” is an anti-underminer hypothesis for one’s
perceptual warrant for “Here is a hand”, for it undermines the hypothesis
that one might be dreaming right now, which, in its turn, would be an
underminer of one’s perceptual warrant for “Here is a hand”. Now, I take
it that the model that is more closely related to the moderate position as I
have characterized it and to the second kind of transmission failure I am
advocating is the second, while I think the first would tally better with,
presumably, some externalist views. For some of these positions hold that
since there is an external world, then one’s current sense experience as of
a hand gives one a defeasible warrant for an empirical belief such as “Here
is a hand”. Therefore, the obtaining of (III) is what makes one’s warrant for
(I) possible in the first place (and it would be a further question whether
such a warrant could or could not transmit to the conclusion of the
argument).44 That Pryor’s discussion is going to engage with moderatism
and TF2 seems further supported by his claim that the two models reflect
“notions of ‘taking for granted’ or ‘presupposing’ that we already [have]
an intuitive grasp of” – I would say one which is more externalist/realist
in spirit and another more internalist/anti-realist in fashion. Moreover, he
claims that “it can be natural to worry whether arguments can be effec-
tive whose conclusions are already ‘taken for granted’, or ‘presupposed’”,45
since that can seem like a covert form of question-begging.
Nevertheless,46 in what follows I wish merely to draw on one of Pryor’s
problem cases in order to consider a possible counterexample to my
view. For, in effect, he presents a little argument where, according to him,
moderates should say that warrant does not transmit – that TF2 occurs,
because assuming the conclusion is necessary in order to have warrant
for the premises – but where, intuitively, warrant does transmit from
the premises to the conclusion. This, in turn, would refute moderates,
at least insofar as they maintain a second notion of transmission failure
(but notice that it would then be a further issue whether they would
also be mistaken with respect to their conception of the architecture of
empirical warrants47). Here it is:
RAT
A guy working at a zoo says that he has just given all animals a potion
which makes them invisible and so he says “All the animals in the pen
are invisible”. His friend rebuts him with the following argument:
(1) [Occurrence of an experience as of rats];
(2) I see rats in the corner;
(3) There are some visible animals in the pen.
According to Pryor, a moderate should say that assuming (3) is needed

in order to have warrant for (2). Hence, that warrant cannot transmit to
(3), since that would be a case of TF2. Pryor, however, thinks it intuitive
that warrant does transmit, although assuming (3) would be needed in
order to have warrant for the premises in the first place. Therefore, RAT
would be a counterexample to, at least, TF2.48
In response to Pryor, I see three possibilities for a supporter of the
moderate position and of TF2. The first one is to say that (3) needs to
be assumed in order to have warrant for (2), especially in light of the
previous information, coming from the zoo guy, that he has given all
animals a potion that makes them invisible. For, given that information,
your visual experience as of rats is unimportant. After all, based on that
information, you could actually take your current visual experience as
a sign of your hallucinating rats or whatever, maybe because of a high
temperature, or some poisoned food you might have ingested, and so on.
Thus, you need to assume (3) in order to have an underminer for what
would otherwise undermine your perceptual warrant, namely the infor-
mation that all animals have been given a potion that makes them invis-
ible. If so, RAT would fail to transmit warrant because of TF2. Furthermore,
the impression – which I personally find it difficult to share – that it may
seem intuitive that RAT does not exhibit any kind of transmission failure
could be explained by saying that it would depend on assuming (3) and
on then implicitly reverting to the liberal conception of the architecture of
warrant. For the assumption of (3) undermines the information coming
from the zoo guy and then the liberal construction of warrant allows
one to have an immediate warrant for (2) which would then transmit to
(3). Of course, one would need to share Pryor’s sympathy for the liberal
construction of warrant in order to find this view plausible – a sympathy
that moderates obviously do not share.
The second possibility open to a supporter of the moderate view and

of TF2, which I favor, is to deny that (3) needs to be assumed in order
to have warrant for (2). In the moderate view as I have been charac-
terizing it, the kind of presuppositions that make perceptual warrant
for ordinary empirical beliefs possible are very general ones. That is to
say, that there is an external world, that one’s sense organs are usually
working reliably, that one is not victim of a lucid and sustained dream,
or suffering from forms of gross sense deception. To put it differently:
moderatism gives you a recipe for what “ingredients” are needed in order
to have ordinary perceptual warrants. Hence, the propositions assumed
must be extremely general ones and the role of their assumption is,
once more, that of making perceptual warrants for ordinary empirical
beliefs possible in general. This excludes the possibility that moderatism
should traffic in specific propositions for the role of collateral assump-
tions; or else that assuming them should only play a contingent role in
forming a specific empirical warrant. Thus, background assumptions
cannot be as specific as (3) – that there are some visible animals where
one is looking – for, in that way, it would never be legitimate to use
one’s current sense experience to acquire warrant that, through valid
inference and holding closure, would give one warrant for propositions
entailed by the ones for which one is perceptually warranted. On this
reading, RAT would not constitute a counterexample to TF2, because it
would not meet the conditions for the moderate position in the first
place.
Finally, another possibility is to say that RAT exhibits TF1. For inde-
pendent (of one’s current sense experience) warrant for “There are some
visible animals in the pen” can be acquired and (3) seems to be a neces-
sary piece of information in order to take one’s current sense experi-
ence as warrant for (2), given the further information coming from the
zoo guy that he has given all animals that particular potion. Hence, RAT
would fulfil the conditions for TF1, but not those for TF2 and would
obviously not be a counterexample to the latter.
Therefore, once the details of RAT are spelled out, at least in the
three directions I could think of, it seems that it would not constitute
a problem for the idea that there can be a second kind of transmis-
sion failure. In the case of perceptual warrants, such a kind of trans-
mission failure would pair with a moderate conception of the structure
of perceptual warrants, alternative both to Wright’s conservative view
and to Pryor’s liberal one. No doubt the possibility of counterexamples
deserves further investigation as, I think, does the notion of TF2 with its
attendant moderate conception of empirical warrants.
9 Summary
In this chapter, we have seen how a new kind of transmission failure,

different from Wright’s and early Davies’, should be countenanced as a
possible consequence of embracing the moderate architecture of empir-
ical warrant.
Transmission failure 2: An argument fails to transmit warrant from

its premises to the conclusion (in way 2) if and only if this very
conclusion needs already to be assumed in order to have warrant for
its premises in the first place.
We have used TF2 to diagnose what is wrong with arguments that are
intuitively odd, like MOORE, and philosophically significant. We have
pointed out how this kind of transmission failure is compatible with the
one recognized by Wright and early Davies,
Transmission failure 1: An argument fails to transmit warrant from

its premises to the conclusion (in way 1) if and only if this very
conclusion needs already to be warrantedly assumed in order to have
warrant for its premises in the first place.
as it applies to different arguments, that is, only to those for which it

is arguably impossible to provide an independent justification for their
conclusions. We have stressed that these are indeed the arguments that
are more interesting from a philosophical point of view. Hence, our kind
of transmission failure is more fundamental and philosophically more
relevant than the one discussed by Wright and early Davies. We have
explored its consequences with respect to the validity of the principle
of closure for epistemic operators under known entailment, and found
out that the latter does not hold unconditionally. We have also used
TF2 to diagnose what goes wrong in arguments that give rise to easy
knowledge and that display problematic forms of bootstrapping. Finally,
we have defended it from criticisms that can be elicited from Pryor’s
recent work. By so doing we have shown the considerable fruitfulness of
embracing the moderate view regarding the structure of perceptual justi-
fication. This lends support to that view or, at the very least, to the fact
that it should be properly taken into account as one of the significant
positions in the on-going debate about the architecture of perceptual
justifications.
4
The Extended Rationality View
In the previous chapter we saw, among other things, how moderatism

could embrace the idea that the principle of closure for warrant does not
hold unconditionally. In particular, it fails when the conclusion of the
argument is a very general proposition whose assumption is necessary
to have warrant for some more specific premises that figure in the argu-
ment. Now, Humean skepticism in the form of an argument based on
closure would therefore be blocked. For, even if there is no warrant for
the existence of an external world, it does not follow, in the moderate
conception of warrant, that we cannot have evidential justification for
an ordinary empirical proposition, such as “Here is my hand”. Indeed,
according to moderatism, given that assumption and an appropriate
course of experience, absent reasons for doubt, we would have a percep-
tual warrant for that proposition.
If so, it would appear that our job should be over. Alas, it is not. For
there is another more threatening challenge moderatism has to face
in conjunction with Humean skepticism about the existence of the
external world. The challenge, in a nutshell, is this: if we cannot have
epistemic warrants for those basic assumptions upon which ordinary
empirical justifications rest, these assumptions cannot be rational.
However, if they are not rational, they are arbitrary assumptions and, at
least in principle, on a par with any other assumption we could make
in their place.1 Thus, let us concede that justification always takes place
within a system of assumptions, as moderates would have it. Still, it is
entirely compatible with the outcome of moderatism that there could
be many such systems, all on a par, if not in practice at least conceiv-
ably (or virtually) as far as their different, yet incompatible assumptions
are concerned. Furthermore, each of these systems could then certify
as justified (and potentially known) very different, ordinary empirical
119
propositions. To illustrate with a ridiculous example that, however,

makes the problem quite vivid: suppose your spouse betrays you and
you believe it; suppose you believe in fortune telling and the fortune
teller tells you that your spouse has betrayed you; your true belief would
be justified and would thus amount to knowledge; while it would not,
if you did not to believe in fortune telling (absent other reasons to hold
it). Justification and knowledge would thus be situated – that is to say,
dependent on the system of assumptions which is accepted and from
which warrants derive. Therefore, the devastating consequence of this
kind of Humean skeptical challenge is that it seems to lead quite natu-
rally to epistemic relativism.
In this chapter, the moderate conception of warrant will be consid-
ered in connection with this threat. The challenge, as should be clear
from the formulation just given, involves two basic steps. Firstly, the
claim that since basic assumptions are not warranted – indeed precisely
because they are unwarrantable, as moderates hold – they are not epis-
temically rational. Secondly, the claim that just because they are not
epistemically rational, these assumptions are on a par with any other
possible assumption. Most of this chapter will be devoted to the first step
of this kind of Humean challenge.2 Hence, §1 shows why some defenses
of the moderate architecture of warrant, vis-à-vis this part of the skep-
tical challenge and inspired by Peter Strawson’s form of naturalism and
by pragmatism of a Jamesian kind, do not work. The former is claimed
to be the result of succumbing to that very challenge, while the latter
is found wanting because it provides pragmatic warrants for our basic
assumptions, whereas skeptics ask for epistemic ones. Wittgenstein’s
account in On Certainty of hinges as norms will also be briefly discussed
and criticized in §1 on the grounds that it imposes a normative concep-
tion of what seem to be entirely contingent empirical propositions.
The extended rationality view is thus proposed (§2), according to
which our basic assumptions, while neither warranted nor warrant-
able, are seen as constitutive of the very notion of epistemic rationality
shared by both skeptics and non-skeptics alike. If so, although Humean
skeptics are right to say that these assumptions are not warranted/able,
they are wrong to conclude that they are not epistemically rational.
For these basic assumptions are, in fact, part of our shared notion of
epistemic rationality, if only as its conditions of possibility. Hence, the
Humean skeptic’s mistake is diagnosed as due to an unduly restric-
tive notion of epistemic rationality, which comprises only warranted
beliefs, when in fact epistemic rationality extends to its very conditions
of possibility. Actually, if Humean skeptics concede that we may have
The Extended Rationality View 121
empirical warrants for ordinary empirical propositions by condoning

the moderate architecture of warrants, they too will be committed to the
extended rationality view. For if it is the case that epistemic rationality
comprises epistemic warrants – as these Humean skeptics are supposed
to concede – and epistemic warrants are possible only on the basis of
certain assumptions, which are part of epistemic rationality, they are
implicitly committed to the extended rationality view. The extended
rationality view is then contrasted with Wright’s recourse to the idea of
rational entitlements.
In §3 the issue of its consequences for Cartesian skepticism, the
closure principle and the problem of “abominable conjunctions” raised
by DeRose (see Chapter 3, §6) is taken up. In §4 the extended view is
defended against the charge of leading to epistemic relativism. That is to
say, to the view that there may be – either in practice or at least conceiv-
ably – several different and incompatible systems of justification, each
defined by adherence to different basic assumptions and determining
diverse, incompatible yet equally legitimate, notions of epistemic
rationality. In §5 the Oblomovian challenge consisting in claiming that
one could opt out of the “game” of epistemic rationality is addressed.
Finally, the relationship between the extended rationality view and anti-
realism about truth is briefly explored (§6), since that kind of metaphys-
ical outlook provides a natural pendent to the account of the moderate
structure of empirical justification presented in this book.
1 Naturalism, pragmatism, and Wittgenstein vis-à-vis

Humean skepticism
As I claimed in Chapter 1 (§4) moderates are in fact legion, though it

may be novel to call them that. In particular, I pointed out how moder-
atism can be seen as the outcome of certain forms of naturalism which,
in a Humean fashion, hold that our ordinary empirical beliefs are justi-
fied within a system of assumptions that it is natural for us to make,
which, however, are not in turn justified or, for that matter, justifiable.
Following Strawson (1985), we could think of this naturalism as either
due to our psychological constitution, in a Humean fashion, or, in a
more Wittgensteinian spirit, as due to our upbringing within a certain
community. In the former option, the belief in the existence of an
external world comes from some psychological connections that are
both involuntary and inevitable to us, given our cognitive structure. In
the latter, by contrast, such a belief would be inevitable because it is a
characteristic trait to our form of life. According to Strawson’s reading
of On Certainty, by being raised within our community we are drilled

to take for granted those beliefs that constitute the foundations of our
language games. Among them there is also a belief in the existence of an
external world, which plays a key role in our language games and epis-
temic practices, since it allows us to take our sensory experiences as due
to the causal interaction with material objects, to provide warrants for
ordinary beliefs about mid-size physical objects. In Strawson’s reading
of On Certainty Wittgenstein is proposing a kind of naturalism that we
could call of “second nature”, to use McDowell’s terminology in Mind
and World. That is to say, a naturalism not due to innate factors but
to our upbringing and acculturation within the human community, as
opposed to the kind of first nature naturalism Hume supported.
The naturalist response to the skeptical challenge usually consists in
pointing out that a skeptic asks for reasons and grounds where there
cannot be any. The doubts raised by a skeptic would therefore be
unnatural in that they go against what it is natural for us to believe (for
the reasons just seen). Furthermore, although these doubts would be
legitimate, as they are logically, metaphysically, and even epistemically
possible, they would be ineffectual. That is to say, they could not shake
our natural, somewhat instinctual faith in the existence of an external
world.
This reply, however, is not satisfactory. For it is agreed from the start
that skepticism cannot shake off our deep-seated convictions. Yet this
does not show that we are nevertheless epistemically within our rights
in making certain characteristic assumptions and that they are not
somewhat arbitrary.3 That is to say, none of this shows that we are not
making assumptions which other forms of life could not make while
still being epistemically rational. After all, it is not mere chance that natu-
ralism is what Hume himself ended up embracing after developing his
skeptical arguments. That is to say, he did not see it as a response to his
own skepticism, but as the only available option once no warrant could
be provided in favor of some of our deep-seated beliefs. He supported his
form of naturalism even though embracing it clearly provoked a tension
at the reflective level between reason, which leads to skepticism, and
nature, which forces us to make ungrounded assumptions. For, in his
view, skepticism correctly showed that we cannot have either empirical
or a priori warrants for “There is an external world” (or any other kind
of warrant). If so, we can only conclude that at the foundation of all our
allegedly justified, even known, beliefs there are a-rational assumptions.
That is to say, assumptions we make because of our cognitive structure,
with no justification for them. Hence, if the aim is to find rationally
secure grounds for our epistemic practices, it should be concluded that it

is unattainable, for the foundations are only a-rational. Thus, naturalism
of a Humean kind could hardly be a solution to the skeptical problem.
Yet, naturalism of a Wittgensteinian brand would not be a solution
either. For it would just be a notional variation on its Humean ancestor.
A variation in which, instead of appealing to our first nature, we appeal
to our second nature. That is to say, we appeal to our upbringing within
a community that shares certain assumptions – a certain Weltbild – char-
acteristic of a common form of life.
In Chapter 1 (§4), I also claimed that moderatism can be seen as a tenet
of certain pragmatist positions, according to which we have pragmatic
warrants for the assumption that there is an external world. While it is
difficult, from a historical point of view, to pinpoint exactly who, among
the various pragmatists, held such a view, the general idea would be that
that assumption is necessary for us to engage in our epistemic practice
of acquiring, assessing, and withdrawing from ordinary empirical beliefs
in light of our sensory experiences.4 That practice, in its turn, is indis-
pensable to us. Therefore, we are practically justified in making it. Yet,
on these pragmatist readings, we would have no epistemic warrant for
such an assumption – that is, we would have no warrant that speaks to
its likely truth. Hence, even if the kind of practice that is indispensable
to us is an epistemic one, committing to its assumptions for the sake of
having that practice would not provide an epistemic warrant for them.
Like the forms of naturalism we encountered at the onset of this
section, this kind of pragmatism would be unsatisfactory when taken in
connection with skepticism. After all, it would actually be giving in to
it if it acknowledged that we do not have epistemic warrants for these
basic assumptions and that we actually fail to be epistemically rational
by holding them, even though we may be perfectly practically rational
in doing so. Furthermore, it does not offer a diagnosis of the mistake
skeptics are possibly making by asking for epistemic warrants for our
basic assumptions. Hence, it does not look promising to develop moder-
atism along pragmatist lines.
Let me now dwell a bit on Wittgenstein’s position as presented in
On Certainty. For, as is well known, there is no consensus on Strawson’s
naturalist reading of it. Two further interpretations have been espe-
cially influential lately: the, so-called, framework reading, proposed
in various fashions by, among others, Marie McGinn (1989), Danièle
Moyal-Sharrock (2004) and myself (2010a); and the kind of epistemic
reading proposed by Crispin Wright (2004a) and Michael Williams
(2004a, 2004b).
Starting in reverse order let me immediately get rid of the latter. For,
by seeing Wittgenstein as supporting the view that we are non-eviden-
tially entitled to our basic assumptions because we cannot investigate
everything and thus have to rely on an assumption, they are obviously
going to rally Wittgenstein to their own conservative cause. I have dwelt
extensively elsewhere5 on the details of why I think this interpretation
is misleading as a construal of Wittgenstein’s notes, notwithstanding
its intrinsic merits and interest. As I pointed out in Coliva 2010a, this
reading is based on a patent misunderstanding of some key passages
in On Certainty. First, in OC 343 Wittgenstein actually denies that since
“we just cannot investigate everything ... we are forced to rest content
with assumption”.6 Second, Wright’s and William’s readings are based
on a misunderstanding of Wittgenstein’s use of the term “logic” in On
Certainty. For it is a tenet of Wittgenstein’s later philosophy that that
word can, unless it is used to indicate the discipline called “logic”, safely
be replaced by the word “grammar”. The latter comprises, in On Certainty,
both semantic rules, such as “An object cannot be of two different colors
at once all over its surface”, and rules of evidential significance, such as
“There is an external world” or “The Earth has existed for a very long
time”. That is to say, norms that allow certain kinds of evidence to bear
on the relevant class of beliefs: in the case of perceptual evidence, onto
beliefs about material objects; and, in the case of geological or historical
evidence, to bear on the likely age of the Earth.
According to the former kind of reading, in contrast, Wittgenstein held
the view that our hinges – namely, our most deep-seated certainties – are
not grounded in reasons, but must stay fixed for us to be able to acquire
our ordinary warrants. More specifically, they cannot be epistemically
justified for any alleged warrant for them would actually presuppose
them. Therefore, on such a reading of On Certainty, hinges are proposi-
tions that, in context, play a normative role. In particular, they play
the role of norms of evidential significance, for they determine what
can count as evidence for our ordinary empirical beliefs. Assuming that
“There is an external world” were a hinge for Wittgenstein,7 it would
then turn out that its assumption, with no warrants in its favor, would
be necessary in order to have ordinary perceptual warrants. Hence,
Wittgenstein would also be a moderate.
When confronted with the skeptical challenge, this view has a better
chance than others looked at so far. For it points out the categorial
mistake of calling norms into doubt; or, at any rate, the mistake of
calling into question propositions which, at least in context, play a rule-
like role. After all, it does not make sense to doubt “Stop at traffic light,
when red” since it does not make sense to wonder whether it comports
with reality or not. Hence, it should not make sense to call into doubt
“There is an external world” either, on the assumption that it is a norm
or proposition that plays a rule-like role (OC 95). More precisely, the
mistake would lie in considering these assumptions as ordinary empir-
ical propositions which, in order to be rationally held, would need to
be warranted somehow. Rather, by having a normative role, the issue
of their warrantedness and rationality could not even be raised.8 For
it does not make sense to see them as accountable to already existing
states of affairs and to the empirical evidence that would make them
more likely to be true and thus epistemically justified. Furthermore, it
is a tenet of Wittgenstein’s position (which he shares with pragmatists
such as Peirce9) that all empirical doubts presuppose reasons on their
side. If these reasons could only be produced by taking for granted that
there is an external world, then skepticism would border on nonsense.
For it could not be supported by reasons and could not, therefore, be
rational.10 More precisely, skeptical alternatives to our own assumptions
(as well as Cartesian skeptical scenarios) could hardly be considered
rational.11
Yet, the idea that hinges are norms is obviously problematic. For, in
a widely held view, norms are not semantically assessable, in the sense
of being comparable to a pre-existing reality, while Wittgenstein’s
hinges seem, on the face of it, to be susceptible to semantic evalua-
tion. “Nobody has ever been on the Moon” is a hinge for Wittgenstein
and many would be inclined to say that, although it was true at the
time On Certainty was composed, it is now false. As pointed out in the
first chapter, a Wittgensteinian would have ways of responding to this
problem. For instance, he could maintain that there are doppelgangers –
that is, sentences that look the same even though they play completely
different roles. Therefore, “Nobody has ever been on the Moon” did
express a rule at the time On Certainty was composed and could not
be semantically assessed. However, its doppelganger “Nobody has ever
been on the Moon”, as we may utter or write it now, looks exactly like
the previous sentence, yet it expresses an empirical proposition that is
actually false.12 The problem with this reading is that it is hard to believe
that a hinge like “Nobody has ever been on the Moon” has no descrip-
tive content.
In response, one can retain a more nuanced idea in which it is one
thing to consider the content of a sentence and another to determine
what role it plays in context. In this view, “Nobody has ever been on the
Moon” would have a descriptive content, and thus describes a possible
state of affairs, like any other ordinary proposition. Yet, in context, it

could be removed from doubt and actually play a rule-like role. Varying
contextual factors, it could cease to play such a role and be re-immersed
in the flux of ordinary, semantically assessed, empirical propositions and
it could actually turn out to be false.13 Yet, it is clear that these strategies
are interesting as ways of making sense of Wittgenstein’s position, but
it is much less clear whether they are viable accounts. Yet, without the
endorsement of the normative nature of hinges, it is difficult to see how
a Wittgensteinian anti-skeptical strategy would play out. For one could
no longer claim that a skeptic is somehow making a categorial mistake
by calling norms into doubt, or questioning their warrantedness and
rationality, if the relevant propositions are not in fact rules, or if they do
not play, at least in context, a rule-like role.
Notice, moreover, that the full Wittgensteinian anti-skeptical package
also comprises a strong adherence to the conception of meaning as
use. Hence, any skeptical doubt, which does not conform to ordinary
ones as it is not backed by any reasons and does not have consequences
in practice, would be literally nonsensical. It would retain an appear-
ance of sense – that is, of meaningfulness – only because, unaware of
it, we project meaning onto it from those ordinary contexts in which
“to doubt” is used meaningfully. Furthermore, the Wittgensteinian anti-
skeptical strategy also includes a conception of philosophy, and thus of
the philosophical context in which skeptical doubts or worries arise, as
nonsensical. That is to say, in Wittgenstein’s view sceptical doubts would
produce a context in which words seem to make sense, while they actu-
ally do not.14 Since one should be wary of such a view of philosophy
and of such a strong application of a use theory of meaning, which
bans philosophical talk by deeming it nonsensical, in the following the
moderate position will not be developed along strongly Wittgensteinian
anti-skeptical lines.
Finally, as I have already remarked in Chapter 1 (§4), externalists might
also endorse moderatism and hold that, given that it is in fact the case
that there is an external world, our experiences can be brought to bear
on specific mind-independent entities. However, I also remarked that
such a version of moderatism would be unsatisfactory precisely because
it looks dogmatic vis-à-vis the skeptical challenge. For a skeptic may
grant that there is an external world while asking the very simple ques-
tion “How do we know there is?”.15 In such a predicament an externalist
seems obliged either to repeat himself and say “Because there is”, or else
engage in bootstrapping arguments which are extremely problematical,
as we saw in Chapter 3 (§7).
For all these reasons, I think moderates had better find alternative
ways of countering the Humean skeptical challenge. I will now develop
one such alternative.
2 The extended rationality view and Humean skepticism
The Humean skeptical challenge consists of a request to produce epis-

temic warrants for very general propositions such as “There is an external
world”. According to a skeptic, epistemic warrants are exhausted by
empirical (and possibly a priori) warrants. Their characteristic feature
is that, by their very nature, they speak to the truth of the propositions
they are meant to justify. According to moderates, as well as skeptics,
this challenge cannot be met. For perceptual warrants would presup-
pose these very general propositions and could not be used, through
valid pieces of reasoning, to justify them. A priori warrants, in contrast,
would incur the problems reviewed in Chapter 2. In a nutshell, it is
difficult to see how reflections of an a priori kind could have a bearing
on the truth of what appears to be a contingent, empirical proposition,
such as “There is an external world”. Finally, entitlements, by exploiting
the pragmatic utility of engaging in projects that could be pursued only
by taking for granted certain propositions, would fail to speak to their
truth. If, in contrast, one just posited the existence of entitlements and
claimed that certain a priori considerations are merely ways of claiming
or redeeming them, their speaking to the truth of “There is an external
world” would be merely an article of faith. Finally, if that much were not
claimed, we would have been offered nothing, which could speak to the
skeptical challenge.
However, responses to skepticism come in two varieties: direct, and
indirect (or skeptical). The former try to meet the skeptical challenge
head-on by providing epistemic warrants for the target propositions.
The latter, in contrast, accept the basic skeptical point that fundamental
assumptions such as (III) cannot be warranted, but argue that the devas-
tating consequences skeptics draw from such a conclusion can in fact be
blocked. My extended rationality view belongs to this second genre of
answers to skepticism. As stressed at the beginning of this chapter, the
crucial consequence of Humean (and Cartesian16) skepticism is not so
much that our ordinary empirical propositions would not be evidentially
(perceptually) justified. If it were, moderatism could easily answer it. For,
by requiring just the assumption that there is an external world, as opposed
to its warrantedness, in order for a hand-like experience to warrant a
belief like “Here is a hand”, it allows for the latter to be perceptually
justified, despite accepting the basic skeptical point that “There is an

external world” cannot be warranted. Rather, the crucial consequence of
Humean skepticism is precisely that the assumptions on which ordinary
perceptual warrants depend are not epistemically grounded. This seems
to entail both that our knowledge and justifications, such as they are,
do not rest on a secure base, and that those assumptions might have
been other than what they actually are. With this idea in mind, Humean
skepticism could actually open the way to epistemic relativism. For once
the view is established that ours are ungrounded, a-rational assumptions,
it is just a short step to concluding that they may actually be different
from what they are or may be changed at will, did we so wish or find it
convenient to do so, thus giving in to epistemic relativism.
In this section, I do not discuss the issue of epistemic relativism, which
may grow out of Humean skepticism. It should be stressed, however,
that it is neither identical to it, nor is it a view Humean skeptics would
be happy with. For epistemic relativism reinstates a kind of knowledge
and justifiedness, at least within possibly different epistemic systems,
with their characteristic, specific, and incompatible assumptions that
skeptics have traditionally been concerned to deny. In the following,
I am content to argue against the Humean skeptic’s idea that since our
basic assumptions are not warranted they lie outside epistemic ration-
ality tout court. Let me add, however, one last word of caution, to help
stay clear of possibly relativistic drifts at this stage, and to better situate
the proposal I shall presently make. It is important to keep in mind that
the kind of assumptions I talk about in this section are very general and
fundamental ones that, as I claim, are operative in the basic epistemic
practice17 of gathering perceptual warrants for ordinary empirical propo-
sitions, which is itself constitutive of epistemic rationality. Such a prac-
tice, I take it, is at the core of all human life given the kind of creatures
we are. Hence, nothing I say in the following bears on other kinds of
assumptions or on non-basic epistemic practices, which are not them-
selves constitutive of epistemic rationality that can actually be proved
not to be rational, at least in some cases. That is to say, when they are
unsupported by reasons and when there are, in fact, several reasons
against them. I will take up this issue in §4.
Thus, in response to the idea that our basic assumptions are not
rational, or are only pragmatically rational, it should be noted that
there are assumptions – I mean propositions that are the contents of
assumptions – which are constitutive of what we – and, crucially, the
kind of Humean skeptic I consider at this stage18 – take (empirical) epis-
temic rationality to be.19 Let me explain. In a broadly Wittgensteinian
spirit, I take it that the notion of epistemic rationality shared by skeptics

and non-skeptics alike – like many other notions – does not hang in
the air, but depends on our practices. In particular, the notion of epis-
temic rationality depends on the basic practice (or method) of producing,
assessing, and withdrawing from ordinary empirical beliefs, such as
“Here’s a hand”, “This wall is red”, and so on, interpreted as being about
mind-independent objects,20 based on the deliverances of our senses. Now,
if, as a Humean skeptic shows, that practice rests on assuming – with
no warrants – that there is an external world, that our sense organs are
mostly working reliably, and that one is not the victim of a lucid and
sustained dream, or otherwise disconnected from causal interaction with
physical objects, then those assumptions are constitutive of epistemic
rationality itself. If so, there seems to be no cogent reason to hold that
they lie outside its scope. Indeed, their being constitutive of epistemic
rationality21 suggests otherwise and, in particular, that they are part of it,
since they are its conditions of possibility.
We could make the point in a more Kantian fashion by saying that
epistemic rationality extends to its very conditions of possibility,
namely, to the unwarranted assumption that there is an external world
(and others), that allows one to acquire warrants for ordinary empirical
propositions. Therefore, these assumptions are neither irrational nor
a-rational. That is to say, they are held neither against contrary reasons,
nor in an epistemically irresponsible way. Rather, as we could put it,
they are “basically” epistemically rational, since, while being epistemi-
cally unwarrantable, they are constitutive of epistemic rationality itself.
Hence, for a supporter of the extended rationality view, epistemic ration-
ality ought to be defined as follows:
Epistemic rationalityER: it is epistemically rational to believe eviden-

tially warranted propositions and to accept those unwarrantable
assumptions that make the acquisition of perceptual warrants possible
in the first place and are therefore constitutive of ordinary evidential
warrants.
If so, and this is the crucial point, both skeptics and non-skeptics alike
are required by the lights of epistemic rationality itself to assume that
there is an external world. Hence, they are equally mandated by a shared
notion of epistemic rationality to assume it. That is to say, in order to be
epistemically rational, and therefore capable of taking part in the prac-
tice of acquiring, assessing, and withdrawing from ordinary empirical
beliefs based on perceptual evidence, one ought to assume that there is
an external world. This, in turn, means that we cannot – consistent with

skepticism – provide evidential warrants for such an assumption. Yet, it
does not follow that that assumption lies outside the scope of epistemic
rationality either. Hence, a Humean skeptic is guilty of inferring that
since our basic assumptions cannot evidentially be warranted, they are
not epistemically rational, because he confines himself to too narrow a
notion of epistemic rationality. Namely,
Epistemic rationalitySK: it is epistemically rational to believe only

evidentially warranted propositions.
If, in contrast, we appreciate what our shared notion of epistemic ration-

ality requires, we can then see that even though our basic assumptions
are not warranted (indeed warrantable), they are epistemically (basi-
cally) rational and required, as they are constitutive of what we – both
skeptics and non-skeptics alike – take epistemic rationality to be.
Let me dwell on this crucial point further, by exploiting what I think is
a useful analogy. Think of a game and its constitutive rules. Clearly, they
are part – indeed constitutive ingredients – of the game, though they
are no moves within it. Consider burraco, one of today’s most popular
card games. To be allowed to put down all one’s cards and win one must
have at least six cards of the same suit in a row and a wildcard. That rule
(among others) is constitutive of that game and determines which moves
can or cannot be made. Analogously, why restrict epistemic rationality
to warranted propositions only? It would be like restricting a game only
to the moves within it, without considering the rules. Yet, without rules
there would be no game and so no moves within it either. Surely with
epistemic rationality things are more complicated for its constitutive
conditions are, in my view, determined by the practice and do not have
a rule-like form (that is, they neither contain oughts, nor do they come
in the, conditional, imperative form “(If C,) φ!”). Furthermore, as we
shall see in the following, playing games such as burraco is entirely
optional for human beings, while playing the game of epistemic ration-
ality is not. But, even so, we can certainly distinguish between what
plays a rule-like role with respect to epistemic rationality – that is, the
appropriate assumptions – and what, in contrast, plays a move-like role
with respect to it, that is, whatever perceptual warrants we possess for
specific empirical propositions, thanks to those very assumptions and
appropriate sensory experiences. Just as rules and moves are both part of
a game so, I contend, both the assumptions that allow us to have percep-
tual warrants and those very warrants are part of epistemic rationality.
At any rate, the opposite view, which restricts epistemic rationality to,
in fact, epistemic rationalitysk, is much less obvious, to my mind, than
it prima facie seems to be. Especially when it is realized that the eviden-
tial warrants it countenances depend for their existence on those very
assumptions.
Notice, furthermore, that according to the analogy we are exploring
here, the rules that are constitutive of a game such as burraco are not
mandated tout court, that is, for all possible games. Yet, if we changed
them, we would be playing a different game. Conversely, to play that
game, we have to abide by its rules. This also holds in the case of the
“game” of epistemic rationality: if, as they are portrayed as doing,
Humean skeptics want to play it, they have to stick to its constitutive
rules, and so they have to hold on to “There is an external world”.
Recall, moreover, that the Humean skeptics I am considering at this
stage are not epistemic relativists. Thus, they are generally happy to
stop themselves well before embracing the view that there are other,
equally legitimate notions of epistemic rationality (I return to this issue
in §4). Rather, they simply maintain that our shared notion of epistemic
rationality is not grounded in warranted assumptions and thereby take
themselves to have shown that its basic assumptions do not lie within
the scope of epistemic rationality. However, this conclusion is based on
disregarding the fact that the very notion of epistemic rationality they
themselves embrace mandates the assumption of its constitutive presup-
positions, such as “There is an external world”.22 For, without such an
assumption, no evidential warrant for ordinary empirical propositions
could be had (recall that here we are considering a Humean skeptic who
does not contest the existence of this latter kind of warrant).
I do not have any obvious explanation of why skeptics are fixated with
epistemic rationalitysk and, in effect, of why philosophers of no skep-
tical persuasion tend to follow suit. My hunch is that since we normally
traffic in evidential warrants for our ordinary empirical beliefs, it seems
natural, once it is correctly realized that they themselves depend on
certain assumptions, to keep on asking for evidential warrants for the
latter. Absent those, one may then be tempted by a priori warrants (or
perhaps to look for sui generis kinds of warrant such as entitlements).
Indeed this is how Hume himself went about the issue in the Treatise
and other writings (save for exploring entitlements, of course). When he
did find that the search was unsuccessful, he then came to the conclu-
sion that not only did these assumptions fail to be warranted – just as
moderates are happy to acknowledge – but also that they actually lie
outside the scope of epistemic rationality itself. That is how, I surmise,
these basic assumptions were cast out from epistemic rationality and
we – epistemologists, I mean – were infected with the disease of believing
that epistemic rationality is in fact identical to epistemic rationalitysk.
Let me also remark, once more, that it is indeed a mistake to think
that for the assumptions of our basic epistemic practice to be rational
they should be justifiable on the basis of evidence. While such a request
would be perfectly in order for non basic epistemic practices and their
characteristic assumptions (see §4), it cannot be right for the basic prac-
tice of forming justified beliefs about material objects in one’s surround-
ings on the basis of one’s perceptions. For, to repeat, that evidence
would circularly depend on taking those assumptions for granted. This
would prove, once again, that to question their rationality in virtue of
their being evidentially unwarrantable could only be based on disre-
garding the fact that the very notion of epistemic rationality we bring to
bear on them depends on taking them for granted. That is to say, if we
did not take them for granted, we could not have any kind of percep-
tual justification for ordinary empirical beliefs and hence no practice of
going about forming and assessing our perceptual warrants for ordinary
empirical beliefs. With no such practice, there would be no notion of
epistemic rationality either.
Contrary to Wright, moreover, who holds,
Epistemic rationalitycw: it is epistemically rational to believe eviden-

tially warranted propositions and to assume those propositions for
which we possess non-evidential warrants, that is, entitlements,
the diagnosis I have proposed of the skeptical mistake is not that it

ultimately depends on too narrow a notion of warrant, but simply on too
narrow a conception of epistemic rationality. One, that is, which confines
epistemic rationality to warranted (warrantable) propositions only,
while it extends also to those presuppositions which, though unwar-
rantable, make the acquisition, assessment, and withdrawal from
empirical beliefs on the basis of the deliverance of our senses possible
in the first place, and are therefore mandated by it. The reasons why
I think the extended rationality view is to be preferred to the view
about epistemic rationality put forward by Wright should be clear from
Chapter 2 (§2), where we unraveled some of the main problems with
Wright’s entitlements.
Furthermore, it is important to stress that we are mandated by epis-
temic rationality itself to accept that there is an external world. Hence,
we do not do so just because we cannot psychologically or culturally
Not Epistemically Epistemically Epistemically

epistemically rationallysk rationallycw rationallyer
(III) rationallysk held held held held
Evidentially Skeptics
unwarrantable
Evidentially Liberals
warrantable (Pryor)
Evidentially Conservatives
unwarrantable but (Wright)
non-evidentially
warrantable
Unwarrantable Moderates
evidentially or (Coliva)
otherwise
help it, or because we cannot but do so if we are interested in preserving

practices that have proved useful or even inevitable to us. The rational
mandate we have for “There is an external world” does not come from
our psychological constitution or from practical rationality, but from
epistemic rationality itself. Of course, in my view, epistemic rationality
depends on a practice, but that does not make its requirements any
more pragmatic or practical in nature than noticing that mathematics
depends, at least for humans, on a practice would make its rules and
axioms pragmatic in kind. We can thus visualize the positions in play
as follows.
This chart makes it apparent that skeptics and liberals share a common
view of epistemic rationality, while conservatives and moderates depart
from that view – namely, epistemic rationalitysk. Skeptics and liberals,
however, take an opposite stance on its instances. In contrast, skeptics
and conservatives share the same structural conception of epistemic
rationality – it always depends on warrants – but they partially differ on
what they take warrants to be. Finally, skeptics and moderates do not
share that structural view for, according to moderates, propositions that
are not warrantable tout court, yet are constitutive of epistemic ration-
ality, are themselves epistemically rational by the lights of extended
rationality.
Before closing this section, let me dwell a little on a comparison
between the extended rationality view and Wright’s position. Wright
advertises his position as “rationalist”. I do too.23 But his rationalism
depends on a belief in having discovered first-order warrants – albeit
non evidential ones – that attach to those assumptions and make them
epistemically rational. My rationalism is different. It depends on being
within a practice – hence the qualification “internal” that we could
add – that grounds a notion – that of epistemic rationality – and, from
there, see also that the tout court unwarranted/able assumptions that
make it possible lie within its scope, by being constitutive of, and there-
fore rationally mandated by, our very notion of epistemic rationality.
Yet it remains that a rational mandate is not anything that can speak
to the likely truth of what it, in fact, mandates. Therefore, it is by no
means an epistemic warrant. Indeed, it is part of the moderate position
that genuine epistemic warrants, in this case, are just ordinary evidential
ones, generated within a given epistemic practice, resting, in their turn,
on unwarrantable assumptions.
Now, for much the same reason it seems to me that Wright’s entitle-
ments cannot be genuine (first-order) warrants for, as we saw in Chapter 2
(§2), they are not meant to speak to the likely truth of what should be
warranted thereby, namely, “There is an external world”. If that were
right, then, actually and contrary to his official pronouncements, Wright
would be proposing a moderate conception of the architecture of percep-
tual warrant, whereby rationally mandated assumptions – as opposed to
epistemically warranted ones – together with a certain course of expe-
rience, would provide warrant for ordinary empirical propositions. It
thus seems to me that, once his proposal is properly situated, perhaps
by going against his own understanding of it, it turns out to differ from
mine only on matters of detail. That is to say, because we place the
origin of the rational mandate we have for “There is an external world”
in different kinds of consideration.24 To stress, the two proposals would
be similar because his would end up being a form of moderatism on its
best construal and not because mine would be a form of conservativism,
where special warrants such as entitlements are provided for “There is
However, let me clarify things a bit more. In my view, it is a brute fact
of epistemic rationality, once properly understood so as to appreciate
its extent, that it mandates certain assumptions, such as “There is an
external world”. Hence, one should be careful not to think of rational
mandates as (epistemic) goods produced by philosophical investigation,
which attach to assumptions and make them rationally held. Rather,
the philosophical explanation of this brute fact – granting for the sake
of argument that it is correct – provides us with a philosophical argu-
ment, and therefore an a priori warrant for believing “The assumption
that there is an external world is epistemically rational, even though
unwarrantable”. Once in possession of such a warrant one could, of

course, confront a Humean skeptic who claims that such an assumption
lies outside the scope of epistemic rationality altogether. Yet, to have
such a warrant is completely different to having a warrant that speaks to
the likely truth of “There is an external world”, sic et simpliciter.
It is my hunch, furthermore, that Wright’s own entitlements are
better understood along similar lines: not as epistemic goods discov-
ered by philosophical reflection that turn a-rational assumptions into
rational ones; nor as philosophical arguments which provide one with
a priori warrant for “There is an external world”; but as philosophical
arguments that, if successful, provide one with an a priori warrant to
believe that the assumption that there is an external world falls within
the scope of epistemic rationality, which is unwarrantable though
mandated by, in Wright’s view, our conceptual scheme. As we pointed
out in Chapter 2 (§2), notice that if he somehow tried to say that enti-
tlements are ultimately first-order a priori warrants obtained through
philosophical reflection for “There is an external world”, this would
create a tension. For, in that case, they would have to speak to the likely
truth of that assumption, while, officially, Wright has been concerned
to deny that entitlements could achieve that much. Yet, if they did
provide such a warrant, why should we then assume rather than believe
that there is an external world? In particular, Wright’s claim that “There
is an external world” could only be assumed and not believed because
we do not have any warrant that could speak to its likely truth would
be impaired.
Therefore, to sum up this section on Humean skepticism, I think that,
in one sense, a skeptic wins since we cannot produce genuine epistemic
warrants for our basic assumptions – neither perceptual nor a priori
ones. Yet, in another sense, he does not for it does not follow that he can
accuse us of holding assumptions that lie outside the scope of what we
both mean by “epistemic rationality”. That is why the extended ration-
ality view is indeed an indirect response to Humean skepticism. To repeat,
by appreciating that epistemic rationality also extends to those assump-
tions that make it possible in the first place, one blocks the unwanted
consequence that it rests on non- or a-rational assumptions. The latter
lie, rather, within the scope of our notion of epistemic rationality and
are mandated by it. Hence, they cannot be other than what they are.
This is no proof or evidence of their truth, but perhaps the important
lesson to be drawn from all this, which we consider in more detail in §6,
is that epistemic rationality, evidence, and truth, at bottom – namely,
when basic assumptions are concerned – come apart.
3 The extended rationality view, Cartesian skepticism,

closure, and “abominable conjunctions”
In this section, I try to stitch together some loose ends. Firstly, I make
explicit what moderatism and the extended rationality view have to
say in response to Cartesian skepticism. Secondly, I qualify, in light of
the extended rationality view, my earlier rejection of the unconditional
validity of the closure principle. Finally, I use this qualification to meet the
challenge DeRose posed, which we briefly considered in Chapter 3 (§6).
Namely, deniers of the absolute validity of closure will have to face the
problem of, so-called, abominable conjunctions. For they would end up
condoning “I know/warrantedly believe that here is a hand, but I do not
know/warrantedly believe that there is an external world/I am not a BIV”.
As is well known, Cartesian skepticism does not directly target the
assumption “There is an external world” but the presupposition that
we are not victims of sustained delusions or dreams. It is clear how
a moderate should respond to Cartesian skepticism since the basic
Cartesian argument is based on closure. By noticing that we do not know
or have a warrant to discard uncongenial scenarios, such as the possi-
bility that we might be BIVs or victims of lucid and sustained dreams,
the Cartesian skeptical paradox exploits closure to conclude that we do
not know or have a warrant for ordinary empirical propositions such as
“Here is a hand”. By generalizing this reasoning to any possible ordinary
empirical proposition, such as “Here is a table”, “Here is my foot”, “Here
is a PC”, it arrives at the disturbing conclusion that we do not know
or have a warrant for “There is an external world” either.25 Obviously,
a moderate is going to object to the application of closure by noticing
that the general assumption that we causally interact with a world popu-
lated by physical objects is necessary in order to have any empirical
warrant. As such, it cannot be warranted. Yet it does not follow that
other propositions that entail it should similarly be unknown or unwar-
ranted. By then endorsing the extended rationality view, a moderate
will also point out that that assumption is mandated by our notion of
epistemic rationality. For it is a condition of possibility of acquiring
and assessing evidence for or against ordinary empirical propositions
that one be properly causally connected with the physical world and
therefore immune to massive delusions produced by powerful demons,
scientists, or machines. Hence, although we have no warrant for “I am
not dreaming (or the victim of massive deception)”, and thus have no
warrant to discard skeptical uncongenial scenarios, we are epistemically
rationally required to do so.
This, together with what we have been saying in §2 about Humean

skepticism and the assumption that there is an external world, shows
that although a moderate denies closure for warrant (and for knowl-
edge too, so long as knowledge is taken to be justified true belief), he
has another good in stock, for which closure does not fail. Namely,
the rational mandate to assume those very general propositions that
are necessary in order to have empirical warrants in the first place,
which stems from the endorsement of the extended rationality view.
This observation has important consequences with respect to DeRose’s
challenge, according to which, if one denies closure, one is committed
to abominable conjunctions such as “I know/warrantedly believe that
here is a hand, but I do not know/warrantedly believe that there is an
external world/I am not a BIV”.
As already remarked on in Chapter 3 (§6), and as Sosa (2000) has
noticed, it is one thing to correctly describe our epistemic situation and
quite a different one to explain the legitimacy or even the pragmatic
felicity of certain assertions. To put it differently, it is one thing to clarify
the conditions in which we have knowledge and justification and quite
another to elucidate the conditions in which it is pragmatically appro-
priate to make certain epistemic assertions. According to Sosa, only the
former pertains to epistemology, while the latter belongs to the philos-
ophy of language. Although I sympathize with this kind of response, it
is obvious that we owe an explanation of why it would sound odd – or
even abominable – to make assertions that the very epistemic theory one
embraces actually predicts to be true. Furthermore, it must be acknowl-
edged that it would be embarrassing to be in possession of a truth that
cannot be appropriately asserted.
The reader will recall from Chapter 3 (§6) that Pritchard (2005a) has
maintained that the assertion sounds odd because it violates the Gricean
maxim of quality, which prescribes that one should assert only what
one has at least a justification for (or even what one knows to be true).
Now, if it is true that we have no justification to believe that we are not
dreaming, then it is obvious that to assert “I am not the victim of a lucid
and sustained dream” would be pragmatically infelicitous. However, I
did point out in that chapter why I think the problem is a little more
difficult than Pritchard makes it seem. For the assertion that, according
to DeRose, would be abominable is not “I am not the victim of a lucid
and sustained dream”, but, rather, “I know (or am warranted to believe)
that there is a hand here, but I do not know (or have no justification
to believe) that I am not the victim of a lucid and sustained dream”.
This is a true assertion for, according to deniers of the absolute validity
of closure, both conjuncts are true. Furthermore, they also think that
there are very good reasons to believe both conjuncts. Hence, it is not
clear how an appeal to the Gricean maxim of quality (or to any other
Gricean maxim) explains why the assertion sounds odd. Finally, the
mere judgment “I know (or am warranted in believing) that there is a
hand here, but I do not know (or have no justification to believe) that I
am not the victim of a lucid and sustained dream” would be extremely
weird. Such an impression, however, cannot (easily) be explained by
appealing to conversational maxims that regulate the pragmatic felicity
of the speech-act of assertion. Therefore, it is not clear how a Pritchard-
inspired approach could generalize to counter the non-assertoric version
of the objection.
However, we are now in a position to respond to DeRose’s objection.
What we are committed to, given our denial of the absolute validity
of closure, is somewhat mitigated by our endorsement of the extended
rationality view. Hence, what our theory allows (and requires) us to
assert (or judge) is not “I know (or am warranted to believe) that there is
a hand here, but I do not know (or have no justification to believe) that
I am not the victim of a lucid and sustained dream”, but is “I know (or
am warranted to believe) that there is a hand here, and although I do
not know (or have no justification to believe) that I am not the victim
of a lucid and sustained dream or that there is an external world, I am
rationally mandated by epistemic rationality itself to accept it”. This is
not – at least to my ear and mind – an abominable conjunction at all.
I think this remark is in keeping with what Harman and Sherman
(2011) propose against DeRose’s objection (even if they do not actu-
ally refer to him, but only to the alleged abominable conjunctions he
proposed against deniers of the absolute validity of closure). Their point
is that these abominable conjunctions can be proposed whenever there
are concepts that are “suitably subtle” (ibid p. 139) and therefore can
hardly be distinguished from others in their vicinity. According to them,
this would be the case with the concepts of knowledge and assump-
tion. In my view, in contrast, this would be the case with the concept
of warranted belief and of rationally mandated assumption. Since it is
hard to keep them apart and doing it requires philosophical analysis,
they argue that it is relatively easy to produce odd assertions such as “I
do not know whether I am dreaming/I have no warrant that I am not
dreaming, but I do know/am warranted in believing that I am in the
kitchen”. However, they notice that the relevant philosophical theories
merely license “I do not know (or have warrant to believe) that I am not
dreaming, yet I do justifiably assume I am not (Harman&Sherman)/I am
rationally mandated to assume that I am not (Coliva) and I do know (or

have warrant to believe) that I am in the kitchen”.
Let me also note, in closing, that DeRose’s objection would go through
if one simply denied the absolute validity of closure and had no story to
tell about the relationship we have with the very general assumptions
that make the acquisition of empirical warrant and knowledge possible
in the first instance. However, as noted, we have such a complemen-
tary story and this is what allows us, I submit, to meet the challenge.
Therefore, what is abominable is to deny the absolute validity of closure
as an ad hoc move meant merely to confront the skeptical challenge.
However, as we saw at length in this and the previous chapter, this is not
what we are proposing here. In Chapter 3 we explored the possibility
of a notion of transmission failure, different from Wright’s, which is
independently motivated and would lead to the denial of closure only
when very general assumptions, entailed by more specific propositions
and constitutive of epistemic rationality, are at stake. In this chapter,
we have noticed how these assumptions are rationally mandated in
light of the appropriate conception of epistemic rationality. In turn,
in Chapters 1 and 2, we saw how these assumptions are needed, in a
moderate conception of the architecture of empirical warrant, for which
we have provided ample motivation, in order to have our usual empirical
warrants. All this has given us the means and the rationale to qualify the
alleged abominable conjunctions raised by DeRose, thus rendering them
quite harmless. In fact, all this entails that both conjuncts of the alleged
abominable conjunction would indeed be justified and this would, of
course, make them quite naturally assertible together.
4 The extended rationality view and epistemic relativism
Let us now turn to the problem of epistemic relativism – better, to a

special case of epistemic relativism. Namely, one that countenances
the possibility of different basic assumptions, characterizing our way of
forming beliefs about our surroundings based on perception. Therefore,
we are not going to consider the more familiar cases usually discussed in
connection with relativism, such as the adoption of different (scientific)
theories, or the alleged adherence to different explanatory or logical
principles.26 As we saw before, the possibility of different basic assump-
tions on which our ability to form empirical beliefs thanks to the deliv-
erances of perception should be based, is not necessarily the outcome of
the Humean skeptical challenge. As I pointed out, a Humean skeptic will
simply claim that our basic assumptions are not warranted and therefore
are not rational. By relying on closure for warrant he is likely – in his

more traditional fashion and if he also subscribed to the conservative
conception of the structure of empirical justification – to deepen this
result by claiming that lack of warrant would then propagate so as to
infect our ordinary beliefs about material objects. The kind of epistemic
relativist who concerns us here, in contrast, can be seen as both radical-
izing and as mitigating the (traditionally) Humean position. He miti-
gates it because he allows that warrant and even knowledge are possible,
at least within a system of basic assumptions, and, in so doing, he joins
the less traditional kind of Humean skeptic we have been concerned
with so far in this chapter. However, he also radicalizes such a skeptic’s
position because, from the fact that these assumptions turn out to be
unwarrantable, he concludes that there may be – either as a matter of
fact or in principle – many different systems of assumptions, which are
mutually incompatible and yet are on a par, that give rise to different
and equally valid systems of justification.27
We also saw how a moderate would respond to Humean skepticism in
its various fashions. By denying closure for warrant, due to transmission
failure 2 (see Chapter 3, §6) he denies the propagation of lack of warrant;
and by endorsing the extended rationality view he denies that our basic
assumptions are arbitrary. For they are rational by the very lights of epis-
temic rationality and actually mandated by it, once epistemic ration-
ality itself is properly understood. Insofar as a moderate is willing to
endorse the extended rationality view, he would then seem to be able to
respond to epistemic relativism. For, by denying the arbitrariness of our
basic assumptions, he would then seem to be in a position to deny the
starting point of epistemic relativism.
Things, however, are not that simple. For an epistemic relativist could
insist that our practice determines our notion of epistemic rationality
with its characteristic assumptions, which are even mandated by it. Still,
he could claim that there are – either in fact or in principle – other prac-
tices which determine different notions of epistemic rationality, with
their distinctive presuppositions, which would be mandated by the very
lights of these alternative notions.
Some preliminaries are necessary before tackling this problem. Firstly,
it has to be kept in mind that we are not dealing with the simple-minded
kind of epistemic relativism we encountered at the beginning of this
chapter as an example to make the problem immediately vivid to the
reader. Namely, the case of a person who believes in fortune-tellers and
of the one who does not. Nor are we dealing with more traditional forms
of epistemic relativism, such as those that depend on the endorsement
of different scientific theories. Rather, we are dealing with the much

more worrying and controversial case of the assumptions of our basic
epistemic practices and the possibility that there may be different and
incompatible such practices with their respective assumptions.28
Secondly, it must be clarified that an epistemic practice is basic when
it does not presuppose other instances of itself and is necessary for
other epistemic practices. Therefore, for instance, to form a belief about
objects in one’s environment based on perception, absent defeaters, is
such a basic practice. By contrast, to form the belief, based on astrology,
that the skies predict that one will undergo a year of bad luck requires
the exercise of observation. The latter is needed both to see the configu-
ration of stars and planets in the sky and to read the texts that contain
the theory by means of which one is in a position to draw such a conclu-
sion. Hence, astrology is not a basic epistemic practice.
Arguably, we have other basic epistemic practices, such as, for instance,
deductive inference in accord with basic logical laws.29 Of course, the
premises of a deductive inference often come from some other source –
that is, “Socrates is human” comes from observation and “All humans
are mortal” comes from observation, testimony, and generalization. Yet,
the conclusion “Socrates is mortal” would depend merely on the applica-
tion of deductive reasoning, and, in particular, of modus ponens. Thus,
that inference is basic and it can enter a number of other, non-basic epis-
temic practices, for instance, relying on theories which predict that one’s
soul will face a divine trial and will be punished or rewarded accordingly,
to generate the conclusion that Socrates’ soul will also face that fate. If so,
modus ponens can be seen as a basic rule of inference of what we may call
logical or deductive rationality (more about this in Chapter 5).
Finally, the extent of basic epistemic practices can of course be a matter
of discussion. There may be a consensus about considering the practice
of φ-ing, ceteris paribus, if one desires that P and believes that φ-ing will
bring about P, as constitutive of what we can call practical rationality.
Equally, there can be a consensus on induction and even inference to
the best explanation as constitutive practices of what may be called
predictive, abductive or explanatory rationality. However, the status of
memory and testimony, that is, of the practices of forming beliefs based
on the deliverances of memory or of testimonial evidence, is less clear.
Nevertheless, we shall now focus on the basic practice of forming beliefs
based on one’s perceptual experiences, absent defeaters, as constitutive
of epistemic rationality.
Hence, the relativist challenge must take the following form: can there
be alternative basic epistemic practices of forming beliefs about objects
in our surroundings, with different characteristic assumptions? This

problem can be interpreted in two different ways. As (i) the problem of
whether there may be other ways of forming warranted beliefs about
physical objects in our surroundings, taken as mind-independent enti-
ties, by means of methods other than observation, with different and
incompatible characteristic assumptions. Or else (ii) as the problem of
whether observation can lead to warranted beliefs about objects in our
surroundings, on an understanding of the notion of object different
from ours, thus doing without the very assumption that we are inter-
acting with a world of mind-independent entities. We will look at these
in reverse order.
The second version of the relativist challenge is tantamount to
considering the relationship between common sense and idealism
or, perhaps more accurately, phenomenalism. Various things must be
noted. First, that by adhering to a different notion of object, idealism
actually poses a challenge of commensurability between our notions
and practices and those licensed by that position. In order to be a chal-
lenge worth taking seriously, however, it must be fully intelligible and
actually compatible with our experience. Is it? I doubt it. For, if we
consider the very content of our perceptual experiences, it has to be
remarked that we see hands, tables, and so on.30 What this means is
that both developmental psychology and the science of vision have
largely shown how newborn infants do not experience the manifold
of colors and shapes that have somehow to be arranged and referred
to external objects, perhaps by means of concepts, to give rise to the
perception of hands and tables. Rather, they actually have visual expe-
riences with representational contents as of external objects with
specific properties. Furthermore, the science of vision has revealed
the existence of perceptual constancies. No matter how near or far we
move from an object, subsequent perceptions obviously differ but the
visual system still refers them to a single object.31 What this means is
that our perceptual experience is objective. That is to say, it represents
not just subjective variations, like changes in temperature, or from
pleasant to unpleasant states. Rather, it gives us representations as of
objects and properties “out there”.
Thus, phenomenalists cannot rescue the very content of our percep-
tual experience. For, in their view, subjects just have sensations of colors,
and so on, without objectifying them – that is, without somehow refer-
ring them to something external to themselves. These phenomenal
experiences then have to be grouped together, presumably by means
of the exercise of our conceptual repertoire and inference to the best
explanation, to give rise to representations as of hands, tables, and so

on – that is, as of objects out there, with their respective properties. Yet,
such a view of perception is totally at odds with our scientific knowledge
of perception. Furthermore, it deprives creatures without concepts, such
as newborn babies and many animals, of the ability to have perceptual
representations as of objects and their properties in their surroundings,
which they do actually have.32
Furthermore, what would warrant, in such a framework, the inference
from one’s phenomenal experiences to their likely objective causes?
That is, what does allow phenomenalists to hold that a certain bundle of
perceptions is to be referred to one single external cause, such as a hand,
or a table? For, after all, each sensory experience is temporally local-
ized and if there are (or had been) no mind-independent objects, which
exist even when unperceived, or in between various subsequent experi-
ences, how could our perceptual system take these subsequent experi-
ences to depend on the very same external cause? Even if multi-modal
perceptions went on at the same time, what would allow one to take
each of them as dependent on the same external cause? For, after all, if
all we have to go by are those experiences themselves, considered in a
phenomenalist fashion, they cannot represent their being dependent
on the same external cause on which other simultaneous experiences
would allegedly depend. Nor would things go any better if phenomenal-
ists supposed the existence of a mysterious and obscure “substratum”
to which all those experiences somehow relate. In such a predicament,
it would be better simply to acknowledge the existence of mind-inde-
pendent objects, which exist even when we do not perceive them and
that retain their properties throughout.
Notice, furthermore, that although the science of vision and develop-
mental psychology are united in taking the perceptual system as capable
of representing objects without the aid of concepts and through the
workings of its subpersonal, partially encapsulated, mechanisms, this
does not prove externalism right tout court.33 It only proves, at most, that
our perceptual faculties must have evolved in such an environment to
give rise to objective representations of objects and/or their properties.34
Yet they could be operating in much the same way in an uncongenial
environment and thus give rise to objectified representations even if
there were no physical objects to causally interact with. However, the
relevant anti-idealist point remains: it seems that phenomenalists could
not even explain the very content of our sensory experiences, which, to
repeat, is as of objects, independently of the exercise of concepts, even
in such uncongenial scenarios.
Therefore, despite their respected credentials as philosophical theo-

ries, I do not think idealism and phenomenalism actually pose a serious
challenge. For, if what we have been seeing is correct, they cannot offer
a suitable account of our perceptual experience, which could be taken as
a sound alternative explanation. This, of course, is not to say that their
views do not represent possible scenarios, logically and even metaphysi-
cally. It only means that they would be unsuitable as alternative accounts
of our human experience. Yet, any form of relativism that is not compat-
ible with the distinguishing traits of human experience would be quite
toothless and not worth taking seriously, at least within a project – like
the present one – of understanding the human condition.
Hence, the remaining relativist challenge to consider is whether there
are other ways of forming warranted beliefs about physical objects in our
surroundings, taken as mind-independent entities, by means of methods
other than observation, with different and incompatible characteristic
assumptions. Now, if we think about creatures like human beings, or of
creatures who are relevantly similar to human beings, it is hard to see
how else they could form beliefs about material objects if not through
the deliverances of their senses. Therefore, in order to engage with this
challenge, we will have to consider completely alien creatures who need
not rely on their senses to form beliefs about physical objects. How would
they accomplish such a task? It is not clear what to say in positive terms. It
seems that the best we can do is to speak of them negatively, as creatures
who would not fulfill this task in the way we would. Perhaps angels or God
would have some kind of rational intuition of there being hands, tables,
human bodies, and so on. Yet, on closer examination, these hypotheses
defy our conceivability powers. For rational intuition, in their case, would
not be like seeing, only with the mind’s eye (what would such an eye
be?); or intuiting the presence and features of material objects through an
analysis of the concepts involved to entertain those concepts. Therefore,
although we should remain open to these hypotheses, they would make
relativism utterly ineffectual. For a relativist challenge to be worth taking
seriously the alternative should at least be intelligible to us.
On second thoughts, it is evident that the faint hypothesis we are
trying to entertain here is surely compatible with our epistemic practice.
For, ex hypothesi, it would lead largely to the same beliefs (or perhaps
only to true ones, in the case of omniscient creatures like God), simply
in a different way. Thus, it is unclear in what sense it would give rise to
a relativist challenge. Finally, one could pose the hypothesis of creatures
who not only get to know truths about physical objects without having
recourse to the workings of their senses, but who also come to believe
(and know) different things about them to what we believe (and take
ourselves to know). Maybe so, but as long as they were to form beliefs
about them, taken as mind-independent objects, it is unclear in what
sense they are adhering to an epistemic practice which runs contrary to
the assumption that there is an external world.
To recap and conclude this section on epistemic relativism. We
have seen how the relativist challenge can take two forms: the former,
idealist/phenomenalist version, turns out to be incapable of explaining
the very content of our experiences; the latter, purely rationalist idea,
while almost defying imagination, does not seem to pose a threat to
the assumption that there is an external world. A relativist might press
further with less and less conceivable hypotheses, up to the point of
positing an inconceivable one, which should be compatible with our
experience but which forsakes the assumption that there is an external
world, or even holds its negation or something else incompatible with
it. In this case, however, I think we should not lose sleep over it. For any
relativist challenge worth taking seriously must be such that we can at
least cognitively engage with it. Ex hypothesi, that would not be the case.
Therefore, although it remains metaphysically possible, it would not be
cognitively relevant and would make relativism ineffectual.
5 The extended rationality view and the Oblomovian

challenge
Whenever constitutive theories are proposed, eventually a characteristic

challenge is raised.35 It usually takes the following form: you constitu-
tivist say that X is constitutive of Y, for instance, you say that winning
is constitutive of playing a game, which countenances winning and
losing; yet you can play the game without wanting to win and play it
nonetheless; therefore, either X is not constitutive of Y, or X is constitu-
tive of Y, but you should show, in addition, why we should care to play
that particular game, rather than any other one, which is sufficiently
similar to it, but that does not aim at winning.36
I think this objection is only partially correct. Of course, you may play
burraco without wanting to win. Yet, in order to play it with the intention
not to win, you need to know what winning at burraco would consist of.
Moreover, if applied to epistemic rationality, the objector would have to
say something like this. You, constitutivist, say that “There is an external
world” is constitutive of epistemic rationality. However, you can play
that game without endorsing that assumption. Hence, one is not forced
to hold that there is an external world while providing reasons for or
against specific propositions about physical objects. When looked at this

way, however, the objection looks weird. If the moderate conception of
perceptual justifications is correct, then one cannot provide ordinary
empirical warrants for beliefs whose contents are about mid-size phys-
ical objects if that assumption is not in place. Therefore that assumption
is not optional with respect to the game of epistemic rationality in the
same way as winning – in fact wanting to win – may be vis-à-vis playing
burraco.
At this stage, the objector could respond as follows. Let us grant that
“There is an external world” is constitutive of epistemic rationality as
you propose. However, one might opt out and decide not to play the
game at all, just as one may opt out and decide not to play burraco at all.
Let us call this “the Oblomovian challenge”, after Goncharov’s character
who wanted to opt out of any decision-making responsibility.37
Can we opt out of epistemic rationality, that is, of the practice of
forming, assessing, and withdrawing from ordinary empirical beliefs
based on perceptual reasons? Of course, even if the costs involved in that
were significant, for we would end up in a state of randomized decisions
and actions with respect to the outer world, which could actually be
very dangerous. Think of crossing the street without playing the game of
epistemic rationality. It would be like crossing it blindfolded. You may be
lucky and get away with it. Still, you would be significantly less likely to
survive than if you crossed the street with your eyes wide open, paying
attention to the traffic and forming a justified belief about the presence
and the distance of cars and other means of transport. Therefore, I think
there are perfectly legitimate practical reasons for wanting to play the
game of epistemic rationality.
However, that is not the point. For, as epistemologists, we are not
looking for something that should psychologically or practically compel
us to play the game of epistemic rationality. After all, people may decide
to take the risk and cross the street blindfolded, not caring much about
the result. What we are trying to do, rather, is to provide us with reasons
to think that the rules of the game – indeed the very game of epistemic
rationality – are kosher from an epistemic point of view. The extended
rationality view does just that. By pointing out that these assumptions
are constitutive of epistemic rationality itself, it tells you why they
are epistemically rational in their turn. It does not provide you with a
reason to be epistemically rational – if there were a need to provide one.
That may well come from altogether different considerations, of a much
more practical nature. Even so, the extended rationality view would
not tell you, mistakenly, that its assumptions are true because they are
constitutive of a game we have practical reasons to play. Rather, it would

tell you that to play that game, for whatever reason you may want to
play it, or to whatever degree it may even be indispensable to play it,
exactly those assumptions have to be in place and that they are there-
fore part of epistemic rationality and do not fall outside its boundaries.
6 The extended rationality view and truth38
In this section, I wish to consider briefly the philosophically deep and

difficult problem of the relationship between skepticism, rationality, and
truth. Supposing for the sake of argument that what I have said so far
about the status of epistemic rationality is correct, there is still a form of
skepticism, usually associated with certain radically Cartesian scenarios,
which could raise the following kind of worry. Let us grant that epis-
temic rationality works the way it has been characterized in this chapter.
Still, what about the possibility that, after all, there is no external world,
and that rationality and truth come apart? That is, what about the possi-
bility that, despite all our epistemic practices that certify as true certain
empirical propositions thanks to certain background assumptions, the
world is actually different from what we take it to be? Suppose that the
possibility of inconceivable, or at least extremely far-fetched error possi-
bilities beyond our recognition, were to obtain. Could it not therefore be
the case that we behave rationally, from an epistemic point of view, yet
are systematically wrong?39
I think this kind of worry is worth taking seriously because it actually
raises the issue of how we can combine a certain view of rationality – the
one presented so far – with the idea that after all it tracks the truth, at
least in some sense. Now, the problem is that if truth is conceived in a
strongly realist fashion as totally mind-independent, it is difficult to see
why there should be such a fit between what we consider as rational and
how things are independently of us. To put it emphatically: why should
we be so optimistic about actually tracking the truth, while behaving in
an epistemically kosher way, once reality is conceived of as being what
it is, irrespective of how we represent it as being?
Obviously epistemic externalism tries to meet this challenge by
requiring that knowledge and justification be intimately tied to truth.
However, insofar as they are happy with a realist, evidence-transcendent
notion of truth, they cannot exclude that, unbeknownst to us, things
are actually different from how they seem to us. Consequently, we either
end up having no knowledge and/or justification, after all, or at least no
really truth-tracking justification for our beliefs.
In order to secure the match between epistemic rationality and truth,

however, it does not seem to be a viable possibility to opt for a tradi-
tional anti-realist conception of truth either. For, if what I have been
maintaining so far is correct, although nothing speaks against certain
background presuppositions, nothing is proof of their truth either, since
all empirical evidence owes its status to taking them for granted. Hence,
we seem to be faced with one last problem. Let it be that we are being
epistemically rational, but there is no guarantee that rationality and the
world, so to speak, are made for one another and that they cooperate to
guarantee that whenever we are being epistemically rational we are actu-
ally, or at least mostly, correctly representing the world in our beliefs.
Notice, moreover, that in a passing remark towards the end of §2, I said
that having a rational mandate for certain basic assumptions is no proof
or evidence of their truth. Hence, the important lesson to draw is that
epistemic rationality, evidence, and truth, at bottom, that is, when basic
assumptions are concerned, may actually come apart. Yet, throughout
this study, I have also repeatedly remarked that, in my view, these basic
assumptions are true and that this, apart from having certain advantages
over other positions, was what differentiated mine from Wittgenstein’s
in On Certainty, given the, so-called, framework reading of that work I
have endorsed. Is there an inconsistency between these remarks? I do
not think so, because what I meant in the passage from §2 was that if
epistemic rationality is extended in the way proposed its basic assump-
tions are epistemically rational. Yet, they are not justified either a poste-
riori or a priori. Furthermore, they are epistemically rationally mandated
all right, but this does not make them likely, or more likely, true. Simply,
they stay put and allow us to acquire evidential warrants for our beliefs
about material objects in our surroundings. Of course, we are implicitly
committed to their truth, and we reflectively treat them as such when-
ever we draw out the consequences of our ordinary beliefs. Therefore,
we are happy to infer that since there is one hand here, then there is
an external world. Such an entailment would not make sense if we did
not consider its consequent true, or at any rate, truth-evaluable. That is
(partly) why we were not content with Wittgenstein’s account of hinges
in terms of rules in On Certainty. For it would then be difficult to explain
how these assumptions could be embedded in such a kind of context.
Yet, it remains that we have no epistemic justification for them. More
precisely, we have no evidence that speaks against them and everything
we consider as evidence in favor ordinary empirical propositions, which
entail these assumptions, presupposes them. Given such a dependence,
no evidence can be proof of their truth, although the lack of contrary
evidence can boost our confidence in them. Yet, we cannot be realists

about truth either and somehow posit the truth of these assumptions
beyond our possibility of recognition. For, as we saw a few paragraphs
back, endorsing such a conception of truth could actually cut both ways.
That is, it could equally support the idea that, after all, things are not at
all how we take them to be, despite our behaving rationally. Is there a
way out of this apparently insurmountable problem?
My suggestion is that the only option left to us is to characterize the truth
of these assumptions in a minimal way. That is to say, by renouncing any
robust metaphysical implication and by acknowledging that their truth
depends merely on the kind of role they play in our epistemic system. For
they are the assumptions that make it possible and that thus allow us to
acquire evidence for other propositions so that we can actually consider
the latter justified, true, in an evidentially constrained way, and even
known. The truth of these assumptions, therefore, is neither of a robustly
realist kind, nor of a traditional anti-realist brand. For, according to tradi-
tional anti-realists, truth depends on justification (no matter whether at
the limit of inquiry, or in ideal circumstances, or when no increment in
the state of information would make a difference).40 Since these assump-
tions are not warrantable, they cannot be true in such an anti-realist
fashion. Yet, they need not be true in any robustly realist sense either.
Certainly, to say that these assumptions are rationally mandated does
not mean to say that they are mandated because they correspond to how
things are, independently of us. To predicate their truth, rather, should
be seen simply as tantamount to being prepared to act on their basis
and even to judge and assert them, thus being disposed to present them
as true. It means you need to think that they have a semantic content,
which can be meaningfully negated, or inserted within conditional state-
ments. Finally, and more importantly, to say that they are true is equiva-
lent to holding that what they state is how things are, given our overall
Weltbild. This latter qualification, I think, suffices to make us appreciate
that the kind of minimal truth we accord to these assumptions is, after
all, of an anti-realist nature. Not because it depends on evidence, but
because it is seen as dependent on how we, human beings, experience
the world around us and conceive of such an experience as bringing to
bear on mind-independent objects. To put it in the classic terms of the
Euthyphro contrast:41 these assumptions are not held true by us because
they hold independently of us and are accordingly represented. Rather,
things are as they state they are because we hold these assumptions and
take the world to be the way they present it as being, and so we cannot
but consider them as true.
On this conception of truth, we can reconcile epistemic rationality

and truth and show why no skepticism of the kind we consider in this
section with respect to our basic assumptions is justified. For they are
minimally true and mandated by epistemic rationality itself. To think
that they may, after all, be false, would thus simply depend on relying
on a form of robust realism, according to which, even if a proposition is
held true by us, all justifications depend on holding it true and therefore
nothing could speak against it, and even if it is actually mandated by
epistemic rationality itself, it could still be false. Yet, opting for such a
realist conception of truth would be understandable only if there were
no other available options. I have tried to show why this is not so and
thus to make available an alternative picture of our place in the world.
No doubt, it would need development, but I think that the preceding is
enough to motivate a partial re-orientation of the debate about skepti-
cism to make it tackle more directly the issue of the relationship between
rationality and truth. That is to say, the issue of the relationship between
epistemology and metaphysics.42
7 Summary
In this chapter we have considered the challenge of Humean descent,

of explaining how our basic assumptions, such as “There is an external
world”, can be epistemically rational while not being warranted or
warrantable. The key move has been to notice that our notion of epis-
temic rationality extends to all presuppositions that make the acquisi-
tion of perceptual warrants possible in the first place and are therefore
constitutive of epistemic rationality. For the notion of epistemic
rationality does not hang in the air but it actually depends on the
basic practice of acquiring, assessing, and withdrawing from empirical
beliefs on the basis of perceptual evidence. This has allowed us to
provide an indirect – namely, a diagnostic – answer to skepticism.
That is to say, a kind of response that does not meet the skeptical
challenge head-on by providing warrants for these basic assumptions;
but, rather, a reply that shows why asking for them is inappropriate.
It is inappropriate, to repeat, because it depends on too narrow and
unmotivated a conception of epistemic rationality. A conception, that
is, which confines epistemic rationality only to warranted beliefs,
while it actually extends to those assumptions which, while unwar-
rantable, make the acquisition of perceptual warrants possible in the
first place. Thus, we have claimed that although these assumptions
are unwarrantable, they are mandated by the very lights of epistemic
rationality itself. Hence, they are epistemically rational, even though

unwarrantable.
We have then applied this result to counter Cartesian forms of skepti-
cism based on the application of the principle of closure for epistemic
operators, such as warrant and knowledge under known entailment.
Accordingly, we have remarked that from the fact that there are no
warrants for basic assumptions such as “There is an external world” or
“I am not the victim of a lucid and sustained dream”, it does not follow
that there is no warrant for “Here is a hand” or “I am in the kitchen
right now”. The denial of the absolute validity of closure motivated in
Chapter 3 allows us to block this result. However, the strategy developed
in the present chapter allows us to take the issue one step further. For it
has become possible for us to hold that these assumptions are rationally
mandated and that at least rational mandates are closed under known
entailment.
In a similar vein, the endorsement of the extended rationality view
has allowed us to counter the objection from abominable conjunctions
raised by DeRose and briefly presented in Chapter 3. The key move has
been to notice how our account licenses the following kind of conjunc-
tion, “I am not warranted in believing that I am not presently dreaming
(or that I am not a BIV), but I am mandated by epistemic rationality
to assume that I am not and I am warranted in believing of having a
hand (based on my current perceptions)”. Such a conjunction is, I have
claimed, far from being abominable. It usually escapes notice because it
takes quite a lot of philosophical theorizing to see the difference between
epistemic warrants and rational mandates.
We then moved on to consider a possible development of Humean
skepticism in the form of a specific kind of epistemic relativism. That
is to say, a type of relativism which raises the possibility that there
might be, at least in principle, different notions of epistemic rationality,
defined by basic epistemic practices (or methods) each depending on
assumptions incompatible with ours. We have pointed out how, in order
to pose a serious challenge, that is, to present a serious alternative to our
notion of epistemic rationality, this form of epistemic relativism should
meet two constraints. Firstly, the alternatives should be intelligible to
us and, secondly, they should be able to account for the datum that the
very content of perceptual experiences is as objective as ours is, that is,
as of objects, properties, and states of affairs out there. We have seen
how the latter requirement is not met by phenomenalist conceptions
of experience and the former is not met by far-fetched hypotheses that
propose a way of acquiring knowledge of physical objects in ways other
than those mediated by perception, and thanks to a hinge like “There is

We then moved on to consider the Oblomovian challenge, which
points out that playing the game of epistemic rationality is not inevi-
table or compulsory. Although, clearly, not playing it would be disad-
vantageous, we have agreed that the extended rationality view does
nothing to convince someone who does not want to play it. Yet, that
is not its business, nor, generally, is that the business of epistemology.
Rather, what we aim at in epistemology and what the extended ration-
ality view delivers is to show how its constitutive assumptions fall, after
all, within the boundaries of epistemic rationality itself.
Finally, we considered the deep and often neglected issue of the rela-
tionship between epistemology and metaphysics in the form of the rela-
tionship between the extended rationality view and truth. The account
we have proposed is one according to which our basic assumptions are
minimally true. In particular, they are not true because they correspond
to how things are independently of how we think of them, but they
are true because we do have a certain Weltbild that presents things as
being that way. That in turn allows us to consider ordinary empirical
propositions true because there is evidence that speaks in favor of them.
It further allows us to dismiss the skeptical hypothesis that, after all,
things might be other than what we take them to be. Such a hypoth-
esis – we have claimed – is due to sticking to a realist conception of
truth that, paradoxically, is the strongest ally of skepticism itself and
seems inescapable once it is acknowledged, as we do acknowledge, that
truth, when it comes to fundamental assumptions, cannot be eviden-
tially constrained. Since, however, there is an alternative to such a realist
conception, I claim that we had better take advantage of it, in order to
offer a yet deeper analysis of the sources of the skeptical mistake and of
the origins of our fascination with it.
5
The Extended Rationality View
Extended
In this chapter, we extend the extended rationality view presented in

Chapter 4 to the case of inductive reasoning (§1). In particular, to the
principle on which enumerative inductions are based, that is the one
that Hume called “the principle of the uniformity of nature”. We then
propose its extension to the case of memory-based and testimonial justi-
fications and their characteristic assumptions, or hinges, such as “There
is a past”, “Informants are generally reliable” and, perhaps surprisingly,
“There are other minds” (§2–3). It is argued that each of these kinds of
justification is central to some aspect of epistemic rationality, consid-
ered in its full-blown version. That is to say, as involving not just percep-
tual justifications for present-tense propositions about mid-size objects
currently perceived, but also inductive, diachronic, and social elements.
Namely, justifications for generalized empirical propositions based on
induction, for past-tense propositions based on memory, and justifica-
tions for propositions that are obtained through testimony.
We then close by showing how the extended rationality view can be
profitably employed in connection with those basic rules of inference
(§4) central to another aspect of epistemic rationality, which may be
called “deductive epistemic rationality”. Its characteristic is not that it
allows us to extend our knowledge to new cases in ways which may fall
short of certainty, as induction does. Nor is it based on characteristic
assumptions or hinges. Rather, certain basic forms of inference, such
as modus ponens, allow us to draw out the consequences of our other
beliefs in such a way that, if the propositions that figure as premises are
true, the conclusion will necessarily be true too.
By so doing, we show how fruitful the extended rationality view is and
we conclude that such a wide survey allows us to maintain that the human
condition is as the extended version of the extended rationality view sees
153
it. Namely, epistemic rationality tout court is possible only thanks to several
general presuppositions, or hinges, and to basic rules of inference that are
constitutive of it. Such hinges and rules are therefore rational in their turn.
1 The extended rationality view and the principle of the

uniformity of nature
Consider the following enumerative inductive inference:
SUN
The sun rose yesterday

The sun rose the day before yesterday
The sun rose the day before the day before yesterday
... ...
The sun will rise tomorrow
Clearly the conclusion is not based just on the list of premises, but also
on the further assumption that what has constantly happened in the
past will keep happening in the future. Hume called this assumption the
principle of the uniformity of nature.
What can justify it? Surely, it cannot be justified through other induc-
tive inferences since they presuppose it. Nor can it be justified a priori
by reflecting on the meaning of the words needed to formulate it, or the
concepts necessary to grasp it. It is a general, yet contingent truth and it
is indeed difficult to see how it could be justified a priori.
Hume held the view that it cannot be justified but that it is inescap-
able, since we find ourselves compelled by Nature to hold onto it. The
precise causal story he put forward to substantiate the claim that we
cannot help abiding by it, given our psychological constitution, will not
be our concern here. Yet, we can see that in the case of the principle of
the uniformity of nature we are confronted with an assumption which –
like “There is an external world” – is general, empirical, and necessary to
perform certain kinds of inference central to our reasoning procedures,
that is (enumerative) inductive ones.
Wittgenstein also mentions it, in a brief remark in On Certainty
(315), which needs to be related to the preceding entry to be fully
understood:
Imagine that the schoolboy really did ask “and is there a table there
even when I turn round, and even when no one is there to see it?” Is
the teacher to reassure him – and say “of course there is!”?
The Extended Rationality View Extended 155
Perhaps the teacher will get a bit impatient, but think that the boy
will grow out of asking such questions.
That is to say, the teacher will feel that this is not a legitimate ques-
tion at all.
And it would be just the same if the pupil cast doubt on the uniformity
of nature, that is to say on the justification of inductive arguments. –
The teacher would feel that this was only holding them up, that this
way the pupil would only get stuck and make no progress. – And he
would be right. It would be as if someone were looking for some object
in a room; he opens a drawer and does not see it there; then he closes
it again, waits, and opens it once more to see if perhaps it is not there
now, and keeps on like that. He hasn’t learned to look for things. And
in the same way this pupil has not learned how to ask questions. He
has not learned the game that we are trying to teach him.
The affinities between Wittgenstein and Hume are clear. The justifica-
tion of the principle of the uniformity of nature does not come from
further inductive inferences, or a priori. Yet, Wittgenstein’s naturalism
is different, in that no appeal to our subpersonal psychological mecha-
nism is made. Rather, we have to learn how to play the game, in this
case the game of making inductive inferences. Once we have learned
it through repeated exposure within our community to instances of it
and are seen as competent inductive reasoners, we will also have “swal-
lowed”, as it were, the basic assumption on which enumerative inductive
inferences rest, that is “What has constantly happened in the past will
keep happening in the future”. If so, it too would be a hinge. Therefore,
it would be neither known nor unknown, justified nor unjustified, true
nor false and it would be like a rule, which has to be in place for the door
to turn, that is, in order for inductive inferences to be possible.
It has to be noted, however, that such a practice is, in fact, constitu-
tive of a specific aspect of epistemic rationality. Namely, what we call
inductive epistemic rationality. Its importance cannot be overestimated.
It allows us to extend our knowledge based on observed cases to new,
unobserved ones. It also allows us to make predictions about the future,
which are fundamental to our ordinary lives and scientific investiga-
tions. Accordingly, it is natural, in my view, to extend the extended
rationality view as follows:
Inductive epistemic rationalityER: it is inductively epistemically

rational to believe inductively warranted propositions and to accept
those unwarrantable assumptions that make the acquisition of
inductive warrants possible in the first place and are therefore consti-
tutive of inductive warrants.
Here too the critical elements are the notion of acceptance used in the
definition. Similarly to the case of “There is an external world”, it will
have to be kept in mind that we are considering propositional justifica-
tions, of an inductive kind, for specific propositions. Hence, the fact
that actual subjects’ abilities may fall short of allowing them to enter-
tain the proposition that what has constantly happened in the past will
keep happening in the future will not impair the treatment of induction
we are proposing. Furthermore, like in the case of “There is an external
world”, subjects can be granted with the acceptance of that assumption
if, while unable to conceptualize it, they are capable of taking part in a
practice that has that assumption as its rational precondition. Finally,
as in the case of “There is an external world”, I consider such subjects
to be committed to the uniformity of nature in their reasoning and
acting.
The principle of the uniformity of nature has to be coupled with
other ingredients to give rise to inductive justifications. For instance,
the number of cases observed has to be large enough to sustain the
generalization. Furthermore, the sample from which one general-
izes has to be appropriately diverse. Finally, it has to be kept in mind
that inductive inferences are never deductively valid. Therefore, even
if one has met these desiderata it will always be possible to find a
counterexample.
Extending the extended rationality view to the principle of the
uniformity of nature allows us to counter those who may think that
it lies beyond rationality because it cannot be justified evidentially or
non-evidentially. Accordingly, we are rationally mandated, by the lights
of inductive epistemic rationality itself, to hold on to it.
A rational mandate is not an epistemic good, of either an evidential
or a non-evidential nature, which can speak to the truth of that prin-
ciple. Indeed, in this case it seems clear that we will certainly transcend
our epistemic rights if we have the pretension of holding it true in a
robustly realist way. After all, what could guarantee us of its truth in
the distant future? Yet, it is a crucial element of our Weltbild. Therefore,
here again, we can think of its truth in a minimal way. That is to say,
by renouncing any robust metaphysical implication and by acknowl-
edging that it depends merely on the kind of role the principle plays
in our epistemic system, for it allows us to acquire evidence for other
propositions. Thus, we can actually consider the latter justified, true,

in an evidentially constrained way, and even known. The truth of
the principle, therefore, is neither of a robustly realist kind, nor of a
traditional anti-realist brand, since it cannot be non-circularly induc-
tively supported. To predicate its truth, rather, should be seen simply
as tantamount to being prepared to act on it and even to judge and
assert it, thus being disposed to present it as true. It means to think
that it has a semantic content, which can be meaningfully negated, or
inserted within conditional statements. Finally, and more importantly,
to say that it is true is equivalent to holding that what it states is how
things are, given our overall Weltbild. This latter qualification suffices
once more, in my view, to make us appreciate that the kind of minimal
truth we accord to this principle is, after all, of an anti-realist nature.
Not because it depends on evidence, but because it is seen as dependent
on how we, human beings, experience the world and conceive of such
an experience as bringing to bear on novel cases. To put it once more
in the classic terms of the Euthyphro contrast: this principle is not held
true by us because it holds independently of us and is accordingly repre-
sented. Rather, things are as it states they are because we hold it and
take the world to be the way it presents it as being, and so we cannot
but consider it as true.
To do without it, in the way an Oblomovian suggests, actually deprives
us of inductive justifications. Consider SUN. Of course, one may infer to
its conclusion, given its premises, without endorsing the principle. That,
however, would be a mere jump in the dark, like a leap of faith. Hence,
for inductive warrants properly regarded, the principle is not optional.
Nor is it easy to see how it could be replaced by any other principle
compatible with the existence of inductive justification. Here again,
however, the problem is not that of explaining why these justifications
are very useful to us – which really needs no explaining – or why we
may even find it psychologically compulsory to form beliefs about new
cases, given what has been observed in a number of similar previous
ones. The problem, like in the case of “There is an external world” is
that of accounting for the epistemic rationality of the principle of the
uniformity of nature, vis-à-vis those critics who may want to consider
it as lying beyond rationality, as it cannot be justified empirically, a
priori, or through entitlements. That is what the extended rationality
view delivers. For in that view “What has constantly happened in the
past will keep repeating itself in the future” will be rationally accepted
because it is constitutive of inductive epistemic rationality.
2 The extended rationality view and the past
Consider the following inferences:
PAST 1
(I) The Earth has existed for a very long time (based on geological and
historical evidence)
(II) If the Earth has existed for a very long time, there is a long-term past
(III) There is a long-term past
PAST 2
(I) In 49 BC Caesar crossed the Rubicon (based on historical

memories)
(II) If in 49 BC Caesar crossed the Rubicon, there is a long-term past
(III) There is a long-term past
PAST 3
(I) Thirty years ago I was living in Bologna (based on memory)

(II) If thirty years ago I was living in Bologna, there is a mid-term past
(III) There is a mid-term past
PAST 4
(I) I was in Modena yesterday (based on memory)

(II) If I was in Modena yesterday, there is a short-term past
(III) There is a short-term past
All these inferences present a common, and by now familiar, structure.

For their respective first premises to be justified, it is not enough merely
to have a certain body of evidence – different in each case – but it is also
necessary that their respective conclusions are in place. For, otherwise, all
the relevant evidence will have no bearing on propositions concerning
the age of the Earth, Caesar crossing the Rubicon, and myself thirty
years ago or yesterday. In each case, the evidence has a certain nature – it
depends on memory, testimony, geology, and history – but it is by itself
compatible with the possibility of its being appropriately crafted and
forged if there were no long-, mid-, or short-term past. In that case, then,
the first premise in each argument will not be justified.
Can these assumptions be independently justified? Clearly not. The
suggestion that some of them could be justified through testimony in
fact presupposes the existence of other subjects who have witnessed and
reported on the relevant assumptions. However, their reports enjoy a
justificatory status only if they did witness events in the past and if it is
granted that other minded subjects have been in a position to do all that
(we will return to that assumption in the next section). A priori justifi-
cations for such presuppositions, or for the more generic and common
one they entail, that is “There is a past”, are also hard to come by. After
all, here we are dealing with (a) general, yet empirical and contingent
proposition(s) and it is difficult to see how they (it) could be justified a
priori. Finally, entitlements for them (it) would be as problematic as the
ones for “There is an external world” we reviewed in previous chapters.
It is therefore tempting to extend the extended rationality view to them
(it) too. For ease of exposition I will therefore concentrate from now on
on the proposition “There is a past”.
To such an end it is key to recognize that “There is a past” is a neces-
sary presupposition of all memory-based justifications, both in the sense
of personal and historical memory. The case of geology is more contro-
versial. For it takes us into the complex terrain of scientific investiga-
tion and evidence. I wish to stay clear of all that, though I am inclined
to think that even in that case the extended rationality view would
prove useful. Therefore, focusing on memory, in the extended sense just
mentioned, it should be clear how it constitutes one of our basic sources
of justification for beliefs whose content concerns the past. As is well
known, there is a complex debate about whether memory can generate
justifications that were not available when a remembered event took
place, or whether it can only transmit justifications that were already
present at that time. What I am going to say is entirely orthogonal to
that debate. For, either way, for memory to generate or simply transmit
justification for beliefs about the past, the existence of the past will have
to be taken for granted.
Now, the centrality of our practice of forming beliefs about the past
based on memory, in the extended sense also involving testimony, can
hardly be overestimated. If we were confined only to justified beliefs
about our own present, we would hardly ever augment our knowl-
edge. Furthermore, as we saw in §1, inductive rationality depends on
the possibility of extending our justifications to new cases, starting
with beliefs about a number of past events, justified either through
one’s own memory or testimony that allow for generalizations to new
cases, either in the present or the future. Memory, in the extended sense
we are concerned with here, is therefore the diachronic equivalent of
observation with respect to the epistemic rationality we analyzed in the
previous chapters of this book. This suggests the following extension of

the extended rationality view:
Diachronic epistemic rationalityER: it is diachronically epistemi-

cally rational to believe propositions about the past on the basis of
memory-based warrants and to accept those unwarrantable assump-
tions that make the acquisition of memory-based warrants possible
in the first place and are therefore constitutive of those warrants.
Here too one critical element is the notion of acceptance used in the
definition. Similarly to the case of “There is an external world”, it will
have to be kept in mind that we are considering propositional justifica-
tions, of a memory-based kind, for specific propositions. Hence, the fact
that actual subjects’ abilities may fall short of allowing them to entertain
the proposition that there is a past will not impair the treatment of the
memory-based justification we are proposing. Furthermore, as in the case
of “There is an external world”, subjects may be granted with the accept-
ance of the relevant assumption if, while unable to conceptualize it, they
are capable of taking part in a practice that has that assumption as its
rational precondition, such as the one of forming justified beliefs about
the past on the basis of memory. Finally, as in the case of “There is an
external world”, I consider such subjects to be committed to the exist-
ence of the past in their reasoning and acting, when they act and reason
in ways which only make sense if that assumption is taken for granted.
The proposition that there is a past will have to be coupled with other
ingredients to give rise to memory-based justifications. For instance, in
the case of personal memory, it will have to be coupled with memories
or, at least, memory experiences, absent defeaters. In the case of testi-
monies, it will have to be coupled with the relevant statements to be
found in books or other supporting material, or from living subjects (we
will return to the case of testimony in the following section). Still, in all
these cases it is possible for the ensuing justifications to be defeated if
new information comes in.
The extended rationality view extended to the case of the existence of
the past allows us to counter those who may think that that assumption
lies beyond rationality because it cannot be justified evidentially – either
empirically or a priori – or non-evidentially. Accordingly, we are ration-
ally mandated, by the lights of diachronic epistemic rationality itself, to
hold on to it.
However, a rational mandate is not an epistemic good, of either an
evidential or a non-evidential nature, that can speak to the truth of that
assumption. Therefore, as in previous cases, “There is a past” will have to

be considered minimally true and required by our Weltbild.
Finally, to do without it, in the way an Oblomovian suggests, actually
deprives us of memory-based justifications. Once more, the problem is
not that of explaining why they are very useful to us or why we may
even find it psychologically compulsory to form beliefs, at least on the
basis of personal memories, given the kind of creatures we are. Nor is it
to explain why we think of them as justified. The problem, as in the case
of “There is an external world”, is that of accounting for the epistemic
rationality of its constitutive assumption, vis-à-vis those critics who may
want to consider it as lying beyond rationality, as it cannot be justified
empirically, a priori, or through entitlements. That is what the extended
rationality view delivers. From that viewpoint “There is a past” will be
epistemically rationally accepted for it is constitutive of diachronic epis-
temic rationality.
3 The extended rationality view, testimony,

and other minds
In the previous section, we partly hinted at the relevance of testimony to

diachronic epistemic rationality through the case of historical evidence.
We saw how testimony plays a crucial role in allowing us to have
justifications for beliefs whose contents concern the past. Diachronic
epistemic rationality, however, considered that way, is a form of social
epistemic rationality. For the justification one has for a given belief, or
for entire classes of propositions in a certain domain of discourse which
may become the objects of one’s belief, will always be due to one’s inter-
actions with other members of one’s epistemic society. We now consider
testimony in more detail and will see how it is a crucial ingredient of
epistemic rationality appreciated in its full extent, as having an impor-
tant social aspect. We then consider its constitutive assumptions and
will therefore claim that they are epistemically rationally accepted, even
if they are not justified or justifiable, for they are constitutive of what
may be called social epistemic rationality.
Consider my belief that I was born in Milan in 1973. If I am justi-
fied in having it, it cannot be through experience and memory of that
event. I am justified in having it, but my justification comes from what
my parents told me, what I found written in various official documents,
and so on. My justification for it is entirely testimonial and there is
nothing I can do that could give me a more direct justification. I cannot
go back in time and check for myself. I cannot appeal to memory for,
as an infant, I had none and certainly not of such a complex nature

(involving reference to geographical places and dates in the Gregorian
calendar). The justification I have for it is, therefore, irreducibly testimo-
nial in nature. That, of course, is entirely consistent with holding that
testimony merely transmits justifications, of a different kind, possessed
by those who passed on the relevant belief, as well as with the view
that testimony can, in fact, produce justification. The debate between
reductionists and anti-reductionists regarding testimony hinges, at
least partly, on that and what I have to say about testimony is entirely
orthogonal to that aspect of their dispute. All that is needed for present
purposes is that for certain propositions and certain subjects, the justi-
fication that those subjects may have to believe those propositions is
irreducibly testimonial.
Now consider the following inference:
BIRTH
(I) My mum tells me “I was born in Milan in 1973”

(II) I was born in Milan in 1973
(III) If I was born in Milan in 1973, my mum’s specific testimony is
reliable
(IV) My mum’s specific testimony is reliable
According to anti-reductionists,1 if there are no defeaters, given (I) one

would ipso facto have a justification for (II). If that were the case, then
BIRTH would give one a justification to believe (IV). Yet, clearly, there
would be something amiss in grounding one’s belief in the reliability of
an informant just on the basis of one piece of testimony coming from
that very informant. Still, it is difficult to believe that one would have
to pile up testimonies from that informant before being entitled to trust
her. That may not be feasible if the informant were no longer available
(think of a case where your informant is someone you had not met
before and whom you will never meet again). Alternatively, the kind of
information passed on by her on different occasions may be such that
one could not check, independent of her testimony, whether things are
(or were) as she said they are (or were). (Consider the case in which your
mum told you that, at birth, you had a red spot on the back of your
head, which, like many such spots, disappeared soon afterwards and no
one else in your family noticed it at the time).2
One suggestion I would like to put forward is that there may be a
midway position between anti-reductionists, who would hold that
the passage from (I) to (II) is safe, epistemically, provided there are no
defeaters, and reductionists who would hold that an independent justi-
fication for (IV) is needed to have a justification for (II), given (I). Such
a suggestion may allow one to counter the allegation of gullibility anti-
reductionists often fall prey to, as well as the allegation, often raised
against reductionists, of making testimonial justifications much harder
to come by than one would be willing to allow.
According to what could be called a moderate conception of the
structure of testimonial justification, the correct account of testimonial
warrants would be:
Moderate account of testimonial warrants: a proposition P is testi-

monially justified iff (i) one has the testimony that P and, (ii) absent
defeaters, (iii) the background assumption that informants are gener-
ally reliable is in place.
It would have to be contrasted with those whom we may count as

liberals regarding testimonial justification, who would stop at (ii) and
with those whom we may consider conservatives with respect to it, who
would require the replacement of (iii) with (iii*), the specific informant
who said that P should be warrantedly believed to be usually reliable.
According to moderates on testimonial justification, “Informants are
usually reliable” would be a hinge of the practice of forming, assessing,
and withdrawing from empirical beliefs on the basis of testimonies.
Of course, with such general assumptions in place, the passage from (I)
to (II) would seem less prone to the charge of gullibility. Moreover, one
would have a justification for (II), which, while clearly very weak, would
transmit to (IV). What would be the epistemic status of that assumption
though? One possibility is to hold that it is justified through repeated
exposure to people’s testimony and inductive generalization. That would
be tantamount to saying that our justification for “Informants are usually
reliable” would come from many instances of arguments such as BIRTH,
together with inductive generalizations. However, as we have seen, in
the moderate conception of testimonial justification, those arguments
presuppose the background assumption they would eventually justify.
Therefore, this cannot be the way that assumption may be warranted.
One could think that independent justifications for conclusions
such as (IV) in several arguments of the kind of BIRTH, could be had
by independently verifying the reliability of the information received.
Hence, the informant would say P, we would verify it independently
and, supposing we succeeded, we could warrantedly infer that the
informant has been reliable with respect to P. We would have to do that

many times, for many informants and many Ps and then generalize to
“People are usually reliable”. Not only would this seem unfeasible, but
it would also impair the very possibility of going from (I) to (II), in an
argument such as BIRTH, in a justifiable way. If we had independently
to check a lot of information received through testimony, to be in a
position, through arguments and generalization, to have a justification
for the hinge “Informants are usually reliable”, one could achieve genu-
inely testimonial justification only after long exposure to testimonies
and independent verification of the information transmitted through
them.
Can there be a priori arguments that would justify that assumption?
Again, “Informants are usually reliable” is a general, empirical, entirely
contingent proposition. It is very difficult to see where such a priori
arguments would come from. Wright-style entitlements for such a
general background presupposition would be problematical too, for the
reasons rehearsed in the course of this book. It is therefore tempting to
apply to it the extended rationality strategy.
To that end, it is important to realize, as we anticipated at the begin-
ning of this section, that epistemic rationality in its full-blown version
comprises synchronic, diachronic, and social elements. Its social dimen-
sion can be accounted for in terms of
Social epistemic rationalityER: it is socially epistemically rational

to believe propositions on the basis of testimonial warrants and to
accept those unwarrantable assumptions that make the acquisition
of testimonial warrants possible in the first place and are therefore
constitutive of them.
Before analyzing the main features of this definition, let me also

note that one constitutive assumption of social epistemic rationality
is that there are informants. Informants are generally other human
beings3 who either directly or indirectly, through records and docu-
ments, pass information on to us, and do so intentionally. A constitu-
tive assumption of social epistemic rationality is, therefore, that there
be other minds, capable of processing information, of encoding it,
and of passing it on to other subjects, who, in their turn, are minded
in the full-blown sense required for processing information received
that way.
Let us move on to analyze the main aspects of the definition. Here too
the critical element is the notion of acceptance used in the definition.
As in the case of “There is an external world”, it will have to be kept in

mind that we are considering propositional justifications, of a testimo-
nial kind, for specific propositions. Hence, the fact that actual subjects’
abilities may fall short of allowing them to entertain the proposition
that there are other minds, or that people (or even more generically,
informants) are usually reliable, will not impair the treatment of testi-
monial justification we are proposing. Moreover, as in the case of “There
is an external world”, I would grant subjects with the acceptance of the
relevant assumptions if, while unable to conceptualize them, they are
capable of taking part in a practice that has those assumptions as its
rational precondition, such as the practice of forming justified beliefs
based on testimony. Finally, as in the case of “There is an external world”,
I consider such subjects to be committed to the existence of other minds
and to the general reliability of informants in their reasoning and acting,
when they act and reason in ways that only make sense if those assump-
tions are taken for granted.
The propositions that there are other minds and that informants
are generally reliable will have to be coupled with other ingredients
to give rise to testimonial justification. In particular, they will have to
be coupled with the relevant statements, as found in books or other
supporting material, or as made by living subjects, which will have to be
understood by the subjects who will eventually be in a position to form
justified beliefs on their basis. Clearly, it will always be possible for the
ensuing justifications to be defeated if new information comes in.
The extended rationality view, extended to the case of the existence
of other minds and of the general reliability of informants, allows us to
counter those who may think that those assumptions lie beyond ration-
ality because they cannot be justified evidentially – either empirically or
a priori – or non-evidentially. Accordingly, we are rationally mandated,
in the light of social epistemic rationality itself, to hold on to them.
However, a rational mandate is not an epistemic good, of either an
evidential or a non-evidential nature, which can speak to the truth of
those assumptions. Therefore, as in previous cases, “There are other
minds” and “Informants are usually reliable” will have to be considered
minimally true and required by our Weltbild.
Finally, doing without them, in the way Oblomovians suggest, actually
deprives us of testimonial justifications. Once more, the problem is not
that of explaining why they are very useful to us, or why we may even
find it psychologically compulsory to form the relevant beliefs, upon
receiving a certain piece of testimony, given the kind of creatures we are.
The problem, as in the case of “There is an external world”, is that of
accounting for the epistemic rationality of its constitutive assumptions,

vis-à-vis those critics who may want to consider them as lying beyond
rationality, as they cannot be justified empirically, a priori, or through
entitlements. That is what the extended rationality view delivers. For in
that view “There are other minds” and “Informants are usually reliable”
are epistemically rationally accepted for they are constitutive of social
epistemic rationality.
4 The extended rationality view and basic logical laws
It is a familiar problem in philosophy of logic how basic logical laws,

such as modus ponens, can be justified. Looked at more closely the
problem ramifies in several sub-problems: (1) what is an inference? (2)
which inferences are basic? (3) how does justification transmit from the
premises of an inference deploying a basic logical law to its conclusion?
(4) how is the very general law justifiably believed or even known to be
valid? These problems are related, so no wonder that discussions in this
area can oscillate between these problems, but surely the impression
one gets by reading the recent literature on the issue is that it is difficult
to pinpoint what the real problems are. Just to give an example: when
we ask what an inference is, are we talking about what it takes for a
subject to engage in an inference, or are we talking about a derivation
in a formal system? Are we concerned with the psychological nature of
the movement of thought, usually called inferring, or with an inference
in the abstract space of reasons and the various relations which hold
between the propositions that figure as its premises and its conclusion?4
Indeed, it is hard to shake off the impression that contemporary philo-
sophical literature in the epistemology of logic proceeds mostly by not
engaging at all with logic as a formal discipline. This impression is rein-
forced because, as we shall see, the issue of the justifiedness of basic
logical principles does not reduce to the problem of their validity, that is
to the classical problem addressed in logic of explaining their property
of being necessarily truth-preserving.
It is not the aim of this section to attempt to correct this trend. Given
the overall purpose of this book, I will concentrate on only two issues,
namely questions (3) and (4), and take for granted that there are basic
logical laws or principles, that is belief-forming methods which are
not employed on the basis of other belief-forming methods, and that
instances of modus ponens containing atomic sentences would qualify
as basic. Other basic logical laws could plausibly be conjunction elimi-
nation and disjunction introduction. The aim will be to show that in
this case too the extended rationality view will be of use in accounting
for at least some of these problems.
Hence, the first question on our agenda, which corresponds to a
possible precisification of (3) is: by virtue of what is a subject justified
in believing a conclusion, reached by deploying a modus ponens infer-
ence, starting from justified premises?5 To give an example: by virtue of
what is my first-year student Giorgio justified in believing “Anna will
take the umbrella” by reaching it through a reasoning that starts with
justified premises such as “It’s raining” and “If it rains Anna will take
the umbrella”?
The second question, in contrast, which corresponds to one possible
interpretation of question (4), is: by virtue of what, as theorists, can we
say that modus ponens in general – or, in other words, the very prin-
ciple – is epistemically justified? When we raise that question, it should
be kept in mind that we are looking for a justification that a theorist can
deploy to vindicate the claim that modus ponens is justified.6 That is to
say, we can grant that modus ponens is a valid rule of inference; we can
also grant that it is indeed a basic logical principle; but, we can still raise
the question of what makes it the case that modus ponens, in general,
is justified. We will come back to the relevance of this question in the
following.
The answer I want to propose to our first question – by virtue of what
is a subject justified in believing a conclusion such as “Anna will take
the umbrella”, upon inferring it from the justified premises “It’s raining”
and “If it rains Anna will take the umbrella”? – is, very crudely, nothing.
Assuming, of course, that he has reached the conclusion as a result of
having entertained and understood the premises and of having taken
them to be justified, at least pro tempore, if only for the sake of argu-
ment. Therefore, apart from having inferred it via an application of
modus ponens, a subject is not required to know anything about modus
ponens, not even that there is such a rule of inference, let alone have a
notion of its being a valid rule of inference. Nor does he need to possess
anything like a justification for it. That is to say, he need not have an
intuition of the validity of modus ponens, if such a thing existed.7 Nor
should he be able to provide an argument in favor of modus ponens, for
this would prevent many subjects from ever having such a justification.
Moreover, as we shall see at length in the following, such a justifica-
tion would indeed be circular, as it would have to rely at some point on
modus ponens (or on other basic rules of inference for which it would
then be an open question how a subject could be justified in employing
them). Finally, if one were not happy with crudely reliabilist notions of
justification,8 one would have to recognize that a subject could perfectly

well reason in accord with modus ponens and thereby reach a justified
conclusion, without having any justification for that rule. Alternatively,
one would have to at least admit that, even if a subject could have such
an externalist justification, he would not have one to offer in response
to the question of how he could be justified in holding a belief reached
via an application of modus ponens. Either way, his way of operating
in accord with modus ponens can be as “blind” as one wishes it to be,
even though, surely, it will genealogically depend on his possession of
the relevant concepts at least – in particular the concept IF THEN.9 That
is to say, since we are considering reasoning on certain propositional
contents and their relations, all concepts needed to grasp those contents
and logical relations will have to be assumed to be at the subject’s
disposal. Therefore, the epistemological issue moves to the second ques-
tion: Can we provide a justification for modus ponens?
Recall that the problem is not that of explaining the validity of modus
ponens but only why using it is epistemically justified: why is it the
right thing to do, epistemically speaking, to use modus ponens, in our
reasoning? Now, the problem is that, on a first-order reading of that
question, it seems that we are looking for an argument to show that
modus ponens is a right belief-forming method, as opposed to other
belief-forming methods, which are not epistemically kosher, such as
affirming the consequent. However, as repeatedly stressed, this cannot
be the main issue we are going to address. For, presumably, the answer to
that question is that modus ponens, as opposed to affirming the conse-
quent, say, is a valid rule of inference and, moreover, it is basic, inas-
much as it is presupposed by all other forms of reasoning. Therefore,
surely we cannot go astray by using it.
There is, however, a second-order reading of that question in the offing.
Namely, let us grant that modus ponens is a valid rule of inference, and
that we have formal means to prove that it is. How can we claim to
possess that knowledge? The trouble here seems to be that any reason
we might want to provide to that end will itself rely on the application
of modus ponens. Therefore, it would be circular and unsuitable as an
account of how we can claim to know that modus ponens is valid. To see
the point more clearly, consider the celebrated example of Achilles and
the tortoise by Lewis Carroll. Since any proof of the validity of modus
ponens we might give will presuppose reasoning in accord with it, how
could we use such a proof to convince someone who was not already
disposed to infer according to modus ponens? That proof would only
seem capable of convincing the converted.
There is, finally, a third possible interpretation of question (4),

according to which, once it is granted that modus ponens is valid and
is a basic belief-forming method, it is still an open question why we are
required to reason in accord with it. Or, to put it in hopefully clearer
terms, what would be wrong with someone, like the tortoise in Lewis
Carroll’s story, who recognized that P is justified (at least pro tempore
and for the sake of argument), who also admitted that “If P then Q” and
yet didn’t conclude that Q? Where the answer we were looking for was
not something that would convince such a stubborn creature, that is
something which should psychologically move her to conclude that Q.
Indeed, it is clear from the way the dialogue between her and Achilles
develops that she will never be moved to infer Q. Rather, the answer
we are searching for would be a diagnosis of what is wrong with her (if
anything at all), from an epistemological point of view, if she were to
persist in her behavior.
Hence, bearing in mind these important qualifications on the under-
standing of the problems we are addressing when we consider the issue
of the justifiedness of modus ponens, let us move on to some current
answers. Notice, however, that in the existing literature there is always
an oscillation between these various issues – that is, to prove that modus
ponens is epistemically kosher, that we know that it is, and to explain
why we are required to abide by it.
Recent attempts in this field have supported the view that its justifi-
cation depends on meaning-constitutive considerations regarding the
concepts involved in basic inferences deploying modus ponens, such as
IF THEN (Boghossian 2003).10 At first approximation (more on the issue
will follow), the idea would be that such a belief-forming method is
justified because the ability to reason in accord with it is constitutive
of the understanding of the concept IF THEN, which figures in it. So a
grasp of the latter concept requires being prepared to reason in accord
with modus ponens. What is wrong with the tortoise, therefore, is that,
insofar as she possesses the concept IF THEN, she is required to infer in
accord with modus ponens. By contraposition, if she does not, this
shows that she does not really have that concept in the first place.
These considerations have been opposed by several theorists who
have pressed the point that a grasp of the concept of the conditional is
not sufficient to provide a justification for modus ponens because one
can grasp the former while sensibly wondering whether modus ponens
is indeed valid, at least globally (McGee 1985 and Williamson 2003).
Hence, one could possess that concept while not being willing to reason
in accord with modus ponens. Another line of attack, pressed by the
late Paolo Casalegno (2004), is to say that there could be someone who
could be said to grasp the concept IF THEN and yet be prevented, by what-
ever causes, from drawing an inference in accord with modus ponens.11
In my view, these are not fatal objections to Boghossian’s proposal
since, arguably, the cases invoked by Williamson and McGee are not
actually basic instances of modus ponens, for they involve embedded
conditionals and depend on complex contextual information, which
could explain why we think it intuitive to reject a conclusion reached
via an application of modus ponens.12 Presumably, a supporter of a
meaning-constitutive account of the justifiedness of modus ponens
should qualify his claims so as to be able to single out those instances
of reasoning in accord with modus ponens that are actually constitu-
tive of having the concept IF THEN. Just to help see the point better:
those inferences could not plausibly be ones that require one to enter-
tain indefinitely long premises. Therefore, such a theorist would surely
be within his rights to confine his claim to basic instances of modus
ponens involving atomic sentences, once all potentially confusing
contextual elements have been avoided or disambiguated. Then it
would certainly become much more plausible that, unless one were
prepared to infer Q, given P and “If P then Q”, one would not have the
concept IF THEN.13
As to the objection raised by Casalegno, I think it remains an entirely
open question, one which is hard to see how to settle, whether a subject
unable – for whatever reason – to infer Q, given “If P then Q” and “P”,
could really be said to possess the concept IF THEN. For, ex hypothesi, such
a subject would be able to utter sentences containing “if then”. However,
since he would never be in a position to use it as a premise in an actual
chain of reasoning, it would be unclear what evidence there could be to
show that he does indeed have the corresponding concept.14
I find it a potentially more worrying objection that there are concepts
whose introduction and/or elimination rules give rise to invalid infer-
ences, such as TONK.
If A, then A TONK B If A TONK B, then B
The problem would arise when B = not-A, for from A its negation would
follow. We therefore seem to have concepts that license inferences, which
would have the remarkable consequence of leading us to unwarranted
conclusions. Moreover, there is no intuitive sense in which we should be
compelled to reason in accord with them, even though, in conceptual-
role semantics, possessing those very concepts would require us to have
the disposition at least to infer in accord with their constitutive rules of

inference.
As is well known, some theorists, like Christopher Peacocke (1992),
would say that TONK is not a genuine concept and would thus be able to
defuse this possible counterexample. However, there are cases that give
rise to equally invalid inferences and yet where it is difficult to deny
that genuine, albeit obnoxious, concepts are at issue. A case in point is
BOCHE.
If A is German, A is BOCHE If A is BOCHE, A is cruel
From which it would be inferred that all Germans are cruel. Hence,
the fact that reasoning in accord with those rules is needed in order to
possess the relevant concepts (for example, BOCHE) does not guarantee
that the form of inference utilized is valid. Furthermore, we face the
problem that to possess that concept we ought to be disposed to make
the relevant inferences, while, intuitively, non-racists could have the
concept BOCHE while not being disposed to conclude that all Germans
are cruel.
This poses the problem of better clarifying under which conditions
a rule of inference is justified, according to meaning-constitutive
accounts. Yet, to solve the problem posed by BOCHE, it is not enough
to require that the rule of inference under scrutiny should be neces-
sarily truth-preserving. The following example, from Boghossian (2003),
clearly illustrates this point.
If x is an elliptical equation, x is FLURG If x is FLURG, x can be correlated

with a modular form.
The Taniyama-Shimura conjecture, proved in 1999, states that all ellip-

tical equations can be correlated with modular forms. Hence, inferences
licensed by FLURG are necessarily truth-preserving. Yet, clearly, inferences
like the one just stated do not seem to be justified. They do not seem to
put a subject in a position to draw a justified conclusion, starting from
the allegedly justified premise that a given equation is elliptical, for they
introduce the non-existing entity flurg which, like phlogiston, can hardly
give rise to warranted conclusions. Furthermore, as in the previous case,
it seems quite intuitive to hold that one could have the concept FLURG
without thereby being disposed to draw the relevant inferences.
It is on the basis of considerations like the ones just explored that
Boghossian proposes that, whenever available, only the conditionalized
versions of the relevant rules of inference would be justified. Hence, the

correct conditional stipulation for FLURG would be as follows (Boghossian
2003, p. 247):
If there is a property which is such that, any elliptical equation has

it, and if something has it, then it can be correlated with a modular
form, then if x has that property, x is flurg.
In the case of IF THEN, however, we cannot conditionalize it without

circularity, for plainly the conditional would be needed to perform such
a conditionalization. Hence, according to Boghossian, the application
of modus ponens is justified because reasoning in accordance with it is
constitutive of possessing the concept IF THEN.
Several important criticisms have been raised against Boghossian’s
account, considered by all parties to be the best developed meaning-
constitutive explanation of the justifiedness of basic logical laws, such
as modus ponens. One very general worry is that the kind of mean-
ing-constitutive considerations advanced to justify modus ponens are
based on a conceptual-role semantics according to which, in order to
possess a given concept, one must be able to engage in certain char-
acteristic inferences. Such a semantics, however, is not rock-solid and,
notoriously, it proves difficult to extend it beyond the case of logical
constants and other concepts for which we have clear definitions, such
as BACHELOR. Indeed, in this section we have already encountered theo-
rists who would not be prepared to endorse such a semantic account for
logical constants. So, even if I am not entirely convinced by the reasons
advanced by detractors of inferential-role semantics in the case of logical
constants, I think that if it were possible to avoid starting from such a
contentious assumption, the ensuing epistemological enterprise would
certainly look more convincing (cf. also Williamson 2003).
Another related worry we have already encountered, is that it seems
that a subject can perfectly well grasp the concept IF THEN and yet
sensibly wonder whether modus ponens is a universally valid principle.
That is precisely the situation Van McGee would be in, according to
Williamson (2003). However, in response to such a worry, I have already
said that, pending a final judgment on McGee’s alleged counterexam-
ples to modus ponens, we should restrict it to its basic instances, such as
those that do not involve embedded conditionals and which do not rely
on potentially problematic contextual information.
Schechter and Enoch (2006) also point out that since not all mean-
ing-constitutive rules provide a justification for the corresponding rules
of inference, meaning-constitutive accounts cannot plausibly give us

the intended epistemological result. Here again, we have seen, at least
partly, how Boghossian might respond. He ought to confine the claim
to valid and basic rules of inference, such as modus ponens, and perhaps
allow it, at most, for other conditionalized, non-basic rules of inference
that are valid.
It has been noted, however, that this would mean forsaking any hope
of giving a uniform account of all basic belief-forming methods, such
as inference to the best explanation, or memory and perception (cf.
Schechter and Enoch 2006). For in the last two cases it is very difficult
to see from which concepts the alleged justification of those very belief-
forming methods should proceed. Clearly, the concepts of EXPLANATION,
MEMORY and PERCEPTION do not seem able to deliver the intended goods.
For they neither help explain why forming beliefs on the basis of one’s
perceptual experiences, memories, and inferences to the best explana-
tion would lead us to form justified beliefs, nor do they provide any
explanation as to why we are compelled by reason itself to use these
methods.
A related worry raised by Williamson (2012), this time applied only to
the epistemology of logic, is that the account proposed by Boghossian,
if viable at all, would hold only for very few basic logical laws. It might
hold for modus ponens and, more plausibly still, for conjunction elimi-
nation, but it already looks problematic for conjunction introduction
and disjunction elimination.
Another problem hinted at by Schechter and Enoch (2006) is that the
kind of account proposed by Boghossian at most explains the permission
to use modus ponens but not our epistemic obligation to do so. Clearly,
at least in cases such as modus ponens, we want to vindicate the latter,
not just the former. That is to say, we would like to be in possession of
an explanation of why the tortoise in Lewis Carroll’s example ought to
infer that Q, once she is prepared to accept P and “If P then Q” and not
merely an account of how it is permissible for her to do so. She knows –
we can assume – that she is permitted to infer that conclusion, but she
does not draw it. Therefore, what we need is an account of precisely why
she is somehow rationally defective in not drawing that inference.
I think that Boghossian might well respond by saying that even though
the possibility of generalizing would be a plus of a putative explanation,
it is no objection to his account that it does not generalize to all cases of
belief-forming methods, however basic they might be. He could actually
insist on the intuitive difference between basic logical principles, which
are closely connected to propositional contents, their relations, and the
conceptual ingredients necessary to grasp them, and other basic belief-

forming methods that take experiences as input, which may be noncon-
ceptual, to produce beliefs as outputs. He could also insist that a grasp of
the concept IF THEN requires – as opposed to merely permitting – reasoning
in accord with modus ponens. Indeed, it seems part of the story he tells us
regarding the justifiedness of modus ponens that it is necessary to apply
it to possess the relevant concepts. What is wrong with the tortoise is
that by refusing to infer Q, given P and “If P then Q”, she shows that she
does not possess the concept IF THEN, after all. Therefore, clearly, there is
room for maneuver here, to defend a meaning-constitutive approach to
the issue of the justifiedness of modus ponens. But Williamson’s point
regarding the difficulty of extending the account to the epistemology of
basic logical laws, beside the one or two for which it might be allowed to
work, still represents a challenge to the plausibility of the view. Hence,
as stressed at the beginning, I think that since Boghossian’s approach
relies on conceptual-role semantics, and that conceptual-role semantics is
problematic and certainly not universally accepted, it would be better to
provide an account that did not rely so heavily on a contentious assump-
tion. Furthermore, if it were possible, it would certainly be better to have
a general account of the justifiedness of basic belief-forming methods, or
at least one applicable to more than one specific case, especially when
confronted with the issue of the justifiedness of several basic logical laws.
For these reasons I think it is instructive to consider another strategy
recently adopted to justify modus ponens, proposed, in slightly different
fashions, by Wright (2004b), and Schechter and Enoch (2006, 2008).
The latter call it the pragmatic strategy – and rightly so, as we shall see –
even though, in my opinion, they do not really take the measure of this
fact. They also rightly point out (2008, p. 548) that, if it works at all, it
merely provides an entitlement for a second-order claim, that is “We
know/justifiably believe that modus ponens is valid” (cf. also Wright
2004b, p. 158), even though – to my mind – they do not seem to take
full account of this crucial fact either.15
Here is the backbone of their proposals. According to Schechter and
Enoch (2008, p. 554):
If a belief-forming method is such that it is possible to successfully

engage in a rationally required project by employing it, and such that
it is impossible to successfully engage in the project if the method
is ineffective, then we are prima facie epistemically justified in
employing that method as basic, even in the absence of a justified
belief concerning the method.
Consider for instance modus ponens, which is a belief-forming method

necessary to engaging in the rationally required project of reasoning. Hence,
we are prima facie epistemically justified in employing it, even if we do
not have any justified belief regarding it. That is to say, even if we cannot
provide a non-circular justification that could prove its correctness.
Schechter and Enoch introduce a number of qualifications, which
clarify and sharpen their proposal. For instance, they tell us that success
in engaging in the project need not involve achieving perfection (ivi,
p. 559). They also remark that it is conceivable that in distant possible
worlds the same project is successfully accomplished by applying very
different methods. Therefore, the notion of “impossibility” appealed to
in the previous definition is to be understood as relative to sufficiently
close possible worlds (ivi, p. 562).
Let us now turn to Wright’s proposal, whose closeness in spirit to
Schechter’s and Enoch’s will be apparent. Say that,
P is a presupposition of a particular cognitive project if to doubt P (in

advance) would rationally commit one to doubting the significance
or the competence of the project. (Wright 2004, p. 163)
Entitlement of a cognitive project can then be defined as follows:
(i) There is no extant reason to regard P as untrue and

(ii) The attempt to justify P would involve further presuppositions
in turn of no more secure a prior standing, ... , and so on without
limit; so that someone pursuing the relevant enquiry who
accepted that there is nevertheless an onus to justify P would
implicitly undertake a commitment to an infinite regress of justi-
ficatory projects, each concerned to vindicate the presupposi-
tions of its predecessor (ibid.)
Wright’s claim is that modus ponens is a presupposition of any cogni-

tive project involving reasoning and that any attempt to justify it
would presuppose it, since it would involve reasoning (cf. ivi, p. 166).
Furthermore, he thinks there cannot be basic counterexamples (ivi,
p. 171), for when P and Q are atomic sentences, if, given P and “If P
then Q”, one is not prepared to infer Q, this “convicts the thinker of
misunderstanding of the conditional” (ivi, p. 170). Hence, we possess an
entitlement of cognitive project.16
Clearly, neither proposal provides us with a first-order justification
for modus ponens. Thus, if the task was to tell apart good and bad
basic inferences, like modus ponens on the one hand and affirming the
consequent on the other, nothing of consequence would follow from
the considerations just advanced.17 The problems rehearsed for Wright’s
entitlements in Chapter 2, in connection with “There is an external
world” hold in this case too. Entitlements of cognitive project do not
speak to the likely truth, or in this case, the validity, of a given rule of
inference, however basic that might be.
One could then think that, at least on the assumption that modus
ponens is valid, the proposals presently on the table allow us to claim
that we know or can justifiably believe – in some sense of “justifiably” –
that modus ponens is valid. The trouble is, as Wright himself recognizes,
that they do not achieve even that much (ivi, pp. 168–9). For all these
strategies attain is to make apparent why it is natural, or even indis-
pensable, for us to apply modus ponens,18 but they do not give us an
epistemic reason to vindicate the second-order claim that we know (or
at least justifiably believe) that it is valid. If we were to propound the
considerations advanced by these authors to someone like the tortoise
in Lewis Carroll’s example, who was not already prepared to infer in
accord with modus ponens, they would certainly not convince her that
modus ponens is a valid rule of inference. They would merely make
it apparent to her why we find it unavoidable to employ it. Hence, it
seems to me that the proposals presently under consideration cannot be
seen as providing an answer to the question “How can we claim that we
know/justifiably believe that modus ponens is valid?”.
Nor do they solve the other problem of providing a diagnosis of what
would be wrong with the tortoise in Lewis Carroll’s story. For surely
they would return the verdict that the tortoise, by refusing to draw the
conclusion that Q, given P and “If P then Q”, would be prevented from
engaging in a valuable cognitive project, that of reasoning. Still, why
would this be a rational deficiency on her part? It is not very informa-
tive to be told, in this regard, that reasoning is a “rationally required”
project (cf. Schechter and Enoch’s definition above).19 In what sense
is that project required by rationality? Why does rationality require it
to be executed on the basis of modus ponens rather than on the basis
of affirming the consequent, say? To make headway with respect to
the diagnosis of what is wrong with the tortoise, we need an explana-
tion that ties reasoning in accordance with modus ponens to the very
notion of logical rationality. This is where the extended rationality view
kicks in.
Recall the move we made earlier in this book in relation to “There is
an external world” and the issue of epistemic rationality. We said that
the latter notion depends on the basic epistemic practice of forming,

assessing, and withdrawing from beliefs about physical objects in our
surroundings, taken as such, on the basis of perceptual evidence, whose
role is precisely that of speaking to the truth of certain beliefs, or their
negations. Such a practice has, as one of its constitutive assumptions,
“There is an external world”. Hence, to assume the latter is constitutive of
epistemic rationality even if it is not justifiable. In fact, we are mandated
by epistemic rationality itself to make that assumption for, without it,
there would be no possible perceptual justification for or against a given
belief about specific material objects. Consequently, there would be no
practice and therefore no notion of epistemic rationality either, at least
as we – and a skeptic – usually understand it.
In connection with modus ponens, I think we can proceed in a
similar fashion and, once again, in a Wittgensteinian spirit.20 First of
all, we know that the notion of deductive rationality does not hang in
the air, but depends on a practice of reasoning by employing certain
basic patterns of inference which are valid, that is necessarily truth-pre-
serving.21 Basic instances of modus ponens are among these patterns
of inference (others may be conjunction elimination and disjunction
introduction). Hence, to reason in accord with modus ponens, at least
in certain basic cases, is itself constitutive of deductive rationality, which
is itself an aspect of epistemic rationality. Namely, the one that allows
us to draw out the necessary consequences of our (hopefully true and
justified) beliefs. Thus, if one did not abide by it, one would not count
as rational, at least on the notion of deductive epistemic rationality we
do have.
Deductive epistemic rationalityER: it is deductively epistemically

rational to believe propositions reached through (deductively) valid
patterns of inference and to accept those unwarrantable basic and
valid inference rules that make the acquisition of deductive warrants
possible in the first place and are therefore constitutive of them.
Notice that the acceptance of modus ponens does not have to be explicit,
but it can be merely implicit and displayed in reasoning in accord with
it. Furthermore, in this view, we are mandated by deductive epistemic
rationality itself to reason in accord with modus ponens and other basic
rules of inference, which are equally valid. This is no proof of the validity
of modus ponens, though. That proof will be provided differently, yet
in a rule-circular fashion, since whatever logical method we apply to
that end, it will presuppose reasoning in accord with modus ponens.
Still, I think we can grant, in a somewhat externalist spirit, that we know

that modus ponens is valid. Yet, we cannot claim that knowledge, for we
cannot prove that we have it to someone who does not already reason
in accord with modus ponens, and would thereby be willing (implic-
itly) to admit its validity. In this respect, we cannot but accept or take
for granted that modus ponens is valid, without being in a position to
prove it in a non-circular way capable of persuading the unconverted.
The characteristic extended rationality move just described, however,
gives us an a priori justification for the proposition “To reason in accord-
ance with modus ponens is rational (by the lights of deductive epistemic
rationality)”, for it tells us that to do so is constitutive of that form of
rationality.22
Now, suppose we met a tribe who, in basic cases, did not reason
in accord with modus ponens, after we have ascertained that they
do mean “if then” the way we do and that no contextual factors
intrude in such a way as to make it understandable why they might
seem to deviate from modus ponens.23 Alternatively, suppose we met
the tortoise from Lewis Carroll’s example. Namely, someone who,
ex hypothesis, understands “if, then” like we do, but who is simply
unwilling to infer “Anna will take the umbrella”, after admitting both
“It is raining” and “If it rains, Anna will take the umbrella” are justi-
fied. Surely, we could not do much to remove their stubbornness,
but we could conclude, with Gottlob Frege, that by doing so, they
would simply be outside the scope of deductive rationality.24 Hence,
the point is not that such a case would be inconceivable. We do not
have to take a stance on that and we can allow that sense could be
made of someone who, while not reasoning in accord with modus
ponens, could still possess the concept IF THEN and be a thinker, in
some sense of the term. The point, rather, is that it would not show
that the notion of “logical rationality” (sub species deductive ration-
ality) is relative – that is, that abiding by different and incompatible
basic logical laws could qualify as being equally logically rational. For,
as long as the meaning of “logical (deductive) rationality” stays put,
that requires people to engage in forms of reasoning governed by basic
rules of inference, which are valid. Thus, if someone refused to apply
them, or even followed different, invalid ones, we could convict them
of logical irrationality. Hence, the diagnosis of what is wrong with
the tortoise, is not that she does not know the meaning of “if then”,
or that she would be merely prevented from taking part in a project
which is extremely valuable or even indispensable to us. Nor is it that
she would be unable to think, in some sense, of that term. Rather, it is
that she would be irrational, as she refuses to take part in an activity

that is constitutive of logical rationality itself.
One could then say that these people would have their own, equally
legitimate, notion of rationality, even though incompatible with ours,
characterized by an appeal to different rules of inference. However, in
order to take this possibility seriously, that is, as a legitimate alterna-
tive to ours, we should look into their rules of inference. The options,
I think, would be as follows: (1) their rules are basic but invalid (like
affirming the consequent); or (2) they are valid but not basic; or (3)
they are basic, valid and yet incompatible with ours. In the first case,
we could still convict them of logical irrationality, for to count as
logically rational in general (or to exhibit a kind of rationality which
could be a serious alternative to ours) these rules should be at least
necessarily truth-preserving. In the second case, in contrast, it could
presumably be shown that these valid but non-basic rules of inference
presuppose valid and basic ones, and, in particular, ones we abide by,
such as modus ponens, that are constitutive of our notion of logical
rationality. Finally, in the third case, it is hard to see how the three
desiderata could all be satisfied. For we are asked to conceive of basic
and valid rules of inference, which, however, are incompatible with
ours. I take that to entail that they will have to return different and
incompatible verdicts regarding their basic application. This, however,
would cast doubt on their validity, from our own point of view. Hence,
in none of these cases, I submit, would we have found a notion of
logical rationality, determined by the observance of different rules of
inference, which, while incompatible with ours, could be taken as a
serious alternative.
Finally, with respect to the Oblomovian challenge of doing without
basic rules of inference of deductive logic, it would have to be noticed
that it would deprive us of the means of drawing out the necessary
consequences of our (hopefully justified and true) beliefs. Once more,
the problem is not that of explaining the utility or the compulsori-
ness of such rules of inference, given the kind of creatures we are. The
problem, rather, is that of accounting for their epistemic rationality,
over and above their validity, vis-à-vis those critics who may want to
consider them as lying beyond epistemic rationality, as they cannot be
justified a priori or through entitlements. That, once again, is what the
extended rationality view delivers. For, in that view, modus ponens, and
other basic rules of inference of deductive logic, will be epistemically
rationally accepted because, even if unwarrantable, they are constitutive
of deductive epistemic rationality.
5 Summary
In this chapter, we presented several extensions of the extended ration-

ality view proposed in Chapter 4. We saw how it can be extended to all
aspects of epistemic rationality that go beyond the ability to form, assess
and withdraw from beliefs about mid-size objects in one’s surroundings
based on perceptual justifications. In particular, we saw how it can be
of use in accounting for the status of the principle of the uniformity of
nature, which is crucial for the attainment of inductive justifications.
We then saw how it can account for the status of “There is a past”,
“Informants are usually reliable” and even “There are other minds”,
which can be considered the hinges on the basis of which we can acquire
memory-based justifications about propositions regarding the past, and
testimonial justifications. Finally, we saw how the extended rationality
view can be extended to basic and valid rules of inference, like modus
ponens, which allow us to draw out the necessary consequences of our
(hopefully justified and true) beliefs.
By so doing, we have shown how fruitful the extended rationality
view is. For it allows us to maintain that the human condition is as it
depicts it. Accordingly, human beings are capable of epistemic ration-
ality. Epistemic rationality, in its turn, is possible only thanks to several
general presuppositions, or hinges, and to basic rules of inference. These
hinges and basic rules of inference are thus constitutive of it and are,
therefore, epistemically rational in their turn.
Notes
Introduction
1. This example can be found in Wright 1985, 2002. Another usual example is
the Zebra argument, famously put forward by Fred Dretske 1970, which is
discussed at length in Chapters 1 and 3.
2. See, for instance, OC 183–192.
3. Ibid.
4. We look at some contemporary attempts in Chapter 2.
5. The one proposed by Crispin Wright in a number of writings, which is exam-
ined in Chapters 2 and 4.
6. Recall the citation from OC 105. See also OC 359, 559.
7. The attempt to build on that horn of the trilemma would lead to founda-
tionalism. Both Pryor’s and Wright’s views can be seen as different ways of
defending it. In Pryor we have immediate justification for ordinary empir-
ical beliefs, thanks to perception and in the absence of defeaters, and from
them derive a justification for very general propositions such as “There is
an external world”. In Wright, in contrast, we have an entitlement – that is,
a non-evidential justification – directly for those very general assumptions
and, thanks to it and to an appropriate course of experience, a justification
for ordinary empirical beliefs.
8. The attempt to build on this horn of the trilemma would lead to various
forms of coherentism, whose fault is that they could give rise to maximally
coherent, yet incompatible systems, among which we could not make no
epistemically sound choice. That is to say, we would have no means to deter-
mine which one is the correct one. Or else, we would have to produce locally
circular justifications, that is justifications for general propositions like
“There is an external world” based on specific propositions, such as “Here is
a hand”, which, in their turn, are justified only insofar as we take for granted
those very assumptions. It will be argued at length, especially in Chapter 3,
why such circular justifications would be no justifications at all.
9. See McGinn 1989 and Moyal-Sharrock 2004 for its main predecessors.
10. The details of such a reading are developed differently by McGinn 1989,
Moyal-Sharrock 2004, and Coliva 2010a and 2013a, b, but the main message
is the same.
11. As always, with Wittgenstein, things are not entirely clear. My own reading,
presented in Coliva 2010a and further developed in Coliva 2013a, b, is that
it might be possible to distinguish between the content and the role of a
sentence. Hence, Wittgenstein’s, so-called, “hinge” propositions would be
propositions, being susceptible to truth and falsity at least in a minimal sense,
which, however, have been removed from doubt and inquiry. Therefore,
they would play a normative role, while retaining a descriptive content. In
peculiar, often unforeseeable circumstances, however, they may return to be
subject to doubt and inquiry and would thus re-acquire a purely descriptive
181
182 Notes
role. A case in point are Wittgenstein’s own remarks on “Nobody has ever
been on the Moon”, which is a hinge proposition for him, yet clearly false
for us. However, it would be much more difficult, if not altogether impossible
to imagine circumstances that could actually lead us to doubt that there is
an external world. For Wittgenstein a merely possible doubt, which was not
backed by reasons to entertain it, would be merely apparent and no real doubt
at all. Hence, in his view, skeptical doubts are only apparently meaningful.
12. Following Boghossian 2006, I hold that an epistemic practice is basic iff it
does not presuppose instances of itself and is presupposed by all others.
Forming beliefs on the basis of perceptual evidence is one such practice, as is
reasoning in accordance with basic laws of inference, such as modus ponens,
while both are presupposed in order to form and verify scientific theories. In
contrast, to form beliefs about one’s future by casting oracles or by means of
horoscopes is not a basic epistemic practice, for it requires both observation
and inference. Notice, moreover, that to reason in accordance with modus
ponens is basic when it does not concern embedded conditionals. These
issues will be explored in more detail in Chapters 4 and 5.
13. It is important that the kind of relativistic challenge considered here were
compatible with our experience for it to represent a serious alternative for
us, as we are interested in explaining the fundamental traits of the human
epistemic condition.
14. Cf. also Burge 2010.
15. Cf. Wright 1985, 2002, 2004a, 2014.
1 Moderatism about Perceptual Warrants

1. I will not discuss reliabilist conceptions of justification for they fall within
the externalist camp. Furthermore, they have been extensively criticized, at
least in their crudest forms, by several theorists working within the exter-
nalist framework, such as virtue epistemologists, anti-luck epistemologists,
and more.
2. One might think that even more extreme views would be possible, beside
Pryor’s and Wright’s. For example, views that hold that, even if there are
defeaters, one retains a perceptual justification for, say, “Here is a hand”
just by having a hand-like experience. Or else, theories that hold that one
must be certain that there is an external world, in order to have a perceptual
warrant for such a specific belief, once the relevant experiences are also in
place. I do not deny that they are possible positions in the logical space, but
they certainly do not look promising. The former would implausibly allow
for justifications even when one does have good reasons to suppose that
one’s experience has not been produced by causal interaction with physical
objects. The latter, in contrast, would make it impossible to possess perceptual
justifications, as all contemporary epistemology has forsaken the characteris-
tically Cartesian quest for certainty, deemed to be basically unattainable.
3. Pryor 2001 and 2004. Notice that Pryor qualifies the view by saying that it
would hold only for perceptually basic beliefs. That is to say, for beliefs about
material objects that do not go beyond the very content of one’s experience.
This qualification, however, opens the way to several criticisms concerning
Notes 183
the individuation of the relevant class of beliefs. For present purposes, we can
ignore the point and concede that we may individuate such a class of beliefs.
For the sake of argument, I will assume that “Here is a red table” or “Here is
a hand” would be instances of perceptually basic beliefs. To the best of my
knowledge, in contemporary epistemology Pryor has been the main advocate
of the liberal position, often dubbed dogmatist as well. However, not even
G.E. Moore, surely a dogmatist in many ways, whose views are often associ-
ated with Pryor’s, had the conception of experience and perceptual justifica-
tion put forward by the latter. Moore, in fact, believed in sense data, even
though he was never sure how to think of them. Still, in most interpretations
of their nature, he held that they are not identical to parts of physical objects.
Hence, by his lights, it could not be the case that experiences, just by them-
selves could provide a justification for a specific belief about a given material
object. I discuss Moore’s conception of experience and perceptual justification
in Coliva 2010a, chapter 1.
4. Pryor 2005 allows also for the possibility that experiences with merely
phenomenal content, such as pains, could directly justify beliefs such as “I
have a headache”. In the perceptual case, however, it seems overwhelmingly
plausible that the relevant experience should have a given representational
content. For if we started out with experiences whose content were merely
sensations it would be difficult – to say the least – to attain an immediate
justification for beliefs about specific mind-independent objects.
5. A justification may be internalist even if it is not, intuitively, an internal state
of a subject. It would be enough for it to be accessible by a subject and articu-
lable by him. A case in point would be a proof of a theorem.
6. Especially John Turri, who gives the following account of propositional
warrant: “Necessarily, for all S, p and t, if p is propositionally justified for S at
t, then p is propositionally justified for S at t because S currently possesses at
least one means of coming to believe p such that, were S to believe p in one of
those ways, S’s belief would thereby be doxastically justified” (2010, p. 320).
7. As to Turri’s reservations, a careful discussion of his paper would take us too
far afield. However, let me just mention two worries about his criticism and
positive proposal. He is right to notice that the “basing relation” which is
used to characterize the notion of “doxastic warrant” is problematical. Yet, his
alleged counterexamples are more a reason for people interested in defining
that notion to try and make it as precise as possible, than refutations of that
very idea. In particular, it seems that the basing relation will have to include
not only propositional warrants but also the rationally correct procedures
by means of which those propositional warrants should be used, by specific
subjects, in specified conditions, to form the relevant beliefs, so as to actu-
ally provide them with doxastic warrants for their actual beliefs. As to Turri’s
positive proposal, according to which propositional justification should be
defined in terms of doxastic justification, leaving aside any perplexity about
its details, it should be noted that it offers merely necessary conditions for
propositional justification but nothing like necessary and sufficient explana-
tory conditions for it. Therefore, it actually falls short of providing a viable
account of propositional justification. Hence, given the extant state of the art,
I do not think there is any compelling reason to be suspicious of the notion of
propositional warrant.
184 Notes
8. Cf. Pryor 2004, 2012.

9. Cf. Burge 2003.
10. Hence, in Burge’s notion of entitlement a subject affected by blind-sight
could, in principle, have a perceptual entitlement for his beliefs, while having
no consciousness of the perceptual experience on which they are (allegedly)
based.
11. Cf. Pryor 2005.
12. Cf. Burge 2010.
13. Pluralist accounts of justification and/or knowledge can be found in Coliva-
Zanetti, forthcoming, Pedersen, forthcoming, Gerken, forthcoming, Olsson,
forthcoming.
14. Such as, for instance, Burge 2003, Silins 2007, Wedgwood 2011a.
15. It has been pointed out to me that Stephen Schiffer makes a similar point.
16. Silins 2007 offers two more reasons to prefer the liberal position with respect
to perceptual warrants. Firstly, the fact that if there are immediate justifica-
tions, perceptual ones seem the right candidate, beside introspective ones for
one’s self-ascriptions of introspectively available mental states. Secondly, that
the only plausible answer it offers to the non pre-theoretical question “Why
does Moore have any reason to believe that he has hands?” is “Because he
sees them”. For the conservative would have to say something, in response
to that question, which is beside the point, i.e. “He sees them and he has
good reasons to discard skeptical hypotheses”. I think both arguments are
not convincing. The first one is conditional on there being immediate justi-
fication at all and, in particular, in the perceptual domain – a presupposition
that obviously begs the question against the conservative. The second one,
in contrast, is based on several misconceptions. For the liberal, insofar as he
is also an internalist, cannot say “Because he sees them” but only “Because
he seems to see them”. This clearly renders the liberal answer to the question
much less convincing than Silins makes it seem.
17. The term occurs in Wright 2004a, although the explanation and the use of
this idea in this context need not coincide with his.
18. Notice that, nowadays, the term basic knowledge is usually associated with
externalist, mostly reliabilist positions, which, roughly, hold that the belief
that P amounts to knowledge if (and only if) it is the result of a reliable process
(or of a safe or sensitive one, on more recent renditions) and that percep-
tion is such a process. These views too are problematical, but for different
reasons. They are not the object of the present study, although some argu-
ments against them, in the way of exposing their undesirable consequences,
are discussed later (see Chapter 3 in particular).
19. Notice that one could also try to carry out an investigation to the effect that
no defeaters occur. Yet, it is consistent with Pryor’s account of experience
and perceptual warrants that such an inquiry would appear the same to one
no matter what its causal origin might be. If asked, then, how one knows
that the results of the investigation were correct, the answer would again be
“Because it looked that way”.
20. Gilbert Harman (1986) puts forward a view that he dubs “conservativism”.
Accordingly, “one is justified in continuing fully to accept something in the
absence of a special reason not to.” (1986, p. 46) There is an important differ-
ence between his view and Wright’s though. Namely, for Wright the absence
Notes 185
of defeaters is not enough to produce a justification for what is assumed.

Rather, a positive warrant is needed to that end.
21. His position is often dubbed as skeptic. I think this is misguided for surely
he is no skeptic. However, what the label may rightly hint at is that skep-
tics share his conception of the architecture of perceptual warrants and
end up being skeptics precisely because they deny that we could ever
attain the warrant for general assumptions that figure as collateral, neces-
sary elements of perceptual justification, beside an appropriate course of
experience.
22. Until his 2014 paper, Wright would have been happy to add that these argu-
ments cannot reinforce any previous justification one might have for (III)
either. The point is contentious, however. For a discussion, see McGlynn
2014 and Wright 2014. We will come back to transmission failure at length
in Chapter 3.
23. Vogel 2008 raises both these objections (cf. in particular, pp. 539 and 541).
However, he takes them to offer prima facie motivation to explore his expla-
nationist account of how (III) in MOORE could be justified (cf. also Vogel
1990). My main objection to explanationism, of which I will consider
Peacocke’s version in Chapter 2, is that it is not at all clear that, if an expla-
nation is simpler than another one and not ad hoc, it is more likely to be
true. Vogel himself considers this counter (p. 549, and fn. 39, 67–68) and,
to my mind, does not give any convincing response to it. Connectedly,
if Cartesian skepticism is allowed to raise doubts about (III) based on the
metaphysical possibility that we might be dreaming or be BIVs, the fact that
explanations of our experience that appealed to elements present in those
scenarios would be more complex than the ones we would normally appeal
to, would do nothing to show that skeptical scenarios are more likely to
be false. Of course it will appear thus to us but, in terms of objective prob-
abilities, there is simply no definitive argument in favor of the simpler is
the most likely.
24. We will consider some of them at length in the following chapter.
25. We will consider some recent developments of these ideas in the next
chapter.
26. Of course, the problem is not that we somewhat hanker after an indefeasible
warrant. As said, the role of this general assumption is merely to allow us to
justifiedly form beliefs about specific material objects, even if we may well be
mistaken about the kind of object we are facing, or even about the fact that
in those very circumstances there is such an object.
27. Wright’s notion of an entitlement is definitely different from Burge’s. The
latter thinks that entitlements are epistemic goods which subjects need not
be in a position to articulate, that, however, bear on the truth of what they
provide a warrant for. Furthermore, Burge’s entitlements are supposed to
warrant type-I propositions in MOORE-like arguments, while Wright’s would
warrant type-III ones. Finally, Burge’s entitlements are externalist types of
warrants, while Wright’s are internalist.
28. One word of caution about my terminology. I talk about assumptions, which,
in ordinary parlance, we may also refer to as beliefs. I have no qualms with
that, provided one does not build into the notion of belief the fact that it
must be based on some kind of evidence (as Wright 2004a does, for instance).
186 Notes
If one did, then the terms “assumption” and “belief” would no longer be
interchangeable.
29. In this doxastic sense what I call assumptions are relevantly similar to Gilbert
Harman’s “implicit commitments” (Harman 1986, p. 44). I do not, however,
subscribe to Harman’s conservativism, according to which “one is justified in
continuing fully to accept something in the absence of a special reason not
to” (Harman 1986, p. 46). For the absence of defeaters is not enough for me to
produce a justification for a given proposition one is implicitly committed to.
30. Mikkel Gerken raised the following objection. Suppose that a person had
just opened his eyes for the first time and made no assumption about the
existence of an external world. From a moderate viewpoint he could not
warrantedly believe that he had a hand in front of him (supposing, for the
sake of argument, he had those concepts). Now, contrast him with a subject
who assumes that there is an external world and could then warrantedly
believe that there is a hand in front of him, given the moderate concep-
tion of warrant. Clearly, however, the two seem to be epistemically on a
par, while moderatism predicts they are not. I think this objection is useful
because to answer it allows me to further clarify the moderate view. If we are
concerned with propositional justification, both are equally justified. If we
are concerned with doxastic justification, so long as the first subject has the
concept of a hand as a mind-independent entity, he could be granted the
relevant assumption, even if he had never entertained the proposition that
there is an external world.
31. In this sense assuming would be similar to Harman’s “tentative assump-
tions” (1986, pp. 46–47), which he thinks may be corroborated by future
investigation and later on turned into full acceptances. Full acceptances for
Harman, however, are things for which one has collated enough evidence
to stop inquiring into them. According to him, this is enough to enable
a subject to take oneself to know that P is true (cf. p. 47). I do not follow
Harman’s latter suggestion though, for, in my view, assumptions are not
knowable, properly speaking. For the putative justifications we could have
for them would actually depend on already taking them for granted. As I
will argue at length in Chapter 3, this form of bootstrapping justification is
no good. That is to say, it involves us in a vicious circle. Hence, in my view,
whatever evidence we may have for them (that is everything we do know
does in fact speak in favor of them) does not actually play a justificatory role
with respect to them.
32. In Italian there is the expression “ammesso ma non concesso” – roughly trans-
lated as “assumed but not conceded” – which would perfectly fit this sense
of “assuming”. To give but one example, one might assume that pigs can fly
and then wonder what would follow from that, knowing that they cannot.
33. Think, for instance, of someone who strongly believes that the defendant is
innocent, even if there is evidence against him. Hence, it may be false that
the defendant is not guilty, yet the subject is committed to considering him
innocent.
34. This distinction is crucial in my view, as we will see in Chapter 4 (§2–ff).
35. See Chapter 3 (§7).
36. This problem originates from a remark made by Stephen Schiffer and
discussed in Wright (2004a, p. 177). It originally concerned Wright’s notion
Notes 187
of entitlement, in which the worry was that if one merely has an entitlement
for “There is an external world” – as opposed to an evidential warrant for it –
in the conservative view of the structure of empirical justification it would
turn out that one merely has an entitlement for “Here is a hand” too, rather
than an evidential warrant for it. I have therefore slightly modified the objec-
tion to make it fit moderatism.
37. One could think of Hume as a conservative. I would agree but add the quali-
fication that he is so only in his skeptical mood. For he can equally be seen
as a moderate, as I explain in the text when he proposes his views about the
human condition. Someone might think of G.E. Moore as a conservative,
but this is dubious for Moore thought he could actually derive a justification
for, in fact knowledge of, “There is an external world” from his justifica-
tion/knowledge for/of “Here is a hand”. True, given his ambivalent attitude
towards the nature of sense data, he may have favored, at least at times, the
view according to which experiences by themselves cannot directly warrant
“Here is a hand”. Yet, it is not clear that he also held that the general assump-
tion that there is an external world should be independently warranted in
order to have a perceptual warrant for “Here is a hand”. Bertrand Russell was
a conservative, while Thomas Reid’s position cannot easily be placed within
the conservative camp, since he supported a form of direct realism, whereby
minds directly apprehend reality.
38. Arbitrary, here, is not to be contrasted with universal. Indeed, if it is part
of our human nature/form of life to make these assumptions as opposed to
different ones, they would be universal, at least as far as human beings are
concerned. Still these assumptions could be epistemically arbitrary. A way of
bringing out their epistemic arbitrariness would be to think that it is possible
that other forms of life could find it natural to make different and incon-
sistent assumptions.
39. If Pritchard 2005b and Jenkins 2007 were right about Wright’s entitlements,
ironically, Wright’s position too would be a kind of pragmatism, hence it
would be committed to moderatism. There will be more on this issue in
Chapter 2 (§2).
40. Cf. McGinn 1989, Moyal-Sharrock 2004 and Coliva 2010a.
41. Williams 2004a denies that for Wittgenstein “There is an external world” is a
hinge. He holds that it is plain nonsense. Were he right in this interpretation,
the point would remain that Wittgenstein endorsed a form of moderatism
whereby the assumptions that are necessary to bring all sorts of evidence to
bear on the class of appropriate beliefs would be more specific and context-
dependent than the version of moderatism presented in this book. In conver-
sation though, Williams, mentioning OC 152, has recently backtracked and
suggested that “There is an external world” would indeed be a hinge for
Wittgenstein. In fact, that it would be even more so than other contextually
determined propositions usually said to play that role in Wittgenstein’s work.
42. This, in broad strokes, is the view put forward by Moyal-Sharrock 2004.
43. I myself have proposed this interpretation of hinges in Coliva 2010a, 2013a,
b, and Coliva forthcoming-a.
44. This would be another way of showing that skepticism is concerned with
epistemological worries that would persist even if one granted the kind of
realist metaphysics that backs the externalist view.
188 Notes
45. McDowell 1994.

46. Sellars 1956.
47. See for instance Davidson 1986.
48. I use small caps for concepts, quotation marks for words, and underlined
characters for perceptual contents.
49. McDowell 2009 maintains the idea that in perception there is a passive
exercise of conceptual abilities but he renounces two further theses which
were characteristic of his Mind and World. Namely, the idea that perception
is propositionally structured and that the concepts deployed in it match the
ones that figure in the corresponding beliefs. In particular, it is maintained
that the concepts passively deployed in perception are much more generic
than the ones used in one’s beliefs. Both revisions are problematical vis-à-vis
the issue of how perceptions can, as such, justify the corresponding beliefs.
For it is less clear than it is on the view presented in Mind and World how
a non-propositionally structured perception could actually justify a belief
based on it. Moreover, the underdetermination problem we will consider
shortly would arise for McDowell too.
50. I agree, for instance, with Burge’s criticism of it in Burge 2010. Accordingly,
our best account of perception to date has that perceptions are objective
representations, that is, they are representations as of objects and properties
“out there”, i.e. as external to the individual and as independent of varia-
tions in proximal stimuli. Yet they could be replicated by means of appro-
priate stimulations of the brain. Hence, there is nothing in the perceptual
representation itself that differentiates it from an illusion or a hallucination.
Moreover, according to Burge, the constituents of perceptual representations
are neither objects nor properties, but their perceptual modes of presenta-
tion. Hence, we do not take in facts in perception. I also agree with Wright’s
criticism of disjunctivism vis-à-vis skeptical challenges, in Wright 2002.
According to him, we may even grant for the sake of argument that percep-
tions and hallucinations or illusions are different kinds of mental state. The
question remains, however, how one can tell which state one is in, based
on merely subjective factors, when ex hypothesi, one would be unable to
discriminate between the two cases. Thus, even if disjunctivism were right
at first order, it would still face the skeptical challenge of providing a reason
for thinking that we are indeed perceiving a hand, say, rather than having a
hallucination as of a hand in front of us.
51. The most thorough and, to my mind, convincing account of the width of
perception is in Burge 2010. Powerful criticisms against the idea that we
should attribute concepts to infants and animals who are capable of percep-
tions can be found in Burge 2010 and in Bermúdez 1998.
52. See Peacocke 1992 and Bermúdez 1998. McDowell is aware of this objec-
tion and tends to discard it in its concept-possession-determination version.
However, the objection is insurmountable in its acquisition version, as
Bermúdez has convincingly argued.
53. Coliva 2003. Cf. also Brewer 1999 and Kelly 2001.
54. Peacocke 1992, chapter 3. Protopropositional content would, for instance,
discriminate between the experience of a square and that of a diamond-
shaped object having the same dimensions and other perceivable properties
as the square.
Notes 189
55. Notice that McDowell’s recent position (cf. note 49) would incur similar
problems if the propositional structure were necessary in order to justify
beliefs. In that case, given that McDowell’s new position forsakes the prop-
ositional structure of perceptual content, perceptions could not, as such,
justify the corresponding beliefs. They could do so only if they were propo-
sitionally structured through judgment. However, at that point, the justi-
ficatory work would not be done by the perception but by the judgment
instead.
56. See Note 20 for the differences between Burge’s and Wright’s entitlements.
Another supporter of entitlements is Fred Dretske. See Dretske 2000. A useful
discussion of this notion of entitlement can be found in Casullo 2007.
57. See Burge 2003, p. 544.
58. There will be more about Peacocke’s position in Chapter 2 (§3.2). Notice that
Burge assigns a much more marginal role to evolution than Peacocke and
strongly criticizes those attempts at reducing perceptual representations to
functional states, which would be useful to survival. However, it is not clear
that Peacocke is actually proposing anything like that. So the comparison
between their positions in this respect is surely quite slippery.
59. Cf. Peacocke 2004, p. 176.
60. In general, I fully agree with the spirit of Burge’s criticism, in Burge 2010, of
this kind of move.
61. Notice that McDowell’s new position would incur a similar problem, albeit
mutatis mutandis, since, for him, perception would nevertheless involve the
passive exercise of at least some generic concepts.
62. Siegel 2010.
63. See Millikan 1998.
64. This goes against McDowell’s last pronouncements on the issue.
65. In my opinion, a subject should be sensitive to defeaters at least in the sense
of refraining from forming beliefs should he have some awareness that condi-
tions might not be normal. This does not mean that each time he is unaware
of defeaters he should form his belief only by having made sure that that is
in fact the case.
2 Further Motivation for Moderatism

1. Silins 2007, Wedgwood 2013.
2. Similar considerations would apply if one were considering Moore’s argu-
ment while having doubts about the liberal conception of the structure of
empirical justification.
3. See White 2006, Wright 2007a, Silins 2007, Wedgwood 2013.
4. Coliva 2004 and 2007.
5. Although nowadays probabilistic renditions of the notion of warrant or
justification are quite fashionable, it should be kept in mind that they also
introduce contentious assumptions; for example, that one’s hand-like experi-
ence be construed as a belief which makes the corresponding empirical belief
more probable; that justification and warrant may be understood in probabi-
listic terms and equated to degrees of confidence; finally, that one’s degrees
of confidence and therefore of justifiedness evolve in response to evidence
190 Notes
along Bayesan lines. None of these further assumptions is watertight, though.

On this, see also Pryor 2012 and Wedgwood 2013.
6. See White 2006, also Wright 2007a.
7. Cf. also Vogel 2008, p. 541.
8. Or has doubts about the liberal conception of the structure of empirical
justifications.
9. We will consider what kind of circularity actually afflicts it in the next
chapter.
10. Coliva 2010b.
11. Another problem for the liberal position is that it gives rise to bootstrapping
arguments (cf. White 2006). For expository reasons, I will consider them in
the next chapter (§7).
12. We will consider in the following a possible attempt at divorcing liberalism
from Mooreanism (§3.1).
13. Cartesian forms of skepticism are those that, by appealing to the metaphys-
ical possibility of uncongenial scenarios, claim that we are not justified in
believing ordinary empirical propositions about mid-size physical objects.
By generalizing from that, they argue that our belief in the existence of
an external world is unjustified as well. Humean forms of skepticism, in
contrast, do not play with uncongenial and far-fetched scenarios. Rather,
they exploit the circularity involved in providing justifications for back-
ground assumptions of our perceptual justifications, such as “There is an
external world”, via logically valid arguments, such as Moore’s, that start
with warrants for ordinary empirical beliefs. They usually claim that all
other forms of reasoning aimed at justifying those assumptions would fail.
Hence, they would turn out to be unjustified and unjustifiable. They may
also, though not necessarily as we shall see in Chapter 4, conclude that since
we have no justifications for these background presuppositions, which are
needed to have perceptual justification for ordinary empirical beliefs, the
latter are not justified either.
14. There will be more about this principle in the following chapter.
15. Wright 2004, pp. 174–175.
16. This is the reading variously defended by, most prominently, McGinn 1989,
Moyal-Sharrock 2004, and Coliva 2010a, 2013a, 2013b, forthcoming-a.
17. As we will see in Chapter 4, the point is not that we have a conception
of experience as bearing onto mind-independent objects, but, rather, that
perception itself provides us with the experience of an objectified world,
that is, with an experience of objects and properties, which are external
to our minds and that displays perceptual constancies. Notice, however,
that Wright’s entitlement of substance is entirely silent on the very nature
of perceptual experience and, for this reason it is not effective against the
pretensions of an idealist.
18. To anticipate, we will see (Chapter 4, § 4) that the very content of perceptual
experiences is objective in the sense of representing objects and properties
“out there”, which remain invariant despite changes in proximal stimuli.
Since Wright does not consider studies in the theory of perception, he does
not have the means to rescue the validity of his entitlements of substance
against an idealist opponent.
Notes 191
19. We will discuss at length the possibility of rationality without warrant-

edness in Chapter 4. It is the gist of the extended rationality view I will
put forward that our basic assumptions are rational yet unwarranted and
unwarrantable.
20. Notice, moreover, that such a proposal would be dangerous because warrants
for our most basic presuppositions would be hostage to some philosophical
theory or other. Hence, we had better be sure of having the right philosoph-
ical theory, for otherwise it would still be doubtful that such assumptions
as “There is an external world” are a priori warranted. However, the record
of accomplishment of philosophical arguments should not make us terribly
confident that we have hit on the right account.
21. As stressed in a previous footnote, the extended rationality view I put forward
in Chapter 4, capitalizes on the distinction between being rational and being
warranted.
22. I was pleased to find a similar objection in Avnur 2011a. Wright 2014 (p. 214)
opens instead with the claim that entitlements are non-evidential “warrant(s)
to accept a proposition as true”. However, the only reason given to that effect
is that trust presupposes trusting in the truth of a certain proposition. This
may well be the case but it is not an argument to show that the proposition
trusted upon is indeed (likely to be) true.
23. We take up Wright’s position again in Chapter 4, in connection with Humean
skepticism and in comparison with our proposed solution to it.
24. Of course the moderate position merely says that we do not need a warrant,
empirical or otherwise, for (III) in order to have warrant for (I), given an
appropriate course of experience. One may then think that moderatism is,
after all, compatible with the view that there could be such independent
warrants for (III), though they are not needed to have a perceptual warrant
for (I). Yet, I think that if there were independent, especially a priori, warrants
for (III) then it would be unclear – at least to me – why we should favor the
moderate architecture of empirical warrant rather than the conservative. I
am grateful to Yuval Avnur for raising this objection.
25. This chart can be found in Silins 2007.
26. Silins 2007, to which this middle-ground position is partly indebted,
claims that a cognate argument fails to transmit warrant even though this
is not explained by reverting to the conservative view. He bases his claim
on Bayesian considerations, of the kind exploited by White 2006, which
we have been rehearsing throughout this chapter. It should be stressed,
however, as we saw in §1, that the argument that is thereby shown to fail to
transmit warrant is not really Moore’s argument, but a cognate one with a
more specific conclusion, namely, “I am not a BIV presently hallucinating
a hand”. Silins also acknowledges that he has no explanation to offer as to
why the real Moorean argument should fail to transmit warrant. Indeed,
he actually goes so far as to allow for the possibility that, after all, it does
not transmit warrant for the reasons put forward by Wright. In contrast,
Wedgwood (2012) denies that there are non-trivial cases of warrant trans-
mission failure. For he thinks that the justification is never transmitted from
one enduring belief to another, while it is always transmitted from the
mental states that figure as premises in an on-going inference to the ones
that figure as its conclusion.
192 Notes
27. Wedgwood could maintain that Moore’s proof is dialectically ineffective.

Yet a better account of its dialectical ineffectiveness than the one offered by
Pryor (see §1) would be owed.
28. Recall that Wedgwood works with rationally available warrants, rather than
propositional ones, due to his adherence to a strong form of internalism.
There will be more about this issue in the following.
29. Notice that, as we shall see in the next chapter, the moderate position does
offer an alternative account of why Moore’s argument fails to transmit
warrant.
30. This use of the term “a priori” is at best confusing and, at worst, wrong.
For if “a priori” means, in this context, just “unmediated”, then use of the
latter term would have been more perspicuous. If, in contrast, Peacocke
wants to say that these transitions are unmediated and a priori, then clearly
what he is saying is false. For the fact that a transition from a given experi-
ence to a given belief is unmediated does not turn the justification for the
eventual judgment into an a priori one. Rather, it remains thoroughly a
posteriori, since it is ultimately based on one’s experience. What he really
wants to say, I think, is that unmediated, truth-conducive transitions from,
in this case, an experience to the corresponding belief, are explainable,
in his view, according to a priori principles, as opposed to physical, or at
least empirical, regularities that rationalize them and are for theorists to
discover.
31. Notice that, if that were true, a BIV could not have experiences with the same
representational content as a normal subject raised in a normal environment.
I personally find this view untenable, but this is not the place to discuss it
further, because I would like to concentrate on Peacocke’s position taken in
connection with the liberal–conservative debate and Moore’s proof.
32. Peacocke considers the counterexample to his view posed by the Müller-Lyer
illusion and argues that the relevant experience is only derivatively instance-
individuated.
33. This is so because the entitling state would be an experience, which is not
only a mental state of an individual but also a state he is consciously aware
of. For a criticism of this idea in connection with the justification of one’s
self-ascriptions of propositional attitudes, see Coliva 2008, for a defense of
Peacocke’s position, see McHugh 2012.
34. This argument resembles Vogel’s (1990, 2008) in many respects, but is said to
provide a priori justification for (III), while Vogel seems to think that being
part of the best explanation of our visual experience provides (III) with an
empirical justification. I have already briefly addressed Vogel’s argument in
Chapter 1, fn 23.
35. I am grateful to Yuval Avnur for pointing out to me that Alston 1993 also
lists several skeptical hypotheses that do not involve intentional states of
deceivers.
36. Further qualms with this argument can be found in Tennant 2005, who
mostly objects to Peacocke’s notion of a priori and to the fact that this argu-
ment is really a priori, given that it must ultimately rely on the truth of
evolution by natural selection. Wedgwood 2007 also objects to it and ques-
tions its a priori status.
Notes 193
3 The Bearing of the Moderate View: Transmission Failures,

Closure, Easy Knowledge, and Bootstrapping
1. Wright 2014 will now entertain the idea that they may enhance one’s
previous justification.
2. Wright 1985, 2000, 2002, 2004a, Davies 1998, 2000, 2003, 2004, 2009.
3. Davies 2009.
4. I discuss them in Coliva 2010b.
5. Pryor 2012.
6. Besides the contributions by Wright, Davies and Pryor 2000, 2004, which
gave rise to, and developed the debate, see Beebee 2001, Peacocke 2004,
pp. 112–115, Schiffer 2004, Brown 2005, Silins 2005, White 2006, Wright
2007a, Coliva 2008, 2012b, d, e, Pryor 2012.
7. Made famous by Dretske 1970, p. 1016 as a putative counterexample to the
principle of closure of knowledge across (known) entailment.
8. We will discuss closure at length in §6.
9. As mentioned in Chapter 1, it is important to keep in mind that the experi-
ence of a zebra together with the information that it is not a cleverly disguised
mule is not enough to give one an indefeasible warrant for (I): it could still be
a cleverly disguised pony, for instance.
10. As we saw at length in previous chapters, it is important to stress that the
kinds of warrant at issue for Wright (and Pryor) are propositional ones. We
will return to this issue in the following section.
11. In general, a warrant for the conclusion of the argument is needed to bring
one’s current sense experience to bear on a class of possible beliefs. As
mentioned in Chapter 1, in the case of Moore’s proof a warrant for the exist-
ence of an external world is needed for one’s experience as of a hand to bear
on a belief about a material object, whether or not the former is actually
caused by the interaction with a hand, rather than with any other material
object that somehow resembles a hand.
12. Davies 2009, Coliva 2010b, §1.
13. In contrast, the issue of the nature of propositional warrants for empirical
beliefs such as (I) is seen by Davies as impinging on another kind of epistemic
project, which he calls “deciding what to believe”. I discuss the legitimacy of
this kind of project in Coliva 2010b.
14. Pryor 2004, pp. 364–365, 369 also discusses the issue of the rational unavail-
ability of warrant to a given subject. He calls it “rational obstruction” and, in
2012, “hypothetical undermining”.
15. One could say that Davies has shown that the property of a warrant of being
rationally available fails to transmit, which is not to say that warrant itself
fails to transmit. I owe this observation to Jesper Kallestrup.
16. I am using this term here in such a way as to avoid any commitment to the
warrantedness of a given piece of evidence.
17. I am not saying that these propositions cannot be derived from themselves.
What I am saying is that there is no procedure within the theory that can
improve their epistemic pedigree. I am also assuming, for the sake of argument,
that the formulation of these theories is tidy enough not to contain redundant
axioms, which could obviously be non-circularly proved from other ones.
194 Notes
18. However, in order to avoid confusion, the reader should keep in mind that
what I mean by “theorem” is a proposition whose epistemic warrant derives
from other warranted propositions.
19. I think this view is in keeping with Wittgenstein’s Remarks on the Foundations
of Mathematics, and with his idea in On Certainty that hinges are norms. This,
however, is not the place to pursue either an exegesis of Wittgenstein’s writ-
ings or an analogy between the two cases.
20. The fact that at second order, as it were, TF1 and TF2 would return a similar
verdict on Moore’s proof should not make their difference at first order irrel-
evant. We will come back to this in the next chapter.
21. This may invite a reformulation of both TF1 and TF2, in which the right-
hand sides of the biconditionals, which state the conditions for an argument
to fail to transmit warrant from the premises to the conclusion, in ways 1 and
2 respectively, should read thus:
(TF1 arises iff) the following conjunction obtains: (i) the conclusion must
be assumed in order to have a warrant for the premises in the first place,
and (ii) such a conclusion is (independently) warrantable;
(TF2 arises iff) the following conjunction obtains: (i) the conclusion must
be assumed in order to have a warrant for the premises in the first place
and (ii) such a conclusion is not (independently) warrantable.
These reformulations should dispel the worry that TF1 might entail TF2.
Notice, however, that even if TF1 ultimately entailed TF2, this would not be
a problem for my overall position, but could be for supporters of TF1 at the
expense of TF2.
22. The suggestion has been put to me that this example is misleading because
no one would have such a warrant when going to the zoo and yet could form
a warranted belief in (I). That is right but it is no objection to the view. For
what I am saying is that in order to diagnose what kind of transmission failure
is at stake in the ZEBRA argument such as it is, one should evaluate whether
independent warrant for its conclusion could be obtained. This does not at all
entail that on a normal visit to a zoo, in order to have perceptual warrant for
(I), one should have that independent warrant. To put it differently, this does
not at all entail that ZEBRA correctly represents the structure of one’s warrant
for (I) on normal zoo visits. In fact, I think that what provides us with a
perceptual warrant for (I) on those occasions is simply a zebra-like experience
together with the much less specific, inductively supported assumption, that
zookeepers do not usually fool visitors by disguising animals.
23. Of course the story is a little bit more complicated than that for one may
hold that the DNA test gives one warrant for (III) (in ZEBRA) only by courtesy
of one’s experience while reading the results of the test, for instance, and that
will introduce further assumptions, which may ultimately involve “There
is an external world”. For present purposes, we may ignore this complica-
tion, because even if ultimately the warrant for (III) in ZEBRA may depend
on further arguments involving such an assumption, the specific argument
arranged to provide warrant for it, would not.
24. The suggestion has been put to me that one might have testimonial warrant
for “There is an external world”. In that case, one’s warrant for it would
be neither perceptual, nor a priori or non evidential in Wright’s sense.
This suggestion is problematical, though. For, in order to be warranted in
Notes 195
believing “S said that P” or “It is written on this piece of paper that P”, where
P is “There is an external world”, the assumption that P should already be in
place. So how could one possibly get a first warrant to believe that there is an
external world through testimony?
25. Furthermore, I take it that testimony would be a non-starter, because in order
to take someone’s words as testimony we should already take it for granted
either that the informant is an intentional being or that an intentional being
is the source from which the piece of testimony derives. Hence, the existence
of other minds will have to be presupposed.
26. Cf. Luper 2006, Dretske 2005 and Hawthorne 2005. For criticism, see Harman
and Sherman 2011.
27. Dretske 1970, 2005, Nozick 1981.
28. It is then an open issue whether, given closure, one could also acquire a
second, as it were, ordinary evidential warrant for (III) via the entailment.
Of course, this is disputable because it may lead to the alchemical result of
producing an ordinary warrant out of an entitlement. A supporter of entitle-
ments concerned with avoiding alchemy may argue that closure for warrant –
that is for evidential warrant – fails in the case of Moore’s proof, while it
holds for entitlements, that is, for non-evidential warrants, as does Wright
himself 2004a, p. 178. Wright 2014 is more receptive to the idea that we can
enhance our previous warrant, in the form of entitlement, for (III) by means
of such an argument.
29. However, it is worth noticing that, as we shall see in the next chapter, in my
view these assumptions are epistemically rationally mandated, although they
are neither warranted nor warrantable (either evidentially or by means of
non-evidential warrants such as entitlements, or indeed of a priori ones). One
could then suggest that what would transmit is, at least, a rational mandate.
30. This is the same conclusion recently reached, in independent ways, by Avnur
2011a. Kallestrup has also pointed out to me that clearly closure fails for
very specific epistemic properties. For instance, I have a visual warrant that
the animal is a zebra but not a visual warrant that the animal is a disguised
mule.
31. A similar defense of the conditional validity of closure for knowledge, as
opposed to warrant, is put forward by Harman and Sherman 2011. However,
they do not connect their defense with TF2 and seem to see it as naturally
entailed by their own view about the structure of knowledge. According to
them, knowledge rests on assumptions, which are not themselves known,
though, as they say (p. 132), they are “justifiably (and truly)” made. Clearly,
their view, even though proposed for knowledge rather than perceptual
justification, resembles in some respects the moderate position I have been
proposing, inasmuch as they countenance assumptions that are not known.
However, it should be noted that they talk about assumptions justifiably and
truly made, and this seems to put them more in the conservative camp than
in the moderate one. Furthermore, they seem to think that merely acknowl-
edging the existence of knowledge based on unknown assumptions would be
enough to provide some counterexamples to closure for knowledge. In fact, I
do not think this entailment is straightforward. For it seems at least logically
possible to hold the view that while unknown assumptions can generate
knowledge of other propositions, once the former figure as conclusions of
196 Notes
inferences starting from the latter, they do get to be justified. Indeed, in

Harman 1986 that seems to be the case. Therefore, I think that beside a
moderate or moderate-like conception of perceptual justification or knowl-
edge one needs TF2 to produce possible counterexamples to closure.
32. DeRose 1995.
33. Sosa 2000 and Pritchard 2005a.
34. This seems to me enough to show that Pryor cannot defend himself simply
by saying that he is merely interested in defeasible warrants, for, no matter
how defeasible a warrant might be, it should, in this case, make it more likely
that one is facing a red wall, rather than a white one bathed in red light.
35. Vogel put them forward in order to make trouble for reliabilism; Cohen
(before 2010) took them to be effective against any account which allowed
for basic knowledge and therefore also against internalist ones, such as the
liberal view. Vogel 2008 argues that the problem with bootstrapping argu-
ments has to do with a circular application of the rule for forming a given
class of beliefs one would aim to justify by means of these very arguments.
Cohen 2010 contests this. I will offer yet another different diagnosis.
36. Were one uneasy with the content of this parenthetical remark, one could
think of Pryor’s theory as a form of basic justification (or warrant)-view.
Accordingly, bootstrapping arguments directed against it would put pressure
on the theory by noticing that one may get a justification for the conclu-
sion – rather than knowledge of it – if that account is in place.
37. Kallestrup 2012. A similar remark can be found in Cohen 2010, who, nowa-
days, thinks that the problem with bootstrapping arguments is not really
the endorsement of basic knowledge or justification. This is not to say that
Cohen is happy with the latter. Indeed, he argues that “the very idea of
basic justification is incoherent” (Cohen 2010, p. 150). However, if this is
the case, he continues, we should look for an account of the faultlessness of
bootstrapping arguments that accepts that “we cannot have justified percep-
tual beliefs without having a prior justified belief that perception is reliable”
(2010, p. 141). As we shall see, I am not convinced by the latter claim. For it
seems to me to proceed from ignorance of the possibility that moderatism is
the right account of perception, and a midway course between views such as
Pryor’s, which countenance basic justification, and conservative views such
as Wright’s, which require prior justification for very general presuppositions
like “My perceptions are generally reliable”. According to Cohen, then, we
should recognize that we are a priori defeasibly justified in holding that if
something looks red, then it is red. Yet, bootstrapping arguments are infelici-
tous because they cannot raise the degree of justification one already has for
that conditional and therefore for their conclusion. As said, I think that we
are not forced to enter into the complex issue of a priori defeasible justifica-
tion of contingent propositions like “If something looks red, then it is red”
we have already briefly addressed in the last chapter (§3.1) in connection
with Wedgwood. For, as I argue, we can diagnose the fault with bootstrap-
ping arguments differently. Let me simply register that I am dubious that the
one under consideration could be rightly taken to be an example of a priori
justifiable empirical proposition.
38. In essence, this is the position put forward by Bergmann 2004.
Notes 197
39. In RED TABLE one could get independent warrant for (3), for example, by
means of testimony.
40. Pryor 2000, 2004, 2012.
41. We shall waive here any worry about whether there is such a class of empir-
ical beliefs, or about how to individuate them.
42. Pryor 2012.
43. Pryor 2012, §7.
44. In his very early writings on the topic Davies seemed attracted to this view
which, however, differs from what he later came to accept as transmission
failure, that is, TF1. This is not the place to rehearse Davies’s views, although
in the interest of completeness it might be useful to check whether, after all,
and contrary to Wright’s and Davies’s own later opinion, it could really count
as a case of transmission failure.
45. Pryor 2012, pp. 298–299.
46. Actually, it might be a bit more complicated than that because, according
to the moderate position, as I have characterized it, though the obtaining
of (III) is not a constitutive element of one’s perceptual warrant for (I), the
assumption of (III) is. Therefore, it might be that moderatism as I think of it
falls in between Pryor’s two notions.
47. Let me repeat, however, that divorcing the moderate view from TF2 seems
to me to pre-empt the motivation for the former view. For it would then be
possible to earn warrant for the conclusion of an argument even if assuming
that very conclusion were needed in order to have warrant for its premises in
the first place. Furthermore, it would involve one in what seems to me a very
overt form of question beggingness, viz. the one I represented in the main
text as
[Q]
P
P→Q
———
Q
Still such a position is part of the logical space in this area. In what follows,
I will ignore this possibility and will thus take transmission failure 2 to be
entailed by the moderate view.
48. (though not necessarily to the moderate view). See previous footnote.
4 The Extended Rationality View

1. This is indeed one of the horns of Agrippa’s trilemma. Namely, that in the
quest for justification we may end up with propositions which are not them-
selves justified and are thus arbitrarily assumed.
2. A word of caution may be useful here. Usually, Humean skepticism endorses
the quite radical position that, since we have no warrant for general proposi-
tions such as “There is an external world” and such a warrant is needed to
have ordinary evidential warrant for empirical propositions like “Here is a
hand”, we do not have warrant for the latter either. Now, it is not my aim
to dispute that, historically, this is Humean skepticism. However, a kind of
198 Notes
skepticism of Humean descent could grant, at least for the sake of argument,
that no warrant for “There is an external world” is needed to have an eviden-
tial justification for “Here is a hand” and still point out the a-rationality and
arbitrariness of our basic assumptions. This understanding of the challenge
posed by a possible version of Humean skepticism is common in certain quar-
ters, such as those inspired by Wright’s writings on the topic. Whether or not
it really deserves the epithet “Humean” is not that important for present
purposes. Hence, should a reader feel uncomfortable with it, my advice
would be to think of the discussion that will occupy us for the best part of
this chapter as a response to a possible objection to moderatism. Namely, one
that pointed out the a-rationality and arbitrariness of the basic assumptions
that are constitutive ingredients of our perceptual warrants and the skeptical
and potentially relativistic consequences of such a predicament.
3. Yuval Avnur has suggested that if we adhered to some “ought/must implies
can/may” principle, the fact that one cannot help believing P would entail
that one is within one’s rights to believe it, even if one does not have a justifi-
cation for P. However, the problem with this way of fleshing out moderatism
is that belief in P would be compelled only by our psychological constitution
or by our acculturation within a certain form of life that shares a certain
Weltbild. So it is hard to see how this could result in making one epistemically
blameless in believing P.
4. For a complex reconstruction of pragmatist positions with respect to this
issue, see Tiercelin 2005.
5. Coliva 2010a, chapter 3.
6. The incipit of that passage is, in fact, the following “But it is not (my emphasis)
that the situation is like this”, followed by the words quoted in the main text.
7. As remarked in Chapter 1 (Note 20) some Wittgensteinian scholars deny
that for LW “There is an external world” is a hinge (Williams 2004b) and
hold that it is plain nonsense instead (but, as remarked in note 20, Williams
has backtracked in conversation). Were they right in their interpretation,
the point would remain that Wittgenstein endorsed a form of moderatism
whereby the assumptions that are necessary to bring all sorts of evidence to
bear on the class of appropriate beliefs would be more specific and context-
dependent than the version of moderatism presented in this book. However,
such a view would obviously sit badly with the kind of anti-skeptical
proposal about the existence of an external world I think can be elicited
from On Certainty.
8. Thanks are due to Yuval Avnur for pressing me on this point.
9. Cf. Tiercelin 2005, chapter 3.
10. Tiercelin 2005, chapter 3 also insists on this point, although she takes
Wittgenstein to be a sui generis kind of pragmatist. While I agree with her
on the relevance of Wittgenstein’s insistence on the fact that meaningful
doubts must be based on reasons, I would resist her pragmatist reading of On
Certainty. Notice, moreover, that Wittgenstein actually claimed that skepti-
cism would be utterly nonsensical, for it violates the conditions of sense. So
skeptical doubts would seem to retain a meaning, while in fact they lack it,
simply because we project it onto them from the ordinary contexts of the use
of “to doubt”. There will be more about this issue in the following.
11. Thanks are due to Yuval Avnur for pressing me on this point.
Notes 199
12. This, broadly speaking, is the view put forward by Danièle Moyal-Sharrock
(2004).
13. I myself have proposed this interpretation of hinges in Coliva 2010, 2012a.
14. Cf. Coliva 2010a, ch. 3, 2013a, b.
15. This would be another way of showing that skepticism is concerned with
epistemological worries, which would persist even if one granted the kind of
realist metaphysics that backs the externalist view.
16. Cartesian skepticism does not target directly the assumption “There is an
external world”, but the presupposition that we are not victims of sustained
delusions or dreams. For the time being, I will focus only on the kind of skep-
ticism whose target is the former assumption – that is, Humean skepticism. I
take up the issue of Cartesian skepticism later (§3).
17. Or belief forming method.
18. Recall that the Humean skeptic I argue against here is someone who accepts
that our ordinary empirical beliefs are warranted through our usual epis-
temic procedures and yet claims that they ultimately rest on unwarranted
and therefore non-rational assumptions. I would like to thank Yuval Avnur
for pressing me on this point.
19. I confine myself to considerations pertaining to empirical epistemic ration-
ality – namely, the rationality produced by empirical and, as we shall see in
a moment, basic epistemic practices such as forming, assessing, and with-
drawing from beliefs about objects in our surroundings on the basis of the
deliverances of our senses. I do not take into account other aspects of epis-
temic rationality. For this reason, and for ease of exposition, in the following
I drop the qualification “empirical”.
20. The latter, I take it, is common ground among skeptics and non-skeptics
alike, since skeptics are no idealists. I touch on idealism in §4.
21. Recall from Chapter 1 (§3) that being constitutive of epistemic rationality does
not amount to merely being a necessary condition for being so rational. Rather,
it amounts to the idea that certain assumptions are needed, together with
appropriate courses of experience, to have perceptual warrants for ordinary
empirical beliefs such as “Here is a hand”. This should respond to an objection,
raised by Yuval Avnur, according to which having a mind, say, would be consti-
tutive of epistemic rationality and yet would not fall within the scope of epis-
temic rationality itself. In my view, having a mind is a necessary condition for
engaging in epistemic rationality (based on perception), but it is not a constitu-
tive condition of it. For it is not in itself part of the nature of perceptual justifica-
tions, though it is a necessary condition for acquiring them since, presumably,
it is only by having a mind that one could have certain experiences.
22. In analogy with certain debates in logic, for example between classical and
intuitionistic logicians, one may worry that though certain assumptions
could be seen as constitutive rules, they could be criticized in light of some
aspects of our practice, or in light of some desirable principles. In the case
of the assumptions under consideration here though, it is hard to see what
reasons there could be to criticize and revise them. For, as we have repeatedly
pointed out, all reasons would seem to presuppose them.
23. In so doing I think we both do not follow Wittgenstein, at least not the letter
of On Certainty, despite the fact that Wright (2004a, p. 189; cf. also Wright
2004c) advertises his own views as Wittgensteinian in spirit. In contrast,
200 Notes
I explicitly acknowledge that I depart from the letter of On Certainty, and

develop some of its elements in directions that would not have been endorsed
by its author. See Coliva 2010a, Introduction and chapter 3 on this, as well as
Coliva forthcoming-a and -b.
24. As the reader will recall from Chapter 2 (§2), he thinks it is mandated by our
conceptual scheme which countenances mind-independent objects, whereas
I think it is mandated by considerations concerned with our notion of epis-
temic rationality. For what it is worth, Wright himself (2004a, pp. 203) seems
to be dubious of the prospects of success of his “entitlement of substance”.
I discuss them critically in that chapter (see also Coliva 2007, 2012b, e and
2014b).
25. See Coliva 2008.
26. I discuss all these cases in other writings, in particular Coliva 2009, 2010c
and I consider the case of basic logical laws in the next chapter.
27. In Coliva 2009 I proposed, through an examination of several specimens
of relativism, five conditions which should be met in order to give rise to
epistemic relativism. I will simply list them here for convenience, and refer
the reader to that book for their defense (pp. 50–67) and specific criticisms
(pp. 129–146). (i) There is no objectively valid or epistemically superior epis-
temic system; (ii) there are different epistemic systems (either theories, or
practices of justification, or explanatory principles); (iii) which are all equally
legitimate; (iv) but mutually incompatible; (v) and such as to give rise to
different sets of justified (or even known) beliefs.
28. Boghossian 2006 also distinguishes between basic and non-basic epistemic
practices. In Coliva (2009) I discuss other interesting cases concerning non-
basic epistemic practices, such as the one regarding different theories about
the constitution of the skies.
29. I discuss this case and the alleged counterexample posed by the different
logic of the Azande in Coliva 2009. Another important example discussed
in that book is reliance on the principle of efficient cause and the apparent
counterexample posed by the Azande’s adherence to the principle of final
cause. I consider the deductive case in Chapter 5.
30. The reader will recall that we tackled the issue of the very content of percep-
tual experiences in the last section of Chapter 1.
31. The most developed discussion of all these aspects of perception can be
found in Burge 2010 with which I agree. See Coliva 2012a for a critical study
of Burge’s book. It has to be stressed that Burge is absolutely no disjunctivist
about perception and that he is a wholehearted representationalist. The very
content of our perceptual experiences is as of objects and properties “out
there”, but this does not mean that it is necessarily veridical (otherwise it
would be no perceptual experience, but an illusion); or that objects and prop-
erties are actually elements of our perceptions.
32. Again, an impressive discussion of the scientific findings in this respect,
which actually chime with commonsense, can be found in Burge 2010.
33. This is where I differ from Burge. See Coliva 2012a.
34. As Burge 2010 rightly notices, not all creatures perceptually represent objects,
although they are capable of perceiving properties, such as colors and shapes,
as objective, that is, as external to them. Yet human beings and many other
animals do perceive objects as such.
Notes 201
35. For a recent version of this challenge, see Enoch 2006. Here is a telling quote
(p. 179): “Classify my bodily movements and indeed me as you like. Perhaps
I cannot be classified as an agent without aiming to constitute myself. But
why should I be an agent? Perhaps I can’t act without aiming at self-constitu-
tion, but why should I act? If your reasoning works, this just shows that I do
not care about agency and action. I am perfectly happy being a shmagent – a
nonagent who is very similar to agents but who lacks the aim (constitutive
of agency but not of shmagency) of self-constitution. I am perfectly happy
performing shmactions – nonaction events that are very similar to actions
but that lack the aim (constitutive of actions but not of shmactions) of self-
constitution.”
36. Here is another telling quote from Enoch 2006 (p. 186): “If a constitutive-aim
or constitutive-motives theory is going to work for agency, then, it is not
sufficient to show that some aims or motives or capacities are constitutive
of agency. Rather, it is also necessary to show that the ‘game’ of agency is
one we have reason to play, that we have reasons to be agents rather than
shmagents (analogously: that we have a reason to build a house rather than
a shmouse)”.
37. It has been suggested to me that this challenge is similar to the Pyrrhonian
one. Pyrrhonian skepticism is notoriously difficult to represent accurately.
Hence, in order to avoid the charge of scholarly inaccuracy, I prefer to avoid
linking the challenge currently under scrutiny to Pyrrhonism.
38. Thanks are due to Yuval Avnur for pressing me to develop this section far
beyond its initial draft. I am not sure the final version will satisfy him, but at
least it should make clearer why I am opting for what can be seen as a radical
conception of the status of our basic assumptions.
39. It has been pointed out to me that Korman (2014) raises such a worry.
40. These are the views about truth that can be traced back to Peirce, Putnam,
and Wright respectively. For an interesting discussion of how, from a histor-
ical point of view, Peirce’s position could in fact be compatible with a some-
what realist account of truth, cf. Tiercelin 2005, ch. 4.
41. Here I have in mind the Euthyphro contrast utilized by Wright 1992 to differ-
entiate realist and anti-realist areas of discourse. I am simply applying it to
the case of the predicate “true”.
42. Skepticism would thus seem to depend on a realist notion of truth that is
not mandatory. Yet, as an anti-realist about truth, one should then face the
challenge of how to draw the distinction between truth and justification that
is central to our conceptual scheme and that skepticism seems to exploit.
Here is a tentative thought. A given piece of perceptual evidence, say, has
a certain content that P (e.g. that there is a hand). That increases the prob-
ability that P is true and therefore makes one’s belief in P justified. To say
that P is true, in turn, would entail that for any increment of information,
there would be no reason to revise one’s belief that there is a hand. Since P in
this case is a contingent proposition, one’s perceptual evidence could in fact
be defeated by further evidence, which, at least in some cases, would suffice
to prove that P is false. Such a reconceptualization of the distinction would
then offer no purchase to a skeptic. For there would be no room for the idea
that despite the fact that all the evidence at our disposal speaks in favour of P
and none against it, P might be false. This reasoning, however, holds only for
202 Notes
propositions whose truth is given within a system of assumptions. That is to

say, thanks to the hinges that, in this case, make the acquisition of perceptual
warrants possible. It cannot be applied to those hinges in turn, for that would
entail a vicious form of circularity. That is why, for them, we should content
ourselves with the minimal notion of truth canvassed in the main text.
5 The Extended Rationality View Extended

1. Their ancestor is Thomas Reid. Contemporary supporters of anti-reduc-
tionsim are Coady 1992, Burge 1993, McDowell 1998, Audi 1997, Graham
1997.
2. That is what reductionists would claim is necessary in order justifiably to
go from (I) to (II). The ancestor of that position is David Hume. The main
contemporary supporter of that view is Elisabeth Fricker (1998, 2006).
3. I would like to leave open the possibility that some primates might, on occa-
sion, be informants in the sense required by testimony, and that one might
receive testimonies from other kinds of creatures, such as angels or God, if
they existed.
4. I think Harman and Sherman 2011 raise a similar concern, though in connec-
tion with their discussion of the limitations of the principle of closure for
knowledge under known entailment.
5. This is the question at the heart of Boghossian 2003.
6. This is nicely brought out in Schechter and Enoch (2006, p. 687 and 2008,
p. 552) even though, as we shall see, they do not take full account of this
fact.
7. Doubts are forcefully cast on such a possibility by Boghossian 2003 and
Wright 2007b.
8. Objections to crudely reliabilist accounts of justification in connection with
the issues presently discussed can be found in Boghossian 2003 and 2012a,
b, repeated in Schechter and Enoch 2006, 2008.
9. As customary, I will use small caps to mention concepts.
10. Even though Boghossian takes himself to be answering question (3), rather
than (4).
11. In fact Casalegno’s example involves the concept AND, but to stick to our leading
example in this section I have taken the liberty of changing it to IF THEN.
12. I draw the following contrived McGee’s counterexample from Cariani 2013.
We know that Mario, an Italian, does not like travelling at all and, if he
travels, he likes staying as close to home as possible. His parents are trying to
convince him to visit relatives, some of whom live in Paris and some in LA.
So Lucia, Mario’s mother, may reason thus: (1) If Mario does not go to Paris, if
he travels at all, he will go to LA; (2) Mario will not go to Paris; so (3) If Mario
travels at all, he will go to LA. According to McGee, while the premises are
intuitively acceptable, the conclusion is not. Personally, I never found this
claim intuitive. Nevertheless, the point remains that the reasoning involves
embedded conditionals and that the intuitions it should elicit depend on
contextual information about Mario and his psychology.
13. For a similar kind of reply, though applied to the case of AND, see Boghossian
2012b, pp. 232–233. Williamson 2012 takes issue with Boghossian’s reply.
Notes 203
However, he seems to concede that the deviant subject would have to behave
like a non-deviant in non-contentious cases and yet deviate in contentious
ones (cf. p. 243). However, as far as I can see, this would mean conceding the
basic inferentialist point: there are some instances of basic inferences one has
to be willing to make in order to count as having the concept (or the same
concept) AND (or IF THEN), as we do.
14. For a similar reply, but addressed to Casalegno’s original example (cf.
fn. 11), see Boghossian 2012b, pp. 228–229. Williamson 2012 tries to defend
Casalegno’s position.
15. As we shall see, they oscillate between providing a justification for modus
ponens itself and recognizing that it is a second-order problem. An impor-
tant clue to this conflation is the fact that they feel the need for their account
to provide a distinction between valid and invalid rules of inference (cf.
Schechter and Enoch 2008, p. 548, where, for instance, they contrast modus
ponens with affirming the consequent). However, that is not the main
problem, as we have already remarked upon several times. Let us grant that
there are basic rules of inference and that some of them can ostensibly be
proven to be valid, that is necessarily truth-preserving. The issue we face is:
are these belief-forming methods justified and, if so, how? Where should
this be taken in the quest for a reason that we, as theorists, can produce
to vindicate the epistemic legitimacy of our employment of such valid and
basic rules of inference? As we saw, a way of making the problem vivid is to
consider what we might say to someone who was not already inclined to
reason in accord with it. Once we are clear about the kind of question we
are asking, we can, I think, more easily see why there is no justification for
modus ponens itself and all we can do is give ourselves an a priori reason
to believe the following “To reason in accordance with modus ponens is
rational, even if there is no justification of modus ponens as such”.
16. Here I will not consider Wright’s final move in his 2004b paper to boost that
entitlement to get knowledge of modus ponens out of it. It relies on a form
of “alchemy”, which is extremely suspect, as he himself somehow recog-
nizes by allowing that people may well think that his rule-circular account
of our knowledge of the validity of modus ponens would “prove to founder”
(p. 174).
17. Surprisingly, Schechter and Enoch (2008, p. 564), at least on one possible
understanding of the phrase “there must be some substantive criterion
that distinguishes epistemically justified basic methods from the rest”, miss
that much and go on to say “this is where the pragmatic account fits in. It
provides a general, principled explanation of in virtue of what certain basic
belief-forming methods are justified”.
18. Surprisingly, Schechter and Enoch (2008, p. 564) take the indispensability for
us to reason in accord with modus ponens as an epistemic justification for
it.
19. Schechter and Enoch (2008, p. 558) attempt an answer in terms of a project
which is such that “a particular agent rationally ought to engage in it given
the facts of her constitution and general abilities”. They do recognize,
however, the uninformativeness of their characterization.
20. Marconi (forthcoming) stresses the similarity between my proposal and
Wittgenstein’s position in Part I of Remarks on the Foundations of Mathematics.
204 Notes
21. Such a qualification should allow us to dispense with McGee’s alleged coun-
terexamples altogether, for even if they were genuine (but see note 10), they
would clearly not be basic instances of modus ponens. It should also allow us
to dispense with invalid rules of inference, like affirming the consequent.
22. Interestingly, Wright 2012b puts forward the view that to reason in accord
with modus ponens is constitutive of rationality as an explanation of what
an inference is. Yet, he does not endorse this as an explanation of why modus
ponens is justified, which he still thinks can be afforded by means of his
entitlements of cognitive project strategy. To repeat a point worth stressing,
in order to avoid potential confusion, I do not think we can justify modus
ponens, unless by this we mean giving a rule-circular proof of its validity.
Yet, we can surely provide ourselves with a justification for the second-order
belief that to reason in accordance with it is rational. As stated, that justifica-
tion is given by the argument we have proposed in favor of the view that to
reason in accordance with it is constitutive of logical rationality.
23. Notice that the alleged ethnographical counterexample of the Azande, due
to E. Evans-Pritchard 1937, is no such case, for it was based on a mistaken
translation. In particular, the first ethnographers had not paid attention to
the fact that for the Azande only the sons of a witchdoctor who are “hot”,
whatever that might mean, are witchdoctors in their turn, despite the fact
that the Azande believe that witchcraft transmits patrilinearly by means of a
magic substance all sons of a witchdoctor inherit from their father. A similar
mistake in translation can be found in ethnographic early reports about the
tribe of the Kassena (Mangiameli 2010) who live in northern Ghana. They
were reported to believe that baobabs are sacred and therefore intangible,
while they were witnessed cutting them with no specific sense of guilt.
Further studies revealed that the Kassena think that only after a certain age
are baobabs sacred and therefore that they believe that baobabs can inno-
cently be cut before that age.
24. In Grundgesetze der Arithmetik, XVI, Frege writes: “What, however, if beings
were even found whose laws of thought directly contradicted ours, so that
their application often led to opposite results? The psychological logician
could only accept this and say: for them, those laws hold, for us these. I
would say: here we have a hitherto unknown kind of madness”. Personally,
I would not ban them as necessarily mad, in the sense of being insane, but
I would deem them as logically irrational. Let me also note that concurring
with Frege’s judgment in this case does not depend on embracing either his
unshakable faith in classical logic, or his overall conception of logic. To stress,
I do agree with him that logic is normative and non-descriptive. However,
the provenance of these norms, according to the Wittgensteinian perspective
I have been concerned to develop in this book, lies in our communal prac-
tices. It is not written in stone in a third realm of abstract entities.
Bibliography
Alston W.P. (1993) The Reliability of Sense Perception (Ithaca, N.Y.: Cornell
University Press).
Audi R. (1997) ‘The Place of Testimony in the Fabric of Knowledge and
Justification’, American Philosophical Quarterly, 34, 405–422.
Avnur Y. (2011a) ‘An Old Problem for New Rationalism’, Synthese, 183/2,
175–185.
Avnur Y. (2011b) ‘Hawthorne on the Deeply Contingent’, Philosophy and
Phenomenological Research, 83/1, 174–183.
Beebee H. (2001) ‘Transfer of Warrant, Begging the Question and Semantic
Externalism’, Philosophical Quarterly, 51, 356–374.
Bergmann M. (2004) ‘Epistemic Circularity: Malignant and Benign’, Philosophy
and Phenomenological Research, 69/3, 709–727.
Bermúdez J.L. (1998) The Paradox of Self-Consciousness (Cambridge, Mass: MIT
Press).
Boghossian P.A. (2003) ‘Blind Reasoning’, Proceedings of the Aristotelian Society,
Supplementary Volume, 77, 225–248.
Boghossian P.A. (2006) Fear of Knowledge (New York: Oxford University Press).
Boghossian P.A. (2012a) ‘Blind Rule-Following’, in Wright C. and Coliva A. (eds)
Mind, Meaning, and Knowledge: Themes From the Philosophy of Crispin Wright
(New York: Oxford University Press), pp. 27–48.
Boghossian P.A. (2012b) ‘Inferentialism and the Epistemology of Logic: Reﬂections
on Casalegno and Williamson’, Dialectica, 66/2, 221–236.
Brewer B. (1999) Perception and Reason (Oxford: Clarendon Press).
Brown J. (2005) ‘Doubt, Circularity and the Moorean Response to the Sceptic’,
Philosophical Perspectives, 19/1, 1–14.
Burge T. (1993) ‘Content Preservation’, Philosophical Review, 102/4, 457–488.
Burge T. (2003) ‘Perceptual Entitlement’, Philosophy and Phenomenological Research,
67/3, 503–548.
Burge T. (2010) Origins of Objectivity (Oxford: Oxford University Press).
Cariani F. (2013) ‘Modus Ponens’, Aphex, 7, 1–32.
Casalegno P. (2004) ‘Logical Concepts and Logical Inferences’, Dialectica, 58,
395–411.
Casullo A. (2007) ‘What Is an Entitlement?’, Acta Analytica, 22, 267–279.
Coady C. A. J. (1992) Testimony: A Philosophical Study (Oxford: Clarendon).
Cohen S. (2002) ‘Basic Knowledge and the Problem of Easy Knowledge’, Philosophy
Cohen S. (2005) ‘Why Basic Knowledge is Easy Knowledge’, Philosophy and
Cohen S. (2010) ‘Bootstrapping, Defeasible Reasoning, and a Priori Justification’,
Philosophical Perspectives, 24/1, 141–159.
Coliva A. (2003) ‘The Problem of the Finer-Grained Content of Experience.
A Redefinition of Its Role within the Debate between McDowell and
205
206 Bibliography
Nonconceptual Theorists’, Dialectica, 57, 57–70. Reprinted in Coliva A. (2006)

I concetti (Roma: Carocci).
Coliva A. (2004) ‘Proof of an External World: Transmission-failure, Begging the
Question or Dialectical Ineffectiveness? – Moore, Wright and Pryor’, in Coliva
A. and Picardi E. (eds) Wittgenstein Today (Padova: Il Poligrafo), pp. 397–415.
Coliva A. (2007) ‘Lo scetticismo sull’esistenza del mondo esterno’, in Coliva A.
(ed.) Filosofia analitica. Temi e problemi (Roma: Carocci), pp. 255–280.
Coliva A. (2008) ‘The Paradox of Moore’s Proof of an External World’, The
Philosophical Quarterly, 58, 234–243.
Coliva A. (2009) I modi del relativismo (Roma-Bari: Laterza).
Coliva A. (2010a) Moore and Wittgenstein. Scepticism, Certainty and Common Sense
(London/Basingstoke: Palgrave Macmillan).
Coliva A. (2010b) ‘Moore’s Proof and Martin Davies’s Epistemic Projects’,
Australasian Journal of Philosophy, 88/1, 101–116.
Coliva A. (2010c) ‘Was Wittgenstein an Epistemic Relativist?’, Philosophical
Investigations, 33/1, 1–23.
Coliva A. (2012a) ‘Critical notice of Tyler Burge’s Origins of Objectivity’, Disputatio,
4/33, 515–530.
Coliva A. (2012b) ‘Moore’s Proof, Liberals and Conservatives. Is There a
(Wittgensteinian) Third Way?’, in Coliva A. (ed.) Mind, Meaning and Knowledge.
Themes from the Philosophy of Crispin Wright (Oxford: Oxford University Press),
pp. 323–351.
Coliva A. (2012c) ‘Percepire le ragioni?’, Iride, XXV/65, 117–130.
Coliva A. (2012d) ‘Varieties of Failure (of Warrant Transmission – What Else?!)’,
Synthese, 189/2, 235–254.
Coliva A. (2012e) Scetticismo. Dubbio, paradosso, conoscenza (Rome-Bari: Laterza).
Coliva A. (2013a) ‘Hinges and Certainty. A Précis of Moore and Wittgenstein.
Scepticism, Certainty and Common Sense’, Philosophia, 41, 1–12.
Coliva A. (2013b) ‘Replies’, Philosophia, 41, 81–96.
Coliva A. (2014a) ‘Lessons from Pascal Engel. Achilles, the Tortoise and Hinge
Epistemology for Basic Logical Laws’, in Meylan A., Dutant J., Logins A. and
Fassio D. (eds) Hommage à Pascal Engel, pp. 696–715, http://www.unige.ch/
lettres/philo/publications/engel/liberamicorum.
Coliva A. (2014b) ‘Moderatism, Transmission Failures, Closure and Humean
Scepticism’, in Zardini E. and Dodd D. (eds) Scepticism and Perceptual Justification
(Oxford: Oxford University Press), pp. 248–272.
Coliva A. (forthcoming-a) ‘Which Hinge Epistemology? – I’, The International
Journal for the Study of Scepticism.
Coliva A. (forthcoming-b) ‘Which Hinge Epistemology? – II’, in Malachowski A.
(ed.) Later Wittgenstein Now (New York: Blackwell).
Coliva, A. and Zanetti, L. (forthcoming) “Epistemic Pluralism”.
Davidson D. (1986) ‘A Coherence Theory of Truth and Knowledge’, in Lepore E.
(ed.) Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson
(New York: Blackwell).
Davies M. (1998) ‘Externalism, Architecturalism, and Epistemic Warrant’, in
Wright C., Smith B.C. and Macdonald C. (eds) Knowing Our Own Minds (Oxford:
Oxford University Press), pp. 321–61.
Davies M. (2000) ‘Externalism and Armchair Knowledge’, in Boghossian P.A. and
Peacocke C. (eds), New Essays on the A Priori (Oxford: Oxford University Press),
pp. 384–414.
Bibliography 207
Davies M. (2003) ‘Externalism, Self-Knowledge and Transmission of Warrant’,

in Frapolli M.J. and Romero E. (eds) Meaning, Basic Self-Knowledge, and Mind
(Stanford: CSLI Publications).
Davies M. (2004) ‘Epistemic Entitlement, Warrant Transmission and Easy
Knowledge’, Aristotelian Society Supplementary Volume, 78/1, 213–245.
Davies M. (2009) ‘Two Purposes of Arguing and Two Epistemic Projects’, in
Ravenscroft I. (ed.) Minds, Ethics and Conditionals (Oxford: Oxford University
Press), pp. 337–363.
DeRose K. (1995) ‘Solving the skeptical problem’, Philosophical Review, 104/1, 1–52.
Dretske F. (1970) ‘Epistemic Operators’, The Journal of Philosophy, 67/24,
1007–1023.
Dretske F. (2000) ‘Entitlements: Epistemic Rights without Epistemic Duties’,
Philosophy and Phenomenological Research, 60/3, 591–606.
Dretske F. (2005) ‘Is Knowledge Closed Under Known Entailment? The Case
Against Closure’, in Steup M. and Sosa E. (eds) Contemporary Debates in
Epistemology (Oxford: Blackwell), pp. 13–26.
Enoch D. (2006) ‘Agency, Shmagency: Why Normativity Will not Come From
What is Constitutive of Action’, Philosophical Review, 115/2, 169–198.
Evans G. (1982) The Varieties of Reference (Oxford: Oxford University Press).
Evans-Pritchard E.E. (1937) Witchcraft, Oracles and Magic Among the Azande
(Oxford: Oxford University Press).
Frege G. (1893) Grudgesetze der Arithmetik (Jena: Verlag Hermann Pohle). English
Translation in Geach P. and Black M. (eds) The Philosophical Writings of Gottlob
Frege (Oxford: Basil Blackwell, 1960), pp. 137–244.
Fricker E. (1998) ‘Against Gullibility’, in Chakrabarti A. and Matilal B.K. (eds)
Knowing from Words. (Dordrecht: Kluwer), pp. 125–161.
Fricker E. (2006) ‘Second-Hand Knowledge’, Philosophy and Phenomenological
Research, 73/3, 592–618.
Gerken, M. (forthcoming) “Entitlement and Its Limits”.
Graham P. J. (1997) ‘What is Testimony?’, The Philosophical Quarterly, 47/187,
227–232.
Harman G. (1986) Change in View: Principles of Reasoning (Cambridge, Mass: MIT
Press).
Harman G. and Sherman B. (2011) ‘Knowledge and Assumptions’, Philosophical
Studies, 156/1, 131–140.
Hawthorne J. (2002) ‘Deeply Contingent A Priori Knowledge’, Philosophy and
Hawthorne J. (2005) ‘Is Knowledge Closed under Known Entailment? The Case
for Closure’, in Steup M. and Sosa E. (eds) Contemporary Debates in Epistemology
(Oxford: Blackwell), pp. 26–43.
Hume D. (1739), A Treatise of Human Nature (Baltimore: Penguin Books, 1969).
Hume D. (1748), Enquiries Concerning Human Understanding and Concerning the
Principles of Morals (Oxford: Clarendon Press, 1975).
Jenkins C. (2007) ‘Entitlement and Rationality’, Synthese, 157/1, 25–45.
Kallestrup J. (2012) ‘Bootstrap and Rollback: Generalizing Epistemic Circularity’,
Synthese, 189/2, 395–413.
Kelly S.D. (2001) ‘Demonstrative Concepts and Experience’, Philosophical Review,
110/3, 397–420.
Lackey J. and Sosa E. (eds) (2006) The Epistemology of Testimony (New York: Oxford
University Press).
208 Bibliography
Luper S. (2006) ‘Dretske on Knowledge Closure’, Australasian Journal of Philosophy

84/3, 379–394.
Mangiameli G. (2010) Le abitudini dell’acqua (Milano: Edizioni Unicopli).
Marconi D. (forthcoming) ‘Persuading the Tortoise’, Philosophical Investigations.
McDowell J. (1984) ‘De Re Senses’, Philosophical Quarterly, 34, 283–294.
McDowell J. (1994) Mind and World (Harvard: Harvard University Press).
McDowell J. (1998) ‘Knowledge by Hearsay’, in Chakrabarti A. and Matilal B. K.
(eds) Knowing from words, (Dordrecht: Kluwer), pp. 195–244.
McDowell J. (2009) ‘Avoiding the Myth of the Given’, in Having the World in View
(Cambridge, Mass: Harvard University Press), pp. 256–272
McGee V. (1985) ‘A Counterexample to Modus Ponens’, Journal of Philosophy, 82,
462–71.
McGinn C. (1989) Mental Content (Oxford: Basil Blackwell).
McGlynn A. (2014) ‘On Epistemic Alchemy’, in Dodd D. and Zardini E. (eds),
Scepticism and Perceptual Justification (New York: Oxford University Press),
pp. 173–89.
McHugh C. (2012) ‘Reasons and Self-Knowledge’, in Coliva A. (ed.) The Self and
Self-Knowledge (New York: Oxford University Press), pp. 139–163.
Millikan R. (1998) ‘A Common Structure for Concepts of Individuals, Stuffs, and
Real Kinds: More Mama, More Milk, and More Mouse’, Behavioral and Brain
Sciences, 21, 55–100.
Moore G.E. (1925) ‘A Defence of Common Sense’, in Philosophical Papers (New
York: Collier Books, 1970), pp 33–59.
Moore G.E. (1939) ‘Proof of an External World’, in Philosophical Papers (London:
Allen and Unwin), pp. 126–148.
Moyal-Sharrock D. (2004) Understanding Wittgenstein’s On Certainty (London/
Basingstoke: Palgrave McMillian).
Nozick R. (1981) Philosophical Explanations (Harvard: Harvard University Press).
Olsson E. J. (forthcoming) “How Carnap’s Methodology of Explication Leads to
Epistemic Pluralism”.
Peacocke C. (1992) A Study of Concepts (Cambridge, Mass: MIT Press).
Peacocke C. (1999) Being Known (Oxford: Clarendon Press).
Peacocke C. (2004) The Realm of Reason (Oxford: Clarendon Press).
Peacocke C. (2008) Truly Understood (Oxford: Oxford University Press).
Pedersen N. (forthcoming) “Varieties of Epistemic Pluralism”.
Peirce C. S. (1935) Collected Papers of Charles Sanders Peirce. Volumes V and VI
(Cambridge (Mass.): Harvard University Press).
Pritchard D. (2005a) Epistemic Luck (Oxford: Oxford University Press).
Pritchard D. (2005b) ‘Wittgenstein’s On Certainty and Contemporary Anti-
scepticism’, in Moyal-Sharrock D. and Brenner W.H. (eds) Readings of
Wittgenstein’s On Certainty (London: Palgrave), pp. 189–224.
Pritchard D. (forthcoming) Epistemic Angst: Radical Scepticism and the Groundlessness
of Our Believing (Princeton: Princeton University Press).
Pryor J. (2000) ‘The Skeptic and the Dogmatist’, Noûs, 34, 517–49.
Pryor J. (2004) ‘What’s Wrong with Moore’s Argument?’, Philosophical Issues, 14,
349–378.
Pryor J. (2005) ‘There is Immediate Justification’, in Steup M. and Sosa E. (eds)
Contemporary Debates in Epistemology (Oxford: Blackwell), pp. 181–202.
Bibliography 209
Pryor J. (2012) ‘When Warrant Transmits’, in Wright C. and Coliva A. (eds) Mind,
Meaning, and Knowledge: Themes From the Philosophy of Crispin Wright (New
York: Oxford University Press), pp. 269–303.
Putnam H. (1975) ‘The Meaning of Meaning’, reprinted in Mind, Language and
Reality. Philosophical Papers Vol. 2 (Cambridge: Cambridge University Press),
pp. 215–271.
Putnam H. (1981) ‘Brains in a Vat’, in Reason, Truth and History (Cambridge:
Cambridge University Press), pp. 1–21.
Reid T. (1764) An Inquiry into the Human Mind on the Principles of Common Sense
(University Park: Pennsylvania State University Press, 1997).
Schechter J. and Enoch D. (2006) ‘Meaning and Justification: The Case of Modus
Ponens’, Noûs, 40/4, 687–715.
Schechter J. and Enoch D. (2008) ‘How Are Basic Belief-Forming Methods
Justified?’, Philosophy and Phenomenological Research, 76/3, 547–579.
Schiffer S. (2004) ‘Skepticism and the Vagaries of Justified Belief’, Philosophical
Studies, 119/1–2, 161–184.
Sellars W. (1956) Empiricism and the Philosophy of Mind (Minneapolis: University
of Minnesota Press).
Siegel S. (2010) The Contents of Visual Experience (Oxford: Oxford University
Press).
Silins N. (2005) ‘Transmission Failure Failure’, Philosophical Studies, 126, 71–102.
Silins N. (2007) ‘Basic Justification and the Moorean Response to the Skeptic’, in
Gendler T.S. and Hawthorne J. (eds) Oxford Studies in Epistemology 2 (Oxford:
Oxford University Press), pp. 198–140.
Sosa E. (2000) ‘Skepticism and Contextualism’, Philosophical Issues, 11, 1–18
Sosa E. (2010) ‘Armchair Philosophy, with Moore and Wittgenstein’, paper
presented at COGITO Epistemology Seminar, University of Bologna, December
15, 2010.
Strawson P. (1985) Scepticism and Naturalism. Some Varieties (New York: Columbia
University Press).
Tennant N. (2005) Review of Christopher Peacocke’s The Realm of Reasons, The
Journal of Philosophy, 102/3, 155–62.
Tiercelin C. (2005) Le doute en question: Parades pragmatistes au défi sceptique (Paris:
Éditions de l’Éclat).
Tucker C. (2013) (ed.) Seemings and Justification: New Essays on Dogmatism and
Phenomenal Conservativism (New York: Oxford University Press).
Turri J. (2010) ‘On the Relationship Between Propositional and Doxastic
Justification’, Philosophy and Phenomenological Research, 80/2, 312–326.
Turri J. (2011) ‘Contingent A Priori Knowledge’, Philosophy and Phenomenological
Research, 83/2, 247–69.
Vogel J. (1990) ‘Cartesian Scepticism and Inference to the Best Explanation’,
The Journal of Philosophy, 87, 658–66. Reprinted in Williams M. (ed.) Scepticism
(Dartmouth: Aldershot), pp. 111–120.
Vogel J. (2008) ‘Epistemic Bootstrapping’, Journal of Philosophy, 105/9, 518–539.
Volpe G. (2012) ‘Cornerstones: You’d Better Believe Them’, Synthese, 189/2,
1–23.
Volpe G. (ms) ‘Explaining Justification: A Defence of the Orthodox View of the
Relationship between Propositional and Doxastic Justification’.
210 Bibliography
Wedgwood R. (2007) ‘Christopher Peacocke’s The Realm of Reason’, Philosophy

Wedgwood R. (2011) ‘Primitively Rational Belief-forming Processes’, in Reisner
A. and Steglich-Petersen A. (eds) Reasons for Belief (Cambridge: Cambridge
University Press), pp. 180–200.
Wedgwood R. (2012) ‘Justified Inference’, Synthese, 189/2, 273–295.
Wedgwood R. (2013) ‘A Priori Bootstrapping’, in Casullo A. and Thurow J. (eds)
The A Priori In Philosophy (New York: Oxford University Press), pp. 226–248.
White R. (2006) ‘Problems for Dogmatism’, Philosophical Studies, 131, 525–557.
Williams M. (1991) Unnatural Doubts (Oxford: Blackwell).
Williams M. (2004a) ‘Wittgenstein’s Refutation of Idealism’, in McManus D. (e(d.)
Wittgenstein and Scepticism (London/New York: Routledge), pp. 76–96.
Williams M. (2004b) ‘Wittgenstein, Truth and Certainty’, in Kölbel M. and Weiss
B. (eds) Wittgenstein’s Lasting Significance (London/New York: Routledge),
pp. 249–284.
Williams M. (2012) ‘Wright Against the Sceptics’, in Wright C. and Coliva A. (eds)
Mind, Meaning, and Knowledge: Themes From the Philosophy of Crispin Wright
(New York: Oxford University Press), pp. 352–376.
Williamson T. (2003) ‘Understanding and Inference’, Proceedings of the Aristotelian
Society, Supplementary Volume, 77, 249–293.
Williamson T. (2012) ‘Boghossian and Casalegno on Understanding and
Inference’, Dialectica, 66/2, 237–247.
Wittgenstein L. (1956) Remarks on the Foundations of Mathematics (Revised Edition,
Oxford: Basil Blackwell, 1978).
Wittgenstein L. (1969) On Certainty (Oxford: Basil Blackwell).
Wright C. (1985) ‘Facts and Certainty’, Proceedings of the British Academy, 71,
429–72. Reprinted in Williams M. (ed.) Scepticism (Dartmouth: Aldershot),
pp. 303–346.
Wright C. (1992) Truth and Objectivity (Harvard: Harvard University Press).
Wright C. (2002) ‘(Anti-)sceptics simple and subtle: G. E. Moore and John
McDowell’, Philosophy and Phenomenological Research, 65/2, 330–48.
Wright C. (2004a) ‘Warrant for Nothing (and Foundations for Free)?’, The
Supplementary Volume of the Aristotelian Society, 78, 167–212.
Wright C. (2004b) ‘Intuition, Entitlement and the Epistemology of Logical Laws’,
Dialectica, 58/1, 155–175.
Wright C. (2004c) ‘Wittgensteinian Certainties’, in McManus D. (ed.) Wittgenstein
and Scepticism (London: Routledge), pp. 22–55.
Wright C. (2007a) ‘The perils of dogmatism’, in Nuccetelli S. and Seay G. (eds)
Themes from G.E. Moore: New Essays in Epistemology and Ethics (Oxford: Oxford
University Press), pp. 25–48.
Wright C. (2007b) ‘Rule-following without Reasons: Wittgenstein’s Quietism and
the Constitutive Question’, in Preston J. (ed.) Wittgenstein and Reason, Ratio,
20/4, 481–502.
Wright C. (2012) ‘Comment on Paul Boghossian’s “What Is an Inference?”’,
Philosophical Studies, 1, 1–11.
Wright C. (2014) ‘On Epistemic Entitlement (II): Welfare State Epistemology’,
in Dodd D. and Zardini E. (eds) Scepticism and Perceptual Justification (Oxford:
Oxford University Press), 213–247.
Name Index
Achilles, 168, 169 Evans, G., 70

Agrippa, 4, 8, 14, 197 Evans-Pritchard, E., 204
Alonso-Gomez, M., ix
Alston, W. P., 192 Ferrari, F., ix
Ashton, N., ix Frege, G., 178, 204
Audi, R., 202 Fricker, E., 202
Avnur, Y., 77, 191, 192, 195, 198, 199,
201 Gao, J., ix
Gerken, M., ix, 184, 186
Bagnoli, C., ix Goncharov, I., 12, 146
Beebee, H., 193 Graham, P., 202
Belleri, D., ix Grice, P., 107, 137, 138
Bermúdez, J. L., 188
Boghossian, P., 169, 170, 171, 172, Harman, G., 138, 184, 186, 195, 196,
173, 174, 182, 200, 202, 203 202
Brewer, B., 188 Hawthorne, J., 77, 195
Brown, J., 193 Hume, D., 7, 8, 9, 14, 32, 38, 39, 64,
Burge, T., 22, 23, 24, 32, 47, 48, 49, 50, 66, 71, 74, 119, 120, 121, 122, 123,
80, 182, 184, 185, 188, 189, 200, 202 127, 128, 129, 130, 131, 135, 137,
140, 150, 151, 154, 155, 187, 191,
Cariani, F., 202 197, 198, 199, 202
Carroll, L., 168, 169, 173, 176, 178
Carter, A., ix James, W., 120
Casalegno, P., 170, 202, 203 Jenkins, C., 67, 187
Casullo, A., 189
Coady, C. A. J., 202 Kallestrup, J. ix, 110, 111, 112, 113,
Cohen, S., 87, 196 193, 195, 196
Coliva, A., 124, 133, 139, 181, 183, Kant, I. 44, 129
184, 187, 188, 189, 190, 192, 193, Kelly, T. 188
198, 199, 200 Korman, D. 201
Kripke, S. 77
Davies, M., 87, 88, 90, 91, 92, 93, 113, Kusch, M. ix
193, 197
DeRose, K., 106, 107, 121, 136, 137, Leonardi, P. ix
138, 139, 151, 196 Luper, S. 195
Dodd, D., x
Dretske, F., 87, 101, 103, 181, 189, McDowell, J., 43, 44, 45, 46, 50, 70,
193, 195 122, 188, 189, 202
Dutant, J., x McGee, V. 169, 170, 172,
202, 206
Engel, P., x McGinn, M. 123, 187, 190
Enoch, D., 172, 173, 174, 175, 176, McGlynn, A. ix, 185
201, 202, 203 McHugh, C. 192
211
212 Name Index
Mangiameli, G. 204 Schiffer, S., 186, 193

Marconi, D. 203 Sellars, W., 43, 188
Meylan, A. x Sherman, B., 185, 202
Millar, A. ix Siegel, S., 51, 189
Millikan, R. 52, 189 Silins, N., 58, 60, 61, 72, 73, 76, 184,
Moore, G. E. 10, 13, 14, 25, 28, 29, 30, 189, 191, 193
31, 32, 57, 58, 60, 63, 64, 65, 71, 72, Sosa, E., 9, 57, 72, 82, 83, 84, 85, 106,
73, 75, 76, 78, 80, 81, 84, 86, 87, 107, 137, 196
88, 90, 91, 94, 95, 96, 97, 98, 102, Strawson, P., 7, 9, 39, 40, 65, 120, 121,
103, 109, 114, 115, 118, 183, 184, 122, 123
185, 187, 189, 190, 191, 192, 193,
194, 195 Tennant, N., 192
Moruzzi, S. ix Tiercelin, C., 198, 201
Moyal-Sharrock, D. ix, 123, 181, 187, Turri, J., 77, 78, 183
190, 199
Vogel, J., 87, 185, 190, 192, 196
Napoleon, 2, 5 Volpe, G., ix
Nozick, R., 101, 195
Wallbridge, K., ix
Olsson, E. J., 184 Wedgwood, R., 57, 58, 61, 69, 72, 73,
Orlandelli, E., ix 74, 75, 76, 77, 78, 83, 84, 184, 189,
190, 191, 192, 198
Palmira, M., ix White, R., 58, 72, 189, 190, 191, 193
Pasquali, A., ix Williams, M., 67, 123, 169, 187, 198
Peacocke, C., 47, 48, 49, 50, 57, 69, Williamson, T., 169, 170, 172, 173,
72, 74, 78, 79, 80, 81, 82, 84, 181, 174, 202, 203
185, 188, 189, 192, 193 Wittgenstein, L., 1, 7, 8, 9, 10, 14, 20,
Pedersen, N., 184 39, 40, 41, 42, 64, 65, 69, 120, 121,
Peirce, C. S., 125, 201 122, 123, 124, 125, 126, 128, 148,
Picardi, E., ix 154, 155, 177, 181, 182, 187, 194,
Pritchard, D., ix, 67, 105, 106, 107, 198, 203, 204
137, 138, 187, 196, 204 Wittwer, S., ix
Pryor, J., 8, 14, 15, 19, 21, 23, 24, 26, Wright, C., 8, 13, 14, 15, 19, 29, 30,
27, 28, 30, 58, 59, 60, 62, 63, 72, 73, 31, 32, 33, 48, 57, 58, 60, 64, 65, 66,
75, 76, 78, 79, 80, 81, 87, 88, 90, 91, 67, 68, 69, 70, 71, 72, 73, 76, 82, 84,
92, 108, 109, 114, 115, 116, 117, 85, 87, 88, 89, 90, 91, 92, 93, 97, 98,
118, 133, 181, 182, 183, 184, 190, 103, 105, 117, 118, 121, 123, 124,
192, 193, 197 132, 133, 134, 135, 139, 164, 174,
Putnam, H., 32, 69, 70, 79, 90, 201 175, 176, 181, 182, 184, 185, 186,
187, 188, 189, 190, 191, 193, 195,
Reichmann, R., ix 196, 197, 198, 199, 200, 201, 202,
Reid, T., 187, 202 203, 204
Schechter, J., 172, 173, 174, 175, 176, Zanetti, L., ix, 184
202, 203 Zardini, E., x
Subject Index
a priori, 6, 10, 23, 32, 49, 57, 64, dialectically ineffective, 58, 62, 84,
65, 68, 70, 71, 72, 73, 74, 75, 87, 108, 192
77, 78, 79, 80, 81, 82, 83, 84, dreaming, 66
85, 94, 159, 178, 192, 195, EARTH, 86, 98, 99, 101, 158, 108
196, 203 epistemically circular, 31, 62, 63, 89
justification, 6, 32, 72, 73, 74, 77, fingers, 103, 104, 105
78, 83, 84, 159, 178, 192 gauge, 111, 112, 113, 114
relatively, 49, 78 inductive, 155
warrant, 10, 57, 68, 70, 71, 72, 75, McKinsey, 90
94, 97, 98, 100, 102, 122, 127, MOORE, 13, 25, 28, 29, 30, 31, 32,
131, 134, 135, 191 57, 58, 59, 60, 63, 64, 65, 71,
abominable conjunction, 106, 107, 72, 75, 76, 84, 86, 87, 88, 94,
121, 136, 137, 138, 139, 151 95, 97, 100, 102, 109, 118, 185,
acceptance, 20, 65, 71, 113, 156, 160, 189, 191, 192
164, 165, 177, 186 other minds, 86, 97, 98, 100
agnosticism, 63, 84 P, Q, R…N, 113, 114
Agrippa’s trilemma, 4, 8, 14, 197 pages, 104
alchemy, 195, 203 past, 86, 99, 158
anti-underminer model, 115 -1/2/3/4, 158
argument, 2, 13, 14, 22, 25, 28, 29, philosophical, 70, 134, 135, 191
31, 32, 37, 42, 46, 47, 57, 58, question-begging, 89
59, 60, 62, 63, 64, 65, 66, 67, rat, 116, 117
68, 69, 70, 71, 74, 75, 76, 77, red table, 87, 109, 110,111, 112, 197
78, 80, 81, 82, 84, 86, 87, 88, red table*, 113, 114
89, 90, 92, 93, 94, 95, 96, 97, red wall, 62, 63, 108, 109
98, 99, 100, 102, 103, 104, 105, sun, 154, 157
108, 109, 110, 111, 112, 113, testimony, 87, 98, 100, 101
114, 115, 116, 118, 119, 122, uniformity, 99, 100, 101
126, 134, 135, 136, 147, 155, zebra, 29, 30, 31, 86, 87, 89, 96,
158, 163, 164, 167, 168, 169, 100, 102, 103, 104, 114, 181,
181, 183, 184, 185, 186, 188, 194
189, 191, 192, 193, 194, 195, assumption(s) (assuming), 1, 2, 3, 4,
196, 197, 204, 5, 6, 7, 8, 9, 10, 11, 12, 13, 15,
a priori, 32, 68, 80, 164 20, 26, 29, 30, 31, 33, 34, 35,
based on closure, 119 36, 37, 38, 39, 40, 41, 42, 43,
Bayesian, 60, 75, 76 45, 49, 53, 54, 55, 61, 62, 64,
BIRTH, 162, 163, 164 65, 67, 68, 69, 70, 71, 72, 74,
bootstrapping, 14, 37, 42, 72, 73, 75, 82, 84, 90, 94, 95, 96, 97,
86, 88, 109, 110, 111, 112, 113, 98, 99, 100, 102, 103, 105, 108,
114, 126, 190, 196 112, 115, 116, 117, 119, 120,
Cartesian, 136 121, 122, 123, 124, 125, 127,
cogent, 58, 63, 99, 112 128, 129, 130, 131, 132, 134,
213
214 Subject Index
assumption(s) (assuming) – continued 177, 179, 180, 181, 182, 183,

135, 136, 137, 139, 140, 141, 184, 185, 186, 188, 189, 190,
142, 145, 146, 147, 148, 149, 193, 194, 196, 197, 199, 201,
150, 151, 152, 154, 155, 156, 203, 204
159, 160, 161, 163, 164, 165, actual, 13, 16, 22, 90, 183
166, 167, 172, 174, 176, 177, empirical, 4, 7, 8, 9, 11, 13, 15,
181, 185, 186, 187, 189, 190, 18, 19, 20, 33, 34, 38, 39, 40,
191, 193, 194, 195, 197, 198, 41, 43, 45, 46, 47, 53, 56, 61,
199, 201, 202 64, 68, 72, 73, 74, 78, 79, 80,
a-rational, 7, 70, 122, 128, 82, 89, 90, 108, 114, 115, 117,
135 121, 123, 124, 129, 131, 132,
arbitrary, 4, 38, 40, 119, 122, 140, 139, 150, 163, 181, 190, 197,
187 199
background, 9, 49, 53, 54, 108, 117, justified, 11, 14, 65, 76, 132, 137,
147, 163, 190 146, 160, 165, 173
collateral, 20, 29, 30, 33, 34, 40, 43, perceptual, 14, 35, 48, 196
72, 90, 117 warranted, 101, 104, 106, 120, 138,
doxastic, 36, 37, 38, 55 142, 144, 194
mandated, 11, 13, 134, 135, 136, BIV(s), 3, 4, 34, 59, 60, 61, 83,
138, 139, 140, 149, 150, 151, 136, 137, 151, 185, 191,
177, 195 192
propositional, 43 blind-sight, 24, 49, 194
rational, 10, 40, 68, 69, 70, 71, 81, BOCHE, 171
96, 119, 120, 125, 128, 134,
135, 146, 191 cognitive locality, 4, 8, 18, 19, 25, 26,
system of, 1, 3, 4, 8, 10, 38, 39, 42, 29, 31, 32, 33, 34, 38, 43, 54,
119, 120, 121, 140, 202 56, 61, 85
unjustified, 8, 10, 155 coherentism, 47, 50, 181
unwarrantable, 7, 8, 129, 134, 155, condition, 4, 7, 8, 13, 18, 19, 20, 21,
160, 164 23, 26, 32, 39, 43, 45, 47, 48,
axiom, 1, 12, 93, 94, 133 49, 50, 56, 59, 61, 62, 70, 79,
Azande, 200, 204 81, 83, 106, 107, 109, 117, 120,
129, 130, 136, 137, 153, 171,
belief, 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 180, 183, 187, 189, 194, 198,
13, 14, 15, 16, 18, 19, 20, 21, 199, 200
22, 23, 24, 25, 26, 27, 29, 32, epistemic, 182
33, 34, 35, 36, 38, 39, 40, 41, human, 7, 8, 13, 153, 180, 187
43, 44, 45, 46, 47, 48, 49, 50, Humean, 7
52, 53, 54, 55, 56, 61, 64, 65, conservative, 8, 13, 19, 26, 28,
66, 68, 71, 72, 73, 74, 75, 76, 29, 33, 34, 35, 38, 39, 42,
78, 79, 80, 82, 83, 84, 87, 89,90, 43, 55, 56, 57, 60, 61, 63,
91, 92, 95, 97, 101, 104, 105, 64, 71, 72, 73, 84, 85, 97, 105,
106, 108, 109, 110, 112, 113, 107, 108, 114, 117, 124, 133,
114, 115, 117, 120, 121, 122, 140, 163, 187, 191, 192, 195,
124, 129, 131, 132, 137, 138, 196
139, 140, 141, 142, 144, 145, constitutive, 6, 10, 11, 14, 34, 38,
146, 147, 148, 150, 153, 157, 39, 47, 79, 82, 120, 128, 129,
159, 160, 161, 162, 163, 165, 130, 131, 133, 134, 139, 141,
166, 168, 169, 173, 174, 175, 145,146, 147, 150, 152, 154,
Subject Index 215
constitutive – continued dreaming, 3, 19, 25, 34, 54, 59, 60, 61,
155, 156, 157, 160, 161, 164, 64, 65, 66, 68, 69, 70, 72, 73, 74,
166, 170, 172, 173, 177, 178, 75, 76, 77, 81, 82, 95, 107, 113,
179, 180, 198, 199, 204 115, 136, 137, 138, 151, 185
assumption, 10, 11, 129, 131, 134,
146, 152, 161, 164, 177, 198 Earth, 1, 2, 3, 5, 6, 9, 25, 86, 98, 99,
meaning-, 6, 169, 170, 171, 172, 101, 124, 158
173, 174 entitlement, 6, 7, 10, 14, 23, 24, 32,
practice, 141 33, 48, 49, 50, 52, 57, 64, 65,
content, 2, 3, 9, 10, 12, 16, 19, 20, 21, 66, 67, 68, 70, 72, 73, 79, 80,
23, 24, 28, 29, 31, 32, 34, 36, 82, 84, 85, 97, 98, 99, 100, 102,
38, 41, 42, 43, 44, 45, 46, 47, 105, 121, 127, 131, 132, 134,
49, 50, 51, 53, 54, 55, 56, 61, 135, 157, 159, 161, 166, 174,
69, 70, 72,74, 78, 79, 80, 89, 90, 175, 176, 179, 184, 185, 187,
91, 92, 114, 124, 125, 128, 142, 189, 190, 191, 195, 200, 203,
143, 145, 146, 148, 149, 151, 204
157, 159, 161, 168, 173, 181, of cognitive project, 175, 176, 204
182, 188, 189, 190, 192, 196, perceptual, 23, 48, 49, 50, 82
200, 201, 202 rational, 121
descriptive, 42, 125,181 of substance, 66, 190, 200
empirical, 10, 44, 45, 53 Euthyphro contrast, 149, 157, 201
mental, 16 evidence, 1, 2, 4, 5, 7, 11, 37, 41, 54,
perceptual, 9,12, 19, 23, 43, 45, 47, 59, 61, 64, 65, 67, 71, 86, 87,
49, 53, 56, 188, 189 98, 99, 100, 104, 124, 125, 129,
conceptual, 46 132, 135, 136, 141, 147, 148,
nonconceptual, 50, 174 149, 150, 152, 156, 157, 158,
phenomenal, 23, 43, 72, 91, 159, 161, 170, 177, 182, 186,
92, 182 187, 190, 193, 198, 201
propositional, 168, 173 empirical, 64, 65, 125, 148
representational, 19, 20, 21, 43, 46, historical, 124, 158, 161
47, 51, 74, 79, 80, 90, 91, 92, perceptual, 7, 11, 104, 124, 129,
142, 183, 192 150, 177, 182, 201
semantic, 70, 149, 157 testimonial, 86, 87, 99, 100,
141
defeater, 3, 8, 15, 21, 22, 23, 26, 27, experience, 3, 4, 6, 7, 8, 9, 10,
29, 34, 38, 48, 52, 54, 55, 114, 11, 12, 13, 15, 18, 19, 20, 21,
141, 160, 162, 163, 181, 182, 22, 23, 24, 25, 26, 27, 28, 29,
184, 185, 186, 189 30, 31, 33, 34, 35, 36, 38, 39,
Dependence thesis, 46 40, 41, 42, 43, 44, 45, 46, 47,
disjunctivism, 45, 188 48, 49, 50, 51, 52, 53, 56, 58,
dogmatist(s), 183, 208 59, 60, 61, 62, 64, 66, 67, 68,
doubt, 2, 19, 20, 41, 42, 54, 58, 69, 71, 72, 73, 74, 75, 78, 79,
59, 62, 63, 66, 70, 75, 76, 79, 80, 81, 82, 83, 84, 85, 89, 90,
80, 81, 83, 87, 91, 92, 107, 108, 91, 92, 95, 96, 97, 98, 103, 107,
109, 110, 112, 114, 117, 119, 108, 109, 112, 114, 115, 116,
122, 124, 125, 126, 142, 150, 117, 119, 122, 123, 126, 127,
155, 175, 179, 181, 182, 185, 130, 134, 141, 142, 143, 144,
189, 190, 191, 198, 202, 205, 145, 149, 151, 157, 160, 161,
210 173, 174, 181, 182, 183, 184,
216 Subject Index
experience – continued hand, 3, 4, 8, 9, 11, 13, 15, 16, 18, 19,

185, 187, 188, 189, 190, 191, 25, 26, 29, 30, 31, 32, 33, 39,
192, 193, 194, 199, 200 45, 54, 57, 58, 59, 60, 61, 70,
conscious, 23, 24, 48, 53 72, 73, 74, 75, 76, 79, 80, 81,
hand-like-, 3, 25, 30, 31, 59, 60, 61, 82, 83, 84, 85, 86, 88, 90, 92,
81, 84, 95, 103, 127, 182, 189 93, 95, 96, 101, 102, 103, 106,
perceptual, 4, 8, 9, 12, 13, 18, 19, 107, 114, 115, 119, 127, 129,
20, 22, 25, 26, 35, 46, 50, 52, 136, 137, 138, 142, 143, 144,
53, 56, 73, 78, 79, 89, 114, 141, 148, 151, 176, 181, 182, 183,
142, 143, 144, 151, 173, 184, 184, 186, 187, 188, 189, 191,
189, 200 193, 194, 197, 198, 199, 201
content of –, 19, 20, 23, 44, 46, 47, hinge, 2, 9, 10, 11, 14, 20, 38, 41, 42,
50, 56, 142, 145, 151 65, 74, 105, 120, 124, 125, 126,
conceptual –, 44, 46 148, 152, 153, 154, 155, 162,
non-conceptual, 47 163, 164, 180, 181, 182, 187,
phenomenal, 142, 143 194, 198, 199, 202
visual, 3, 51, 52, 80, 85, 116, epistemology, 9, 11, 14, 20, 42
142, 192
external world, 4, 6, 7, 8, 9, 11, 12, 13, idealism, 67, 142, 144, 199
15, 17, 18, 28, 29, 30, 31, 32, IF THEN,168, 169, 170, 172, 174, 178,
34, 35, 36, 39, 40, 41, 42, 43, 202, 203
48, 54, 55, 57, 58, 59, 60, 61, immediate conceptualization thesis,
63, 64, 65, 66, 67, 68, 69, 70, 46
71, 72, 73, 75, 78, 80, 82, 83, inference to the best explanation, 73,
84, 85, 86, 88, 90, 91, 92, 93, 141, 173, 192
94, 95, 96, 97, 102, 102, 106, infinite regress, 4, 8, 66, 175
107, 08, 114, 115, 117, 119, internalism, 16, 49, 69, 76, 192
121, 122, 123, 124, 125, 126, intuition, 6, 8, 44, 47, 58, 59, 82, 85,
127, 128, 129, 131, 132, 133, 93, 103, 144, 167, 202
134, 135, 136, 137, 138, 145, rational, 82, 144
146, 147, 148, 150, 151, 152,
154, 156, 157, 159, 160, 161, justification, 1, 2, 3, 4, 5, 6, 7, 8, 9,
165, 176, 177, 181, 182, 186, 10, 11, 13, 14, 15, 16, 17, 18,
187, 190, 191, 193, 194, 195, 20, 21, 22, 24, 25, 26, 27, 28,
197, 198, 199 29, 34, 35, 36, 39, 42, 43, 45,
proof of an –, 29, 57, 86, 88 46, 47, 48, 49, 53, 54, 55, 56,
skepticism about the –, 17, 28, 58, 84 58, 61, 64, 70, 72, 73, 74, 75,
externalism, 16, 69, 70, 90, 143, 147 76, 77, 78, 79, 80, 82, 83, 84,
epistemic, 147 85, 87, 90, 91, 92, 107, 109,
semantic, 69, 70, 90 112, 118, 119, 120, 121, 122,
128, 132, 137, 138, 140, 146,
FLURG, 171, 172 147, 148, 149, 150, 153, 155,
form(s) of life, 7, 121, 123, 187, 198 156, 157, 159, 160,161, 162,
foundationalism, 181 163, 164, 165, 166, 167, 168,
169, 172, 173, 175, 177, 178,
Given (the), 43, 44, 208 180, 181, 182, 183, 185, 186,
myth of, 43, 44, 208 187, 189, 190, 191, 192, 193,
195, 196, 197, 198, 199, 200,
hallucinating, 3, 19, 25, 54, 59, 60, 61, 201, 202, 203, 204, see also
84, 116, 191 warrant
Subject Index 217
justification – continued intuitive, 83

a priori, 6, 32, 72, 73, 74, 77, 83, 84, self-, 90
159, 178, 192
of basic laws of logic, 14, 172, 174 leaching problem, 39
circular, 6 liberal(s), 34, 39, 42, 43, 73, 88, 107,
doxastic, 183, 186 114, 133, 163, 206
empirical, 42, 64, 75, 121, 140, 187, logic(al), 2, 11, 14, 64, 124, 139, 141,
189, 190, 192 166, 167, 172, 173, 174, 176,
epistemic, 14, 148 178, 179, 182, 197, 199, 200,
evidential, 109, 119, 198 204
externalist, 168 basic – laws, 14, 141, 166, 167, 172,
immediate, 7, 8, 74, 109, 181, 183, 173, 174, 200
184
indefeasible, 4 mandate, 11, 13, 129, 131, 132, 133,
independent, 9, 73, 79, 87, 112, 134, 135, 137, 138, 139, 140,
118, 163 148, 149, 150, 151, 156, 160,
inductive, 14, 156, 157, 180 165, 177, 195
inferential, 6 rational, 11, 13, 129, 133, 134, 137,
internalist, 79 138, 139, 148, 149, 150,160,
memory-based-, 159, 160, 161, 180 165, 177, 195
non circular, 175 maxim, 107, 108, 137, 138
non-evidential, 85, 181 Gricean, 107, 108, 137, 138
perceptual, 3, 8, 9, 10, 11, 13, 15, of quality, 107, 108, 137, 138
17, 18, 21, 25, 26, 28, 29, 34, memory, 14, 99, 141, 153, 158, 159,
35, 43, 53, 54, 55, 80, 84, 118, 160, 161, 173, 180
132, 146, 177, 180, 182, 183, moderate(s), 8, 13, 14, 15, 19, 20,
185, 190, 195, 196, 199 26, 28, 33, 34, 35, 36, 37, 38,
propositional, 16, 21, 22, 76, 156, 39, 40, 41, 41, 43, 54, 55, 56,
160, 165, 183, 186 57, 58, 61, 71, 72, 75, 76, 84,
rationally available, 16, 22, 24, 25, 85, 86, 88, 94, 95, 102, 105,
36, 75, 76, 82, 90 , 92, 192 106, 107, 108, 114, 115, 116,
rationally unavailable, 22, 91 117, 118, 119, 120, 121, 124,
regress of –, 175 126, 127, 131, 133, 134, 136,
testimonial –, 153, 163, 164, 165, 137, 139, 140, 146, 163, 186,
180 187, 191, 192, 193, 195, 196,
197
Kassena, 204 account of empirical warrants, 33,
knowledge, 16, 22, 28, 30, 50, 51, 63, 43, 88, 105, 117, 118, 139
65, 77, 78, 82, 83, 84, 85, 86, account of perceptual warrants, 8,
88, 90, 94, 101, 102, 103, 104, 9, 13, 34, 35, 54, 55, 71, 108,
105, 106, 107, 108,109, 110, 117, 146
111, 118, 120, 128, 137, 138, account of testimonial warrants,
139, 140, 143, 147, 151, 153, 163
155, 159, 168, 178, 183, 184, modus ponens, 14, 141, 153, 166,
187, 193, 195, 196, 200, 202, 167, 168, 169, 170, 172, 173,
203 174, 175, 176, 177, 178, 179,
a priori, 77, 78, 85, 94 180, 182, 203, 204
basic, 109, 110, 111, 184, 196 moon, 41, 125, 182
easy, 28, 63, 86, 88, 108, 118, Mooreanism, 25, 57, 58, 60, 63, 73,
193 75, 84, 190
218 Subject Index
naturalism, 7, 39, 40, 120, 120, 121, practice, 7, 9, 11, 27, 35, 38, 40, 73,
122, 123, 155 74, 119, 121, 122, 123, 126,
norm(s), 23, 24, 41, 42, 48, 49, 65, 128, 129, 130, 132, 133, 134,
120, 124, 125, 126, 194, 204 141, 142, 144, 145, 146, 147,
natural, 23, 24, 48, 49 150, 151, 155, 156, 159, 160,
163, 165, 177, 182, 185, 199,
object(s), 3, 4, 7, 8, 12, 13, 19, 20, 21, 200, 204
23, 25, 26, 31, 32, 33, 34, 35, (basic) epistemic –, 7, 9, 11, 122,
38, 40, 44, 46, 47, 51, 54, 55, 123, 128, 129, 132, 134, 141,
61, 66, 67, 69, 70, 76, 78, 83, 145, 147, 151, 177, 182, 185,
85, 89, 90, 97, 105, 108, 109, 199, 200
124, 129, 132, 136, 140, 141, non basic epistemic –, 128, 132,
142, 143, 144, 145, 146, 148, 141, 200
149, 151,155, 161, 177, 180, pragmatism, 40, 120, 121, 123, 132,
182, 183, 184, 185, 188, 189, 187
190, 193, 199, 200 principle, 13, 14, 28, 44, 64, 75, 87,
material, 19, 20, 21, 25, 26, 29, 34, 88, 101, 102, 104, 106, 114,
35, 40, 41, 54, 55, 61, 67, 76, 118, 119, 121, 136, 139, 140,
108, 109, 122, 124, 132, 140, 151, 153, 154, 155, 156, 157,
141, 142, 143, 148, 177, 182, 166, 173, 180, 192, 193, 198,
185, 193 199, 200, 202, 203
mid-size, 3, 4, 7, 13, 153, 180 a priori, 192
mind-independent, 4, 12, 13, 29, of closure, 13, 28, 75, 87, 88, 101,
31, 35,55 , 129, 143, 145, 149, 102, 104, 114, 118, 119, 121
183, 190, 200 136, 151, 202
physical, 3, 4, 8, 12, 13, 25, 26, 31, of efficient cause, 200
32, 33, 35, 41, 54, 66, 67, 69, for epistemic operators under
70, 89, 95, 122, 129, 136, 142, known entailment, 13, 28, 64,
144, 146, 151, 177, 182, 183, 87, 101, 102, 118, 151, 202
188, 190 of final cause, 200
Oblomovian challenge, 12, 121, 145, logical –, 139, 166, 167, 173,
146, 152, 179 for rational mandate,
other minds, 86, 97, 98, 100, of the uniformity of nature, 14, 154,
153, 161, 164, 165, 166, 180, 155, 156, 157
195 proof, 1, 5, 12, 13, 22, 57, 60, 63, 72,
75, 76, 78, 80, 81, 84, 88, 90,
past, 2, 14, 23, 32, 74, 86, 97, 99, 91, 93, 95, 96, 103, 114, 115,
100, 101, 153, 154, 155, 156, 135, 148, 168, 177, 183, 192,
157, 158, 159, 160, 161, 165, 193, 194, 195, 204
180 Moore’s – of an external world, 57,
perception, 12, 14, 18, 20, 21, 23, 24, 60, 63, 75, 76, 78, 80, 81, 84,
26, 32, 43, 44, 45, 46, 47, 48, 88, 90, 91, 93, 96, 103, 114,
49, 50, 51, 52, 53, 69, 76, 78, 115, 192, 193, 195
87, 90, 109, 113, 114, 132, 139, proposition(s), 1, 2, 3, 4, 5, 6, 7, 8, 9,
141, 142, 143, 151, 152, 173, 10, 11, 14, 15, 16, 20, 21, 22,
181, 184, 188, 189, 190, 196, 29, 30, 31, 32, 34, 35, 36, 38,
200 39, 41, 42, 43, 45, 47, 53, 54,
content of –, 20, 43, 44, 45, 51 55, 58, 59, 62, 63, 64, 65, 68,
phenomenalism, 142, 144 69, 70, 71, 74, 76, 77, 82, 83,
Subject Index 219
proposition(s) – continued extended, 10, 11, 13, 14, 15, 119,

84, 85, 93, 94, 97, 98, 105, 117, 120, 121, 127, 129, 132, 133,
119, 120, 121, 124, 125, 126, 135, 136, 137, 138, 139, 140,
127, 128, 129, 130, 131, 132, 145, 146,148, 151, 152, 153,
133, 134, 136, 137, 139, 146, 154, 156, 157, 158, 159, 160,
147, 148, 149, 150, 152, 153, 161, 164, 165, 166, 167, 176,
155, 156, 157, 158, 159, 160, 177, 178, 179, 180, 191, 197,
161, 162, 163,164, 165, 166, 202
168, 173, 175, 177, 178, 180, inductive, 155, 159
181, 182, 185, 186, 187, 190, logical, 141, 178, 179, 204
191, 192, 193, 194, 195, 196, practical, 133, 141
197, 201, 202 predictive, 141
contingent, 74, 159, 164, 196, social, 164
201 reason(s), 2, 3, 8, 13, 16, 19, 20, 21,
cornerstone, 65 25, 26, 28, 30, 31, 32, 33, 34,
empirical, 2, 3, 4, 5, 7, 8, 9, 10, 11, 38, 40, 41, 42, 45, 48, 49, 54,
14, 20, 41, 42, 43, 59, 71, 77, 55, 57, 58, 59, 62, 63, 65, 66,
83, 85, 119, 120, 121, 125, 126, 68, 69, 70, 71, 73, 74, 76, 77,
127, 18, 129, 130, 131, 134, 78, 79, 80, 81, 82, 89, 91, 92,
136, 147, 148, 152, 153, 190, 99, 103, 104, 106, 107, 108,
197 110, 111, 119, 120, 122, 124,
hinge, 121, 181, 182 125, 126, 128, 129, 138, 145,
mathematical, 1, 22, 93 146, 147, 166, 168, 170, 172,
necessary, 173, 175, 176, 182, 183, 184,
proto-, 47 186, 188, 190, 191, 198, 201,
203
rationalism, 73, 78, 133, 134 a priori, 71, 80, 203
rationality, 10, 11, 12, 13, 14, 15, 37, abstract space of –, 16, 30, 31, 32,
69, 70, 71, 74, 87, 95, 119, 120, 34, 38, 55, 65, 82, 104, 166
121, 125, 126, 127, 128, 129, dialectical, 80, 91
130, 131, 132, 133, 134, 135, epistemic, 110, 176
136, 137, 138, 139, 140, 141, perceptual, 146
145, 146, 147, 148, 150, 151, practical, 146, 147
152, 153, 154, 155, 156, 157, relativism, 11, 58, 120, 121, 128, 139,
158, 159, 160, 161, 164, 165, 140, 144, 145, 151, 200
166, 167, 176, 177, 178, 179, epistemic, 11, 58, 120, 121, 128,
180, 191, 197, 198, 199, 200, 202 139, 140, 145, 151, 200
epistemic –, 10, 11, 12, 13, 14, 28, reliabilism, 24, 49, 110, 196
69, 70, 71, 120, 128, 129, 130, representation, 12, 23, 24, 43, 45, 48,
131, 132, 133, 134, 135, 138, 49, 51, 52, 53, 67, 79, 80, 142,
139, 140, 141, 145, 146, 147, 143, 188
148, 150, 151, 152, 153, 155, perceptual, 12, 23, 24, 45, 52, 79,
157, 160, 161, 164, 165, 176, 143, 188, 189, 192
177, 178, 179, 180, 199 rule(s), 10, 12, 14, 25, 41, 42, 44, 124,
–CW/ER/SK , 132, 133 125, 126, 130, 131, 13, 141,
abductive, 141 146, 148, 153, 154, 155, 167,
deductive, 41, 177, 178, 179 168, 170, 171, 172, 173, 176,
diachronic, 160, 161 177, 178, 179, 180, 196, 199,
explanatory, 141 203, 204
220 Subject Index
rule(s) – continued epistemic, 137

constitutive, 130, 131, 199 of meaning, 6, 126
of evidential significance, 41, 124 of mental content, 16
of inference, 141, 154, 167, 168, of perception, 190
171, 172, 173, 176, 177, 178, of perceptual warrant, 90
179, 180 TONK, 170, 171
-like role, 42, 124 truth, 5, 6, 10, 11, 13, 15, 16, 24, 32,
of meaning, 172 33, 36, 37, 40, 43, 55, 60, 65,
semantic, 124, 125, 126 68, 70, 77, 78, 80, 83, 85, 94,
101, 106, 109, 111, 121, 123,
semantics, 6, 170, 172, 174 127, 129, 134, 135, 137, 144,
conceptual role –, 170, 172, 174 147, 148, 149, 150, 152, 154,
inferential role-, 6, 172 156, 157, 165, 166, 171, 176,
sensation(s), 12, 142, 183 177, 179, 185, 191, 192, 201,
sense data, 25, 43, 183, 187 202, 203
sentence(s), 41, 125, 166, 170, 175, 181 a priori, 80
atomic, 170, 175 anti-realist (about-), 37
content vs. role of, 181 contingent, 77, 94, 154
skepticism, 7, 17, 28, 37, 38, 40, 56, deeply contingent a priori,
58, 64, 70, 71, 81, 84, 101, 102, minimal, 149, 157
119, 121, 122, 123, 127, 128, -preserving, 166, 171, 177, 179, 203
130, 135, 136, 137, 147, 150, realist (about-), 37, 149, 150, 152,
151, 152, 185, 187, 190, 191, 201
197, 198, 199, 201 -tracking, 24, 101, 147
argument for – based on closure,
Cartesian, 64, 101, 121, 136, 151, underdetermination problem, 52, 188
185, 190, 199
direct, 127 warrant, 1, 6, 8, 9, 10, 11, 13, 14, 15,
external world-, 17, 28, 58 16, 18, 19, 20, 21, 22, 23, 24,
Humean, 7, 64, 71, 119, 121, 127, 25, 26, 27, 28, 29, 30, 31, 32,
128, 135, 137, 140, 151, 191, 33, 34, 35, 36, 37, 38, 39, 40,
197, 198, 199 41, 42, 43, 47, 48, 49, 50, 52,
indirect, 127 53, 55, 56, 57, 58, 59, 60, 61,
Pyrrhonian, 201 62, 63, 64, 65, 66, 68, 69, 70,71,
response to –, 127, 81 72, 73, 75, 76, 77, 80, 81, 82,
sceptical, 127 84, 85, 88, 89, 90, 91, 92, 93,
94, 95, 96, 97, 98, 99, 100, 101,
testimony, 2, 14, 87, 98, 100, 101, 102, 103, 104, 105, 106, 107,
109, 141, 153, 158, 159, 160, 108, 109, 110, 111, 112, 113,
161, 162, 163, 164, 165, 195, 114, 115, 116, 117, 118, 119,
197, 202 120, 121, 122, 123, 124, 127,
antireductionism about –, 162, 163, 128, 129, 130, 132, 133, 134,
202 135, 136, 137, 138, 139, 140,
reductionism about –, 162, 163, 202 143, 146, 148, 150, 151, 156,
theorem, 1, 12, 22, 93, 94, 183, 194 157, 160, 163, 164, 177, 182,
theory, 6,8, 16 , 19, 20, 27, 35, 36, 39, 183, 184, 185, 186, 187, 189,
45, 52, 59, 62, 89, 90, 93, 94, 190, 191, 192, 193, 194, 195,
106, 109, 126, 137, 138, 141, 196, 197, 198, 199, 202, see also
190, 191, 193, 196, 201 justification
Subject Index 221
warrant – continued perceptual, 8, 9, 11, 13, 19, 20, 21,

a priori, 10, 57, 68, 70, 71, 72, 75, 23, 24, 26, 27, 28, 30, 32, 33,
94, 97, 98, 100, 102, 122, 127, 34, 35, 38, 39, 40, 41, 43, 53,
131, 134, 135, 135, 191 55, 56, 57, 59, 61, 62, 63, 71,
deductive, 177 72, 76, 82, 85, 88, 89, 90, 91,
defeasible, 18, 25, 26, 58, 61, 95, 96, 93, 94, 97, 102, 108, 114, 115,
115, 196 116, 117, 119, 124, 127, 128,
doxastic, 21, 22, 90, 183 129, 130, 132, 134, 150, 182,
empirical, 19, 20, 21, 25, 28, 29, 33, 184, 187, 191, 194, 197,198,
41, 42, 43, 52, 58, 61, 63, 64, 199, 202
75, 81, 88, 92, 105, 116, 117, pragmatic, 40, 69, 120, 123
118, 121, 127, 136, 137, 139, propositional, 21, 22, 24, 25, 30,
146 31, 32, 34, 35, 90, 91, 92, 101,
epistemic, 6, 8, 10, 11, 39, 40, 68, 183, 193
70, 85, 103, 123, 127, 134, 135, testimonial, 14, 97, 163, 194
151, 194 transmission failure, 93, 94, 96, 97,
evidential, 66, 130, 131, 148, 187, 100, 102, 103, 104, 105, 108,
195, 197 109, 112, 113, 114, 115, 117,
indeafeasible, 30, 31, 52, 77, 194, 197
193 1 (TF1) 93, 94, 96, 97, 100, 102,
independent, 30, 57, 81, 85, 89, 94, 103, 104, 105, 108, 112, 113,
96, 100, 102, 104, 109, 113, 114, 115, 117, 194, 197
194, 197 2 (TF2) 96, 100, 102, 103, 105,
inductive, 156, 157 112, 113, 108, 114, 115, 117,
internalist, 32 194
memory-based, 160 warrant-making factor model, 115
non-evidential, 6, 32, 107, 132, 185, Weltbild, 37, 123, 149, 152, 156, 157,
195 161, 165, 198

A. Coliva - Extended Rationality_A Hinge Epistemology (2015)

Uploaded by

Copyright:

Available Formats

You might also like

A. Coliva - Extended Rationality_A Hinge Epistemology (2015)

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

A. Coliva - Extended Rationality_A Hinge Epistemology (2015)

Uploaded by

Copyright:

Available Formats

Extended Rationality

Palgrave Innovations in Philosophy

Series Editors: Vincent F. Hendricks, University of Copenhagen and

Palgrave Innovations in Philosophy

1. Justification (Theory of knowledge) 2. Knowledge, Theory of. I. Title.

8 A possible counterexample: Pryor’s (?) failure to

4 The Extended Rationality View 119

5 The Extended Rationality View Extended 153

Name Index 211

Subject Index 213

Chapter 1, §5 contains a translation of my paper “Percepire le ragioni?”,

Chapter 3, §3–5 and 8 contain revised and enlarged material stemming

Chapter 4 is an extensively revised and enlarged version of my papers

Chapter 5, §4: an earlier version appeared as “Lessons from Pascal Engel.

OC Wittgenstein L. (1969) On Certainty (Oxford: Basil Blackwell)

It is natural to think that justifications stem from a system of assump-

All testing, all confirmation and disconfirmation of a hypothesis

Earth’s long existence. Rather, it concerns the structure of perceptual justi-

Notice, moreover, that the general propositions I claim must be

(defeasibly) to justify our beliefs about mid-size objects in our environ-

empirical beliefs, and that we do in fact possess, it takes place within

only if those assumptions were in the business of epistemic appraisal

To ban constitutive assumptions from epistemic rationality simply because

be different sets of axioms that would generate proofs for different

practical or even psychological. Therefore, it is only to be expected that

Furthermore, it is because of this kind of transmission-failure that boot-

intrinsically plausible, as it gives us the means to vindicate the intuitive

is a hand”. To put it contentiously, it will make it (more) true that there

to adjudicate between them, at least pro tempore, by assessing how each

We do think of having perceptual justifications for our ordinary empirical

both previous extremes. Accordingly, an empirical perceptual warrant is

1 The architecture of perceptual warrants 1:

Liberal account of perceptual justification: a belief about specific

ones.6 Let me therefore speak to this general concern first. In my view,

regarding, what he labels, “perceptual entitlements”.9 According to

allegedly represents. There can be representations of unicorns even if

many theorists,14 that the issue of the architecture of empirical warrants

of having a hand”, and so on? In such a predicament, it would seem

2 The architecture of perceptual warrants 2: the

The take-home message of the previous discussion of the liberal position

comprising, in particular, the proposition that there is an external world.

(I) Here is a zebra

(I) Here is my hand

Conservative account of perceptual justification: a belief about

there is an external world (and possibly other general propositions,

information contained in the conclusion is needed in order to have

We consider Wright’s entitlements more extensively in the next

3 The architecture of perceptual warrants 3:

The take-home message of the previous section is that it is very hard to

Moderate account of perceptual justification: a belief about specific

A few points of clarification are in order. Moderates, like liberals and

To assume that Q propositionally: in the abstract space of warrants, a

of experience (typically an experience with content that P) and on

A second issue worth considering at this stage is what it means to assume

To assume1 that P: to act or judge as if P were true, even if one knows

Let us recap the main features of the moderate position as I have

4 Moderates are legion (while liberals and

be part of our cognitive endowment that we take it for granted that

turns out to be a moderate.40 For, according to him, our hinges – namely,