Moss - Plato's Epistemology. Being and Seeming

Plato’s Epistemology
Plato’s Epistemology
Being and Seeming
J E S SIC A M O S S
1
1
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Jessica Moss 2021
The moral rights of the author have been asserted
First Edition published in 2021
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2020950722
ISBN 978–0–19–886740–1
DOI: 10.1093/oso/9780198867401.001.0001
Printed and bound in Great Britain by
Clays Ltd, Elcograf S.p.A.
For my mother, Rita Moss, with all my love.
Acknowledgments
I owe many thanks to many people for their help with this book. Cian Dorr,
Toomas Lott, Whitney Schwab, and Damien Storey read and discussed
multiple versions and drafts, and were wonderful interlocutors. Enormously
helpful discussion of various parts came from: Francesco Ademollo, Merrick
Anderson, Cinzia Arruzza, Rachel Barney, George Boys-Stones, Sarah
Broadie, David Bronstein, Qian Cao, Victor Caston, Tim Clarke, Adam
Crager, Carlo DaVia, David Ebrey, Matt Evans, Gail Fine, Emily Fletcher,
Jane Friedman, Natalie Hannan, Stephen Hetherington, Sukaina Hirji, Sean
Kelsey, Gabriel Lear, Sara Magrin, Marko Malink, Susan Sauve Meyer, Alex
Mourelatos, Jacob Rosen, David Sedley, Gabe Shapiro, Allan Silverman,
Nick Smith, David Sosa, Iakovos Vasiliou, Katja Vogt, Heather Whitney,
Jessica Wilson, and more people than I can name in audiences at Austin,
Berkeley, Chicago, Columbia, Indiana, MIT, Northwestern, NYU, Penn,
Pittsburgh, Princeton, Stanford, Wellfleet, and the West Coast Plato
Workshop. Oscar Dorr did excellent eleventh-hour editorial work. Finally,
I am grateful to the exceptionally helpful referees at OUP, and to Peter
Momtchiloff for all his work.
Parts of this material appear in “Plato’s Doxa,” in Analytic Philosophy 61,
and in “Is Plato’s Epistemology about Knowledge?” in What the Ancients
Offer to Contemporary Epistemology (Routledge, 2019), edited by Stephen
Hetherington and Nicholas D. Smith. Thank you to the editors and publishers
for their permission to reprint.
Introduction
What Is Plato’s Epistemology About?
That correct doxa is something different from epistêmê is

s omething I do not at all seem to be conjecturing, but, if I were
going to claim that I knew anything—and there are few things
I would claim that about—this one, at any rate, I would include
among the things that I know.
(Meno 98b)1
Throughout the dialogues Plato contrasts a superior kind of cognition with

an inferior one, often calling the superior kind epistêmê and the inferior
kind doxa.2 He studies the nature of each and the differences between them,
he argues for the superiority of epistêmê, and he discusses how to achieve it.
What is Plato doing when he does all this—what kind of project is he
undertaking?
The answer will seem obvious: epistemology. That is, Plato is engaged in
the same kind of investigation philosophers engage in nowadays when they
study knowledge and belief. Indeed, Plato had a large part in inventing the
whole field of epistemology, and gave us some of the problems and views
still central to it today.3
1 ὅτι δέ ἐστίν τι ἀλλοῖον ὀρθὴ δόξα καὶ ἐπιστήμη, οὐ πάνυ μοι δοκῶ τοῦτο εἰκάζειν, ἀλλ’ εἴπερ
τι ἄλλο φαίην ἂν εἰδέναι . . . . ἓν δ’ οὖν καὶ τοῦτο ἐκείνων θείην ἂν ὧν οἶδα. Translations throughout
are my own unless otherwise noted.
2 Or he contrasts gnôsis, phronêsis, sophia, noêsis, or nous with doxa, or contrasts these with
pistis. In treating these groups of terms as synonyms I follow many others: see for example
Crombie, 1962, vol. 2, 35; Moline 1981, 191; Snell 1924, 87 ff. I will try to show that we can very
fruitfully see Plato as using varying terminology to capture a single well-unified epistemo
logical contrast; for convenience I choose ‘epistêmê’ and ‘doxa’ to label the superior and inferior
kinds. (See further discussion below and in Chapter 1.2). I will note where the differences in
terminology seem significant, such as at Republic 511d.
3 Plato is often credited with inventing the “classical analysis of knowledge” as (roughly)
justified true belief: see for example Gettier 1963 and Armstrong 1973, 87–8.
Plato’s Epistemology: Being and Seeming. Jessica Moss, Oxford University Press (2021). © Jessica Moss.
DOI: 10.1093/oso/9780198867401.003.0001
2 Introduction
One major presupposition of this answer, sometimes made explicit, is

that Plato’s epistêmê and doxa are obviously to be identified with the main
players in contemporary epistemology (both folk and philosophical):
knowledge and belief. Indeed our word ‘epistemology’ wears this lineage on
its sleeve: it lays claim to being the study of that very thing Plato discussed,
epistêmê.
Another major presupposition, shared even by some who doubt that
Plato’s epistemological categories should be straightforwardly identified
with our knowledge and belief, is that Plato’s concerns in studying epistêmê
and doxa are broadly similar to the concerns prominent in epistemology
nowadays.4 He is studying epistemology as an enterprise in its own right.
That is to say, his project is to make sense of the epistemically salient f eatures
of various phenomena: for example, that some of our convictions are the
product of reasoning while others are not, that some are stable in the face of
argument while others are not, or that some can be transmitted to others
through speech while others cannot. His goal is to carve epistemic nature at
its joints. To put it apparently tautologically, his aims in doing epistemology
are primarily epistemological.
The main task of this book is to challenge both presuppositions, and to
offer a very different account of Plato’s epistemology instead.
I will argue that Plato’s epistêmê is very different from knowledge as
understood by contemporary epistemology, and his doxa very different from
our belief. I will not try to show that these differences entail non-identity;
that question depends on how one individuates concepts. I do hope to show
however that the differences are radical enough and fundamental enough
that it is not fruitful to study epistêmê and doxa by beginning from the
assumption that they are knowledge and belief; we should not use our
4 This assumption is very widely shared but usually implicit, testified to only by the way
scholars go about assessing Plato’s project and the questions they do not ask about it. It is how-
ever sometimes made explicit. Fine argues for translating epistêmê as ‘knowledge’ on the
ground that his translation “makes it clear that Plato is engaged in a familiar epistemological
enterprise” (2004, 71). See also Everson, who argues that even if Plato’s epistêmê is not best
conceived of as knowledge, and even though his focus is different from that of much contem
porary epistemology, he is nonetheless doing epistemology in ways that can be mapped onto
how we do it now: “The Platonic and Aristotelian concern with describing and mapping the
various cognitive states a subject can be in is part of—indeed initiates—a different tradition of
epistemological method from that which centres on justification. In this respect, they have
more in common with what has come to be called ‘natural’ or ‘naturalistic’ epistemology, in
which the epistemologist’s enterprise is part of, rather than prior to, psychology and cognitive
science”; he adds that Plato (along within Aristotle) also did “normative epistemological
enquiry” since, as evinced by the Republic’s Line, he was “interested not merely in how one
derives one’s beliefs but also in how one structures them” (Everson 1990, 5–6).
Introduction 3
theories of knowledge and belief to guide or constrain our interpretations of

epistêmê and doxa.
I will also argue that to understand what epistêmê and doxa are—to make
sense of how Plato fundamentally conceives them, and thus why he attri
butes to them the properties he does—we have to recognize that his over-
arching epistemological project is very different from ours. It is not
primarily driven by purely epistemological concerns. Instead, his epistemo
logical project is driven by metaphysical and ethical views: views about
what there is, and how one should live. Thus the epistemological properties
and distinctions that are salient to him are so primarily because of their
relation to metaphysical and ethical properties and distinctions, not for
purely epistemological reasons.
Two caveats before I begin. First, this is a book primarily about the epis
temology of the dialogues generally characterized as belonging to Plato’s
middle period: the dialogues which distinguish “Two Worlds,” a perceptible
realm of ordinary things and an intelligible realm of Forms.5 It is here that
the metaphysical concerns that (I shall argue) underlie Plato’s epistemology
are most stark and most developed, and so it is here that the epistemological
distinctions are clearest too. I will argue however that the epistemology that
is at its most developed in these dialogues is at work in the Socratic dia-
logues too, and survives in important ways in the Theaetetus; I touch on the
epistemology of these dialogues briefly throughout, and give an extended
treatment in Chapters 9 and 10.
Second, although I will speak throughout about epistêmê and doxa, this
is not a book about the meaning of those words. I think that Plato is loose
with his epistemological (as well as other vocabulary), sometimes using
several words synonymously while at other times distinguishing their
meanings, and often using a single word ambiguously. My project is thus
not to trace his use of ‘epistêmê’ and ‘doxa’ throughout the dialogues.
I instead want to show that throughout the dialogues he has a strongly uni-
fied, consistent theory about the essential features of and contrasts between
the best kind of cognition, and a kind that is its salient inferior counter-
part—that he is concerned with one central epistemological division,
5 The phrase ‘Two Worlds’ was traditionally used to describe this feature of Plato’s meta-
physics (see for example Cornford 1941), and when I refer to these dialogues as ‘the Two
Worlds dialogues’ that is how I intend it. Fine 1978 popularized the phrase ‘Two Worlds epis
temology’ to refer to the view that epistêmê is only of the intelligible realm, doxa only of the
perceptible—a view that I discuss at length, under the name Distinct Objects, in Chapter 1
and beyond.
4 Introduction
despite describing it with varying terminology, and despite sometimes

using that same terminology to describe more marginal phenomena.6 I will
not defend this approach at the outset, but instead will aim to vindicate it
through my investigations.
1. Epistêmê and doxa versus knowledge and belief
Let us begin then with a look at the assumption that Plato’s epistêmê and
doxa—or more precisely, the salient superior and inferior kinds in his epis
temology, which he presents under various labels—map closely onto con
temporary notions of knowledge and belief.
It is of course a major simplification to speak of the contemporary notions
of knowledge and belief, since there is plenty of disagreement among con
temporary epistemologists. Nonetheless we can easily identify certain views
that are widely shared about knowledge and belief, and track whether Plato’s
views converge with them or diverge. For example: almost everyone now
adays agrees that knowledge is factive and belief is not; therefore it would
constitute a major convergence if we saw Plato countenancing false doxa
but refusing to countenance false epistêmê—as indeed we will. Or: most
people nowadays agree that one can know that something is so without
knowing why, so it would constitute a major divergence if we saw Plato
insisting that epistêmê requires the ability to give an explanation—as indeed
we will.
Certainly there are some striking convergences that make it easy to
equate Plato’s central epistemic distinction with ours. Here are some claims
we see him making, with representative citations:
• Epistêmê but not doxa is infallible or unerring, always true (Republic

477e, Gorgias 454d, Theaetetus 187b).
• Epistêmê is more valuable than doxa (Meno 97d).
• Epistêmê but not doxa confers epistemic authority (Euthyphro 5a).
• Epistêmê is the aim of or norm on doxa: people who have mere doxa
often think they have epistêmê, and want to have it (Republic 476d–e).
6 Compare Crombie: “A contrast between two intellectual levels is very pervasive in Plato’s
writings from the Gorgias to the Laws,” but under various names; “The looseness of language is
of course typically Platonic” (1962, vol. 2, 34–5).
Epistêmê and Doxa versus Knowledge and Belief 5
• Epistêmê—according to some dialogues, although not others—is built

up from true doxa in combination with some extra ingredient, which
is at the least analogous to contemporary epistemology’s justification,
warrant, or the like: an account (logos, Theaetetus 201c–d) or a reason-
ing out of the explanation (Meno 98a).7
In all these ways, Plato’s epistêmê and doxa align naturally with contem
porary epistemology’s knowledge and belief. It is then no surprise to see
‘knowledge’ and ‘belief ’ as standard translations, along with many scholarly
discussions of Plato’s theory of knowledge, or of his thoughts about belief,
and many claims that he is the father of epistemology and the originator of
some of modern epistemology’s most important views.
I suspect however that there is a deeper and more general reason behind
this assimilation of Plato’s views to ours, going beyond specific similarities.
This is a thought that never gets an explicit defense because it seems so
obvious: Plato was clearly doing something in the neighborhood of epis
temology, and the things people think about when they are engaged in that
kind of project are, obviously, knowledge and belief. Compare: Plato was
clearly doing something in the neighborhood of astronomy (he had some
thoughts about the things you see in the sky at night, their natures and
movements and the differences between them), and the things people think
about when they are engaged in that kind of project are, obviously, stars and
planets and moons. Either because knowledge and belief are natural kinds,
or because it is in human nature to develop concepts of them, it is obvious
that they would have been the subject of Plato’s philosophical attention as
much as they are now of ours.
This background assumption makes it easy to fall into a kind of dogma-
tism: Plato must be thinking about knowledge and belief, and therefore any
difficulties we encounter in fitting his views to our own must stem from
oddities in his substantive theories. That Plato had various astronomical
views we now think false is no reason to doubt that he was talking about
anything other than the very stars and planets we study now; that he had
various epistemological views we now think false should likewise be no rea-
son to doubt that he was talking about anything other than knowledge
and belief.
7 Below we will see that in other places Plato seems to reject this theory: epistêmê is not a
kind of doxa; having epistêmê about something is incompatible with having doxa about it.
6 Introduction
The fact is however that quite a number of Plato’s epistemological views

are so hard to construe as views about knowledge and belief that one might
reasonably wonder if the effort is worthwhile. Here are some other claims
we see in certain contexts, again with representative citations:
• Epistêmê requires reasoning out explanations (Meno 98a).

• Epistêmê requires the ability to define or give an account of its object
(Meno 71a–b, Phaedo 76b, Republic 534b).
• Epistêmê cannot be transmitted by testimony, but requires direct
acquaintance with its object (perhaps Meno 81c, 97a–b, Theaetetus 201c).8
• Doxa excludes epistêmê, rather than being an ingredient or condition
of it (Republic V–VII throughout).9
• Epistêmê and doxa are “powers” of the soul, along the lines of sight or
hearing (Republic 477d–e, Charmides 167e–168d).
• Epistêmê and doxa are distinct powers, each with their own distinct
object (Republic 477d–478e, Charmides 168a–d).
• In the dialogues that distinguish an imperceptible realm of Being (the
Forms) from a perceptible realm of Becoming, epistêmê is of the former
and doxa of the latter (Republic 479d–e, 534a; Timaeus 27d–28a, 52a).10
When we focus on these features of epistêmê and doxa, they start to look
radically different from contemporary knowledge and belief. Should we
then question the identification? Or is it so obviously correct as to be
beyond doubt?
One need only step back from recent analytic Plato scholarship to see
that it is not. In fact, the identification of Plato’s epistêmê and doxa with
knowledge and belief, standard as it has become, is only a recent develop-
ment, and has already faced forceful criticism.
8 It is the one who has been to Larissa himself who has epistêmê of the road (Meno 97a–b);
before birth the soul has epistêmê of things in virtue of the fact that it “saw” them (Meno 81c).
The idea shows up also in Plato’s denials that epistêmê can be transmitted through testimony,
even the reliable testimony of someone who has epistêmê themselves: conventional teaching
cannot give us epistêmê, which can only be got through recollection of the things we directly
encountered (Meno 81c–86b); even a jury “justly persuaded” of the truth about some crime
will not thereby get epistêmê, which is available only to someone who was there and saw what
happened (Theaetetus 201b).
9 At Meno 98a and Theaetetus 201c–d Plato seems to reject this claim; I discuss this tension
in Chapter 7.
10 The way I have put these last two claims is controversial. I will defend it at length in the
remainder of the book.
Epistêmê and Doxa versus Knowledge and Belief 7
If we look back before the 1960s, we find that a very common English
translation of epistêmê is ‘Science.’11 This translation picks up on a philosophical
tradition that begins with the widespread use of ‘epistêmê’ in Plato’s time to
denote what we would call sciences: specialized bodies of knowledge held by
experts in particular areas.12 Aristotle elevated this use into a very specific the-
ory: his epistêmai are deductively valid systems grounded in necessary truths
about natures or essences.13 The Aristotelian notion gets preserved and
developed in the medievals’ Scientia, and via this route in the early moderns’
Science, or Knowledge: a grasp of the deep, ultimate truths, which are
necessary and essential.14 When scholars in the nineteenth or early twentieth
centuries translated Plato’s ‘epistêmê’ as ‘Science’ they were taking him to
belong to this tradition—indeed, presumably, to have founded it.15
As for doxa, older interpreters tended to translate not as ‘belief ’ but as
‘opinion,’ or sometimes ‘seeming.’ Moreover, there is a tradition dating back
to ancient times on which doxa is something much narrower than belief as
regarded nowadays: something very closely tied to perception, sometimes
described as assent to perceptual appearance, sometimes described as
empirical cognition.16
Recent scholarship presents alternatives as well. Despite the widespread
tendency to treat epistêmê as knowledge, a number of scholars have argued
that it should be construed as something else: as “logically certain
knowledge,”17 or as expertise,18 or as understanding—a deep, systematic,
explanatory grasp.19 Despite the almost universal equation of doxa with
belief, a few recent accounts have argued for other interpretations: as
11 See for example Bosanquet 1895, Jowett 1908, Taylor 1926, Shorey 1930, and
Hackforth 1945.
12 LSJ (Liddell and Scott, rev. 1940) cites Republic 477b ff. along with the Posterior Analytics
as loci for the meaning “scientific knowledge, science” (s.v. epistêmê, A.II.2).
13 Thus only some of what we call sciences, and of what Aristotle in looser moods calls
epistêmê (e.g. at Nicomachean Ethics 1138b27) are Aristotelian epistêmai in this strict sense:
mathematics counts; medicine does not.
14 For citations and defense of this reading of Spinoza, Leibniz, and Descartes, and a brief
comment on the continuity with Aristotle’s epistêmê, see Carriero 2013.
15 For a recent defense of this line, see Wolterstorff 1996, 220–1: “the medievals were at one
with Plato in their understanding of epistêmê” because in the Republic epistêmê is a grasp of
“what is fully real. . . the necessary, eternal, immutable.”
16 Crombie 1962, Cornford 1941, Gulley 1962, Sprute 1962; in Ancient times see especially
Alcinous and Proclus, quoted and discussed in Chapter 8.
17 Vlastos 1981, 1985. 18 Woodruff 1990.
19 See Moravcsik 1979, Annas 1981, Burnyeat 1980, Nehamas 1985, Benson 2000,
Schwab 2015, 2016.
8 Introduction
inherently deficient cognition,20 or cognition of particular tokens rather

than universal types.21
I will discuss these interpretations in more detail in Chapters 5 and 8. I
think there is much that they get right. Most generally, they are focusing
appropriately on certain features of Plato’s epistemology, features that cast
serious doubt on the identification of its central concepts with contempor
ary knowledge and belief.
2. The plan
Abandoning that identification in favor of any alternative, however, will

raise major questions. There are of course questions about the interpretative
adequacy of the alternatives: for example, epistêmê is much more demand-
ing than understanding as nowadays conceived, and restricted to a much
narrower domain, so that we might raise the same kind of worries about
this interpretation as about the interpretation of epistêmê as knowledge.
More fundamentally, there are questions about why Plato would choose
epistêmê and doxa as the star players in his epistemology if they are not in
fact knowledge and belief. We know lots of reasons to center epistemology
around the study of knowledge and belief; what are the reasons for doing it
some other way? Why would Science, for example, or understanding, be so
salient to Plato? Why would he choose to devote so much attention to defin-
ing one of these? And why would he think that the relevant contrast kind is
opinion, or seeming, or empirical cognition?
Most generally, once we take seriously the possibility that Plato’s central
epistemological concepts are radically different from ours, we are faced with
the task of reassessing his whole project. If he is not studying the same phe-
nomena that we are, perhaps that is because he is not asking the same ques-
tions. If so, then the evolution of epistemology is not after all much like that
of astronomy (different views of the same objects). A closer parallel would
be the evolution of ethics, according to a common interpretation. It is often
held that Ancient ethics asks a different set of questions from those we ask
nowadays, focusing not on moral rightness and moral requirements, but
20 Vogt 2012, although she translates ‘doxa’ as belief and does not in my view go far enough
in distinguishing the two. See also Moss and Schwab 2019.
21 Rowett 2018.
The plan 9
instead on the notion of happiness. A theory developed to address the

question “How should one live?” will be badly misunderstood if it is taken
to be focused on the question “What makes an action right?”; perhaps
Plato’s epistemology is similarly misunderstood when we look at it through
the lens of current concerns.
If so, we need to figure out what his actual concerns are. Is he analyzing
the epistemological concepts current in his day? Articulating the epistemic
ideal, and the corresponding deficiency?22 Introducing new concepts that
serve his non-epistemological philosophical purposes in some way? If Plato
is not doing epistemology the way we do it nowadays, then we need to fig-
ure out what he is doing when he does epistemology, and what his epistem
ology is about.
My aim in this book is to answer these questions—to interpret Plato’s
whole epistemological project, as well as its central concepts of epistêmê and
doxa—by taking as a starting point what is arguably the most radical differ-
ence between his epistemology and ours. I have in mind the last claim in the
list of differences above: epistêmê is of Being, while doxa is of something
distinct from and inferior to Being.
This may seem a perverse place to start. The claim comes from an argu-
ment in Republic V that is famously difficult, riddled with ambiguities, and
the subject of numerous conflicting interpretations; indeed, on some inter-
pretations it is misleading even to state the claim as I have done.23 Surely
then, one might think, whatever claim Plato is making here is best read in the
light of independently established views about his epistemology, rather than
used as a foundation. I will try to show however that there are major textual
and philosophical gains to be made by taking this claim as a starting point.
Taking “epistêmê is of Being, doxa is of something inferior” as bedrock
characterizations, we can construct interpretations of epistêmê and doxa
that make sense of all their peculiarities: interpretations that integrate the
features Plato attributes to epistêmê and doxa across many dialogues,
accommodating their similarities to contemporary knowledge and belief as
well as their differences. We can also construct an account of Plato’s entire
22 For this interpretation of Aristotle’s epistemology, see Pasnau 2013.

23 Smith (2000, 2012, 2019) argues that while epistêmê is specially correlated with being,
and doxa with something inferior, the relation is not the relation of being of or about; cf.
Szaif 2007. Fine (1978, 1990) has an interpretation on which the formulation I have given is
misleading, since epistêmê’s objects (true propositions) are not wholly distinct from doxa’s
(true and false propositions); cf. Gosling 1968. I discuss these interpretations in further detail
in Chapter 1.
10 Introduction
epistemological project which shows it to be not a muddled or nascent

version of our own, but something radically different, very much in keeping
with his overall philosophical concerns, and compelling in its own right.
Here is a brief summary of the claims I will make.
First, Plato’s epistemological projects are motivated by, and must be
understood in the context of, his central ethical and metaphysical views.
These are views that can be found in inchoate forms in the earlier dialogues,
and survive in complicated ways in the late ones, but are most robust, devel-
oped, and explicit in the middle dialogues, and find their most dramatic
expression in the Republic’s allegory of the Cave.24 Namely:
• There is a crucial metaphysical distinction between two levels of

reality: genuine Being on the one hand, and something inferior and
derivative on the other, which can be characterized among other ways
as the realm of what seems.
• There is a crucial ethical distinction stemming from this metaphysical
one: to be in contact with Being is to be living well, while to rest con-
tent with the inferior level of what seems is to hinder oneself even
from aspiring to live well.
If one comes with this metaphysical and ethical background to epistemo

logical investigations, the distinction one will find most salient and import
ant will be that between cognitive contact with what Is and cognitive contact
with what seems. And that is precisely how we should understand the fun-
damental distinction between Plato’s superior and inferior cognitive kinds.
Republic V thus gives us not a muddle or mystery, but an explicit statement
of Plato’s central epistemological doctrine. The main claims of this doctrine
are these:
• Epistêmê and doxa are each essentially to be understood as cognition

of a certain kind of object, since each is adapted to grasp its special
object and indeed inherits its character from its object.25 (Chapter 2)
24 I adopt the standard ordering of Plato’s works, but am not committed to it: for my pur-
poses, as will become clear below, the major division is between dialogues which embrace a
full-blown two-level ontology, distinguishing intelligible Forms from sensible particulars, and
dialogues which do not.
25 In my formulations here I gloss over the distinction between the powers (dunameis) of
epistêmê and doxa, and their activities; for discussion see especially Chapter 2.
The plan 11
• Epistêmê is in its essence the grasp of being, what is. This Plato takes to
be a truism, which he develops and inflates with his metaphysics: the
beings are ontologically privileged items, the real, which in the Two
Worlds dialogues are the Forms. This explains both the features that
make epistêmê resemble contemporary knowledge and those that
make it more resemble Science and understanding. Epistêmê is best
understood as a deep grasp of ultimate reality. (Chapters 3–5)
• Doxa is in its essence a response to what seems. This too Plato takes to
be a truism, which he inflates with a metaphysical view: the things that
seem are ontologically inferior items, namely images and appearances,
a class which in the Two Worlds dialogues includes the entire percep-
tible realm of Becoming. This explains both the features that make
doxa resemble contemporary belief and those that make it look like
empirical cognition. Doxa is best understood as cognition that stays at
the level of readily accessible appearances, and therefore ignores the
imperceptible entities that are, on Plato’s view, truly real. (Chapters 6–8)
In sum, just as on Plato’s view the world contains both colors and sounds,
and we are equipped with special powers of grasping each, so too on Plato’s
view the world contains both being and something inferior, and we are
equipped with special powers of grasping each. Epistêmê is the power by
which we understand what is, doxa the power by which we form impres-
sions of the ontologically inferior things which seem to be. This turns out to
explain all the features of doxa and epistêmê, all the contrasts between them,
and the supreme value of epistêmê.
My account of Plato’s epistemology draws very much on others’ work:
many of the claims I make have been presented already in some version. For
example, the idea that Plato’s epistemology is firmly grounded in his meta-
physics is stressed by Gerson (2009) and White (1992); my characterization
of epistêmê has much in common with what we find in Moline (1981), and
also in various interpretations of epistêmê as explanation-grounded under-
standing (e.g. Nehamas 1985, Schwab 2016); my characterization of doxa is
in large part similar to what we find in Sprute (1962) and Gulley (1962); my
overall picture is strongly indebted to Cornford (especially 1941). In many
cases, however, these authors reach their conclusions through arguments
with which I disagree, or through insufficient argument, and so I wish to
offer new foundations. Moreover, these interpretations are still hotly con-
tested, or treated as marginal, or in some cases evidently outright forgotten.
I aim therefore to offer them new and more systematic defenses, and also to
12 Introduction
bolster them by showing their place in an overall interpretation of Plato’s

epistemology. I also aim to draw out their significance more sharply—in
particular, to show just how radical a difference they entail between Plato’s
epistemology and today’s.
My account will show that Plato’s central epistemological categories are
different enough from ours that it is dangerous to use our own epistemo-
logical intuitions and theories in interpreting his. Perhaps one might, after a
careful comparison, conclude that Plato is indeed getting at the same phe-
nomena that we track nowadays, albeit by a very different method; that
should however be the conclusion of a study of his epistemology, and not a
premise. (For brief attempts, see Chapters 5.4 and 8.4.)
My account will also show that Plato’s epistemological project is very dif-
ferent from the projects epistemologists undertake nowadays. If Plato’s epis
temological categories are different in important ways from ours this should
not be surprising, because his epistemological inquiry is driven by ques-
tions we would not consider purely and properly epistemological. Instead, it
is motivated by metaphysical concerns, and ultimately by ethical ones.
Plato’s epistemology is part of his overall inquiry into how one should live.
1
Plato’s “Two Worlds” Epistemology
If we cannot simply assume that epistêmê and doxa are knowledge and
belief, how should we go about determining their natures? How should we
investigate what Plato’s epistemology is about?
It would be nice if we could do ordinary language philosophy, determin-
ing what the words meant in Plato’s time and taking those meanings as his
central notions. Indeed some interpreters take this approach as if it were
unproblematic.1
There is certainly a good deal to be said for looking to contemporary lan-
guage for guidance, and I will do so in developing my own accounts of
epistêmê and doxa; indeed, I will eventually argue that Plato’s epistêmê and
doxa are theory-laden, metaphysically inflated versions of the conceptions
current in his day.
As a core strategy, however, beginning with contemporary usage is
unpromising. First, different authors in Plato’s time use these terms in dif-
ferent ways in different contexts, so much so that it is an interpretative task
in itself to determine what the terms meant in their mouths, and whether
each term had a single standard meaning or was systematically ambiguous.2
Second, Plato is notoriously willing to twist ordinary language almost
beyond popular recognition: consider for example his treatment of justice
(dikaiosunê) in the Republic, or of wanting (to boulesthai) in the Gorgias
1 See Barnes, countering Burnyeat’s suggestion that Plato’s epistêmê is not ordinary know
ledge but instead understanding: epistasthai, epistêmê, and cognates “are not philosophical
neologisms; they occur frequently in Greek literature from Homer onwards, and they are there
correctly translated by ‘know ’ and its cognates. . . Both Plato and Aristotle talk of epistêmê
without special qualification or apology; they give no indication that they intend the term in a
novel or restricted sense; we are obliged to conclude that they thought they were investigating
the ordinary concept of knowledge” (Barnes 1980: 204–5); cf. Fine 2004, 70. For direct argu-
ments against these claims about ordinary Greek usage, see Schwab 2015.
2 Pace Barnes’ and Fine’s claims about epistêmê, quoted in the previous note. A look at the
LSJ entry is instructive: as the main heading they offer “acquaintance with a matter, under-
standing, skill, as in archery” (citing Sophocles Philoctetes 1057, where the claim that someone
“has this epistêmê” is naturally translated as “is skilled in this art”; cf. their passages from
Thucydides and Lysias, as well as Plato). “Generally, knowledge” comes in as a second heading,
with far more citations for its subheading, “scientific knowledge.”
DOI: 10.1093/oso/9780198867401.003.0002
14 Plato ’ s “ Two Worlds ” Epistemology
(466e–468d). Whether he means literally to redefine certain words, or

simply to present surprising theories about their application, ordinary usage
is often a poor guide to Plato’s.
There is another strategy for understanding Plato’s epistemology, implicit
in much work on the subject: begin with a list of all the various notions in
our own epistemological repertoire—knowledge and belief, but also under-
standing, scientific understanding, a priori knowledge, expertise, opinion,
perception, empirical cognition, and so on—and determine which ones are
the best matches for epistêmê and doxa as Plato characterizes these. This
strategy is inspired, presumably, by the assumption that these are the basic
kinds out there to be studied (whether because they are natural kinds, or
universal human constructs), and therefore that Plato must have been talk-
ing about some of them. The strategy has its strengths, but is plagued by
vagueness: given that Plato’s kinds clearly have some features in common
with various of these, and others diverging from them, how do we decide
which commonalities and divergences are decisive, and which marginal? It is
also subject to charges of anachronism: why should we assume that the con-
cepts central to Plato’s epistemology are ones central to ours, or even ones
that we acknowledge at all? The parallel strategy for understanding other
areas of Plato’s philosophy might well fail: there is room for serious doubt
about whether the kalon, a concept central to Plato’s view of virtue, has any
counterpart in modern ethics (the fine, beautiful, noble, admirable?); most
modern psychologists would deny that they work with any concept much
resembling Plato’s notion of the thumoeides (spirited part of the soul).
Thus we should be open to the idea that the only way to identify Plato’s
core cognitive concepts is to look for unifying features within his own char-
acterizations: to bracket our assumptions about what he is trying to track,
and look only at his actual descriptions. But this would seem to be a dizzy-
ing task. Plato makes so many different claims about epistêmê and doxa in
so many different contexts, many of them varied or even conflicting. How
do we know which give us the core concept, as opposed to marginal or
optional details, or even if there are any core concepts there to be found?
I want to show that there is no need for special inventiveness, nor for
despair. For alongside his various claims and characterizations, Plato offers
us very clear and explicit instructions about how to define and understand
cognitive kinds.
I have in mind the instructions he gives in the course of Republic V’s
famous argument about the difference between epistêmê and doxa. Here
Socrates identifies both epistêmê and doxa as powers (dunameis—one might
Plato ’ s “ Two Worlds ” Epistemology 15
also translate “abilities,” “capacities,” “faculties”), and presents a general

strategy for defining powers:
In a power I look toward this thing only: what it is over and what it accom-
plishes [eph’ hôi te esti kai ho apergazetai], and in that way I call each of these
a power.3 And what is set over the same thing and accomplishes the same
thing I call the same power, while what is over something different and
accomplishes something different I call a different power (Republic 477d)
Epistêmê and doxa are powers of the soul (477e). A power is distinguished
from others only by “what it is over and what it accomplishes.” If you want
to know what makes one power different from another, then, this is where
you should look.
Is this meant as a mere heuristic guide to recognizing powers? Or might
it be something stronger, a claim about what it is to be a given power? If so,
Plato would be providing a manual for answering our questions about
epistêmê and doxa: to understand what each of these is, look to what it is
over and what it accomplishes; that is what yields the essential characteriza-
tion. Of course we would need to figure out what he means by ‘over,’ what he
means by ‘accomplishes,’ and how to fill in the values of each for epistêmê and
for doxa. We would also need to understand what it means for epistêmê
and doxa to be powers, and how this characterization relates to the more
familiar (to us) notion of cognitive kinds as attitudes or states. At least,
however, we would have a method that tells us how to proceed.
Consider the instructions for understanding powers that Aristotle pro-
vides in his work on the soul. Aristotle divides the soul into a number of
things he calls powers (dunameis), such as the nutritive, perceiving and
thinking powers. In a famous passage of De Anima II.4, he explains how
each power is essentially defined:
If one must say what each of these [powers] is, for example what is the think-
ing or the perceiving or the nutritive [power], it must first be said what think-
ing is and what perceiving is. For the activities and the actions are prior in
account to the powers. And if this is so, and prior still to these the corres
ponding objects must be studied, one must first define these, for the same
reason [i.e. that they are prior in account], for example about nourishment
and the perceptible and the thinkable. (Aristotle, De Anima 415a16–22)
3 Or “in that way I name each power” (ταύτῃ ἑκάστην αὐτῶν δύναμιν ἐκάλεσα).
The echoes of Plato are striking.4 Powers are defined by two things. First,
their activities and actions: this is already reminiscent of Plato’s “what they
accomplish,” and elsewhere Aristotle calls these erga, accomplishments or
functions (402b9–16), closer still. Second, their ontological correlates, the
things out there in the world on which they do their work—like Plato’s
“what they are over.” As we will see in Chapter 2, Aristotle elsewhere expli
citly applies this principle to cognitive powers: not just thinking, as men-
tioned here, but also the powers of epistêmê and doxa (Nicomachean Ethics
1139a6–12 with 1140b26–28, quoted in Chapter 2.2).
Aristotle’s claim in De Anima goes further than what we find explicit in
Republic V, in two ways. First, he puts the ultimate work onto the objects
rather than sharing it between objects and activities: the objects are “prior
in account” to the activities. Second, the work he puts on the objects (and,
derivatively, on the activities) is explicitly definitional. The power of sight,
for example, is what it is because the visible is what it is. Thus Aristotle’s
theory of psychological powers is fundamentally objects-based: each psy-
chological power is ultimately defined by its object.
Plato says none of this explicitly in Republic V. Looking back at the argu-
ment, however, it is very easy to see him as closely anticipating Aristotle’s
argument; the differences are superficial or inconclusive. I will defend this
interpretation in greater detail in Chapter 2, but here is a brief defense. First,
although he identifies powers in part by “what they accomplish,” his descrip-
tion of the accomplishments is wholly uninformative: he simply uses the
verbal forms of each noun, as if to say that in each case the power of Φ-ing
has as its accomplishment the activity of Φ-ing;5 this implies that, as in
Aristotle, it is ultimately the objects that do the work. Confirming this, he
goes on to state twice that a difference in powers entails a difference in
objects.6 Second, although he does not explicitly state that his individuating
criteria are also definitional, a look at his treatment of powers elsewhere,
and especially of cognitive powers, strongly suggests that they are (see
Chapter 2). Perhaps then this is a case of Plato leaving implicit or
4 They have not however been noted much in discussions of the Republic’s argument.
Johansen (2012, 94) notes the parallel, and interprets it in a way very congenial to what I will
argue here: “Aristotle’s understanding of dunamis is inspired by Socrates’ account. Like
Socrates, he thinks that in order to determine what a dunamis is we need to determine what it
is about.” The account he goes on to give of how Aristotle understands the priority of objects is
very similar to what I will attribute to Plato in Chapter 2.
5 Doxa is the power by which we doxazein (477e), epistêmê that by which we gnônai (cog-
nate with ‘gnôsis,’ a word this passage uses interchangeably with ‘epistêmê’) (478a).
6 478a–b, quoted in Chapter 2.1.
Plato ’ s “ Two Worlds ” Epistemology 17
undetermined something that Aristotle makes explicit: powers are not only
individuated but in fact defined by their objects.
The powers argument does mention one other difference: epistêmê is
infallible (anamartêton), doxa fallible (477e). It is hard to reconcile this
claim with the earlier claim that powers can only be distinguished by their
objects and accomplishments; perhaps the idea is that these are features of
their accomplishments: doxa accomplishes fallibly, or accomplishes some-
thing fallible, by contrast with epistêmê (Cornford 1941, comment ad loc).
One might think that Plato means here to be identifying the fundamental,
essence-giving feature of each power. Notably however it is Glaucon who
mentions these features, and they are treated as evidence that the two p owers
are different, where this is at least compatible with their being explained by
the essential difference rather than constituting the essential difference. In
the next chapters I will argue that this is indeed Plato’s view: doxa and
epistêmê are each ultimately defined by their objects; because it is of Being,
epistêmê is infallible; because it is of the ontologically inferior, doxa is fallible.
If the Republic’s powers argument really anticipates Aristotle’s objects-
based psychology, then we would have in this argument the key to his epis
temology. Cognitive kinds are defined by their ontological correlates. The
most revealing characterization of epistêmê is that it is the power set over its
particular object, here in Republic V identified as “what is” (478a). The most
revealing characterization of doxa is that it is the power set over its particu-
lar object, here identified as “what is between being and not being” (478d–e).
The project of this book is to conduct an experiment: the experiment of
taking these characterizations as starting points in understanding Plato’s
epistemology. I will develop the idea that the nature of Plato’s cognitive
kinds is determined by their objects: indeed, the nature not just of the cog-
nitive powers, but also of the (more familiar to us) occurrent states or atti-
tudes, which we should understand as accomplishments or exercises of the
powers. I will develop accounts of the objects of both epistêmê and doxa,
and from there will develop accounts of epistêmê and doxa themselves.
I do not claim that there is conclusive evidence that Plato means us to
proceed this way, but I will argue that there is enough evidence to make the
experiment well worth undertaking (Chapters 1 and 2). If it can produce
compelling results with textual support and philosophical payoff—as I hope
to show that it can (Chapters 3–8)—we should take it to be vindicated.
There is however a serious objection to undertaking this experiment at
all. The proposed objects-based interpretation of the Republic’s powers
argument rests on a reading of that argument that has recently become
highly contentious. The past half-century has seen extensive debate over the
interpretation of Republic 476e–480a, with nearly every aspect of the argu-
ment subject to multiple interpretations.7 Many deny that Plato really
means to individuate doxa and epistêmê by ontological correlates that serve
as their objects. Indeed, many regard such an interpretation as so thor-
oughly problematic as to be a total non-starter. And if Plato is not even
individuating cognitive kinds by their objects, how could he mean to be
defining them that way?8 Before I begin my positive account of Plato’s epis
temology, then, I must show that the approach is viable.
The task of the present chapter is to show that it is admissible to read
Plato as holding that cognitive powers are individuated by their objects—
indeed, to shift the burden of proof onto those who reject that reading.
With the ground thus cleared for the stronger interpretation on which
powers are defined by their objects, the task of the next chapter is to provide
positive arguments that this is precisely what we should expect Plato to
hold, given his approach to powers in general, and to cognition in particular.
1. The Two Worlds debate
Let us begin with a brief overview of the Republic’s powers argument.

Socrates wants to show that the ideal city is possible only when the rulers
are philosophers, because philosophers alone have epistêmê, while many
who think they have epistêmê, such as “sight-lovers” (aesthetes), in fact have
only doxa. In order to establish that philosophers alone have epistêmê, he
begins with the following claims:
(1) Epistêmê and doxa are distinct powers (dunameis) (477d–e).9

(2) Distinct powers are “over” (epi) distinct things (478a).
(3) Epistêmê is over what is (478a).
(4) Doxa is over what is between being and not being (478d–e).
So much is explicit. As to the interpretation, however, there is a raging

controversy—particularly centering on the questions of what Plato means
7 See among many others Gosling 1968, Fine 1978 and 1990, Annas 1981, Stokes 1992,
Brown 1994, Gonzalez 1996, Baltzly 1997, Butler 2007, Smith 2000, 2012, Szaif 2007,
Taylor 2008, Kamtekar 2009, Boylu 2011, Vogt 2012, Harte 2018.
8 For more on the connection between individuating by objects and defining by objects, see
the end of Section 2, and Chapter 2.
9 The evidence for their being distinct is that epistêmê is infallible, doxa fallible (477e).
The Two Worlds debate 19
by ‘over,’ what he means by ‘what is,’ and what he means by ‘what is between
being and not-being.’
Here is one interpretation. ‘What is’ refers to the very things Socrates will
shortly identify as the object of epistêmê, namely the Forms (479e). ‘What is
between being and not-being’ refers to the very things Socrates will shortly
identify as the object of doxa, namely the many beautiful things, large
things, and so on (479a–e), i.e. the perceptible (to horaton, 509d). The “over”
relation is the relation of a cognitive power to the objects out in the world
onto which that power directs the mind—in other words, something like
the subject matter with which that power is concerned. The rest of the argu-
ment should thus be understood as follows:
(5) The many perceptible things, including beautiful sights, are between
being and not-being (479c, e).10
(6) The Forms are (479e).
(7) Therefore only those who attend to Forms—that is, philosophers—
have epistêmê; those who attend to the perceptible realm have doxa
(479e–480a).
If this interpretation of the powers argument is correct, then Plato is

individuating cognitive powers by their ontological correlates. Epistêmê is
exclusively exercised on Forms, and doxa exclusively exercised on ordinary
perceptibles. One can no more have epistêmê of a perceptible, or doxa of a
Form, than one can see a sound or hear a color.
What I have just outlined is often called the “Two Worlds” interpretation
of Plato’s epistemology, but the name does not get to the heart of the issue.
Nearly everyone agrees that according to the Republic there are two worlds,
as symbolized so clearly by the Line and Cave images: the perceptible realm
of ordinary objects, and the intelligible realm of Forms. What is at issue
between this interpretation and its rivals is whether each world is the sole
and exclusive object of its own correlated cognitive power—whether
10 The end of the argument makes this explicit: those who attend to sights and sounds have
only doxa, because they are in touch only with what is between being and not-being (479d–e).
Confusion has resulted from Plato’s reference to “the conventions [nomima]” that roll about
between being and not-being (479d). Nomima is often translated as ‘belief ’ (see for example
Shorey 1930, Fine 1990), with the odd result that Plato is now identifying as doxa’s objects—
what is between being and not-being—doxa itself. In the sequel to the argument, however,
Plato uses ‘nomima’ again, to mean the images of Forms that philosopher-rulers instill in the
city (484c). Most charitably, then, nomima in the powers argument too are images of Forms,
the objects of doxa, rather than doxai themselves.
cognitive powers have distinct objects.11 I will therefore refer to the

interpretation instead as the Distinct Objects interpretation. (I use ‘Two
Worlds’ to refer to the metaphysical thesis that there are two distinct realms;
when I speak of Two Worlds dialogues I mean those that put forth that
thesis.)
The Distinct Objects interpretation is the traditional interpretation of
Plato’s epistemology: as I show in Section 2, it was standard and largely
unquestioned for more than two millennia. In the past half-century, how-
ever, it has come to be so widely doubted as to seem to many to be simply
off the table. First came complaints that Plato should not have held it (for
example Cross and Woozley 1964), then a flood of new interpretations,
starting with Gosling 1968 and Fine 1978, on which he in fact did not hold it.
Many now think that Plato cannot mean to separate the objects of epistêmê
and doxa so sharply; indeed such a view is regarded as “outrageous”
(Baltzly 1997, 240), “deeply paradoxical, at best” (Smith 2012, 57). Far more
charitable to think that Plato had something subtler in mind. For example:
- Each power has a privileged relation to its associated objects, but not an
exclusive one. For example: each power can apprehend its own objects
directly but can also apprehend the other’s objects with assistance from
another power (Kamtekar, 2009, 142–3); doxa is “naturally adapted” to
perceptibles but can in a deficient way grasp Forms (Vogt 2012, 55); each
power is “normatively tasked” for its own objects but can with the aid of
those objects cognize other things (Harte 2018).
- Each power has an exclusive relation to its associated objects, but these
objects serve only as material for the power, and not as subject matter,
what particular accomplishments of doxa or epistêmê are about; thus there
can be overlap in the latter (what I will call the Content Overlap view: see
Smith 2000, 2012, and 2019; compare Szaif 2007).12
- The claim of the powers argument is not at all that epistêmê and doxa
have different ontological correlates like Forms and perceptibles, but
instead that they are related to different but overlapping sets of
11 Moreover, as I will argue below, one can hold that epistêmê and doxa have distinct objects
without holding that those objects occupy separate realms: they might be distinct aspects or
properties of the same entities, as colors and sounds are of ordinary physical objects.
12 There are important differences between the views in these works; I will focus on their
common trait of distinguishing between the objects which a power is “over” and the objects
about which one can have doxa or epistêmê.
propositions: epistêmê to the set of true propositions, doxa to the set of

true and false propositions (Fine 1978, 1990).
We can refer to these and many related interpretations as Overlap readings,

since their common claim is that there is some overlap between the objects
of doxa and those of epistêmê.13
Overlap interpretations have been the norm in analytic Plato scholarship
since Fine’s influential paper (although recent years have seen a few compel-
ling criticisms of individual Overlap interpretations, and a few intriguing
developments of Distinct Objects views in their place).14 They were almost
unheard of before it. Why are they now so widely accepted?
Certainly there are text-based motivations for them. If philosophers’ fit-
ness to rule rests on their having epistêmê, surely epistêmê must be applic
able to the perceptible world, telling us which human laws and actions are
just.15 Indeed, Socrates says almost explicitly that philosophers have
epistêmê of such things: when philosophers return from theoretical study to
the Cave of political life they will “know [gnôsesthe] each image, what it is
and what it is an image of ” (Republic 520c; cf. 506a).16 And while the point
is less often noted, Plato also characterizes wisdom (sophia), the virtue that
makes an individual able to rule their own individual life well, as a kind of
epistêmê (Republic 443e); if epistêmê has no practical application this too is
13 The interpretation defended by Smith and Szaif is on one sense of ‘objects’ a Distinct
Objects interpretation, on another an Overlap interpretation.
14 Arguing extensively against Fine see especially Gonzalez 1996; defending either a com-
plete Distinct Objects reading or (more commonly) only the restriction of epistêmê to Forms,
see also Vlastos 1981, Kahn 1996, Gerson 2003, 2009, Sedley 2007, Carpenter 2012–13,
Woolf 2013, and Schwab 2016.
15 On the Distinct Objects interpretation the philosophers live in “a different cognitive
world” from those they rule, and hence “knowledge turns out to be irrelevant to the problems
that inspired the search for knowledge in the first place” (Annas 1981: 194); “the ‘two-worlds’
theory of Plato’s epistemology manages to destroy the very argument Plato takes himself to be
advancing . . . that philosophers would be better rulers” (Smith 2000, 153–4).
16 The verb in both passages is gignôskein, which the powers argument used as correlate
with epistêmê: see especially the characterization of the objects of knowledge as to gnôston at
478b, quoted above. For extensive arguments that the relevance of epistêmê to ruling under-
mines the Distinct Objects reading see especially Fine (1990) and Smith (2000). I have not seen
those who object to the Distinct Objects reading cite as evidence against it what might at first
seem more decisive: various passages in which Plato refers to epistêmê in some particular
material domain, e.g. house building (Republic 438d) or horse riding (Republic 601c–d). I sus-
pect it seems even to opponents of the Distinct Objects reading that in these passages Plato is
using epistêmê loosely or broadly, as equivalent to what he would elsewhere call technê, rather
than in the strict sense intended in Book V. (Compare Aristotle’s occasional broad use of
epistêmê, for example at Nicomachean Ethics 1138b27, to cover what he elsewhere calls technê
by sharp contrast with epistêmê.)
paradoxical. Conversely, Socrates and his interlocutors evidently have many

doxai about the Forms, since they have thoughts about Forms but deny hav-
ing epistêmê;17 indeed, Socrates says almost explicitly that he has doxai
about the Form of the Good (Republic 506c).18 Finally, there is the broader
text-based argument that in other dialogues—the Meno and Theaetetus—
Plato holds an Overlap view: if a true doxa can become epistêmê by being
tied down with an account of the explanation (Meno 98a), or if a true doxa
together with a logos amounts to epistêmê (Theaetetus 201c–d), then a for-
tiori, the argument goes, doxa and epistêmê can share objects.19
These are serious challenges that any Distinct Objects interpretation
must face, and I will address them below, arguing that they can in fact be
met. (A brief preview: epistêmê informs our judgments about the percep
tible realm without directly applying to it, the way that in some areas ideal
theory informs but does not directly apply to practice (Chapter 4.5); mean-
while, thoughts about Forms that fall short of epistêmê are not doxa but
instead something between the two, like what Plato calls dianoia
(Chapter 7.6). The Meno can be read as embracing a Distinct Objects epis
temology without a Two Worlds metaphysics, so that the objects of the
powers are closely related (Chapter 9.3). The Theaetetus is aporetic, in part
to highlight the problems with abandoning a Distinct Objects view
(Chapter 10).)
Here, however, I want to focus on a different kind of question. Why were
these challenges for the Distinct Objects interpretation only noticed so
recently? And why are they now so widely taken as decisive, mandating
creative readings of the passages that correlate cognitions with objects,
rather than showing that the passages which suggest these challenges stand
in need of creative readings themselves? Proponents of the Distinct Objects
interpretation from years past are usually simply silent on the passages that
Overlappers find so significant, or else give readings consistent with the
Distinct Objects interpretation without seeming to notice any problem.20
17 See especially Vogt 2012.

18 Adeimantus asks Socrates to relate his own dogma (etymologically closely related to
doxa) about the good (Republic 506b); Socrates refuses, not saying that he can have no such
thing, but instead implying that he does and that they are deficient: “Haven’t you perceived that
doxai without epistêmê are all shameful? The best of them are blind” (506c), and hence there is
no point talking about “what seems to me” about the Good (506e). I discuss this passage in
more detail in Chapter 7.6.
19 The sharing of objects is particularly explicit at Theaetetus 208e: αὐτοῦ ἐπιστήμων γεγονὼς
ἔσται οὗ πρότερον ἦν δοξαστής.
20 For example, a look at major English-language commentaries (Adam 1902,
Cornford 1941, Bosanquet 1895, Jowett and Campbell 1894) shows straightforward Distinct
When they do notice passages that seem to conflict with Distinct Objects, they
find creative ways to read those passages on which the conflict disappears.21
Why is the situation so different now? What has changed in the philosophical
and interpretative landscape?
Here is a very natural hypothesis: the objections to the Distinct Objects
interpretation stem from reading Plato through the lens of twentieth-
century epistemology. The main motivation for doubting that Plato held a
Distinct Objects epistemology is the thought that such a view is philosoph
ically hopeless, and this thought rests on the assumption that in doing epis
temology, Plato is doing more or less what epistemologists do nowadays. In
particular, it rests on the assumption that epistêmê and doxa are more or less
just like what we now call knowledge and belief—an assumption that often
surfaces explicitly, but is operative through much work on the subject.
When Overlappers insist that it must be possible to have epistêmê of per
ceptibles, they are often obviously thinking of epistêmê as the sort of thing
philosophers once thought well analyzed as justified true belief. That state is
very promiscuous in its subject matter; it can certainly be had about percep
tible things.22 Hence for example Fine’s complaint that it would be absurd
for Plato to deny epistêmê of perceptibles, for it would follow that “no one
can know even such mundane facts as that they’re now seeing a tomato, or
sitting at a table” (1990, 86).
When Overlappers insist that it must be possible to have doxa of Forms,
they are thinking of doxa as belief: either generic belief, the sort of thing
philosophers now analyze as “taking to be true”; or as belief that falls short
Objects interpretations of Republic 477d–80a, and no comments indicating any worries for that
interpretation at 506c or 520c.
21 For example: when Neoplatonist commentators on the Phaedo notice an allusion to doxa of
Forms (Phaedo 66b–67b, discussed below) they take this as evidence that the word ‘doxa’ is
ambiguous between something that arises from perception and is confined to perceptibles—at
issue, presumably, in all the discussions that imply Distinct Objects—and something else, derived
from thought, at issue in the present passage (see the commentaries by Damascius and
Olympiodorus, discussed below and in Chapter 7.6). When traditional interpreters notice that
the Theaetetus implies the possibility of epistêmê of perceptibles, they take this to show that Plato
did not in fact endorse this analysis: see for example Cornford 1935, or the interpreters referred
to by the Anonymous Commentator in the late first century BCE as holding that the Theaetetus
shows “what epistêmê’s objects are not” (viz., perceptibles): see discussion in Sedley 1996.
22 Of course we can only know true facts or propositions, but we can know such truths
about all sorts of objects—e.g. about unicorns we can know that they do not exist, and about
contingently existing objects we can know that they exist contingently (see Introduction). Also:
many skeptics deny knowledge of perceptibles, but it is very difficult indeed to read Plato as
holding that kind of view, and no Distinct Objects interpreters to my knowledge do. (I will say
more about this in the discussion of true doxa, in Chapter 7.1.)
of knowledge, sometimes called ‘opinion.’ Beliefs of both kinds can in

principle be had about any subject matter at all. Generic belief is widely
taken to be a necessary condition of knowledge, and opinion can share its
subject matter with knowledge. Hence for example Vogt’s objection to the
traditional Distinct Objects reading on which there is no doxa of the Form
of the Good:
From the point of view of contemporary ethics, and from the point of
view of ordinary talk about our ethical lives, it seems obvious that we have
beliefs about the good . . . Starting from such beliefs, we might try to for-
mulate a theoretical account . . . It would seem that an ethical theory
according to which one cannot do this must be misguided. Surprisingly,
Plato scholars have been content with an interpretation of the Republic
according to which Plato cannot allow for beliefs about the good [viz., an
interpretation on which there is no doxa about the Form of the Good].
(Vogt 2012, 51)
When Overlappers deny that epistêmê and doxa are distinguished by their
objects they are thinking of cognitive attitudes as topic-neutral attitudes,
distinguished only by degree of justification, or method of acquisition, or
other purely epistemological features. They are motivated by the kind of
thought we find in Cross and Woozley’s complaint that Plato was misled by
his own analogies in the Republic’s powers argument:
[I]f, without any preconceived theories, we consider knowledge and belief,

which are the two powers with which Plato is especially concerned here . . . [we
see] that the difference lies in the way of apprehending, and there is no neces-
sary difference in the objects apprehended. (Cross and Woozley 1964, 164)23
23 Cross and Woozley are not Overlappers: they think that Plato did hold the Distinct
Objects view, but that he was wrong to hold it. Their assumption that epistêmê and doxa are
knowledge and belief is clearly the motive for this disapproval: “Plato draws a distinction
between knowledge and belief, and. . . between two different sets of objects corresponding to
these different states of mind. . . We ought now to ask ourselves whether there is in fact a dis-
tinction to be drawn between knowledge and belief, and if there is, whether Plato’s way of rep-
resenting the distinction is satisfactory;” their answer to the second question is that it is not,
because knowledge and belief can overlap in their objects (1964, 166). Compare Armstrong,
who thinks that in the Republic Plato advances a Distinct Objects view, which as he puts it
“contests, that the one state of affairs, for instance the roundness of the earth, can be the object
of the distinct cognitive attitudes,” and then argues that the falsity of this view—apparent on
the assumption that the attitudes are knowledge and belief—is “one of those obvious facts
which we should appeal to in the course of testing philosophical theses and arguments rather
than allow philosophical theses and arguments to cast doubt upon” (Armstrong 1973, 141).
The thought is frequently echoed; for example:
Now even before looking at the [powers] argument we find it prima facie
odd that an object of belief can never be an object of knowledge. For a
start, it violates our intuition that the same thing can be first believed and
then known, that coming to know is not changing the subject.
(Annas 1981, 194)
[On the traditional reading of the powers argument], knowledge and
belief cannot have the same objects. But this seems puzzling . . . [I]ntui-
tively, it is possible for one person to have knowledge of the very same
thing that another person believes; indeed, in modern epistemological
discussions it is assumed that knowledge entails belief so that whatever a
person knows that same person will also believe. (Harte 2018, 142)
In brief, the assumptions underlying many Overlap readings are that Plato’s
epistemological categories are the same as ours, that his epistemology is
more or less like ours, and in particular that he, like us, recognized the obvi-
ous fact that cognitive kinds are simply not the sort of things to be individu-
ated by their objects.
The task of the present chapter is to undermine these assumptions, in
part by showing that they are very recent indeed, and hence should be sus-
pected of anachronism, and in part by showing how natural it is to read
Plato himself as being innocent of them—as simply laying out a Distinct
Objects epistemology without concern to accommodate any of the worries
such an epistemology would face now.
I am not trying to show that the Distinct Objects reading must be correct,
but instead to show that it is far from a non-starter, and should be put back
firmly on the table. Indeed, I think the arguments I offer here should render
the Distinct Objects interpretation the default, to be abandoned only if we
cannot meet the textual and philosophical objections alleged against it. In
the following chapters I will argue that we can.
A final note before I begin. My purpose in defending the Distinct Objects
reading is to clear the way for an objects-based account of Plato’s epistem
ology; there are however objects-based accounts consistent with various
Overlap interpretations, and indeed several such interpretations explicitly
construe cognitive kinds as tailored to or defined in terms of their objects.
For example, on accounts like Harte’s, Kamtekar’s, and Vogt’s, epistêmê can
be defined by a relation to Forms even though doxa can be had of Forms
too, just as for example a pruning knife is defined by a relation to pruning
even though other knives can be used to prune (Republic 353a).24 On a

Content Overlap account like Smith’s or Szaif ’s, epistêmê as a power can be
defined by its relation to Forms, even though its exercises or applications
can be about perceptibles, and vice versa for doxa. Indeed, it is only Fine’s
propositional reading of the powers argument that is sharply at odds with
an objects-based interpretation. Why not then defend or at least leave open
a moderate Overlap reading, insisting only that epistêmê and doxa each bear
some special relation to an ontological correlate, and proceed to develop an
account of Plato’s epistemology from there?
I have a number of reasons for experimenting with a radical Distinct
Objects reading. First, I believe, and hope to convince the reader, that this is
very much the most obvious way to read Plato, and also the way that he was
read for about two millennia. We simply never see him nor his older inter-
preters working to draw the kind of distinctions on which moderate Overlap
readings rely, nor showing signs of noticing problems with the starker view.
Second, while there could indeed be versions of an objects-based epistem
ology consistent with some overlap, I will argue in Chapter 2 that both
Plato’s characterization of how powers in general are related to objects,
and also his characterization of how cognitive kinds share significant prop-
erties with their objects, point us toward an objects-based epistemology on
which the objects of each cognitive kind are sharply distinct. Third, I will
try to show in the remainder of the book that reading Plato this way yields
compelling accounts of both epistêmê and doxa, ones not only consistent
with the texts but also with Plato’s wider philosophical projects. In short, I
think the only reason to prefer a moderate Overlap reading to a stark
Distinct Objects one is the thought that the latter cannot be made to work—
but I hope to show that it can.
2. The prima facie case for Distinct Objects
Until recently there was no Two Worlds debate: scholars simply assumed
that Plato held the Distinct Objects view, both in the Republic and related
dialogues (see Section 3 for evidence). Why was this reading so widespread?
24 Harte (2018) cites this passage as background for the powers argument; Vogt explicitly
argues that in saying that each power is “by nature” (pephuke) over its objects (Republic 477b),
Plato means that “it is the nature of knowledge and belief to be structured toward their objects”
(Vogt 2012, 63).
The prima facie case for Distinct Objects 27
I think the answer is clear: it is the natural, simple, straightforward reading

of Plato’s discussions of epistêmê and doxa throughout the Republic and in
related dialogues as well.
I cannot hope to establish that claim decisively, and I will not try: so
much has been written in recent years questioning the Distinct Objects
reading, and so many strategies found for re-interpreting the texts that seem
most to support it, that I am sure further strategies can be found to under-
mine whatever further evidence I provide. Nonetheless it may be helpful to
remind ourselves just how obviously right the Distinct Objects interpret
ation can seem when we are not already convinced that it must be wrong.
To be clear about my goal: I want to show that there is copious evidence
that in the Two Worlds dialogues—those that separate Forms from
perceptibles—Plato is working with an epistemological view on which the
crucial divide tracks that metaphysical divide: the superior kind of cogni-
tion is restricted to Forms, and the inferior kind is restricted to perceptibles.
For simplicity’s sake, and to follow Plato’s most prominent usage, I will
generally refer to the Form-confined kind of cognition as epistêmê, and the
perceptibles-confined kind as doxa; as I mentioned in the Introduction,
however, this is a simplification, and in two ways. First, as we will see shortly,
Plato often uses other terminology: he sometimes uses ‘phronêsis,’ ‘gnôsis,’
‘noêsis’ or ‘noûs’ to refer to cognition which he describes as c onfined to
Forms; he sometimes uses ‘pistis’ to refer to cognition which he describes as
confined to perceptibles. Second, he sometimes uses ‘epistêmê’ and ‘doxa’ in
ways that do not fit this picture: for example, the housebuilder has epistêmê
of building houses (Republic 438d); the philosopher has doxa that truth is to
be found only in the intelligible realm (Phaedo 66b–67b). Although I will
argue that these latter are non-standard uses which are nonetheless
explained by Plato’s core uses, as I said in the Introduction my thesis is not
a thesis about Plato’s use of the words ‘epistêmê’ and ‘doxa.’25 It is instead a
thesis about his overall epistemological project: I want to show that he has a
pervasive project of distinguishing superior cognition from inferior by cor-
relating the superior kind with ontologically superior objects, and the inferior
kind with ontologically inferior objects. I hope readers of Plato will recog-
nize that he is often loose enough with his terminology that inconsistencies
in vocabulary do not belie the existence an overall unified view; my aim is to
show that he does have such a view. Now to the textual evidence.
25 I am thus taking a very different approach from what we see in for example Fine 2016, or
Lyons 1963.
First, there is the much-discussed passage from Republic V with which

we began, where Plato argues that all powers (dunameis) are distinguished
by what they are set over and what they do (477c–d).26 His idea must be that
these are interdependent criteria, for he twice stresses that a difference in
powers entails a difference in objects:
[W]e clearly agree that epistêmê is different from doxa?—Different.—So each

is by nature over [epi] something different, since each has the power for
something different?—Necessarily . . . . If different powers are by nature over
different things, and both of these are powers, doxa and epistêmê, and
different from one another, as we say, from these things it follows that the
same thing cannot be gnôston and doxaston [the object of gnôsis, i.e. epistêmê,
and the object of doxa]. (Republic 477e–478b, emphases added)
Why does Plato ignore the possibility that two powers could be set over the
same object, distinguished only by what they do to it?27 The simplest
explanation is that he conceives of a power as the power to do some activity
φ to some type of object f where φ-ing can only be done to fs. In other words,
it would appear that Plato is individuating powers of the soul by their objects:
there is a one-to-one correspondence between objects and powers. I will
develop and defend this reading in detail in Chapter 2, arguing that this is
precisely what he is doing; for now I want simply to point out that that is the
simple reading of the text, accommodating Plato’s repeated claim that a dif-
ference in powers necessarily entails a difference in objects.
Overlappers find plenty of ways to read the powers argument that loosen
the correlation between object and power. For example, to say that a power
is over (epi) an object is to say that it operates at its best or by nature on this
object, not that it can operate only on this object.28 Or, as Fine argues, the
things that doxa and epistêmê turn out to be over—what is, and what is
between being and not-being—are not wholly distinct: Plato means that
epistêmê is over the set of true propositions, doxa over the set of true and
false propositions.
These readings all depend on subtle interpretations of special features of
the powers argument: of the over (epi) relation, or of the phrase ‘what is.’
26 Compare Phaedrus 270d: to define anything we must see what power it is has—“to do
what in relation to [pros] what, or to undergo what by the agency of what?”
27 To take Fine’s example, as butchery and husbandry do different things to the same
objects, animals (1990, 5).
28 For example Kamtekar 2009, Vogt 2012, Harte 2018.
There is much to be said about whether or not these interpretations are

viable on their own terms, as interpretations of this specific argument: I can
see the force of some of the going interpretations of epi, while I am con-
vinced (with what is perhaps a majority these days) that the propositional
interpretation of ‘what is’ is strongly in tension with the text.29 I will not
enter into the debate about this particular passage at more length here. I
discuss the passage in more detail in Chapter 2, but for now I want to point
out something crucial that discussions of it often ignore. Even if the
Overlapper persuades us of an interpretation of this passage, her work is not
done: although the debate has focused almost exclusively on this argument,
it is only one among many pieces of strong prima facie evidence for the
Distinct Objects thesis.
Within the Republic, the powers argument is only the beginning: as Plato
elaborates his ontological and epistemological divisions through the famous
passages on the Sun, Line, and Cave, he makes repeated claims that correl
ate distinct cognitive kinds with distinct objects—claims not easily tackled
by the particular strategies Overlappers have developed to explain the
powers argument.
First, when Plato first returns to the topic of levels of reality and of cogni-
tion, in the simile of the Sun, he makes what looks like an unqualified cor-
relation between powers and objects:30
Whenever [that part of the soul by which we think] is fixed on that on which
truth and being shine [the Forms], it understands and knows it and appears
to have intelligence,31 but whenever it [is fixed on] that which is mixed with
darkness, that which comes to be and passes away, it has doxa and is weak-
sighted, shifting its doxai back and forth.32 (508d, emphasis added)
29 For extensive arguments see Gonzalez 1996. I will not repeat his arguments here; to my
mind the most decisive is that throughout the argument, and especially toward its end (479c–e),
there is a clear identification of what is with the Forms, and of what is between being and
not-being with perceptibles.
30 He introduces this passage with a claim that is to serve as an analogy for epistêmê and
doxa: when our eyes fix on things illumined by the sun we see clearly, but when they turn to
things in darkness we do not. It is natural to think that here we have one set of objects, not two:
the very same trees, for example, can be the object of good vision (analogous to epistêmê) or
bad (analogous to doxa). But the passage I quote shows that in the epistemic case Plato clearly
intends two distinct sets of objects. The Cave allegory yields a cleaner parallel: one set of
objects (the statues and shadows) is by nature always in the dark, and it is these that symbolize
the object of doxa.
31 ἐνόησέν τε καὶ ἔγνω. The related nouns (noêsis, gnôsis, gnômê) show up as apparent syn
onyms for epistêmê throughout the middle books (see citations below); in our Republic V pas-
sage, as we saw, Plato uses gnôston as the objective correlate of epistêmê.
32 δοξάζει τε καὶ ἀμβλυώττει ἄνω καὶ κάτω τὰς δόξας μεταβάλλον.
Strategies for deflating the ‘over’ of Republic V as describing a mere norm or

primary function will founder on the ‘whenevers’ we find here: one will
need a bold new strategy for denying that Plato is saying what he seems to
be saying, namely that every cognition in which the mind is directed toward
Forms is an instance of epistêmê, and every cognition in which the mind is
directed toward perceptibles an instance of doxa. Of course some such
strategy may be found: one might suggest for example that the “fixed on”
relation is a very strong one; if the soul is merely attending to Forms with-
out being fixed on them, it may have doxa, and mutatis mutandis for
epistêmê of perceptibles. My point is not that any such strategies are hope-
less, but that Overlappers will need to employ a hodgepodge of them to
tackle different texts—to the point where their strategy becomes quite
strained.
Proposition-based strategies like Fine’s will founder on this text too. Fine
interprets Republic V’s “over” talk as being about the relation of cognitions
to propositions, and thus as making no claims about distinct subject mat-
ters, but the present passage clearly claims that epistêmê is the result of
attention to Forms, doxa the result of attention to things that come to be
and pass away—perceptibles. Whether or not Plato conceived of doxa and
epistêmê as propositional attitudes, the claim here seems to be that they
have distinct objects precisely in the sense of ontological correlates, not
propositions. Indeed all the passages I cite below name the relata of doxa
and epistêmê as perceptibles and intelligibles, or as Becoming (genesis) and
Being (where it is clear, as we will see, that this is a distinction between two
different types of object: things that change and things that remain the
same). I will therefore set aside the propositional reading from here on,
influential as it has been in this debate.
Next comes the famous simile of the Line, which distinguishes sharply
between two realms, the visible and the intelligible (noêton) (509d), and
then refers back to these as the objects of doxa and gnôsis (the doxaston and
the gnôston, 510a). This seems to assume a straightforward identity claim:
what can be seen and what can be the object of doxa are one and the
same; what can be the object of epistêmê is identical with what can be
accessed by intellect instead of by perception.
Indeed, the Line strongly reinforces the impression that Plato held a
Distinct Objects view by compounding the distinctions, for now we have
four levels of reality, and a different cognitive state over (epi again) each one.
There are two subsections of the doxaston, each presided over by its own
species of doxa, and two subsections of the gnôston, each presided over by
its own species of epistêmê.33
In the Cave allegory the correlation is even stronger: at each of four stages
the ascending prisoner achieves a higher cognitive state because she is
attending to a higher class of object (see especially 515d). Cognitive states, it
seems, are determined by their objects, as we saw at 508d: if you are focused
on perceptibles, things in the Cave, you have doxa; if you are focused on
intelligibles, things outside of the Cave, you have epistêmê. At the end of the
discussion Plato sums up the correlations with a new formula:
Doxa is about [peri] becoming, noêsis34 is about being. (Republic 534a).
Overlap strategies for handling the ‘over’ (epi) of Republic V will now have
to extend to the ‘about’ (peri) of Book VII, but the Cave allegory makes it
particularly natural to take the correlation to be strict.
Along the way, moreover, Plato has made several statements about the
highest cognitive state in particular that seem to confine it very starkly to
intelligibles:
[the many beautifuls, many goods, etc.] are seen, but not understood
[noeisthai], while the Forms in turn are understood, but not seen.
(507b–c, emphasis added)
. . . if anyone tries to learn something about perceptibles, by gaping up or
blinking down, I deny that he would ever learn—for such things admit of
no epistêmê.35 (Republic 529b–c, emphasis added)36
33 “Over [epi] the four sections take there to be four conditions [pathêmata] arising in the
soul: noêsis over the highest, dianoia over the second; render pistis for the third, and over the
last one eikasia” (511d–e). Recapping these divisions at 534a Plato refers to eikasia and pistis as
both being (species of) doxa. He never explicitly calls noêsis or dianoia species of epistêmê (or
gnôsis), but at 533e–4a, in what seems to be a blatant shift of terminology, calls the very top
section epistêmê instead of noêsis, and says that both epistêmê and dianoia are (species
of) noêsis.
34 Noêsis here names the highest stage of thought; see previous note.
35 ἐάν τέ τις . . . τῶν αἰσθητῶν τι ἐπιχειρῇ μανθάνειν, οὔτε μάθειν ἄν ποτέ φημι αὐτόν -
ἐπιστήμην γὰρ οὐδὲν ἔχειν τῶν τοιούτων…
36 Schwab (2016) draws attention to this passage as powerful evidence for his case that
epistêmê is restricted to Forms; I return to the passage in Chapter 4.4, arguing that the ensuing
lines both confirm this reading and provide Plato’s rationale for the view. Readings have how-
ever been proposed on which these lines make a weaker claim: it is impossible to gain epistêmê
by focusing solely on perceptibles, but if you have already studied the Forms then you can come
to have epistêmê of the perceptibles. (See Rowe’s (2012) comment ad loc).
These look like outright denials that epistêmê of perceptibles is possible—

not mere denials that it is the norm, or natural, or best.
Looking outside the Republic, we find that other dialogues with Two
Worlds metaphysics—dialogues that distinguish a realm of Being from a
realm of Becoming—also seem to correlate each ontological realm with a
distinct cognitive power, again without any overt qualification or apology.
The Timaeus is a particularly clear case. Here as in the Republic Plato distin-
guishes “that which always is” from “that which is always becoming,”
and adds:
The former is grasped by noêsis with reason [logos], always being the same;
the latter in turn is the object of doxa [doxaston] through doxa with unrea-
soned perception, becoming and passing away, but never really being
(Timaeus 27d–28a)
Later in the dialogue we get a distinction between Forms:
which it falls to noêsis to investigate, and the other kind . . . per

cep
tible . . . grasped by doxa with perception. (Timaeus 52a)
These claims doubtless could be interpreted as describing mere norms, but

Plato mentions no exceptions; his evident interest is in correlating distinct
cognitive kinds with distinct objects.37 Starker still is this:
Whenever logos arises that is about [peri] the perceptible, and the circle of
the different going straight proclaims it to the whole soul, doxai and
pisteis38 arise stable and true; but whenever in turn logos is concerned with
the calculating39 and the circle of the same, circling smoothly, reveals
these things, nous and epistêmê are necessarily accomplished.
(Timaeus 37b–c, emphases added)
37 Indeed this last passage compounds the impression that powers are correlated with
objects by adding a third ontological kind and assigning a third power to grasp it: place or the
receptacle is a different genos, and thus cannot be grasped by either noêsis or doxa; instead it is
“apprehended by a kind of bastard reasoning with non-perception, and is scarcely an object of
pistis” (52b).
38 Plural of pistis.
39 λογιστικόν. This must here mean the object of reasoning, rather than the faculty (see dis-
cussion in Cornford 1937, 95 note 3): in other words, the intelligible, identified in the two pas-
sages quoted previously as Being, i.e. the Forms.
Cognition of intelligibles is by nature different from cognition of percep

tibles—it proceeds through a different “circle”—and it always yields a different
cognitive state.
Then there is the evidence of the Phaedo. Here as in Republic V Plato
contrasts stable Forms with shifty perceptible particulars; here too he uses
Beauty as his exemplar (Phaedo 78d–e). And here too he distinguishes two
modes of cognition corresponding to two ontological realms:
These [particulars] you could touch and see and perceive with the other
senses, while those [beings] that remain always the same it is not possible
to grasp in any other way than with the reasoning of thought40…So should
we posit two species of things that are, the one visible, the other invisible?
(Phaedo 79a–b)41
This looks to be a statement of a Distinct Objects view. Here the contrast is

not between epistêmê and doxa, but instead between reasoned thought and
perception; as we read on, however, it becomes clear that the two distinc-
tions align. Here is a striking echo of the Republic’s Sun analogy (508d,
quoted above):
[T]he soul, whenever it makes use of the body for investigating something,
either through seeing or through hearing or through some other sense . . . is
dragged by the body toward the things that are never the same, and itself
wanders and is confused and is dizzy as if drunk, because it is having con-
tact with things of that sort . . . But whenever it investigates itself by itself, it
passes toward what is pure and always being and deathless and always the
same, and since it is akin to it, it always is with it, whenever it is itself by
itself and is able to, and it has ceased its wandering and remains always in
the same state, because it is having contact with things of that sort.42 And
this affection of it is called wisdom [phronêsis]. (Phaedo 79c–d)
As at Republic 508d, whenever the soul attends toward stable Forms the
result is a stable psychic state (here called phronêsis, but with obvious affi n
ities to the Republic’s epistêmê or noêsis or nous: cognition of the Forms).
40 τῷ τῆς διανοίας λογισμῷ.

41 Socrates goes on to argue that there is a further similarity: soul is like the divine in being
a ruler, body like the perceptible in being ruled (80a).
42 ἅτε τοιούτων ἐφαπτομένη.
Whenever it attends instead to shifty perceptibles, investigating things

through the body, i.e. through perception (79c), the result is a confused and
shifting psychic state. That state is not here named, but shortly below Plato
distinguishes again between the condition of someone who pays attention
to perceptibles and that of someone who does not, now characterizing the
former as doxa. The soul of a person who experiences much bodily pleasure
and pain in visible things (ta horata, 83c):
has doxa that the very things that the body declares true are true…[and]
shares its doxai43 . . . [While by contrast the soul of the philosopher] fol-
lows reasoning and is always with it, studying the true and the divine and
the adoxaston and nourished by that. (Phaedo 83d–84a)
Attending to perceptibles yields doxa, and can never yield a higher cogni-
tive state because perceptibles confuse the soul. Attending to Forms yields
phronêsis, and can never yield doxa because Forms are adoxaston—not pos-
sible objects of doxa.44 This is a very stark and explicit Distinct Objects view.
Moreover, it is put forth in passages with clear echoes of Republic V, encour-
aging us to think that that is Plato’s view there too. Epistêmê is cognition of
Forms, doxa of perceptibles. (Note that the Phaedo’s emphasis on the role of
perception in doxa is echoed in the other Two Worlds dialogues: in the
Republic doxa is introduced as the mindset of “lovers of sights and sounds,”
and Plato later refers to the lower half of the Line interchangeably as the
doxaston and the visible (509d–510a); in the Timaeus, as we just saw, the
lower ontological order is grasped by “doxa with perception” (Timaeus 52a).)
In the Phaedrus, the “true kind of epistêmê” is about “the colorless, shape-
less, and intangible being that really is,” namely the Forms (247c).45 In the
Philebus, the crafts concerned with things that come to be and pass away
rely on doxa (59a), while true epistêmê and nous are about (peri) “what is
and what really is and is always the same” (58a; cf. 59c–d).46
I have presented all this not as decisive evidence for the Distinct Objects
reading, but as a reminder of why it seemed so natural for so long. Looking
43 δοξάζουσαν ταῦτα ἀληθῆ εἶναι ἅπερ ἂν καὶ τὸ σῶμα φῇ. ἐκ γὰρ τοῦ ὁμοδοξεῖν τῷ σώματι…
44 For this interpretation see for example Gallop’s commentary ad loc (Gallop 1975). One
could of course argue that Plato means the claim about Forms to be more restricted: they are
not merely objects of doxa, but also of epistêmê, or they are not objects of doxa for the philoso-
pher (see Fine 2016).
45 ἡ γὰρ ἀχφώματός τε καὶ ἀσχήματιστος καὶ ἀναφὴς οὐσία ὄντως οὖσα . . . περὶ ἣν τὸ τῆς
ἀληθοῦς ἐπιστήμης γένος.
46 For a more nuanced treatment of the Philebus’ epistemic categories see Chapter 2.3.
A brief history of the Distinct Objects interpretation 35
at Republic V’s powers argument, at the rest of the Republic, and at other
dialogues that distinguish two ontological realms, we find Plato consistently
drawing one-to-one correspondences between cognitive kinds and onto-
logical kinds, with the higher cognitive kind confined to intelligibles, the
lower to perceptibles. The evidence is strong and copious. As we will now
see, it was for over two millennia taken as unambiguous.
3. A brief history of the Distinct Objects interpretation
Suppose that you are learning about Plato’s epistemology in the fourth cen-
tury bce, by reading his student Aristotle. Here is what you will hear about
epistêmê:
Plato . . . having in his youth become familiar with Cratylus and with the
Heraclitean doctrines that all perceptibles are always flowing and there is
no epistêmê about them,47 continued to hold these doctrines in later years.
(Metaphysics 987a32–b1)48
Aristotle evidently thinks that Plato holds both an extreme metaphysical

view: perceptibles are in radical, Heraclitean flux.49 He also thinks that Plato
holds on that basis an extreme epistemological view: there is no epistêmê
about perceptibles. Whether or not Aristotle is correct in these attributions,
it is clear that he made them: according to Aristotle, Plato countenanced no
epistêmê of perceptibles. Nor does Aristotle show any perplexity as to how
Plato could have held such a view; he simply asserts that he does. (He makes
no parallel statement about Plato’s doxa, but we will see in Section 4 that he
himself endorses a version of a Distinct Objects epistemology on which
doxa is confined to its own realm, and that when he does so he claims to be
saying what “everyone believes.”)
47 ἁπάντων τῶν αἰσθητῶν ἀεὶ ῥεόντων καὶ ἐπιτσήμης περὶ αὐτῶν οὐκ οὔσης.
48 Compare Metaphysics 1078b12–17: proponents of the Forms posited them in the first
place because they believed that perceptibles were “always flowing,” and thus needed to find
some other object for epistêmê, since “epistêmê cannot be of the flowing [οὐ γὰρ εἶναι τῶν
ρεόντων ἐπιστήμην].”
49 Plato discusses, but does not endorse, views about the extreme instability of the percep
tible world in the Cratylus (439c–440d) and Theaetetus (152d–160c); Aristotle may also have in
mind Plato’s contrasts in the Republic and other Two Worlds dialogues between the changing,
contradictory nature of perceptibles and the permanence and stability of Forms (see for
example Republic 479a–b, 524a, 529d–530b, Phaedo 74b–c and 79c–d, Symposium 210e–211b,
and Philebus 59a–b).
Might Aristotle have intended a subtle claim compatible with an Overlap

reading—for example, that there is epistêmê of perceptibles in an indirect or
derivative or atypical way, or that while epistêmê as a power takes only
Forms as inputs, exercises of epistêmê can be about perceptibles? Certainly
we see him making no such qualifications. Moreover, the argument he
appeals to very strongly suggests a stark, unqualified account: perceptible
things are so radically unstable that no thoughts about them will have the
kind of abiding truth required for epistêmê.
I will not try to offer a comprehensive account of Plato scholarship over
the ensuing two millennia, but a brief look at a few influential interpreters
will confirm this pattern: commentators repeatedly attribute to Plato a stark
Distinct Objects view, without raising any of the concerns that motivate
Overlap readings.
Suppose you are learning about Plato’s epistemology in Roman times,
from Cicero’s description of Plato’s Academic followers. You will hear that
the Forms are the province of scientia (the standard Latin translation of
epistêmê), while perceptibles are opinabile, the object of opinio (the standard
Latin translation of doxa).50 No doubts or qualifications, just a simple report.
Or suppose that you are learning about Plato’s epistemology the way
people did for many centuries, by reading the second century ce “Handbook
of Platonism” (Didaskalikos), attributed to Alcinous.51 You would get an
unqualified Distinct Objects claim:
[Human logos (reason)] has two elements, one concerned with [peri] intel-
ligibles, and the other with perceptibles, of which the one concerned with
intelligibles is epistêmê and epistemonic logos, and the one concerned
with perceptibles is doxastic [logos] and doxa . . . Of epistêmê . . . and
doxa . . . the principles [archai] are thought [noêsis] and perception.
(Didaskalikos, 4.2–3; cf. 4.7–8)
Again there are no qualifications or subtleties. Perhaps ‘concerned with’

(peri) could be read consistently with an Overlap account, but we see the
50 All perceptibles, which are in constant flux, “they called objects of opinio [opinabilem].
Scientia however they thought existed nowhere except in the notions and reasoning of the
mind” (Cicero, Academica I.31–2); just above we were told that the mind (mens), by contrasts
with senses, “discerns that which is always simple and uniform and such as it. This they called
Form…” (I.30). Hence this is a Distinct Objects view: opinio is of the perceptible, Scientia of
the Forms.
51 The work was particularly popular from the Byzantine period up until the nineteenth
century; for discussion of its influence, and authorship, see Dillon 1993, preface and
introduction.
author making no efforts to give such an account, nor raising any concerns
that might motivate one.
Moreover, the author gives an account of doxa that seems constructed to
confine it to perceptibles. That doxa has its principle or origin (archê) in
perception suggests that all doxai are derived from perception, and so
cannot be of purely intelligible things. In the next lines, we get a definition
of doxa as the “interweaving of memory and perception” (4.5). This is a
definition we find nowhere explicit in Plato, but it has echoes of the Philebus:
doxa “comes on each occasion from memory and perception” (Philebus
38b). In the Philebus, memory has earlier been defined as the preservation
of perception (34a), and so is itself only of perceptibles. Moreover, the
Didaskalikos supports the definition with an example based on the wax
block discussion in the Theaetetus (193c), where the memory in question is
clearly derived from perception.52 According to the Didaskalikos’ defi nition,
then, every episode of doxa is derived from an episode of perception. Doxa
is, in effect, perception-based cognition, and hence there is no doxa of
Forms. (Some will protest that a doxa about perceptible Fness is, given
Plato’s metaphysics, ultimately about the Form of F; I address this view in
Chapter 7.)
Suppose you turned instead to another middle Platonist, Plutarch. You
would learn that Plato, following Parmenides, distinguished an impercep
tible realm from a perceptible, and made the first the object of noêsis
(noêton), the second the object of doxa:
And Plato (and Socrates even earlier) understood that nature has one
element that is doxaston, and has another that is noêton. And the doxaston
is unstable and wandering. . . (Adversus Colotem 13)
A few lines below the doxaston is identified with Parmenides’ aisthêton, per
ceptible realm. Once again, there are no qualifications or concerns.53
52 When we perceive an object which we remember from past perception, “we put together
the previous memory with the second sensory perception, and we say within ourselves, for
instance, ‘Ah, Socrates!’…and it is this that is termed ‘doxa’” (Didaskalikos 4.5). The Didaskalikos
shows no sign of concern about other passages in the Theaetetus which imply that one can have
doxa of imperceptibles such as numbers (195e–196a): the author is perhaps simply ignoring
these examples, or somehow interpreting them in a way consistent with this thoroughgoing
empiricism about doxa.
53 Plutarch strongly implies a similar view in his interpretation of Plato’s doctrine of recol-
lection (Fr. 215, Sandbach): epistêmê comes from pre-birth encounters with Forms, while what
sense-experience gives us is something inferior (see Scott 1987). For a detailed account on
Or suppose you turned for illumination to the Neoplatonists. Far from

raising any problems for the Distinct Objects interpretation, they seem to
embrace it with gusto, correlating a proliferation of cognitive kinds each
with its own realm, in ways that would seem to leave no room for Overlap.54
Proclus interprets Plato as holding that there are various kinds of gnôsis,
cognition, each “defined by their objects”55 (In Timaeum 2.249.8–9, com-
ment ad Timaeus 27d–28a, with reference to the Republic; cf. 2.254.24–27).
Indeed, each kind of cognition is not only correlated with its own type of
object, but somehow stems from it:
The things [pragmata] are of two kinds, what is and what has become, and
the cognitions are of two kinds, noêsis and doxa . . . For from where else do
cognitions come, but from the objects of cognition?56
(In Timaeum 2.339.14–17)
Moreover, each kind of cognition is similar in character to its correspond

ing objects:
For we say that everywhere the like is cognized by the like: for obviously
by perception the perceptible [is cognized], and by doxa the object of doxa
[doxaston], by dianoia the object of dianoia [dianoêton], and by nous the
object of nous [noêton].57 (Theologica Platonica, I.15.17–20)
I will argue in Chapter 2 that Proclus is right to find in Plato the doctrine
that like is cognized by like, and furthermore that, given Plato’s metaphys-
ics, the view entails Distinct Objects. A cognitive kind that cognizes eter-
nally self-same intelligibles must be similar to them, and therefore unable
to cognize anything dissimilar to them such as the shifting objects of
doxa—and vice versa.
which Plutarch takes himself to be following Plato in denying noêsis of the perceptible realm,
see Boys-Stones 1997. See Section 4 for discussion of Plutarch’s own Distinct Objects view.
54 With one qualification: Plotinus may hold that dianoia as well as noêsis is of Forms.
I consider the special case of dianoia in Chapter 7.6.
55 διορίζων τὰς γνώςεις τοῖς γνωστοῖς.
56 πόθεν γὰρ ἀλλαχόθεν αἱ γνώσεις, ἢ ἀπὸ τῶν γνωστῶν.
57 Τῷ γὰρ ὁμοίῳ πανταχοῦ φαμὲν τὰ ὅμοια γινώσκεσθαι· τῇ μὲν αἰσθήσει δηλαδὴ τὸ αἰσθητό
ν, τῇ δὲ δόξῃ τὸ δοξαστόν, τῇ δὲ διανοίᾳ τὸ διανοητόν, τῷ δὲ νῷ τὸ νοητόν. On the difference
between doxasta and aisthêta, see main text below.
Proclus also makes it explicit that each cognitive power is activated only
when operating on its particular set of objects, with no overlap. There is a
general cognitive (kritikê) power, namely logos, which:
when it directs itself toward the contemplation of intelligibles uses both

itself and noêsis . . . and when judging the objects of doxa also [in addition
to itself] sets in motion doxa . . .58 (In Timaeum 2.255.3–12)
Moreover, like Alcinous but even more starkly, Proclus offers very specific
accounts of each cognitive kind, accounts which entail that they are by
nature restricted each to their own realm. Nous is direct, non-discursive,
intuitive grasp of intelligible objects (see for example In Timaeum
2.243.18–22). Doxa meanwhile is explicitly is defined as “cognition [gnôsis]
of perceptibles” (In Timaeum 2.249.2–3 comment ad Timaeus 28a). Its func-
tion is to discern the whole object underlying various proper perceptibles:
in Proclus’ example, while sight says that a perceived object is red, smell that
it is fragrant, and so on, doxa is the power (dunamis) by which we connect
these proper perceptibles and say what the perceptible object is—an apple—
“and for this reason Plato has called the perceptible the object of doxa
[doxaston]” (2.249.13–27).59 This seems simply to rule out any doxa of
intelligibles: doxa operates only on objects we can perceive. (I return to this
characterization of doxa in Chapter 8.) Clearly this is not an interpretation
designed to leave room for Overlap of any kind.60
Much the same views are to be found a millennium onwards in the
Renaissance reviver of Platonism. Ficino argues that on Plato’s view Scientia
58 ὁ λόγος . . . ἐπὶ μὲν τὴν τῶν νοητῶν θέαν στελλόμενος ἑαυτῷ τε χρῆται καὶ τῇ νοήσει . . . τὰ δὲ
δοξαστὰ κρίνων κινεῖ καὶ τὴν δόξαν. Logos uses both itself and dianoia to discern the “inter
mediates,” itself and phantasia to discern the phantasta, and itself and perception to discern
perceptibles.
59 For discussion of Proclus’ view of doxa see Lautner 2002, which in passing attributes to
Proclus a thoroughly Distinct Objects reading of Plato, as well as a Distinct Objects epistem
ology of his own.
60 Might the Neoplatonists have in mind a Content Overlap reading of the kind proposed
by Smith or Szaif? Answering that question would require a very careful study, but my inclin
ation is to say no. Reading their detailed, thorough, inquisitive commentaries on passages that
trouble modern readers, we simply see no signs of these concerns. See for example Proclus’
lengthy commentary on Timaeus 27d–28a, a passage I quoted above where Plato assigns Being
to nous and Becoming to doxa (In Timaeum 2.240–64). Proclus raises questions about whether
or not Plato means to be offering definitions of being and becoming; he also takes the occasion
to state the Neoplatonist doctrine of intellect’s identity with its objects and sense’s identity with
its objects; but he says nothing at all elaborating let alone qualifying the relation between epis
temological kinds and ontological.
(the standard Latin translation for epistêmê) can be had only of “things
which are always the same, and are in the same way,” namely “the highest
principles of things, and the eternal causes of all things.”61 Indeed, this fol-
lows from its very nature:
What then is Scientia? The comprehension of divine things by certain rea-

son [divinarum rerum certa ratione compraehensio.] (Summary of Plato’s
Theaetetus, in Divini Platonis Opera Omnia, 93).
Opinio (the standard Latin translation for doxa), meanwhile, is confined to

perceptible things—it is “the general comprehension” (notionem commune)
of the perceptible by contrast with the intelligible, and again this is due to its
very nature: as on Proclus’ view, its role is “to judge [iudicare] the things
which the common sense perceives through the five organs, not only what
they are like, but what they are” (ibid, 92).
The Neoplatonic interpretation of Plato kept its hold for centuries, and
with it a consistent embrace of Distinct Objects. Let us jump ahead then to
another major era of Plato scholarship, the extensive German scholarship of
the nineteenth and early twentieth centuries. Here we do find attention to
one worry that motivates some modern Overlappers: how to reconcile the
Distinct Objects view of the middle dialogues with the Meno’s implication
that doxa and epistêmê share objects.62 Nonetheless, we also find widespread
commitment to the view that those middle dialogues do indeed embrace
Distinct Objects. Here for example is Zeller, commenting on the Timaeus:
[Plato agrees with Parmenides that] only Being as such can be

known . . .Therefore what is thought must be as distinctly separated from
what is represented [dem Vorgestellten – i.e. the doxaston] as thinking is
from representing [der Vorstellung – i.e. doxa]. (Zeller 1920, 152)
Finally, suppose you are learning about Plato by reading the great flowering
of commentaries and discussions in the late nineteenth and early twentieth
centuries in the Anglophone world. You will find stark, unqualified
61 “Habetur autem Scientia de iis quae semper eadem, et eodem modo sunt: huius modi sunt
summa rerum principia, et aeternae rerum omnium rationes.” (Summary of Plato’s Theages,
Divini Platonis Opera Omnia, 6.)
62 Because a true doxa about something, once tied down, becomes epistêmê—plausibly
about this very same thing (98a). For discussion of this passage, and my own attempt to recon-
cile it with the Distinct Objects interpretation, see Chapter 9.
statements of the Distinct Objects view. Here is Cornford, using the term
‘Two Worlds’ to characterize the Republic’s metaphysical view, and then
moving unapologetically to a matching epistemology:
Corresponding to the two worlds, the mind has two faculties: Knowledge
of the real and Belief in appearances [doxa]. . . . Knowledge, by definition,
is of unique, unchanging objects. . . . Doxa and its cognates denote our
apprehension of anything that seems: . . . . what seems to exist, sensible
appearances, phenomena [and only these, i.e. doxa does not apprehend
Forms]. (Cornford 1941, 180–1)
Or, typically of the period, here is Adam commenting on the powers

argument:
The εἴδη [Forms] provide objects for Knowledge, as opposed to Opinion,

and they are capable of being known. (Adam 1902, comment ad 476a)
He adds nothing about the Forms being opinable in an indirect or deriva-

tive way, nor anything about perceptibles being knowable in any way.
Moreover, the statement is quite casual: he does not treat it as in need of
special explanation or excuse. Nor does he, nor do others of his era, make
any note of the various textual obstacles to the Distinct Objects reading
flagged by Fine and others: he is silent about the apparent reference to doxa
of Forms at 506c–e, and equally about the apparent reference to epistêmê of
perceptibles at 520c. We find the same pattern in other commentaries of the
era.63 In multiple commentaries, alive to all kinds of textual ambiguities and
philosophical questions, the issues that motivate the Overlap reading go
almost entirely unnoted. (I have found one exception: Murphy (1951, at
123–4) argues that there must be doxa of Forms, using the same kind of
arguments we find in Vogt 2012.)
Moreover, in these commentaries, as in the others surveyed in this sec-
tion, we also find very specific interpretations of epistêmê and doxa as spe-
cial states by nature confined to a particular kind of object. I detail some of
these interpretations in Chapters 5 and 8, but briefly: epistêmê, often trans-
lated as ‘Science,’ is treated as something like a priori knowledge of ultimate
reality; doxa, often translated as ‘Opinion,’ is treated as something like
63 See Jowett and Campbell 1894, Bosanquet 1895, Shorey 1930.

experience-based, atheoretical thought about particular perceptibles. Far

from treating the implications of Plato’s Distinct Objects as something to be
explained away, these commentators seem to regard it instead as a revealing
guide to what epistêmê and doxa in fact are.
On the other hand, suppose you are learning about Plato’s epistemology
by reading Plato scholarship in the analytic tradition in the 1960s and
1970s. For the most part, you will still be innocent of any doubts that Plato
held a Distinct Objects view. You will however encounter doubts about the
viability of the view. In 1962 Crombie discusses disagreement among
scholars about whether Plato had any good reason to hold the view, where
all parties to the debate assume that he did in fact hold it (vol. 2, p. 34).
Scholars are starting to see the view as problematic.
From there the ground is laid for new interpretations of familiar texts, on
which the view is not in fact Plato’s (Gosling 1968, Fine 1978). By 1997 an
article can refer to the Distinct Objects interpretation as “outrageous”
(Baltzly 1997, 240).
Thus doubts about the Distinct Objects interpretation, so pervasive now,
are a very modern phenomenon indeed. Why does the Distinct Objects
view suddenly start to seem problematic in the middle of the twentieth cen-
tury? I have already suggested an explanation above: our contemporary
epistemology is an Overlap epistemology. Its star players, knowledge and
belief, are not distinguished by their objects. Thus we are imposing our own
epistemological views on Plato.
As confirmation for this explanation, it is worth realizing that Overlap
views in epistemology in general are a fairly modern innovation. If older
scholars found no problem attributing a Distinct Objects view to Plato, that
is at least in part because—I will argue in Section 4—such a view seemed to
them perfectly sensible and familiar.
4. Distinct Objects beyond Plato
On a popular interpretation, Parmenides holds a Distinct Objects epistem

ology—and indeed one very like the one I will attribute to Plato in this book.
There are two main cognitive kinds, nous and doxa. Nous is of Truth, or Being.
Doxa is of something quite different: the realm most mortals concern them-
selves with, which is in fact merely what seems (ta dokounta, B1.32).64
64 See for example Cornford 1933 and Palmer 1999. For discussion and defense of this
interpretation of Parmenides’ doxa, see Chapter 8.
Distinct Objects beyond Plato 43
More explicitly, we find a Distinct Objects view in Plato’s student, Aristotle:
The object of epistêmê [to epistêton] and epistêmê differ from the object of
doxa [to doxaston] and doxa, because epistêmê is universal and comes
through necessities. . . . [while] doxa is concerned with what is true or false
but can also be otherwise (Posterior Analytics 88b30–89a3)
Let it be laid down that there are two parts of the soul that have logos, one
by which we contemplate those things whose first principles do not admit
of being otherwise, and one by which we contemplate those [whose prin-
ciples] do admit of being otherwise: for in relation to things different in
kind, so the part of the soul related to each is different in kind, since
indeed cognition [gnôsis] belongs to the parts by virtue of some likeness
and kinship. Let one of these be called the part capable of epistêmê [epistê-
monikon] and the other the part capable of calculation [logistikon]…[The
second part is also called] the part capable of doxa [doxastikon], for both
doxa and practical wisdom [phronêsis] are about what can be otherwise.
(Nicomachean Ethics 1139a6–12; 1140b26–28)
According to the passage from the Ethics, there are two different types of
objects in the world, and these are contemplated by different powers. Things
that are necessarily as they are (such as universals, essences, or God) are
contemplated by a part of the soul responsible for epistêmê. Things that are
contingently as they are, and can change (particulars, accidental proper-
ties), are contemplated by a different part of the soul, the part responsible
for doxa. This looks to be a robust Distinct Objects view: that Aristotle has
in mind two classes of object, rather than for example two types of propos
ition, is confirmed by his redescription below of what can be otherwise as
things that can be done or made, and what cannot be otherwise as what is
eternal, ungenerated and imperishable.65 There is less clarity about how to
interpret the claim in the Posterior Analytics: perhaps the distinction here is
between types of propositions (necessary and contingent) rather than types
of objects.66 Even if so, however, Aristotle’s metaphysics entails that epistêmê
65 “Everyone believes that what we have epistêmê of [ἐπιστάμεθα] does not admit of being
otherwise . . . Therefore the object of epistêmê [ἐπιστητόν] is of necessity. Therefore it is eternal,
for the things that are of necessity without qualification are all eternal. And eternal things are
ungenerated and imperishable” (Nicomachean Ethics 1139b20–4). “In the class of what admits
of being otherwise, there is both what can be made and what can be done [ποιητὸν καὶ
πρακτόν]” (Nicomachean Ethics 1140a1–2).
66 Aristotle goes on to define (non-demonstrative) epistêmê as “belief [ὑπόληψις] in an
immediate proposition” and doxa as “belief in a proposition that is immediate and not neces-
sary” (Posterior Analytics 88b37, 89a3–4).
will therefore only be about certain kinds of objects—objects about which

there can be true substantive necessary propositions, such as essences and
universals.67
Aristotle thus clearly presents a version of the Distinct Objects view that
scholars have been so reluctant to attribute to Plato. What is more, he tells
us that these Distinct Objects views are widely held:68
Nobody thinks that they have a doxa [doxazein] when they think it impos-
sible for it [the thing under consideration] to be otherwise, but rather that
they have epistêmê [epistasthai]. (Posterior Analytics 89a6–8)69
Everyone believes that what we have epistêmê of does not admit of being
otherwise (Nicomachean Ethics 1139b19–21)
Many people deny that there is epistêmê of perceptibles. (Topics 114a23)
Why is there no “Two Worlds” debate about Aristotle, as there is about

Plato? Surely in large part because he so clearly characterizes epistêmê as a
very specialized state, and one that by its nature cannot be of contingent
facts or truths. Epistêmê is achieved when we deduce necessary truths from
necessary first principles (see Nicomachean Ethics VI.6 and Posterior
Analytics I.2).70 Given this explicit account, there is no temptation to
assume that he has in mind ordinary knowledge, and likewise no tempta-
tion to insist that he must mean to allow epistêmê of perceptibles. (His
account of doxa is less clear and also less studied; the fact that he treats doxa
as one species of a wider genus, hupolêpsis, where the latter clearly has much
in common with belief, has perhaps taken some of the pressure off of his
doxa to play that role and made interpreters happier to think of it as some-
thing more specific.)71
What about Aristotle’s claims that these Distinct Objects views are wide-
spread? We have some independent evidence to support these. Consider an
67 As Fine acknowledges in a recent paper on the Posterior Analytics (Fine 2010). Proclus
interprets Aristotle the way I suggest: epistêmê is of the stable unchanging essences in an object,
doxa of the shifting particular properties: In Rem publicam 1.260.29–264.20.
68 Some hold that in the passages quoted below Aristotle means “nobody” and “we all” to
refer only to members of the Academy; this would still establish that he finds the view in Plato,
but the passages from Isocrates presented below support unqualified readings.
69 This claim on its own leaves room for doxai about necessary matters that one does not
recognize as necessary, although the other passages just quoted seem to rule out even this.
70 Aristotle does often use ‘epistêmê’ in what is clearly a looser sense, on which for example
paradigmatic crafts like medicine count: see for example Nicomachean Ethics 1138b26–32.
71 See Moss and Schwab 2019.
argument we find in Plato’s contemporary, Isocrates. There is no epistêmê of

political affairs, nor indeed of any practical matters (“what we should do or
what we should say”), because in realms where individual cases vary widely,
and so success depends on finding the “right occasion” (kairos), there can
only be better or worse doxai:
Since is not in human nature to attain an epistêmê by having which we can

know [eideimen] what should be said or done, for the rest I consider wise
those who are able by their doxai to hit on what is best for the most part, and
I consider those who occupy themselves with the studies from which they
will most quickly attain that kind of wisdom [phronêsis] to be philosophers.
(Isocrates, Antidosis 271; cf. 184)
Isocrates certainly does not share Plato’s metaphysics, but he too argues that
the domain of epistêmê is sharply restricted, with the contingent realm left
to doxa. Certainly however there is no “Two Worlds” debate about
Isocrates—not just because he is less famous, but also because it is clear that
he has in mind by epistêmê a specialized state, rather than ordinary know
ledge: something like expertise. (Arguably he also has in mind by doxa
something different from ordinary belief: developed theory.)72
Evidently then Aristotle was right that his own Distinct Objects view was
not a radical one, at least as far as the restriction on epistêmê is concerned.
Moreover, it was a view with lasting appeal for the Greeks: three centuries
later we find Dionysius of Halicarnassis saying that no-one has defined a
technê of the right occasion (kairos), for this “cannot in general be captured
by epistêmê, but by doxa.”73
Distinct Objects epistemologies are also widespread among Plato’s self-
declared followers, the Platonists.
We see this already in the first Platonists. Xenocrates, head of the
Academy toward the end of the fourth Century bce, is reported to have
divided all reality into three kinds of objects: the objects of epistêmê, those
of doxa, and those of perception (Sextus, Adversus Mathematicos 7.147–9).
We do not have enough resources to understand the view in detail, and in
particular to rule out that he held a moderate Overlap view, but we also
have nothing to suggest it.
72 For this view of doxa see Poulakos 2001.

73 οὐδ᾽ ὅλως ἐπιστήμῃ θηρατός ἐστιν ὁ καιρὸς ἀλλὰ δόξῃ, De Compositione Verborum 45.17.
Or consider Plutarch, a middle Platonist. Above we saw him endorse a dis-

tinction between the unchanging, imperceptible noêton and the shifting,
perceptible doxaston, in attributing that distinction to both Parmenides and
Plato (Adversus Colotem 13). He defends this kind of view elsewhere too: see
for example De Iside et Osiride 383c–d, which contrasts the “pure and simple
noêton” with the “mixed and varied doxasta.” Plutarch scholars take such
passages to reveal a stark Distinct Objects view: as one puts it, it is one of
the “basic assumptions” of Plutarch’s epistemology that “sense-perception . . .
results in mere opinion [δόξα] about the always changing world, and real
knowledge or intellection on the other hand is restricted to the world of
Forms” (Sierksma-Agteres 2015; compare Boys-Stones 1997 and Scott 1987).74
It seems that Plutarch did not hold even a moderate Overlap view: the claim
is not that we can have noêsis of perceptibles in a derivative or deficient way,
but simply that we could not have noêsis of perceptibles at all.
As for the Neoplatonists, their intricate and complex epistemology is very
clearly concerned to separate the objects of the highest cognitive kind from
those of doxa. Proclus not only attributes to Plato a Distinct Objects epistem
ology in the Republic and Timaeus, but endorses the view himself: see the
passages quoted above, and the entire discussion of Timaeus 28a in his In
Timaeum.75 Plotinus tells us that “the name of doxa fits” the study of percep-
tibles, while epistêmâi “of intelligibles, which are really epistêmai, come from
nous to the rational soul and do not think any perceptible thing” (Ennead
V.9.7; cf. VI.9.3).76 As for nous itself, it is not only of the intelligible Forms
but in fact identical with them (see especially Ennead V.5.8). There is no
more exclusive relation possible between cognitive kind and object. We saw
some Overlappers argue that Plato’s highest cognitive kind is typically of
Forms but atypically of perceptibles, or by nature of Forms but derivatively
or deficiently of perceptibles, but no such move is possible on this view: a
thing cannot be identical to something other than itself, not even atypically
or derivatively or deficiently; thus there can be no nous of perceptibles.
74 Boys-Stones argues that Plutarch follows Plato in thinking that “the ontological reliance
of the sensible world on the realm of the divine means that the former can be used as an epis
temological ‘tool’ in the quest for knowledge of the latter—but is not an object of true know
ledge itself . . . the sensible world is just intrinsically unknowable…[by contrast with] the divine
realm, where alone true knowledge is to be found” (Boys-Stones 1997, 228, 230).
75 For discussion see Lautner (2002) 260–1.
76 Αἱ δὲ ἐπιστῆμαι ἐν ψυχῇ λογικῇ οὖσαι αἱ μὲν τῶν αἰσθητῶν—εἰ δεῖ ἐπιστήμας τούτων
λέγειν, πρέπει δὲ αὐταῖς τὸ τῆς δόξης ὄνομα—ὕστεραι τῶν πραγμάτων οὖσαι εἰκόνες εἰσὶ τούτων·
τῶν δὲ νοητῶν, αἳ δὴ καὶ ὄντως ἐπιστῆμαι, παρὰ νοῦ εἰς λογικὴν ψυχὴν ἐλθοῦσαι αἰσθητὸν μὲν
οὐδὲν νοοῦσι.
Thus Plato’s followers not only take him to hold Distinct Objects views,
but also develop such views of their own. They carve up reality into distinct
classes and correlate these with distinct cognitive kinds; in so doing they
pay no attention to the worries that motivate Overlap readings.
Moreover, they give specialized accounts of the cognitive kinds which
make them look quite different from modern knowledge and belief. In
Plotinus, for example, noêsis or nous is direct union with intelligible objects,
while doxa is passive reception (Ennead V.5.1.63–66).
Nor are such views confined to Platonists. The restriction of the highest
cognitive kind to intelligibles survives, with modifications, through the
medievals’ theory of Scientia; it retains its hold throughout the early mod-
ern period, with Aquinas, Spinoza, Leibniz, Locke, Descartes and others
explicitly confining the highest kind in their system to necessary truths.77
We much more rarely find anything like the restriction of doxa to percep
tibles, but Hume is a notable exception: he embraces a full-blown Distinct
Objects view, on which knowledge is restricted to necessary truths (“rela-
tions of ideas”), and belief to the wholly distinct class of contingent affairs
(“matters of fact”).78
In short, there is a long philosophical tradition according to which a
superior cognitive kind—nous, epistêmê, scientia, scientific knowledge,
knowledge—is sharply distinguished from an inferior kind, doxa, opinio,
opinion, or belief; it is a firm part of this tradition that the superior kind is
restricted to a special kind of object, and it is a less prevalent but significant
part of this tradition that the inferior kind is restricted in its objects too, so
that there is no overlap between their objects.
My aim in this brief historical survey has been to show just how unsur-
prising it would be to a reader of Plato in his own day, or of any day up until
very recent ones, to find him distinguishing epistêmê and doxa by their
objects. The modern view that superior and inferior cognitive kinds can
obviously overlap in their objects turns out to be very modern indeed, and
77 Locke’s “sensitive knowledge” is arguably an exception, although his view of science fits
this description. For citations and defense of this reading of Spinoza, Leibniz, and Descartes,
see Carriero 2013; for the path from Aristotle to Hume see Morris and Brown 2001, Section 5,
and more generally for the path from Aristotle to early modern Scientia see Pasnau 2017
Chapter 2.
78 For Hume’s distinction (sometimes referred to disparagingly as Hume’s Fork), see espe-
cially Treatise I.iii.1. His Distinct Objects view is probably subtler than Aristotle’s, or the one I
am imputing to Plato; it would be an interesting project to see in detail how they compare. For
one thing, while Hume clearly distinguishes the objects of knowledge and belief, he is very
loose with the corresponding verbs. Then there is the question of exactly what sort of overlap
he will allow.
this should make us suspicious of the assumption that Plato shared it.
Indeed, the arguments of this section should open us to the possibility that
Plato’s epistêmê and doxa have considerably less in common with know
ledge and belief as nowadays conceived than they do with some of the states
we have seen canvassed in this section—that Plato’s epistemology is much
more like Aristotle’s or Hume’s than like our own.
5. A new starting point
Only around 1960 did scholars start to agree that there are glaring problems
for the Distinct Objects interpretation of Plato’s epistemology; before that it
was the default, usually accepted without argument. Moreover, many philo
sophers held similar epistemologies themselves. A dramatic shift occurred:
an epistemological view that was once widespread in its own right, and
almost unquestioningly attributed to Plato, is now regarded as beyond the
pale. The idea that a Distinct Objects epistemology is a non-starter is a thor-
oughly modern idea; the idea that Plato cannot have held one is radically
ahistorical.
Moreover, we can identify one major driver of the shift. The dominant
approach to Plato’s epistemology in recent years begins from the assump-
tion that epistêmê and doxa are more or less what we call knowledge and
belief, uses our views about these to constrain our interpretation of Plato,
and concludes that the Distinct Objects interpretation is absurd. We have
seen some evidence already, however—and will see more in Chapters 5
and 8—that the same scholars who accepted the Distinct Objects interpret
ation without question were working with accounts of cognitive kinds
which look quite different from contemporary knowledge and belief.
Perhaps then it is not that for two millennia no-one noticed the oddness of
distinguishing knowledge and belief by their objects, but instead that
scholars took Plato to be investigating something else.79
79 Or more precisely, something understood so differently from how people nowadays
understand knowledge and belief that even if they are in fact identical, modern intuitions are a
poor guide. See again my caveat in the Introduction: on some theories of concept identity, or
some theories of knowledge and belief, these older interpreters were still talking about know
ledge and belief but giving very different accounts of them than most philosophers give today;
on other theories, they were talking about something else. I return to this question in Chapters
5.4 and 8.4.
A new starting point 49
Were older interpreters on the right track with their accounts of Plato’s
epistêmê and doxa? I will argue that in many cases they were. We need how-
ever to develop a systematic way of understanding Plato’s epistemology.
What we have seen in this chapter suggests that we have good reason to
reverse the current dominant approach: rather than assuming our own epis
temo logic
al views as a starting-point, we should assume the Distinct
Objects interpretation, and use it to guide our investigations into the nature
of Platonic doxa and epistêmê. That is the plan of this book.
2
Plato’s Objects-Based Epistemology
What happens when we read Plato’s epistemology without in any way trying
to avoid a Distinct Objects interpretation? What we find, I argue in this
chapter, is ample evidence not only for that interpretation, but also for a
stronger one.
Looking both at Republic V’s powers argument and at other discussions of
powers, we find evidence for three crucial claims about the relation between
powers, their accomplishments, and their objects. First, powers and their
accomplishments are individuated by their objects (Section 1). Second, the
rationale for this thesis comes from a stronger one: powers and their accom-
plishments are defined by their objects: it is the objects that make them what
they are (Section 2). Finally, in the case of cognitive p
owers, the accomplish-
ments—understood as occurrent cognitions—inherit their character from
their objects: cognition works on the principle Aristotle calls “like-by-like”
(Section 3). Thus Plato’s epistemology is thoroughly objects-based. Epistêmê
is, in essence, the kind of thing one can have only in relation to a special kind
of object; doxa is, in its essence, the kind of thing one can have only in rela-
tion to a different special kind of object.
A few notes are in order before I begin.
First, although I will be discussing Plato’s notion of powers, I do not aim
to give a full account of this notion, but instead focus on the relation
between powers and their objects; thus I leave open important questions
when they do not bear directly on this relation. (For example: Does he have
a consistent doctrine of powers across the dialogue? Are some powers sub-
ordinate to others, and if so how? What is the relation between powers and
parts of soul? In particular, how are the powers of epistêmê and doxa as
described in Republic V’s powers argument related to the single inborn
power mentioned later on (518c), which looks like a more general cognitive
faculty?1)
1 I discuss this last issue briefly in Section 3. Szaif 2007 takes Republic 518c to show that
there is just one cognitive power, which can be used either to cognize Forms or to cognize
perceptibles, with the result that doxa and epistêmê are not genuine powers on a par with sight
DOI: 10.1093/oso/9780198867401.003.0003
Plato’s Objects-Based Epistemology 51
Second, as throughout this book, I use the English words ‘cognition’ and
‘cognitive’ to characterize both epistêmê and doxa. I do not mean to beg the
question of what epistêmê and doxa are, nor—when I go on to say that
both powers’ accomplishments are cognitive—to violate the principle I will
ascribe to Plato below, that different powers must have different accom-
plishments. Instead, I use ‘cognition’ to refer to something for which Plato
has no single term: the genus of which both doxa and epistêmê are species.2
Third, the main evidence for the “like-by-like” interpretation of Plato’s
epistemology comes from the Two Worlds dialogues, those that distinguish
a realm of intelligible Forms from the realm of perceptible particulars. Thus
here, as throughout the book, I am trying to show that Plato’s epistemology
is best understood as taking full shape in the Two Worlds dialogues, with
anticipations and echoes in the earlier and later dialogues.
Finally: the strategy of this chapter is to assume the viability of the Distinct
Objects interpretation, in order to clear the way for the main project of this
book: an objects-based account of epistêmê and doxa. Nonetheless, many of
the arguments I give serve as further evidence for the Distinct Objects inter-
pretation. Against various Moderate Overlap views, including the Content
Overlap view, we will see Plato using objects to distinguish the cognitive
accomplishments just as much as the powers themselves, and we will also see
strong suggestions that the relevant objects are subject matters. Against any
form of Overlap view, the evidence that cognitions resemble their objects
entails that doxa can only be of things that resemble doxa, while epistêmê
can only be of things that resemble epistêmê; given how different the two
cognitive kinds are, on Plato’s view, this means that they can never share
objects. Thus this chapter will indirectly support the Distinct Objects
interpretation; a full defense of that interpretation must wait however until I
have addressed the putative evidence for epistêmê of perceptibles (Chapter 4.5)
and doxa of Forms (Chapter 7.6).
or hearing. Plato might however hold that doxa and epistêmê, while subordinated to a more
general power, are genuine powers themselves. Perhaps there is one main cognitive faculty,
which in turn has the powers of epistêmê and of doxa; which of these it exercises depends on
the available objects (compare Aristotle’s distinction between the common and special senses,
on some interpretations—e.g., Everson 1997). Or perhaps the power to cognize in general is
inborn, but must be developed to become the power of epistêmê; what results is nonetheless a
genuine power, just as crafts are powers.
2 The powers argument classifies both as powers (dunameis, 477b), while the Divided Line
passage speaks of affections (pathêmata, 511d), but these terms are both far too broad to get at
what specially unites epistêmê and doxa—for example, Plato will elsewhere call strength a
power, and anger an affection.
52 Plato ’ s Objects-Based Epistemology
1. Powers and their accomplishments are

individuated by their objects
We can begin with the famous powers argument of Republic V. The conclu-
sion of the argument is that the sight-lovers, since they attend to perceptible
beautiful things but not the Form, have only doxa, not epistêmê. I gave a
brief summary of the argument in Chapter 1.1; here are the crucial lines on
the relation of powers, accomplishments, and objects:
Do we say that powers [dunameis] are one class of the things that are, by
which we have the power to do the things we have the power to do, and
everything else has the power to do what it has the power to do? I mean
for example sight and hearing are among the powers, if you understand
the kind I mean . . . In a power I see no color or shape or anything of that
sort, as one does in many other things, looking toward which I can define/
distinguish [diorizomai] for myself different ones from one another. In a
power I look toward this thing only: what it is over and what it accomplishes
[eph’ hôi te esti kai ho apergazetai], and in that way I call each of these a
power. And what is set over the same thing and accomplishes the same thing
I call the same power, while what is over something different and accom-
plishes something different I call a different power. . . Do you say that
epistêmê itself is a power, or what class do you put it in? – In that class, he
said, and of all powers the strongest. – And what about this, will we refer
doxa to [the class of] powers or to some other kind? – Not at all to another
kind, he said; for that by which we have the power to form doxa [doxazein]
is no other than doxa. (Republic 477c–e, emphases added)
What distinguishes one power from all others is (a) “what it is over” and (b)
“what it accomplishes.” What does Plato have in mind?
We will have to determine what it means for cognitive powers to be
“over” things, but for now let us leave the question open while following
convention in using the term ‘object’ for a power’s relatum, whatever kind of
relation Plato has in mind. As for the accomplishments, Plato here explicitly
says that doxa is the power by which we doxazein (the verbal form of doxa—
having or forming doxa); in the course of identifying epistêmê’s object below
he says that epistêmê “is set over what is, to have gnôsis [gnônai] of it as it is”
(478a); he nowhere else names anything that could plausibly be either
powers’ accomplishment. ‘What a power accomplishes’ is thus evidently
Powers and their accomplishments 53
simply a way of referring to what it does (compare the equation at Hippias

Minor 373d–e of what something does (poiei) with what it accomplishes
(ergazetai)), and Plato’s thought seems to be that what these particular
powers do is simply get exercised. Certainly in some cases he thinks that a
power’s accomplishment is a distinct product. See especially Charmides
165c–d: medicine, defined as “epistêmê of the healthy,” “accomplishes a fine
accomplishment, health.”3 But his idea in the Republic’s powers argument is
a plausible one: epistêmê and doxa without qualification (by contrast with a
technê like medicine, although Plato does sometimes call this ‘epistêmê’)
accomplish in the first instance nothing but their own exercises.4 One might
think that Plato should distinguish between the activity of cognizing on the
one hand, and the resulting cognitive state on the other, and reserve “accom-
plishment” for the latter, but I see no evidence that he finds that distinction
important in this context, or even tries to track it. In what follows, then,
I will speak indifferently of powers’ second individuating criterion as
accomplishments and as exercises.
Thus each power is individuated by two things: its object and its accom-
plishment. One might think that these two criteria are independent of one
another, and therefore that two powers could share the same object, while
being distinguished by their accomplishments. One might even argue that
Plato’s sole examples support this reading: sight and hearing are set over the
same things—material objects—but accomplish different things, sight see-
ing them while hearing hears them. There are however clear signs that this
is not what Plato intends. (Indeed, this is widely recognized, and the argu-
ment is sometimes accused of invalidity on these grounds.)5 Not only does
he refer to the distinguishing criteria as “this thing [singular] only [ekeino
monon]” (477d), as if the two criteria are effectively one, but in the ensuing
argument he twice stresses that a difference in powers entails a difference in
3 τὴν γὰρ ὑγίειαν καλὸν ἡμῖν ἔργον ἀπεργάζεται. Compare Hippias Minor 375d–376a, where
justice, defined as a power, is that through which a soul accomplishes (ἐργάζεται) fine and
shameful things.
4 This distinction between different kinds of accomplishment is one that Aristotle makes
explicit: some activities have a function or accomplishment (ergon, the root of the verb I trans-
late ‘accomplish’) distinct from the activity, while in other cases the ergon is the activity itself
(Nicomachean Ethics 1094a4–5).
5 See for example Balztly 1997. Fine takes this worry to be insurmountable for the reading
on which powers are set over objects, rather than propositions (1990).
object, clearly ignoring the possibility of distinct powers doing different

things with respect to the same object:
We agree that epistêmê is different from doxa? – Different. – So each is

by nature over [epi] something different, since each has the power for
something different?6 – Necessarily. . . . If different powers [dunameis]
are by nature over different things, and both of these are powers, doxa
and epistêmê, and different from one another, as we say – from these
things it follows that the same thing cannot be the object of gnôsis and
of doxa.7 (Republic 478a–b, emphases added)
Evidently Plato is assuming that there is no accomplishing different things

with respect to the same object.
We can make sense of this assumption, and render the argument valid, if
we construe the notion of object very narrowly: for example, the object of
sight is color, rather than material objects.8 Indeed, when at 507c–e Plato
again refers to sight and hearing as powers, he explicitly identifies the object
of sight, the visible (to horaton), as color—what Aristotle will call sight’s
special (idion) object, the one exclusive to it. At 524a he refers to the objects
that each power is “set over” (tetagmenê epi)—the same language he used
for the relation between power and object in our passage—and again has in
mind special objects: touch is over the hard, soft, heavy and light (524a). In
the Timaeus hearing is defined as a motion caused by sound (67b; cf. 67c on
sight and color); the Theaetetus goes further, arguing explicitly that the only
perceptibles are special objects like color and sound (184e–5a). In the
Charmides, Socrates lists sight and hearing as powers (dunameis), and says
that “hearing is the hearing of nothing other than sound,” while sight can
only be of color (168d). Each of these powers, then, has its own object; there
is no crossover between them.
Note that on this view it is perfectly acceptable to speak, as Plato
sometimes does, of one and the same physical object as being both visible
and audible: the idea would be that it achieves this by having both a color
and a sound.9 The claim is that the objects of different powers are distinct,
6 ἐφ’ ἑτέρῳ ἄρα ἕτερόν τι δυναμένη ἑκατέρα αὐτῶν πέφυκεν.

7 ἐκ τούτων δὴ οὐκ ἐγχωρεῖ γνωστὸν καὶ δοξαστὸν ταὐτὸν εἶναι.
8 Crombie suggests that Plato is thinking of objects as “internal accusatives”: “the internal
accusative of ‘see’ is ‘sights’, of ‘smell’ is ‘smells’; and it is true that sight and only sight sees sights
and nothing but sights” (1962, vol. 2, 57). I hope to show that my reading is better supported by
Plato’s examples.
9 There are further questions about how to apply the objects-individuation theory to per-
ception, which Plato leaves under-determined. First, might he instead think that substances
on-identical; this does not entail that they will always be separate. In
n
principle we might find that the objects of epistêmê and doxa can be com-
present in the same objects too; indeed I shall argue in Chapter 9 that Plato
hints at this view in the Meno, although he sharply rejects it with his Two
Worlds metaphysics.10
What about the accomplishments of powers: can they share objects, as on
various Overlap interpretations canvassed in Chapter 1? Plato strongly indi-
cates that this is not his view, for the Republic’s powers argument moves
without argument from claims about the objects of powers to claims about
the objects of their accomplishments.11 After first stating that each distinct
power is over something distinct and accomplishes something distinct, he
says that “Epistêmê is over what is, to have gnôsis of [gnônai] it as it is”
(478a): both power and accomplishment are directed toward what is. In
demonstrating that the power of doxa is not over “what is not,” he argues
that “it is impossible to have doxa of [doxasai] what is not . . . Doesn’t the
one who has doxa [doxazôn] bear his doxa over something?” (478b): the
verb doxazein surely refers to the power’s accomplishment (cf. 478c), and
here the noun doxa very plausibly does too.12 Whatever the power of doxa
is over, so too is its accomplishment. Finally, in summing up the powers
argument, he makes Distinct Objects claims about the accomplishments:
“Those who contemplate the many beautifuls . . . have doxa of [doxazein]
everything and have gnôsis of [gignôskein] nothing of the things they have
doxa of,” while those who contemplate the Beautiful Itself and other Forms
“have gnôsis [gignôskein] but do not have doxa [doxazein]” (479e). The
claim seems to be a very stark one indeed: it is not just that the power of
doxa is suited to perceptibles, but that if you attend to perceptibles you will
like Socrates or a finger are not strictly speaking perceived at all, but instead the objects of
another power—perhaps doxa? That is sometimes taken to be the lesson of Theaetetus 184–7;
Proclus for example applies it more widely (see his comment ad Timaeus 28a at In Timaeum
2.248.16–17.) Second, what about the properties Aristotle calls common perceptibles, like size
and shape? At Republic 523e Plato says that sight sees bigness and smallness (to megathos, hê
smikrotatêta), just as touch feels thickness, thinness, hardness and softness. Does he notice that
big and small are commons, while the others are propers? I suspect not. The same issue arises
at Theaetetus 184–7: does he mean to deny that we perceive size, shape, motion? He seems
oblivious of the issue. Were he to try to fit common perceptibles into his theory of powers, he
might do it by saying that we directly perceive the propers, and thereby indirectly become
aware of the commons.
10 Annas 1981 attributes a version of this view to Plato in the Republic: one can have
epistêmê of Socrates’ humanness, but not of his justice.
11 Thanks to Damien Storey for pointing out the significance of these passages.
12 Or perhaps more precisely, to the product of an exercise—the activity of doxazein yields a
doxa—but in any case not to the power.
only accomplish that power’s accomplishment, and mutatis mutandis for

epistêmê and Forms.
There are no other discussions in Plato which explicitly lay out this
objects-individuation theory of powers, but there are several others which
discuss the same phenomena which Republic V refers to as powers, some-
times using that label explicitly, and they fit this objects-individuation
model very well.
Consider first a discussion mentioned above in the Charmides, particu-
larly relevant for us because here as in Republic V Plato classifies epistêmê
and doxa as powers alongside sight and hearing. Socrates is arguing against
the hypothesis that temperance is “epistêmê of epistêmê” by showing that
each power is only of some particular other object, not of itself. Sight is of
color (not of sight), hearing is of sound (not of hearing), and the list con
tinues: appetite is for pleasure, wish is for the good, erôs is for the beautiful,
fear is of the terrible, and doxa is “of the things that doxai are of ” (hôn doxa-
zousin); thus it would be very strange if some epistêmê were of epistêmê
itself, rather than of what-can-be-learned (to mathêma) (Charmides
167e–168a). Summing up the argument a few lines later he uses the term
“dunamis” to label the category he has been discussing (168d). We can then
generalize the point as follows: each power is exercised in relation to (pros)
its special object (168d), or is of (objective genitive) its particular object
(168a).13 Moreover, this includes epistêmê and doxa: although the charac-
terization of their objects is minimally informative, it at the very least leaves
open that these objects are distinct, and the fact that all the other members
of the list have proprietary objects suggests that these do too.
The Charmides’ argument does not explicitly mention anything corres
ponding to the Republic’s “what the power accomplishes,” but it is very
natural to read it as saying that the accomplishments too, not only the
powers, are individuated by the correlated objects: surely Plato thinks not
only that the power of appetite is related to the pleasant but that so too is
what the power accomplishes—the activity of having appetites; likewise
for sight and color, and so on. Indeed in the characterization of doxa, an
accomplishment-term, the verb doxazein, is built into the specification of
the power’s object (168a).
13 Anything to which the power applies must have “that being [ousia] in relation to which
the power is” (τὴν οὐσίαν πρὸς ἣν ἡ δύναμις αὐτοῦ ἦν) (168d); throughout the discussion Plato
usually uses the genitive, e.g. appetite is of pleasure (ἡδονῆς, 167e), fear is “of fearful things”
(τῶν δεινῶν, 168a).
A brief digression is in order here before we move to other passages. The

Charmides’ examples suggest an important distinction between two kinds
of power-object relations which Plato does not discuss. Sight can only be of
what really is colored: if you think mistakenly that the air is colored you do
not thereby see it. We might naturally think that fear is by contrast of what
one thinks terrible, whether or not it really is: if you think that an innocuous
spider is dangerous you will fear it; if you are unaware that a lethal lollipop
is dangerous you will not fear it. There are various ways of theorizing this
distinction: some powers are related to objects de re and others de dicto, or
some powers’ objects are defined extensionally and others intensionally. It is
natural to ask whether Plato recognizes this distinction, and how it bears on
his understanding of cognitive powers’ objects. As to the first question, I do
not have room here for a full treatment, but it is worth noting that he very
much seems to reject the distinction when it comes to desire: wish (boulêsis)
is not for what seems good, but what is good (Gorgias 467b–468d); people
desire and have appetites for what is really good, not for bad things they
think good (Meno 77a–78a). In this case at least, Plato rejects our intuition
that a power’s object should be defined de dicto. He does not explicitly con-
sider the question for other powers such as fear, and perhaps there is some-
thing special about desire, or the good, that makes this case different. What
about epistêmê and doxa? My account of their objects, in Chapters 3 and 6,
will entail that here at any rate the objects are defined de re. There is some-
thing about certain things (Beings, I will argue) that makes them possible
objects of epistêmê, and something about certain other things (perceptible
things, I will argue) that makes them possible objects of doxa. One can
mistakenly think that one is in touch with an object of epistêmê, and thus
mistakenly think that one has epistêmê (as do the sight-lovers in Republic V),
but thinking does not make it so.
Now let us consider another passage which proves a useful source for
Plato’s theory of powers, although he does not here use that term. I have in
mind Republic IV’s discussion of how various phenomena, including ones
the Charmides characterizes as powers, are related to objects. The context is
the argument for the division of the soul. Socrates wants to show that appe-
tite can conflict with rational desire, and his argument depends on the claim
that “thirst itself ” is for drink rather than for good drink. He begins by indi-
viduating species of appetite by their objects: thirst is for drink, hunger for
food, and in general each appetite is for “just what it is of by nature” (houper
pephuken) (437e). Plato does not here use the word dunamis, but appetites
were on the Charmides’ list of powers, and it is fairly easy to see how to
make the examples here fit Republic V’s two-part formula: the power of
thirst is the power to feel thirst (what it accomplishes) for drink (what it is
over). Here too then, both powers and accomplishments are individuated
by their objects.14
Indeed Plato goes on to make the correlation even more fine-grained:
different kinds of drink—hot drink, cold drink, much drink—correspond
each to their own kind of thirst (“If some heat is added to the thirst, it would
render it an appetite for cold [drink]” (437d–e)). Is the idea here that the
power of thirst for cold drink is different from the power of thirst for hot
drink—or instead that there is just one power, thirst, which can be activated
in different ways, yielding different accomplishments set over different
objects? The latter sounds more plausible. This would mean that we here have
a case in which the accomplishments are more finely object-individuated
than the powers. On the other hand, perhaps Plato thinks that to have the
power of thirst entails having other powers too: the power to thirst for cold
drink in the appropriate circumstances, and so on. I will not try to settle this
question, since I think doing so would involve pressing Plato’s theory farther
than he has determined it. What I want to emphasize is that this passage in
Republic IV extends the objects-individuation principle—and indeed the
fine-grained version of it—to epistêmê:
What about the epistêmai [pl.]? Isn’t it the same way? Epistêmê itself is
epistêmê of what-can-be-learned [to mathêma] itself, or of whatever one
should posit epistêmê as being of, while a particular epistêmê of a particu-
lar kind is of a particular something of a particular kind. I mean this kind
of thing: when epistêmê came to be of building houses, didn’t it differ from
the other epistêmai, so that it was called architecture? (Republic 438c–d)
Epistêmê has its own proprietary object, the same one we saw in the
Charmides: what can be learned. Moreover, for each particular kind of thing
14 Does Plato violate the one-to-one principle I’m attributing to him by assigning different
objects to appetite in different contexts: pleasure in the Charmides (167e), and things like drink
and food in the Republic? I think he more likely had varying thoughts about the best way to
specify the objects of appetites: perhaps thirst is for drink (Republic 437d), perhaps it is for the
pleasure of drink—which would accommodate the Charmides’ claim, and also fit well with the
description of the appetitive part at the conclusion of Republic IV’s argument as the “compan-
ion of indulgences and pleasures” (439d); perhaps instead, as he argues in the Philebus, thirst is
a desire for being filled with drink (35a). Plausibly then Plato considered various models, but
all support the view that the appetites do not overlap in their objects. If x is different from y
(whether x is drink, the pleasure of drink, or being filled with drink), the appetite for x must be
different from the appetite for y.
that can be learned there is a distinct kind of epistêmê. As in the powers

argument, different objects are correlated with different cognitive kinds.
Furthermore, these examples point in favor of the starkest reading of
Distinct Objects, and against the moderate view. There is no suggestion that
the accomplishments of distinct powers might share objects: surely an
accomplishment of the power of house-building is directed as exclusively
toward house-building as is the power itself. Moreover, if the object correl
ated with the power of epistêmê is “what-can-be-learned,” where the
examples are things like building houses, it is extremely natural to take the
object of the power to be just what the Content Overlap view denies: its
subject matter, what it is about.
We find a similar principle of objects-individuation at work in Plato’s dis-
cussions of particular epistêmai or crafts (technai). In several dialogues
Socrates argues that some practice is not a craft, or not an epistêmê, on the
grounds that this practice does not have its own special domain. For
example, in the Gorgias Socrates presses Gorgias to say what power rhetoric
has, on the assumption that if it is a craft it must have some power (Gorgias
447c);15 the ensuing discussion eliminates various accounts on the grounds
that they do not specify a unique province for rhetoric—they fail to identify
the unique thing that it is about (peri, 449d, 451d). As Benson puts it, in his
discussion of cognitive powers, “the assumption that underlies this stretch
of the dialogue is that since rhetoric is an expertise and so a dunamis, there
is some special object associated with it—what it is ‘about’ – and by which it
can be distinguished from other dunameis” (Benson 2000, 202). Note that
again the distinguishing object is the object of the accomplishment as much
as of the power (the question shifts from what rhetoric is about to what par-
ticular rhetorical speeches are about (see 451d)), and also that it is very
natural to take the correlated objects as subject-matters, what rhetoric is
about.16 The same theory is suggested by the brief discussion of sophistry at
15 Is it significant that epistêmê and doxa in the passages we have reviewed are powers, while
craft here has a power? I doubt that it is, especially since Plato elsewhere sometimes says that
craft is a power, and sometimes that epistêmê has a power (for good discussion see
Benson 2000). Even in the midst of the Republic’s powers argument, where he identifies doxa
and epistêmê as powers, Plato also speaks of “the power of each” (477b).
16 It is worth noting that we get a somewhat different picture in a passage discussing rhet
oric as a power, in the Phaedrus. Here Socrates claims that in determining the nature (phusis)
of anything, we must investigate “what power it has, to do what in relation to [pros] what, or to
undergo what by the agency of what?” (σκοπεῖν τὴν δύναμιν αὐτοῦ, τίνα πρὸς τί πέφυκεν εἰς τὸ
δρᾶν ἔχον ἢ τίνα εἰς τὸ παθεῖν ὑπὸ τοῦ. Phaedrus 270d)). This looks a good deal like Republic V’s
formula, with “do or undergo” corresponding to “accomplish.” When Socrates applies the for-
mula to rhetoric, however, he implies that its correlated object is the soul (it is the power to
the start of the Protagoras (312c–e). Painters, carpenters and harp-players

each make their students clever at speaking about the particular things
about which they themselves have epistêmê (peri houper epistasthai, 312e):
that is, each species of epistêmê has its own particular domain, and this is its
subject matter—the subject of its typical speeches. The Ion offers a similar
argument about rhapsody, and the same assumptions are prominent here
too: each epistêmê has its own domain, about which it is able to speak well
(Ion 537c–538a)—that is, its own subject matter (cf. Laches 199a). (For fur-
ther discussion see Section 4.)
This is far from an exhaustive survey of Plato’s treatment of powers and
their relation to their objects. I suspect that such a study would reveal that
he does not have a thoroughly consistent theory. In some places he may
countenance overlap between the objects of different powers. Moreover, he
sometimes characterizes the objects of the powers as properties or aspects
of other objects—colors, sounds, the pleasant, the fearful—while at other
times he characterizes the objects as practices (building houses), or as sub-
stances (bodies, for medicine). What I hope to have shown, however, is
that he often works with one particular view of powers on which they are
individuated by their objects. Furthermore, when he makes use of this
theory, he individuates cognitive powers (epistêmai, doxa, crafts, rhetoric)
by the objects in the world with which they are concerned, i.e., their
subject matters.
Returning now to the powers argument in Republic V, we see that when
Socrates ignores the possibility that two powers could do different things to
the same object, he is not making a careless mistake, but instead relying on a
developed theory. Powers are correlated one-to-one with their objects; thus
distinct powers must be the powers to relate in distinct ways to distinct
objects. Plato would not accept that animal husbandry and butchery take
the same object (Fine’s example (1990, 220)), any more than he would
accept that sight and hearing take the same object; instead, he would define
the object of each more narrowly. (Perhaps husbandry is of animal well-
being by contrast with animal flesh.) Each power is the power to do some
activity φ in relation to some object f, where f is defined so narrowly that
φ-ing is the activity uniquely suited to it.
produce persuasion in the soul, Phaedrus 271e)—the thing it affects, rather than as implied in
the Gorgias its subject matter—although see Gorgias 463b for a similar claim: rhetoric is over
(epi) the soul.
Powers and their accomplishments are defined 61
Thus we should positively expect Plato to do just what he does in this

passage, namely identify two distinct kinds of object for epistêmê and doxa.
Just as hearing is concerned with one kind of thing and sight with another,
or just as medicine is concerned with one kind of thing and rhetoric with
another, so epistêmê is concerned with one kind of thing, and doxa with
another.
2. Powers and their accomplishments are defined

by their objects
Why would someone construe powers, their accomplishments, and their

objects as so closely correlated? Perhaps because they are thinking of these
things as somehow inter-defined, essentially related to one another. As I
mentioned in Chapter 1, that is a view we find in Plato’s student, Aristotle.
Here again is Aristotle’s claim:
If one must say what each of these is, for example what is the thinking
[power] or the perceiving or the nutritive [power], it must first be said
what thinking is and what perceiving is. For the activities and the actions
are prior in account to the powers. And if this is so, and prior still to these
the corresponding objects must be studied, one must first define these, for
the same reason [i.e. that they are prior in account], for example about
nourishment and the perceptible and the thinkable.17
(Aristotle, De Anima 415a16–22)
Here, as in the Republic’s powers argument, we can distinguish a power by

looking to (a) something that it does (for Plato, “what it accomplishes”; for
Aristotle, its activity, energeia, later referred to as its erga, products or
accomplishments (402b9–16)), and (b) some object to which it is specially
related (for Plato, “what it is over”; for Aristotle, its correlate, antikeimenon).
On Aristotle’s version, the claim clearly concerns not just how to recog-
nize different powers, but how to define them. Both activity and object are
“prior in account” to the power, and the object is prior in account to the
17 εἰ δὲ χρὴ λέγειν τί ἕκαστον αὐτῶν, οἷον τί τὸ νοητικὸν ἢ τὸ αἰσθητικὸν ἢ τὸ θρεπτικόν,

πρότερον ἔτι λεκτέον τί τὸ νοεῖν καὶ τί τὸ αἰσθάνεσθαι. πρότεραι γάρ εἰσι τῶν δυνάμεων αἱ
ἐνέργειαι καὶ αἱ πράξεις κατὰ τὸν λόγον. . . . ἔτι πρότερα τὰ ἀντικείμενα . . . οἷον περὶ τροφῆς καὶ
αἰσθητοῦ καὶ νοητοῦ.
activity. Aristotle elsewhere explains this relation as follows: if A is prior in

account to B, then the definition of B includes a reference to A, but not vice
versa (Metaphysics 1028a34–6; cf. 1049b12–17 on the priority of activity
over power). The claim is metaphysical, not semantic: what it is to be B is to
be something involving A.18 Thus powers and activities are ultimately
defined by their objects: they are what they are because the object is what it
is. What it is to be a certain power is to be the power to do a certain activity,
φ-ing, and what it is to φ is to do the kind of thing that can be done in rela-
tion to a certain kind of object, f. He later gives us a clear application of this
principle regarding perception:
It is the proper perceptibles [color, sound, etc.]…in relation to which the

being of each sense is by nature.19 (De Anima 418a24–5)
It is not just that sight can (strictly) perceive only color, it is that the power
of sight is in essence something related to color: it is the power to see colors.
Likewise, the earlier passage (415a16–22) implies, the power of nutrition is
in essence something related to nutrition, and the power of intellection is in
essence something related to intelligibles. Each power is what it is because it
is the power to act on, or be acted on by—let us say “relate to” to allow for
both cases—its particular kind of object. (Johansen brings out the implica-
tion of these claims nicely by arguing that the objects are the formal causes
of the activities: “it is colour that makes vision the sort of perceptual suffer-
ing it is . . . and such defining is the proper work of the formal cause”
(2012, 98).)
Thus Aristotle not only directs us to discriminate powers by their objects,
as Plato does, but offers a metaphysical rationale for this procedure. Could
Plato have the same rationale in mind?
Indeed he comes close to an explicit statement of it in the argument we
saw in Republic IV: each appetite is related to “what it is of by nature”
(houper pephuken) (437e). One cannot thirst for food, because thirst is by its
nature an appetite for drink. Appetites—which, as we saw, the Charmides
treats as powers—belong to the class of “things that are such as to be of
something” (toiauta hoia einai tou, Republic 438a–b). The object is built into
18 For one interpretation of priority in account, and its relation to what Aristotle calls “pri-
ority in being,” see Peramatzis 2011.
19 πρὸς ἅ ἡ οὐσία πέφυκεν ἑκάστης αἰσθήσεως. Hicks’ commentary elaborates: “Each sense is
by nature so constituted as to be affected by, and to perceive, its special objects” (1907, 364,
comment ad loc.)
the definition of the power; to be that power is to be the power to relate to

that kind of object. In this way powers are like the greater, lesser, more,
fewer, double, and half. Plato is here clearly sketching out a theory of rela-
tional entities, a theory Aristotle will develop in detail (Categories 7).20
Returning with this in mind to the Charmides’ list of powers and objects,
it is extremely natural to see the same principle at work in their pairings.
Plato’s thought is not that sight simply happens to be of colors, or fear of the
terrible, but instead that sight is by its nature exercised on colors and these
alone, while fear is by its nature a response to the terrible and this alone, and
so on. As Carone puts it, “What Socrates seems to be suggesting for the first
time in the Charmides, and then Plato elaborating in the Republic, is what
we might call some kind of ‘theory of intentionality’ by which every mental
faculty is individuated and defined by its own distinctive object.”21
I can imagine someone protesting that Plato is aiming instead at a merely
formal point about powers and objects: the power of φ-ing is by definition
exercised in relation to something φ-able. This would tell us nothing
important about the power: it is a mere tautology. It is consistent with an
account on which the power can be exercised in relation to any kind of
thing whatsoever; what renders something φ-able is simply that some-
one φs it.
What Aristotle offers in De Anima II.4 is clearly something very different.
An account on which the power and its exercise are the way they are because
the object is the way it is requires a substantive characterization of the
object—a definition that is independent of the power. As Aristotle puts it
elsewhere, “sight is sight of something, but not of that of which sight is
(although it is true to say this), but in relation to color or something else like
that” (Metaphysics 1021a33–b2). Indeed, in the rest of the de Anima he goes
on to give substantive characterizations of each power’s object. Let us call
accounts that explain powers in this way objects-based: they define powers
in terms of objects that can be substantively characterized on their own
terms, independently of the powers.
When Plato individuates powers by their objects, which kind of account
does he have in mind? We have seen that he sometimes gives tautologous
characterizations of the objects, but he clearly thinks there are substantive
20 Plato’s goal here is to argue that unqualified relata are related to unqualified relata, while
qualified relata are related to qualified ones; for example thirst itself is of drink itself, while
much thirst is for much drink. What I wish to emphasize however is the more general point:
some things, evidently including powers of the soul, are essentially relational.
21 Carone 2001, 119–20; cf. Kamtekar 2009.
characterizations to be had. To take an example we saw above, although he

sometimes refers to the object of sight merely as the visible (horaton), he
elsewhere makes clear that there is a particular kind of thing on which sight
is specially suited to act: color.22 Likewise for desire: rather than character-
izing desire’s object as whatever is desired, he holds that there are certain
things whose nature renders them such as to be desired: appetite is only of
the pleasant rather than of anything else; erôs is only of the beautiful
(Charmides 167e); wish is of the good (Charmides 167e, Gorgias 468c).
Plato evidently thinks there is a substantive account to be given of the
objects of perception and of desire.
What about cognitive powers? Again, he sometimes offers tautologous
characterizations of their objects: in the Charmides, doxa is of what doxai
are of (hôn doxai doxazousin), and epistêmê of what-can-be-learned (to
mathêma) (168a); in Republic V, epistêmê is of the object of gnôsis (gnôston),
doxa of its object (doxaston) (478b); later epistêmê is of the object of noêsis
(noêton) (509d with 511d). If this were all he said, those wishing to assimi-
late Plato’s epistemology to the modern way of doing things could take
heart. When he says that epistêmê and doxa are set over different things,
they could argue, Plato means to make a merely formal point: that when
you have doxa about something you thereby render it an object of doxa, and
likewise for epistêmê. On this picture there would be no special quality
some entity or domain needs to be the object of epistêmê, nor of doxa, and
Overlap is possible.
Yet there are ample signs that Plato has in mind something stronger.
Consider—as we will in much more detail in the next chapters—his fre-
quent characterizations of the object of epistêmê as what is or being, and of
the object of doxa as becoming, or what is between being and not-being, or
as the visible. We will have to figure out what these descriptions mean, but
one thing should already be clear: these are substantive characterizations;
they make no reference to the powers.
This encourages us to take the tautologous characterizations as superfi-
cial. When Plato says for example that epistêmê is of the intelligible (to
noêton), the claim is not that whatever we turn our epistêmê toward thereby
becomes its object, but rather that there is something in the nature of certain
things which makes them accessible to intellect by contrast with perception,
and epistêmê is by its nature related to this kind of thing. (Indeed, even the
claim that epistêmê is of “what-can-be-learned” (to mathêma) may sound to
22 Many philosophers nowadays deny that color and sound are metaphysically independent
of sight and hearing, but Plato, like Aristotle, thought they were.
Plato’s ear much more of a substantive-object characterization than it does

to ours. ‘Mathêma’ is a familiar label for highly structured domains such as
geometry or architecture, what we might call sciences. In naming this as the
object of epistêmê, therefore, Plato plausibly means to be saying something
substantive: epistêmê by its nature can only be of structured, systematic
domains.)23
I am arguing that Plato holds that there is something special about cer-
tain objects that suits them to be cognized in certain ways. Just as there is
some feature in the nature of certain things that makes them digestible and
some other feature that makes things visible, so there is some feature in the
nature of certain things that makes them suitable for epistêmê to grasp, and
some other feature that makes things available instead to doxa. I am here
following Gerson, a strong defender of the view that Plato held an objects-
based epistemology: “the question of what is knowable is . . . a question
of . . . finding out what something must be like to be knowable, that is, to be
able to put us in the real state of knowing” (2009, 7).
If this idea sounds too strange to attribute to Plato, it should help to note
that Aristotle certainly seems to have held it. For as I will now show, he very
clearly applies the De Anima’s objects-based account of psychological
powers to the cognitive domain.
In a passage we saw in Chapter 1 (Nicomachean Ethics 1139a6–12),
Aristotle gives a Distinct Objects account of two rational parts of the soul
(where parts are characterized with the same suffix used in de Anima for
powers, -ikon). One is called the part capable of epistêmê (epistêmonikon);
the other Aristotle refers to first as the part capable of calculation (logis-
tikon) or deliberation (bouleutikon), and later as the part capable of doxa
(doxastikon), on the grounds that doxa, like deliberation, is concerned with
“things that can be otherwise.”24 Aristotle motivates this psychological dis-
tinction via a distinction in objects:
Let it be laid down that there are two parts that have logos, one by which
we contemplate those things whose principle do not admit of being other-
wise, one by which we contemplate those [whose principles] do admit of
23 I mean this description of mathêmata only as a rough characterization; we will get much
more information about what sorts of things can be the objects of epistêmê in Chapters 3–4.
24 “Since there are two parts of the soul that have logos, [practical wisdom] must be the
virtue of the second, the doxastikon: for both doxa and practical wisdom are about what can be
otherwise” (Nicomachean Ethics 1140b25–8). What Aristotle here calls the doxastikon must
be the very same second logos-having part that he earlier called the logistikon, the part responsible
for thoughts about what can be otherwise, including practical thought.
being otherwise: for in relation to things different in kind, so the part of

the soul related to each is different in kind.
(Aristotle, Nicomachean Ethics 1139a6–10)
There is a division in nature between necessary things like the truths of

mathematics or theology and contingent things like individual actions or
objects. Corresponding to this division is an epistemological one: the for-
mer are cognized by the epistemonic part of the soul, the latter by the calcu-
lative or deliberative part, also known as doxastic. That is, the former are
objects of epistêmê, the latter of doxa.25 The object of epistêmê differs in kind
from the object of doxa, just as the object of sight differs from the object of
hearing, and therefore the power (and presumably, as in our De Anima pas-
sage above, the activity) of epistêmê differs from that of doxa. Cognitive
powers are tailored to specific objects; the powers are distinct because their
objects are distinct.
We find similar claims repeated at length in the Magna Moralia—
evidence if not of Aristotle’s views then of later Peripatetics’. Here the author
argues for a distinction between the parts responsible for epistêmê and for
phronêsis (practical wisdom) on the grounds that their objects are distinct.
(This time there is no mention of doxa, but the argument is nonetheless a
striking example of a Distinct Objects epistemology, where the distinction
is between theoretical and practical thought.) Here the identity of the
objects is very clear: just as in Plato’s Two Worlds dialogues, the distinction
is between intelligibles and perceptibles.26 Moreover, the author uses the
same analogy we see in the Republic’s powers argument, between the rela-
tions cognitive powers bear to their objects and the relation perceptual
powers bear to theirs:
[That the deliberative and epistemonic parts are distinct] will be evident
from their objects [tôn hupokeimenôn]. For just as color, flavor, sound and
smell are different from each other, so too nature has given different senses
25 Or at least of the doxastikon. If this is the same as the logistikon, does all practical and
technical reasoning result in doxa? This would be an ill fit with Aristotle’s distinctions between
doxa and the good exercise of practical and theoretical reasoning (phronêsis and technê).
26 In Chapter 1 we considered a passage from Posterior Analytics I.33 in which epistêmê and
doxa are distinguished by being about different kinds of proposition. I mentioned that even
there, the implication is that the objects are different. The passage from Nicomachean Ethics is
almost explicit that objects are at issue (see discussion in Chapter 1). The present passage, from
the Magna Moralia, is simply unequivocal.
Powers and Accomplishments 67
for them (for we cognize [gnôrizomen] sound by hearing, flavor by taste,

color by sight), likewise we must suppose it to be the same way in other
cases. Where the objects [hupokeimena] are different, different too are the
parts of the soul by which we cognize them. Now the intelligible [noêton]
and the perceptible [aisthêton] are different, and we cognize these things
through the soul. Therefore [ara] the part of the soul concerned with [peri]
perceptibles [viz., the deliberative part] must be different from the part
concerned with intelligibles [the epistemonic part]. The deliberative and
deciding part is concerned with things that are perceptible and changing,
and simply with all things that come to be and pass away.
(Magna Moralia 1196b15–29, emphasis added)
Nature fits us with different powers tailored to different objects. Because

unchangeable things are different from changing ones, we have two differ-
ent cognitive powers to cognize them.
We have thus seen explicit evidence that Aristotle’s epistemology is
objects-based, and strong indications that Plato’s is as well. Powers and their
accomplishments are individuated by their objects because they are defined
by their objects: each power is, in its essence, the power to relate to a certain
kind of thing. In the next section I will argue that we get confirmation of
this interpretation from the fact that both Plato and Aristotle also held a
stronger view, one that provides a rationale for this one. Cognitive powers
and their accomplishments are tailored to their objects by virtue of resem-
bling these objects: cognition operates on the principle Aristotle calls
“like-by-like.”
3. Powers and their accomplishments resemble their objects:

cognition of like-by-like
In the very next lines of the discussion quoted above from the Nicomachean
Ethics, Aristotle offers an explanation for why there is a different cognitive
power for each kind of objects:
In relation to things different in kind [genei], so the part of the soul related
to each is different in kind, since indeed cognition [gnôsis] belongs to the
parts by virtue of some likeness and kinship [homoiotêta kai oikeiotêta].
(Nicomachean Ethics 1139a8–11, emphases added)
Each cognitive power is uniquely suited to cognize its particular type of

object by virtue of resembling that object in some way.
Although he does not draw the connection here, this view provides a
strong rationale for the objects-based account of powers and activities we
find in De Anima II.4, at least so far as cognition is concerned. The world is
populated with various kinds of objects, which are the way they are inde-
pendently of our cognizing them. Nature endows us with powers to cognize
these objects, tailoring different powers to the different types of object.
Since cognition works through resemblance, each power owes not only its
existence but its particular character to its object. Thus if you want to under-
stand the epistêmonikon, or its exercises, you will have to begin by under-
standing its objects—viz., things that cannot be otherwise—for the power
and activity are the way they are due to their resembling those objects.
Aristotle does not elaborate on the claim that cognition works through
resemblance, but it is clear that he is endorsing some version of a view he
elsewhere attributes to his predecessors: that “like is cognized by like” (De
Anima 409b26–7).27 Aristotle takes this it to be a widespread view of cogni-
tion. There is one extremely clear case in a passage he quotes from
Empedocles:
With earth we see earth, with water water, with air bright air, but ravaging
fire by fire, love by love, and strife by gruesome strife . . .
(Empedocles B 109, trans. R.D. Hicks)28
We cognize each kind of external phenomenon by the action of some quali-

tatively similar phenomenon within us.29 Aristotle attributes the same kind
of view to Diogenes of Apollonia (soul cognizes external objects because
both are made of air, De Anima 405a23), and to Heraclitus (because both
are in motion, De Anima 405a27), among others.
In de Anima II.5 and III.4, Aristotle himself develops a more complex
view that nonetheless accepts the basic principle that like can be cognized
only by like. Vision occurs when something in us that is already potentially
like visible objects—the visual system—takes on the visible form of some-
thing and becomes actually like it, and likewise for the other senses; intel-
lection (nous) happens when something that is potentially like intelligible
objects—the noetic system—takes on the intelligible form of something and
becomes actually like it. He does not tell us how to apply the like-by-like
principle to Nicomachean Ethics VI.1’s distinction between the epistemonic
27 γνωρίζειν τῷ ὁμοίῳ τὸ ὅμοιον. 28 Hicks 1907, 13–15.

29 Aristotle characterizes Empedocles’ view as being strictly materialistic (De Anima I.5),
but the inclusion of Love and Strife casts doubt on this interpretation.
and doxastic parts of the soul, and it is difficult to see how this would work;
perhaps the idea would be that the epistemonic part is potentially like
unchanging things, and thus capable of taking on their intelligible forms,
and likewise for the doxastic part with changing things (a view similar to
what I attribute to Plato below). At any rate, the account he gives in de
Anima shows that Nicomachean Ethics VI.1’s remark about cognition work-
ing through likeness and kinship is far from casual.
Thus both Plato’s predecessors and his student held versions of the like-
by-like view of cognition. Did Plato himself? Evidently Aristotle thought so.
He argues in de Anima that psychologists who want to account for cogni-
tion and perception (to ginôskein kai to aisthanesthai) say that the soul is
composed of the first principles (archai)—that is, the things of which every-
thing else is composed. His first example is Empedocles (the passage given
above). He continues:
In the same way, Plato in the Timaeus makes the soul from the elements,
for like is cognized by like [gignôskesthai gar tôi homoiôi to homoion], and
[external] things are made up of the first principles. Similarly, in the work
On Philosophy he defined the world-animal as made up of the idea of the
One and of the primary length, primary breadth, primary depth, and
similarly with all the rest. And furthermore he put it in another way: nous
is the One; epistêmê the Two…; doxa is the number of the primary surface,
Three; and perception the number of a solid, Four . . . Some things are judged
by nous, some by epistêmê, some by doxa, some by perception . . .
(Aristotle, De Anima 404b16–27, translastion based on Hicks)
What should we make of this? The claim about the Timaeus is clear enough:
in this dialogue, the character Timaeus puts forth a straightforward
Empedoclean account of sensory perception on which it is through fire in
the eyes that we see external fire. Whether this is meant as Plato’s own view
of perception is unclear, but certainly it is a like-by-like theory. Of more
relevance for our purposes, Aristotle may have had in mind another passage
from the Timaeus too, where commentators from Proclus to the present day
have found the like-by-like doctrine:30
30 See citations in Cornford, who agrees (1935, 64–65 and 94). Hicks, in his commentary on
De Anima 404b17, refers to this passage of the Timaeus as evidence for the like-by-like view in
Plato: “Plato with his immaterial principles gives a wholly original application to the maxim
‘like is known by like’ ” (Hicks 1907, 222).
Seeing, then, that the [World-] soul had been blended of Sameness,
Difference, and Being . . . whenever she is in contact with anything . . . she is
set in motion all through herself and tells in what respect precisely, and
how, and in what sense, and when, it comes about that something is quali-
fied as either the same or different with respect to any given thing, what-
ever it may be . . . Whenever logos . . . is about that which is perceptible, and
the circle of the Different, moving aright, carries its message through-
out all its soul – then there arise doxai and pisteis stable and true. But
whenever logos is concerned with the calculating, and the circle of the
Same, running smoothly, declares it, the result must be nous and
epistêmê. (Timaeus 37a–c, based on Cornford’s translation)
The argument is difficult to decipher, but commentators have taken the idea
to be as follows: because the World-soul is composed of the same basic
elements as Forms and perceptibles—Being, the Same, and the Different—it
can cognize these things, cognizing the Different in them by means of the
Different within it, and the Same in them by means of the Same within it.
Moreover, the result when the objects are perceptibles is doxa, when they
are intelligible, epistêmê. Here then we have evidence for a Platonic like-by-
like view of epistêmê and doxa, although arguably one confined to the
Timaeus, premised as it is on the dialogue’s mythical cosmogony.
As to Aristotle’s claims about On Philosophy, we have very little to go
on—this is an otherwise unattested work, and the Pythagorean doctrine
here described is found nowhere else in Plato—but we can discern a basic
picture. Just as in Empedocles’ or Timaeus’ theory of perception, cognition
works because the same “first principles” compose both the soul and exter-
nal objects: not now fire and air, but numbers. It is by the One in the soul
that we cognize the One in the world, and so on.
What is significant about this doctrine for our purposes is that, like that
of Timaeus 37a–c, it applies the like-by-like theory of cognition to doxa and
epistêmê. Just as vision cognizes what is vision-like out in the world, so nous
cognizes what is like nous-like out in the world, and the same for epistêmê
and doxa. Thus whatever his evidence, Aristotle finds in Plato a view of cog-
nitive powers broadly similar to the one he himself endorses in Nicomachean
Ethics VI.1: cognitive kinds are correlated with distinct objects because cog-
nition is of like-by-like.
I want to show that Aristotle was on the right track. Although the idea
has received relatively little attention in studies of Plato’s epistemology, there
is ample evidence—and much of it clearer and more systematic than what
we have just seen—that Plato thinks epistêmê and doxa work on the
principle of like-by-like.31 Moreover, the doctrine applies both to powers

and to their accomplishments. Here is a rough sketch, to be filled in below.
There is a sharp qualitative distinction between Forms and perceptibles.
Forms are what I will call, for shorthand, clean: they are stable, clear, and
precise; perceptibles by contrast are what I will call messy: unstable, obscure,
and imprecise.
Moreover, there is something in us—either the soul as a whole (Phaedo
79a–d) or a special component of it (Republic 490b)—that resembles the
Forms in being clean, and therefore can cognize them. There is also something
in us that resembles perceptibles in being messy, and therefore can cognize
them. (This latter claim is explicit in the Phaedo, where the job goes to the
body (79a–c); the Republic makes no explicitly equivalent claim, but it is
natural to supply the power of doxa for the role; I return to this issue below.)
Call this Stage 1.
Finally, when we actually cognize the Forms our souls become even
cleaner; when we actually cognize perceptibles our souls become messier;
actual contact with the objects increases the resemblance. The clean condi-
tion that results from cognizing Forms is occurrent epistêmê; the messy
condition that results from cognizing perceptibles is doxa (where Plato per-
haps conceives of these as activities, perhaps as states). Call this Stage 2.
Now for the evidence. That Plato characterizes both Forms and epistêmê
as clean, and both perceptibles and doxa as messy, is beyond question.
Forms are always in every way the same, while perceptibles change and are
infected with contradictory properties (Republic 479a–c, 524a, 529d–530b,
Phaedo 74b–c, 78e, 79c–d, Symposium 210e–211b; cf. Philebus 59a–c, where
Plato sums up the point by saying that perceptibles have no stability
(bebaiotêta)). Perceptibles are more obscure than Forms (skotôdestera,
Republic 479c); Forms have more clarity (saphêneia) (509d); Forms are most
precise (akribestata) (implied at Republic 529d). Likewise, epistêmê is stable
or steadfast (bebaios, monimos) while doxa is unstable (Meno 98a, Philebus
59b, and the Platonic Definitions 414b–c; cf. Republic 484b, and Protagoras
356d–e);32 epistêmê is clear (saphês) and doxa is unclear (Republic 511e,
Philebus 58c–59d); epistêmê is precise (akribês) and doxa imprecise
31 A handful of commentators on Aristotle have noted that Plato holds some version of the
like-by-like view: see for example Hicks’ commentary on De Anima 404b17–27 (noted in the
previous note), and Joachim’s on Nicomachean Ethics 1139a. David Ebrey (forthcoming) has a
very good discussion of the kinship between the soul and the Forms.
32 Although at Timaeus 37b, true doxa is stable. I take the claim to be that this doxa, belong-
ing as it does to the World-Soul, has as much stability as doxa can have—which, given its
objects, is not perfect stability, but relative stability. See discussion in Chapter 4.5.
(Philebus 58c–59d). What Plato means by these characterizations is less

clear. I will have a bit more to say in Chapters 4 and 7; for present purposes
I wish only to emphasize that Plato does attribute cleanness and messiness
both to cognitive kinds and to their objects.
Indeed he sometimes explicitly emphasizes this similarity, and spells out
to some extent how it allows for cognition. Consider a passage we saw in
Chapter 1 as evidence for Distinct Objects: Phaedo 79c–80b.33 The Forms
are invisible, immortal, uniform, and unchanging; perceptibles are the
opposite. Our bodies, the part by which we cognize perceptibles,34 is “more
like and akin” to them than it is to Forms (homoioteron kai sungenesteron,
79b), while our soul, the part of us that grasps the Forms, is “more like and
akin” to them (79e). These are the key terms that describe cognition of
like-by-like at Nicomachean Ethics 1139a11, and this is clearly a similar doc-
trine: we cognize each type of thing by means of some part of us akin to that
type.35 This is Stage 1.
Moreover, when we are actively cognizing, the resemblance grows
stronger (Stage 2):
[T]he soul, whenever it makes use of the body for investigating some-
thing . . . is dragged by the body toward the things that are never the same,
and itself wanders and is confused and is dizzy as if drunk, because it is
having contact with things of that sort36…But whenever it investigates itself
by itself, it passes toward what is pure and always being and deathless and
always the same, and since it is akin to it, it always is with it, whenever it is
itself by itself and is able to, and it has ceased its wandering and remains
always in the same state, because it is having contact with things of that sort.
(Phaedo 79c–d, italics mine)
Not only does the soul already resemble the Forms, but by attending to
them it comes to resemble them still more, becoming more stable; not only
33 In his De Anima commentary Shields recognizes this passage as evidence for the doctrine
of like-by-like (comment ad 409b24–5); cf. Ebrey (forthcoming).
34 Or more precisely, that through which the soul cognizes perceptibles; the soul cognizes
Forms all by itself.
35 See also Gallop’s commentary ad loc; on his interpretation Plato is also appealing to the
converse of the doctrine, that unlike cannot be cognized (well) by unlike: “The soul’s know
ledge of the unvarying Forms shows its likeness to them; its confusion when it is in contact
with varying sensibles shows that it is unlike them” (1975, 140).
36 ἅτε τοιούτων ἐφαπτομένη. ἅτε has clear causal force; it is often translated “inasmuch as”
or “seeing that.”
does the body already resemble perceptibles, but by attending to them

through the body, the soul is infected with the resemblance, becoming more
unstable. Moreover, the stable condition inherited from contact with Forms
is phronêsis (79d)—the word here seems to play the role of the Republic’s
‘epistêmê’ or ‘noêsis,’ denoting an intellectual grasp of Forms—and the
unstable condition, Plato strongly implies, is doxa (see Phaedo 83c–84a, dis-
cussed in Chapter 1.2). So here we have a full like-by-like theory: as on the
classic Empedoclean view, cognition happens by something in us acting on
external objects which it resembles (Stage 1), and the resulting cognitive
conditions further resemble those objects (Stage 2).
We find another version of the view in the Republic. Here too there is
something in us that has a special resemblance to Forms, and for that rea-
son enables us to cognize the Forms. It is not this time the soul simpliciter,
but one aspect of it:
It is the nature of the real lover of learning to struggle toward what is, and
not to remain over [epi] each of the many things that are opined to be
[doxazomenois],37 as he moves on . . . until he grasps the Being of each
nature itself with that of his soul which it befits to grasp this kind of thing –
it befits that which is akin38 . . .[and begets] nous and truth.
(Republic 490a–b, translation based on Grube/Reeve, emphases added)
This is clearly a Stage 1 like-by-like view, although the passage leaves open
several questions about how it works. First, which part of the soul is Plato
here describing as akin to the Forms: the whole rational part of the soul
(logistikon), or some specific component of it? Second, and related, is this
the same thing by which we cognize perceptibles, or is that some other part
which is akin instead to them—perhaps the body, as in the Phaedo, or the
non-rational parts of the soul, or the power of doxa? Certainly Plato implies
that the very same part of us can either be turned toward Forms and gener-
ate epistêmê, or toward perceptibles and generate doxa: see Republic 508d,
where doxa and epistêmê are both operations of “that of the soul by which it
thinks [noei],” and Republic 518c, where the “eye of the soul” can be turned
either toward perceptibles or toward Forms. Indeed, the Phaedo view is not
so different: it is one and the same thing, the soul, that cognizes both Forms
37 Or, “each of the things that are opined to be many.” I use ‘opined’ here from failure to find
a better idiomatic way to translate.
38 ᾧ προσήκει ψυχῆς ἐφάπτεσθαι τοῦ τοιούτου, προσήκει δὲ συγγενενεῖ.
and perceptibles; the difference is that it cognizes Forms by itself, and

perceptibles through the body (Phaedo 79c).
Given that Plato has earlier in the Republic argued that epistêmê and doxa
are each the operation of a different power (478a–b), I suggest the following
interpretation: Plato thinks that we have something in us, the rational part
of the soul, which he sometimes refers to as a power, but which itself has
several powers. When it is turned toward perceptibles it exercises its power
of doxa; when it is turned toward Forms it exercises its power of epistêmê.39
It is as a whole the part of us most similar to the Forms, and therefore its
true or best function is to cognize these. But insofar as it has the power of
doxa it also resembles perceptibles. (Perhaps, as in the Phaedo, it has the
power to cognize perceptibles only because it is embodied, and it is the
body that is the direct source of resemblance to perceptibles.) This is how-
ever a speculation; I think the text leaves it underdetermined, and suspect
that Plato had not fully worked out his view here.
Second, the Republic is far less explicit than the Phaedo about the basis of
resemblance to the Forms. Is the claim again that the relevant part is stable,
immortal, divine? Arguably this is what Plato has in mind: toward the end
of the dialogue he strongly implies that the rational part of the soul alone is
“akin” (sungenês) to the divine, immortal, and eternal (Republic 611e). At
any rate, this passage clearly presents Stage 1 of a like-by-like doctrine of
cognition: there is a part of us specially suited to grasp the Forms because it
resembles them.
The Republic is very explicit about Stage 2. When we actually contemplate
Forms, we take on a condition that resembles them in being stable and clear,
and that condition is called epistêmê. When we actually attend to perceptibles,
we take on a condition that resembles them in being unstable and murky,
and that condition is called doxa. We have seen evidence of this already in
Chapter 1, from a claim Plato makes in the course of the Sun analogy. Just
as when our eyes are turned on dim things they cannot see well, but when
they are turned toward well-lit things they see clearly (saphôs), so too:
Whenever [that part of the soul by which we think] is fixed on that on

which truth and being shine [the Forms], it understands and knows it and
39 (For a roughly similar view see Szaif 2007.) Although Plato’s tripartite theory of the soul
is not my focus in this work, I will briefly note that he seems to attribute doxa to all three parts
of the soul in the Republic (see Moss 2008 for one defense). Just as each part has its own desires,
each part has its own doxai; the rational part is however unique in also having the power of
epistêmê.
appears to have intelligence, but whenever it [is fixed on] that which is
mixed with darkness, that which comes to be and passes away, it has doxa
and is weak-sighted, shifting its doxai back and forth.40 (Republic 508d)
Attention to obscure and shifty perceptibles makes our souls engage in the
activity of having doxa, which activity makes the soul itself obscure (“weak-
sighted”) and shifty (changing its views). Attention to the truth-illuminated
Forms makes us instead engage in the activity of noêsis and gnôsis; the ana
logy with vision, and the contrast with doxa, strongly imply that this
involves our minds themselves being stable and clear rather than shifty and
obscure. At Stage 2, our cognitive condition inherits its properties directly
from its objects.
This is confirmed shortly afterward in the Divided Line passage’s correl
ation of cognitive conditions with objects: as the clarity of objects increases,
so does the clarity of the cognition.41 More elaborately: the upper section
has more clarity (saphênia) than the lower (509d), and also more truth
(alêtheia, 510a); after listing the four conditions (pathêmata) of soul, each
set over (epi) a section of the line, Socrates says “to the extent that the things
they are set over share in truth, so do these share in clarity” (511e).). The
implication is that, as at 508d, the cognitive conditions have the degree of
clarity they have because their objects do. (This causal claim is confirmed by
the Cave allegory: because one moves from attending to shadows to attend-
ing to statues and so on, one’s cognitive condition improves from eikasia to
pistis and so on.)
We find a very explicit statement of this Stage 2 inheritance view in the
Philebus’ ranking of cognitive kinds. (Confusingly, Plato here treats various
crafts (technai) as lowly species of epistêmê, describing what he would else-
where call epistêmê as the highest kind. This is not however such a depart
ure from his usual practice: we have seen that even in the Republic he
sometimes joins popular usage in using ‘epistêmê’ to label crafts like carpen-
try, house-building, or flute-playing (428b–c, 438d, 601e–602a).) Here too
he distinguishes a lowly messy cognitive condition from a lofty clean one,
and here too he traces this difference to a corresponding difference in their
objects. After stating that true epistêmê is clear, stable, and precise (Philebus
40 ἐνόησέν τε καὶ ἔγνω. . . . δοξάζει τε καὶ ἀμβλυώττει ἄνω καὶ κάτω τὰς δόξας μεταβάλλον.
41 I am here treating the cognitive terms in the Line passage as referring to exercises rather
than powers—something at Stage 2. I think this is warranted by the text; if it is wrong, however,
then we can take these like-by-like claims as referring to Stage 1.
58a–c), Socrates argues that one can have no such epistêmê about perceptibles,
which admit only of doxa:
Most arts, and the men who devote themselves to them, make use of doxai
and persistently investigate things which have to do with opinion [ta peri
doxan]. And even if they think they are studying nature, they are spending
their lives in the study of the things of this world . . . Such thinkers, then,
toil to discover, not what always is [ta onta aei], but what becomes [gig-
nomena] and has become and will become?…And can we say anything
clear [saphes] about these things, with most precise truth, since none of
them ever was, will be, or is in the same state?. . . How could we ever get
anything stable [bebaion] about things that have no stability?…Therefore
one can have no nous nor epistêmê possessed of the highest truth about
these things. (Philebus 58e–59b, emphases added)
The study of things that come to be and pass away (ta gignomena), rather
than of what is (ta onta), yields only doxa rather than (the highest) epistêmê.
This is due to a principle of like-by-like: if you attend to unstable objects,
your cognitive condition cannot be a stable one; the stability of the cogni-
tive condition depends on that of its objects.
We find further confirmation in a much-discussed passage from the
Timaeus. The explicit claim here is that there is a similarity between objects
and the accounts (logoi) one can give of them, but there is a strong implica-
tion that the similarity extends to the cognitive conditions of people who
give these accounts:
Accounts are akin to the things they expound.42 Accounts of that which is
fixed and stable and discoverable with the aid of nous will themselves
be fixed and unchangeable, so far as it is possible…; while accounts of
what is made in the image of that other but is only a likeness [eikôn], will
themselves be only likely [eikotas], standing to accounts of the former
kind in a proportion [ana logon]: as Being is to Becoming, so is truth
[alêtheia] to pistis. (Timaeus 29b–c)
When you grasp the stable world of Being, your cognitive condition, along
with the accounts that issue from or secure it, are inevitably stable; when
you are in touch instead with the world of Becoming, your cognitive state,
42 τοὺς λόγους, ὧνπέρ εἰσιν ἐξηγηταί, τούτων αὐτῶν καὶ συγγενεῖς ὄντας.
along with the accounts that issue from or secure it, will resemble that
world. Plato here characterizes this latter resemblance in terms of “likeli-
ness,” rather than in terms of the familiar properties of messiness, but he
may well mean to imply them; a similar passage in the Phaedo speaks
explicitly of accounts in which there is nothing stable (bebaios, 90c).
Moreover, although the present passage is about accounts, these are clearly
expressions of cognitive conditions, and the claim here is quite similar to
what we have seen just above: attention to qualitatively inferior things yields
a qualitatively inferior cognitive condition, while attention to qualitatively
superior things yields a qualitatively superior one.
As a final piece of evidence that Plato endorses a view with both Stage 1
and Stage 2, consider the Timaeus’ discussion of becoming like the divine
(90a–d). The rational part of the soul is that in us by which we can cognize
Being, and once again this part is itself akin to the divine (90a), while the
other parts are more akin to the mortal (Stage 1). Moreover, active attention
to lowly perceptibles makes us more like them (here, more mortal (90b)),
and also gives us doxa (here, dogmata, 90b) (Stage 2); if instead we engage
in love of learning (philomathia) and true wisdom (phronêseis), we ourselves
become as immortal as possible (90b–c). More specifically, if we nourish
our intellect on the things that are most familiar and akin to it, it will
become even more like them (Stage 2):
There is one way tending to anything—namely, to supply each with its

own proper [oikeias] food and motion; and for the divine part within us
the kindred [sungenneis] motions are the intellections [dianoêseis] and
revolutions of the Universe. These each one of us should follow, rectifying
the revolutions within our head, which were distorted at our birth, by
learning the harmonies and revolutions of the Universe, and thereby mak-
ing the part that thinks be like the object of its thought, in accordance
with its original nature.43 (Timaeus 90c–d, translation based on Lamb)
The best thing in us, which is already by nature akin to the best external
objects, becomes even more like that by contemplating them.
Two notes are in order before we sum up this view. First, in the case of
epistêmê, the resemblance will not be complete, for surely not even the best
human epistemic condition could be as stable as the Forms. (Indeed, one
might protest that since we are embodied, changeable creatures, even our
43 τῷ κατανοουμένῳ τὸ κατανοοῦν ἐξομοιῶσαι κατὰ τὴν ἀρχαίαν φύσιν.

most stable achievements are more akin to certain perceptible things: the
stability of human epistêmê is more like a rock’s than a Form’s.44) Plato
argues however not for exact resemblance between cognitive kinds and
objects, but instead for a proportionate analogy: in cleanliness, as Being is
to Becoming, so epistêmê is to doxa (Republic 511d–e; cf. 534a).
Second, Plato clearly holds that contact with Forms makes us resemble
them not only in our cognitive condition but in our whole selves: we
become more immortal (Timaeus 90c, Phaedo 84a–b), more ordered and
more unchanging (Republic 500b–d). My focus here is only on cognitive
conditions, but of course for Plato one’s cognitive condition has conse-
quences for one’s psychological condition as a whole. I leave this issue for
another occasion.
What we have seen from these passages is that Plato holds a two-fold
doctrine of cognition of like-by-like: cognition is enabled by something in
us already resembling the object, and the exercise of cognition involves
increased resemblance. I will now briefly show how this bears on an issue
we have kept in mind throughout the chapter: the difference between cogni-
tive powers and their exercises or accomplishments.
Plato does not draw this distinction explicitly in any of the passages we
have reviewed, but the two-fold nature of his like-by-like doctrine fits nicely
with it. We have something in us which already resembles the Forms, and
which enables us to cognize them—the soul (Phaedo), or a part of it
(Republic, Timaeus). We also have, according to the Phaedo at least, some-
thing which already resembles perceptibles, and enables us to cognize them:
the body (Phaedo 79a); perhaps Plato has in mind a similar view elsewhere.
Thus there is something Form-like in us all the time that enables us to cog-
nize Forms, and something perceptible-like in us all the time that enables us
to cognize perceptibles (Stage 1). These things are naturally understood as
cognitive powers, or parts that have powers: Plato after all defines a power
(dunamis) as what makes us able (dunametha) to do whatever we are able to
do (Republic 477c). Meanwhile the occurrent cognitions (Stage 2), the
activities of actively attending to Forms or perceptibles, are well understood
as accomplishments of those powers.45
44 I owe this objection to Gail Fine.

45 Thus Adam on Republic 508d: “The faculty of νοῦς is suddenly actualized into νόησις by
being turned upon its proper object. . . Reason has hitherto lain dormant within” (1902, com-
ment ad loc).
Distinct Objects confirmed 79
As to what the resemblances consist in, Plato seems to hold that the
power of epistêmê, or perhaps more precisely the part of us which has that
power, is divine and immortal like the Forms; in the Phaedo at least he holds
that the part with the power of doxa, namely the body, is lowly and mortal
like perceptibles, and perhaps he has some such view in the Republic too. It
is much clearer that and how resemblance works in the case of the accom-
plishments. We have seen in Phaedo 79c–d, Republic 508d, and Philebus
59a–b descriptions of what happens during an actual instance of attention to
perceptibles or Forms. Something in us takes on the character of the objects
to which it is actively attending: when we contemplate Forms our soul takes
on a clean condition, i.e. it has occurrent epistêmê; when we attend to percep-
tibles our soul takes on a messy condition, i.e. it has occurrent doxa. I will
examine the nature of these qualities further in Chapters 4 and 7; for now, I
simply emphasize that Plato repeatedly asserts that our cognitive accom-
plishments have these qualities, and inherit them from their objects.
Perhaps Plato would accept Aristotle’s view that the powers potentially
resemble their objects, while the active cognitions (accomplishments) actu-
ally do. At any rate it is clear that his like-by-like doctrine of cognition
applies to both the powers of epistêmê and doxa and to their accomplish-
ments. In examining the relation between these and their objects, then, I
will from here on in mostly bracket the power/accomplishment distinction,
and simply speak of cognitive kinds resembling their objects.
4. Distinct Objects confirmed
I have devoted so much attention to the like-by-like doctrine because Plato’s

embrace of it confirms that he also held both the views we attributed to him
above: cognitive kinds are individuated by their objects (the Distinct
Objects view), and also defined by them.
First, if cognition works on the principle of like-by-like, then cognitive
kinds must be individuated by their objects: the like-by-like doctrine entails
Distinct Objects. We saw this in Aristotle, who introduces the like-by-like
doctrine to justify a Distinct Objects claim: different kinds of objects
(“things whose principles cannot be otherwise” by contrast with things
whose principles can) must be cognized by different things within us,
“because cognition belongs to the parts by virtue of some likeness and kin-
ship” (Nicomachean Ethics 1139a10–11). If Plato holds that cognitive kinds
resemble their objects, and if he holds—as he so clearly does in the Two
Worlds dialogues—that the available objects are of two qualitatively

different kinds, then he is committed to Distinct Objects. One cannot have
epistêmê of perceptibles because there is nothing epistêmê-like there: noth-
ing clear, precise, and stable. One cannot have doxa of intelligibles because
there is nothing doxa-like there: nothing unclear, imprecise, changing.
More elaborately: powers are able to cognize only what they resemble, so a
clean power like epistêmê can get no grip at all on messy objects like percep-
tibles; cognitive accomplishments in turn resemble their objects, so the
result of attending to perceptibles will be messy, and a fortiori will not be an
instance of epistêmê. (Mutatis mutandis for doxa and Forms).46
This rules out Moderate Overlap views on which each cognitive kind has
a natural, or normative, or constitutive domain (the power is set over different
objects), but is able to act in some derivative or indirect way on the other’s
domain (the accomplishments can overlap) (e.g. Kamtekar 2009, Harte 2018).
It also begins to rule out Smith’s Content Overlap view, on which cogni-
tive powers have distinct relata, understood as the objects they are exercised
on, but the resulting accomplishments—occurrent cognitions—can be
about the same things (see especially Smith 2000). For this view too involves
drawing significant distinctions between the way powers are related to
objects and the way that their accomplishments are related to objects. What
we have seen in this chapter however undermines that strategy. When
describing the relation between cognitive kinds and their objects, as indeed
throughout most of his epistemological discussions, Plato makes no men-
tion of the power/accomplishment distinction, nor does he clearly mark
when he is talking about one by contrast with the other. I suggest then that
in considering the relation between cognitive kinds and their objects, Plato
did not think this distinction—let alone any finer-grained distinctions—
significant.47 Cognitive kinds, both powers and their accomplishments,
resemble their objects; therefore, both are individuated by their objects.
46 It is worth recalling that this view leaves open the possibility that the objects of epistêmê
and doxa might be co-instantiated in the same things, as sight and color are: the same thing
might have some properties that are grasped by epistêmê, others that are grasped by doxa. That
is, the objects of epistêmê and doxa might be distinct without being separate. Arguably this is
Aristotle’s view, as manifest especially in the Posterior Analytics’ claim that in a way the same
thing can be the object of both (89a23–37). Perhaps in some places it is Plato’s too (see my dis-
cussion of the Meno in Chapter 9). In the Two Worlds dialogues, however, Plato’s metaphysics
entails that the objects are not only distinct, but wholly separate.
47 Smith 2000 argues that powers have distinct objects (Forms for epistêmê, perceptibles for
doxa), while the “cognitive states” they produce—roughly, beliefs—can share the same con-
tents, i.e. have the same subject matter. His 2012 adds a new element: epistêmê and doxa as
powers, operating on their distinct objects, each allow us to form “conceptions” of the same
Distinct Objects confirmed 81
Content Overlappers might argue that the accomplishments can be about

different objects from those by which they are individuated: an episode of
doxa, in which one confines one’s attention to perceptibles, is nonetheless
about the underlying Forms. This is undermined by what we saw in
Section 1: Plato applies his objects-individuation thesis to phenomena like
crafts (technai), where the objects at issue seem to be subject matters.
Medicine not only has health as its correlated object but is about health;
health figures in the content of judgments that issue from the power of the
craft. More g enerally, we never see Plato working to draw a distinction, or
even recognizing the need to argue that there is or is not a distinction,
between the domain of cognitive powers—what they are “over” or “set over,”
in Republic V’s language—and the subject matters of their accomplishments,
what they are about.
There is more to say about the Content Overlap view, and I return to in
my detailed discussions of doxa and epistêmê in Chapters 4 and 7, but here
I will note that, despite my disagreements with it, I find it worth taking very
seriously. It aims to respect all the evidence that cognitive kinds are distin-
guished by their objects, while also accommodating all the concerns that
motivate typical Overlap readings: the alleged philosophical necessity of
and textual evidence for epistêmê of perceptibles and doxa of Forms. I do
think that it gets the most important point right: Plato’s thought is that one’s
cognitive condition is determined by the objects with which one is in con-
tact. I disagree only with the further claim that Plato wanted to distinguish
what cognitive powers are over from what cognitions are about.
If we reject any such distinction, then we are left with a stark Distinct
Objects interpretation, and there remain worries about this. Why is epistêmê
needed for ruling if there is no epistêmê of the perceptible realm; how can
Plato deny doxa of Forms; and isn’t there textual evidence that he acknow
ledges overlap? I will address these worries in Chapters 4 and 7, arguing that
Plato gives us resources to solve them, without embracing Content Overlap.
I take the arguments of this chapter to show however that whether or not he
was consistent in its application, Plato’s understanding of the relation
things (e.g. beauty), which are then applied in judgments about the same subject matter (e.g.
beauty). Thus the activities of the powers have distinct objects, but the products of those activ
ities—conceptions and judgments—do not. As I mentioned above, although there is plausibly
a distinction to be drawn between cognitive exercise and cognitive product—the activity of
cognizing produces a cognitive state—I see no signs that Plato was concerned to draw this
distinction.
between cognitive kinds and their objects commits him to a Distinct

Objects epistemology.
5. Objects-Based epistemology
Turning back now to the thesis that cognitive kinds are defined by their
objects—the thesis central to the argument of this book—we find that the
like-by-like view of cognition supports this too. There are two different
types of object in the world; we are endowed with powers to cognize both;
the powers, and also their exercises, resemble their allotted objects. Thus
epistêmê exists, is distinct from other powers, and has the qualities it has,
because its objects are the way they are; doxa exists, is distinct from other
powers, and has the qualities it has, because its objects are the way they are.
Each cognitive power and accomplishment is the way it is because the
object is the way it is: Plato’s epistemology is objects-based.
One important objection to consider, before we consider the conse-
quences of this interpretation: there is a tradition going back to Aristotle
that appears to interpret Plato’s views as proceeding in precisely the reverse
order, from epistemology to metaphysics. Here is a passage we saw briefly in
the last chapter:
The theory [doxa] about the Forms48 occurred to those who state it on
account of being persuaded of the truth of the Heraclitean theories that all
perceptibles are always flowing, so that if there is going to be epistêmê of
something, and phronêsis, there must be some other, abiding [menousas]
natures beyond the perceptible ones. For epistêmê cannot be of the
flowing. (Metaphysics 1078b12–17)
The implication is that Plato began from various convictions about the
nature of epistêmê, and the difference between it and doxa; given these
views, he needed to come up with an ontology that would furnish these
kinds with suitable objects, and this is what led him to posit the existence of
Forms. This picture has real textual support: on a number of occasions Plato
argues from the fact that we are able to achieve a good cognitive condition
48 Certainly a use of ‘doxa’ that would seem to go against the evidence for Distinct Objects
in Aristotle (Chapter 1.3)! Aristotle is however here using ‘doxa’ the way later writers will use
dogma; it is aptly translated ‘theory.’
Objects-Based epistemology 83
to the conclusion that there must be Forms. Most explicit is a passage from
the Timaeus:
If then nous and true doxa are two different kinds, then these things do by
all means exist, themselves by themselves, the Forms that are impercept
ible to us and objects of nous only [nooumena]. If however, as it appears to
some, true doxa and nous differ not at all, then whatever we perceive
through the body must be assumed to be the most stable things there are.
(Timaeus 51d)
When properly understood, however, such claims do not conflict with the
objects-based reading of Plato’s epistemology. The key is to distinguish
between Plato’s evidence for the intrinsic qualities of cognitive kinds, and
his explanation for them. Plato clearly takes himself to have excellent reason
to hold that there are two different cognitive kinds, one far superior to the
other. (Recall Socrates’ claim we began with in the Introduction: that there
is a difference between correct doxa and epistêmê is one of the few things he
would claim to know (Meno 98b).) He also takes various features of epistêmê
and doxa to be obvious and uncontroversial: for example, epistêmê is stable
where doxa is not (an unargued assumption at Meno 98a), or epistêmê is
infallible where doxa is not (an unargued assumption attributed to “anyone
with any sense” at Republic 477e), or that epistêmê is clear, precise, always
true, brings with it the ability to give a logos, and so on. He does not argue
for these claims, nor does he say why they need no argument; we can con-
jecture that he thinks of them as conceptual truths, or empirically evident,
or culturally accepted, but any rate he clearly thinks them obvious.
This is not however to say that he takes these to be brute facts. Here is an
alternate hypothesis: he holds (a) that epistêmê differs from doxa, in stability
and clarity and all the other ways just listed, and (b) that cognitive kinds are
defined by, individuated by, and resemble their objects (objects-based epis
temology). Therefore he holds that (c) there must be suitable objects to
explain the nature and possibility of epistêmê. (He can already observe,
through perception, that there are suitable objects to explain the nature and
possibility of doxa).
By way of analogy, imagine a zoologist examining a new terrain who
finds both big rounded footprints and tiny pointed footprints. She recog-
nizes (a*) that the two kinds of footprint differ, in size and in shape. She also
has (b*) a (very reasonable!) animals-based theory of footprints on which
footprints are defined by, individuated by, and resemble their makers.
(Footprints are, in essence, the prints left by the feet of animals; moreover
footprints are like-by-like—big footprints are caused by big creatures, small
ones by small creatures, and the shape of the footprint is inherited from
the shape of the foot). Therefore she concludes that (c*) there must be big
round-footed animals around, and also small pointy-footed ones. I am arguing
that Plato proceeds the same way: whatever the order of his evidence, he
takes the explanatory order to be metaphysics-first; his epistemology is
objects-based.
If we want to understand Plato’s epistemology, therefore, we now have a
clear directive about how to proceed. First, we should abandon the assump-
tion that Plato’s cognitive kinds map cleanly onto contemporary epistemol
ogy’s knowledge and belief. These do not fit the objects-individuated model:
unlike sight and hearing, knowledge and belief are object-neutral (I can
have opinions about the same things you know or the same things I once
knew). They do not fit the like-by-like model: one can stably know about
changing things; one can have murky beliefs about clear things. Therefore
they do not fit the objects-defined model: neither knowledge nor belief are
nowadays characterized as, first and foremost, the kind of cognition one has
of a particular type of object. Notice that this is not merely an issue about
whether or not epistêmê and doxa are propositional attitudes. Indeed I think
that that question, prominent as it has been in debates about Plato’s epis
temology, is largely orthogonal to the question of how epistêmê and doxa
relate to contemporary philosophy’s knowledge and belief. Plato could per-
fectly well hold that epistêmê and doxa are propositional attitudes while still
holding that they are objects-based: he could hold that in order for one to
have epistêmê of the proposition that x is F, it must be the case that x is a
certain kind of object or F a certain kind of property, and likewise for doxa.
Thus even if epistêmê and doxa are propositional attitudes, their objects are
essential to their nature, and this distinguishes them sharply from contem-
porary epistemology’s conceptions of knowledge and belief.
Instead, we should go about understanding Plato’s epistemology by iden-
tifying the relevant objects of each cognitive kind, and showing how these
objects explain the nature of the kinds. That is the project of the remainder
of this book. Here is a brief preview of the results.
Plato’s ontology—most fully in the Two Worlds dialogues, but to some
extent throughout—is built around the idea that there is a fundamental dis-
tinction between two levels of reality, as illustrated most dramatically in the
Republic’s Cave allegory. The best thing is Being, but there is also an inferior,
derivative level, sometimes called Becoming, sometimes referred to in other
Objects-Based epistemology 85
ways. I will argue that Plato’s epistemology is built around that same
division. Just as on Plato’s view the world contains both colors and sounds,
and we are equipped with special powers of grasping each, so too on Plato’s
view the world contain both Being on the one hand, and something
ontologically inferior on the other, and our souls are equipped with special
powers of grasping each.
Epistêmê is the power by which we understand Being: the ultimate real
ities. When we exercise this power we have cognitions that are like Being:
stable, precise, and clear. Epistêmê is thus best understood not as knowledge
in the ordinary sense, but as a deep grasp of ultimate reality. Doxa mean-
while is the power by which we form impressions about the world’s inferior
ontological kind: perceptible particulars that seem—are apparent to us, and
invite us to believe in them—but fall short of Being. When we exercise this
power on its objects we have cognitions that are like those objects in being
unstable, imprecise, and unclear. Doxa is thus best understood not as belief
or opinion in the ordinary sense, but rather as superficial thought that fails
to penetrate beyond appearances.
3
Epistêmê Is of What Is
If Plato’s epistemology is indeed objects-based, then he conceives of epistêmê

first and foremost as the cognition of a certain kind of object. Epistêmê has
the features it has, and differs from doxa in the ways that it does, and most
generally is what it is, because it is the cognition of that kind of object.
Identifying this defining object is therefore central to the project of under-
standing Plato’s epistêmê. This will explain why Plato attributes to epistêmê
the particular features he does, such as the definitional requirement or the
special relation to Forms; it will also be the key to understanding what kind
of thing epistêmê itself is—whether it is knowledge, or understanding, or
something else altogether.
What then is this special object? What corresponds to epistêmê as color
does to sight, the pleasant to appetite, or the beautiful to erôs?1
We find a very clear answer in our central text on the objects-based
nature of epistêmê, namely Republic V’s powers argument. Epistêmê is set
over what is (to on) (477a–b, 478a). Just as sight is the power to see the
visible, epistêmê is the power to have epistêmê of what is.
To be sure, Plato sometimes suggests other answers: epistêmê is of the
objects of gnôsis or noêsis, or is of what-can-be-learned (to gnôston, to
noêton, to mathêma). As we saw in Chapter 2, however, these are merely
formal characterizations of epistêmê’s object, not able to play the substantive
definitional role required in an objects-based epistemology. Moreover, as I
will argue in Section 1, the context of the powers argument strongly suggests
that it is identifying the defining object of epistêmê, and this suggestion is
reinforced by other texts. Let us for now then take this as our working
hypothesis: the defining object of epistêmê is what is.2
A worry might immediately arise that this is a very uninformative
answer, one that advances our project of understanding epistêmê barely at
1 Charmides 167e–168d, discussed in Chapter 2.1.

2 At Parmenides 134a, epistêmê is of truth itself. I will argue that this is a sense of ‘truth’
which coincides very closely with the relevant sense of ‘being.’
DOI: 10.1093/oso/9780198867401.003.0004
Epistêmê and What Is 87
all—at best, only by showing that that project depends on answering very
vexed metaphysical and linguistic questions. For the meaning of ‘being’
throughout Plato’s texts, and in the Republic’s powers argument in particu-
lar, is famously ambiguous, and much contested. Some interpret the powers
argument’s claim that epistêmê is “set over what is” as the claim that epistêmê
is of what exists, others as the claim that epistêmê is of what is true, or of
what is F for any predicate F, or of what is really F; some argue that Plato is
appealing to some fusion of these senses.3 Surely only an overall interpret
ation of the argument, or more broadly of Plato’s epistemology, can decide
which of these is correct. Hence it may seem a hopeless interpretative task
to begin with an account of being, and use that to construct an account of
epistêmê. Furthermore, looking outside this one text, the problem becomes
even worse. Very plausibly Plato’s notion of being changes and develops
from text to text; trying to construct an epistemology based on that notion
will give us a highly disunified account of epistêmê.
In what follows I will argue that there is a general characterization of
being which allays these worries. While it is not always clear how Plato is
using ‘being,’ and while the world may in fact have different senses in differ-
ent relevant contexts, the features of being relevant to its role as epistêmê’s
defining object turn out to be reasonably simple, very general, and strongly
unified across the dialogues. The key notion is of that which has superior
ontological status, by contrast with things that have some derivative or infer
ior status. This notion of ontological superiority—of what really is, by con-
trast with what is in some lesser way—is a notion Plato relies on in his
metaphysics, and we can use it to explain his epistemology.
The contrast between superior and inferior ontological levels plays out
differently in different contexts: for example, as the contrast between
essences and qualities, as the contrast between stable existence and unstable
generation and destruction, or as the contrast between things themselves
and their images. It is a job for the metaphysicians to analyze each of these
contrasts and tell us how, if it all, they are unified—that is, whether Plato
has a single theory of what makes things ontologically superior. There are
also important questions as to whether Plato is committed to a substantive
theory of what ontological superiority amounts to—for example,
fundamentality—or whether his notion is largely undertheorized, waiting
to be developed by Aristotle and others. I will make a few remarks about
3 See citations in Section 2.

88 Epistêmê Is of What Is
these issues here (Section 4), but I hope to show that for our purposes, the
general characterization of being as the ontologically superior turns out to
be sufficient. For this general characterization yields an account of epistêmê
which fits well with Plato’s texts, and explains all of epistêmê’s other features
(Chapter 4).
Identifying epistêmê’s defining object as being will thus yield an answer
to our starting question: what is Plato’s basic conception of epistêmê? I con-
sider the resulting picture in Chapter 5. We will find that Plato’s epistêmê is
not fundamentally to be conceived of as knowledge, nor as Science, nor
understanding, nor other familiar cognitive kinds. Fundamentally, it is to be
conceived as a deep grasp of ultimate reality. In the book’s final chapter, after
the discussion of doxa, I will turn to the question of why Plato would build
his epistemology around such a concept.
1. Epistêmê and what is
The powers argument in Republic V gives us strong reason to take being as

the defining object of epistêmê. First, as we saw in Chapter 2, Plato is here
explicitly proposing a method for identifying powers by their objects, and
so the fact that he here names being suggests that this is the most inform
ative characterization of epistêmê’s object. Second, Plato presents the claim
that epistêmê is of being not just as true, but as obviously true—indeed, as
something that everyone will accept.
This second point emerges from the context. Socrates is taking as his
imaginary interlocutor a “lover of sights,” an aesthete who considers himself
an expert about beauty. The sight-lover’s views about epistêmê sharply
oppose Socrates’ own: the sight-lover thinks that he has epistêmê, while
Socrates will argue that only philosophers do; the sight-lover thinks that
epistêmê comes from experience with perceptible things, while Socrates will
argue that it instead requires a grasp of intelligible Forms. In arguing for his
positions, however, Socrates wants to begin from premises that are accept
able to his opponent: the plan is to “soothe and persuade him gently, hiding
from him that he is unhealthy [in mind]” (476e).4 Here are the premises he
chooses:
4 Fine interprets this as “the dialectical requirement”: Socrates must argue from premises
acceptable to his interlocutors (Fine 1990; cf. Fine 1978). Whether or not this is a generally
promising interpretation of Socratic dialectic—and I suspect, given Socrates’ tendency to use
Does the one who has gnôsis have it of something or nothing?5 . . . – He has
gnôsis of something. – Something that is [on] or that is not? – Something
that is. For how could what is not be the object of gnôsis [gnôstheiê]? – Are
we then sufficiently assured of this, even if we examine it in many ways,
that what completely is is completely an object of gnôsis [to men pantelôs
on pantelôs gnôston], and what is in no way is in every way not an object of
gnôsis? – Most sufficiently . . . – Gnôsis is over what is [epi men tôi onti] . . .
Epistêmê is by nature over what is, to have gnôsis of what is as it is [gnônai
hôs esti to on] (476e–477b)
Socrates is claiming that epistêmê is of or over what is, and he is claiming

this almost without argument. He does give substantive arguments that
epistêmê has a different object from doxa (477c–478a), and that doxa’s
object is what is between being and not-being (478b–479d), recognizing
both of these as controversial claims; that epistêmê is of what is, however, he
presents as obvious. Indeed, he assumes that this claim will be readily
accepted even by interlocutors who disagree with him radically about other
features of epistêmê, and therefore can be used as a premise they accept in
an argument that will undermine their view. The sight-lovers think that
they themselves have epistêmê (476d); Socrates assumes that they must
therefore think that the objects to which they attend, beautiful sights, are
what is. If he can show them that beautiful sights are not in fact what is, as
he purports to do with his arguments about the contradictory nature of per-
ceptibles, they will thus have to abandon their claim to epistêmê.6 That
epistêmê is of what is thus functions as a shared premise throughout the
debate over who has epistêmê.
Why is Socrates so confident that ‘epistêmê is of what is’ is both true and
uncontroversial? It seems that he is taking this to be a generally accepted
constraint on any account of epistêmê. Socrates can assume that all parties
to a debate about epistêmê will agree to this claim, because anyone who
words in ways that his interlocutors misunderstand, that it is not—it does seem that he is here
going out of his way to emphasize that his first premises are ones his interlocutor will accept.
5 ὁ γιγνώσκων γιγνώσκει τὶ ἢ οὐδέν. As we have already observed, Socrates in this argument

seems to use gnôsis interchangeably with epistêmê, and to use gignôskein as the verb correlate with
both: note especially the wholly unflagged shift from talk of gnôsis to talk of epistêmê at 477a–b.
6 Indeed, while we never get an explicit statement of the sight-lovers’ ontology, we do get
one for their representatives in the Cave allegory: cave-prisoners think that the shadows are
what is (onta: 515b, following Slings).
denies it would have a very dubious claim to be talking about epistêmê at all.
We might call this a basic principle about epistêmê: something assumed as
not only true but beyond dispute. (If you think Plato is a believer in concep-
tual truths, you can say that he treats this as one of those. Alternately, you
might think that he is treating it the way Aristotle treats the claims he takes
as endoxa: commonly accepted views that should be respected by theory.
I will stick with ‘principle’ as less theory-laden than either of these.)
Notably, the claim that epistêmê is of what is shows up in another dia-
logue playing just this same role, as an unargued premise treated as beyond
dispute, in an argument between opponents with sharply differing views of
epistêmê.
The dialogue is the Theaetetus.7 This makes the parallel particularly strik-
ing, for Socrates’ approach to epistêmê here differs sharply from his
approach in the Republic. There he puts forth a radical epistemology of his
own, on which epistêmê is the province only of philosophers, because it is of
Forms rather than perceptibles; in the Theaetetus, he carefully considers the
hypothesis that epistêmê is perception rather than rejecting it out of hand,
and even after that hypothesis is refuted frequently uses as examples of
epistêmê’s objects particular perceptible objects or facts.8 If we find Socrates
in both contexts treating as indisputable that epistêmê is of what is, that is
further evidence that Plato takes this claim to be included in any account of
epistêmê—a basic principle. Therefore, while I will treat the Theaetetus’ epis-
temology in more detail in Chapter 10, I will give an initial discussion of
this aspect of it here.
The claim that epistêmê is of what is emerges in the first section of the
dialogue. Theaetetus suggests that epistêmê is perception (aisthêsis);
Socrates argues that he is thereby expressing Protagorean relativism,
understood as the view that as things appear to each person, so they are
7 Cornford recognizes this similarity between the two arguments, and identifies a second,
by equating the Theaetetus’ claim that epistêmê is without falsehood (apseudes) with the
Republic’s claim that it is unerring (anamartêton). He takes both claims to name attributes that
“any candidate for the title [of epistêmê] must possess on Plato’s views” (Cornford 1935, 29).
Indeed it is widely recognized that the Theaetetus treats these two features of epistêmê as
beyond dispute. See among others Burnyeat: “It has been agreed from the start (152c) that any
candidate for knowledge must pass two tests: it must be always of what is and it must be unerr-
ing” (Burnyeat 1990, 8).
8 There is much to be said about the relation between the two dialogues, and in particular
about the significance of the Theaetetus’ silence about Forms; here I want just to note that given
these apparent differences, it is particularly noteworthy that both treat “epistêmê is of being” as
a basic principle.
for that person. What is the connection between the two views? Here is
Socrates’ argument:
[On Protagoras’ account], in matters of hot things and all things like that,
as each person perceives things, so they are for that person . . . Perception
therefore is always of what is, and without falsehood, as befits epistêmê.9
(152c, emphases added)
We can make sense of the argument as follows:10
(1) ASSUMPTION: Epistêmê is always of what is, and without falsehood.

(2) Therefore if perception is epistêmê, perception must always be of
what is and without falsehood.
(3) Therefore if perception is epistêmê, my perceptions are true-and-of-
what-is for me, your perceptions are true-and-of-what-is is for
you—that is, Protagorean relativism holds.
ASSUMPTION is never explicitly stated, let alone defended. Given the

context—Socrates is drawing out the presuppositions or consequences of
Theaetetus’ own account of epistêmê—he must be assuming not only that
the claim is true, but also that Theaetetus accepts it.
This is confirmed when the claim resurfaces in Socrates’ final refutation
of Theaetetus’ account. The refutation begins with a difficult and much-
discussed argument that perception does not grasp being (184b–186e).11
There is a good deal of contention in the secondary literature as to what
Socrates here means by being (ousia); I will discuss the question briefly in
Chapter 10. What is important for my purposes here however is neutral
between all interpretations: whatever being is, Socrates assumes that epistêmê
9 αἴσθησις ἄρα τοῦ ὄντος ἀεί ἐστιν καὶ ἀψευδὲς ὡς ἐπιστήμη οὖσα. The hôs clause is confus-

ing, since it looks like it treats the equation of perception with epistêmê as a premise in the
argument rather than its conclusion; for an argument that it should be deleted, see White 1972.
If we retain it, Levett’s translation, “as befits” works best. (Some translations limit the applicant
of this clause to the second feature: “Perception, then, is always of something that is, and, as
being knowledge, it is infallible” (Cornford; cf. McDowell), but it is widely recognized that the
argument demands that we take both features as features perception must have if it is epistêmê
(see quotations from Cornford’s commentary below).
10 For a different construal see Burnyeat (1982); our readings coincide on the point crucial
to my discussion here, the uncontroversial status of what I call ASSUMPTION.
11 Perception cannot grasp or apprehend or hit on or lay hold of being: λαμβάνειν, 185b,
τυχεῖν, 186c, ἅψασθαι, 186d, 186e). Plato seems to use these terms interchangeably; I will use
‘grasp’ to cover them all.
must grasp it. For having established that perception does not grasp being,
he proceeds as follows:
Perception has no share in grasping [hapsasthai] truth, since it has no [share

in grasping] being [ousia], and therefore neither [has it any share in]
epistêmê.12 And therefore perception is not the same as epistêmê. (186e)
Socrates is appealing again to ASSUMPTION, and again assuming that

Theaetetus accepts it. That is, he is arguing:
(4) Perception is not “of what is” (it does not grasp being).13
(5) Therefore it is also not “without falsehood” (it is not always true).
(6) Therefore, by ASSUMPTION, perception is not epistêmê.
Throughout the discussion, then, Socrates holds fixed this claim about
epistêmê and assumes that even proponents of the radical theory accept it:
whatever epistêmê is, it must be of what is. Thus just as in Republic V, the
claim that epistêmê is of what is appears with the status of an unargued
principle.14 (As does the claim that epistêmê is always true; I will discuss the
relation of these claims below.)
Thus we have seen that Plato presents the claim that epistêmê is of what is
as undisputed common ground between Socrates of the Theaetetus, Socrates
of the Republic, Theaetetus, and the Republic’s sight-lovers.15 Moreover, it
is likely that he presents the claim this way not because he thinks there is
something special uniting the views of these specific characters, but because
he thinks the claim will be common ground between proponents of any
theory of epistêmê. He treats it as a basic principle or perhaps even concep-
tual truth, as he does claims like “fear is of the terrible” or “sight is of color.”
12 οὐ μέτεστιν ἀληθείας ἅψασθαι: οὐδὲ γὰρ οὐσίας. . . οὐδ᾽ ἄρ᾽ ἐπιστήμης.
13 Is it significant that Socrates here uses ‘being’ (ousia), whereas at 152c he used ‘what is’ (to
on)? I think not: in leading up to Premise (4) Socrates has used the noun ousia interchangeably
with the verb einai (see especially 186b: “their being [οὐσία] and that they are [ὅτι ἐστὸν]….”
Moreover, the present argument is clearly offered as a refutation of Theaetetus’ hypothesis; on
the reading I have given, the argument is valid, while if ‘being’ in Premise (4) is distinct from
‘what is’ in Premise (1), it is not.
14 In Chapter 10 I will argue that it retains this status in the rest of the Theaetetus.
15 There is no reason to think that Theaetetus and the sight-lovers have the same views of
epistêmê, although we learn too little about either to be sure. Certainly there is no reason to
think the sight-lovers would accept the Protagorean relativism on which all perceptions are
true and thus every perceiver is an epistêmôn (possessor of epistêmê), for they think themselves
epistemically superior to others on the subject of beauty.
Which Sense Of ‘ being ’ ? 93
This is confirmed, as I will show below, by his unargued reliance on the

claim in many other contexts too.
If all this is right, then we should expect Plato to treat this as a fundamen-
tal feature of epistêmê, one he can rely on in developing his theories about
epistêmê even when he does not explicitly appeal to it. More specifically, if
Plato’s epistemology is objects-based, we should expect this to be the rele-
vant object. Here then is a proposal: Plato takes the defining object of
epistêmê to be being.
We are faced now with two tasks: to make sense of the proposal by figur-
ing out the relevant notion of being, and to test the proposal, seeing if it is
borne out by Plato’s characterizations of epistêmê throughout the dialogues.
I take on the first task in the remainder of this chapter. Once we see how
Plato is using ‘being’ in the relevant contexts, I will show that we can indeed
see this notion of epistêmê operative throughout the dialogues, making
sense of his characterizations of epistêmê (Chapter 4).
2. Which sense of ‘being’?
If you are not steeped in scholarly controversies about Plato’s notion of

being, you might think we have discovered something fairly simple, and
furthermore something that strongly supports the view that epistêmê is
knowledge more or less as we conceive it now. You might interpret the ‘what
is’ in the phrase ‘epistêmê is of what is’ as meaning the facts, what is the case,
and take the claim to be that epistêmê is veridical or factive. (Or you might
even take it to mean what is true, i.e. true propositions, and reach the same
result.)16 Thus you would take the claim we have identified as a basic prin
ciple about epistêmê to amount to a claim nearly all philosophers now rec-
ognize as a basic principle about knowledge: knowledge is always true.
If you are steeped in scholarly controversies about Plato’s notions of
being, however, you will know that there are many, many competing
interpretations. The secondary literature on the meaning of ‘being’ in our
Republic and Theaetetus passages alone offers an embarrassment of options.
Indeed, you may think that Plato’s notion of being is so ambiguous, so
obscure, and so contested, that the project of using it as a starting-point for
understanding anything else is deeply misguided.
16 Thus Fine 1978 and 1990, and Gosling 1968.

Much of the literature presents three options as exhaustive:17 when Plato

says that epistêmê is of being, he intends either (i) an existential sense of
‘being,’ so that epistêmê would be of what exists (e.g. Stokes 1992, on the
Republic), (ii) a veridical sense, so that epistêmê would be of true proposi-
tions (Fine 1990 on the Republic) or of actually obtaining states of affairs
(Kahn 1981, 112–13), or (iii) a predicative sense, so that epistêmê would be
of what is F, for any predicate F (among many others Annas 1981 on the
Republic, and Burnyeat 1976 on the Theaetetus). There are however many
more interpretations out there. Perhaps Plato intends (iv) a special genuine-
predicate sense, so that epistêmê would be of what is genuinely or really F,
by contrast with imitations (Vlastos 1981 on the Republic). Or (v) an essen-
tial sense, so that epistêmê would be of the essence of Fness (Silverman 2002).
Or (vi) a specially Platonic metaphysical sense, so that epistêmê would be of
the Forms (Cornford 1935 on the Theaetetus). Or (vii) an objective sense, so
that epistêmê would be of what is objectively the case rather than what
merely appears so (Cooper 1970 on the Theaetetus). Or (viii) some “fused”
sense which incorporates two or more of these (Bosanquet 1895, Kahn 1981,
Brown 1994).18
This is an important and difficult controversy about which an enormous
amount has been written. If we have to settle it for each dialogue in order to
understand epistêmê, we have a truly daunting task ahead of us. Indeed, we
might be strongly inclined to follow Vlastos in suggesting the opposite
course: in the Republic at least, we should take Plato’s theory of being to
derive from his notion of epistêmê: what it is to be is to be well-suited to be
the object of epistêmê, where Plato is starting from an independently fixed
conception of epistêmê.19
17 For the three-item menu see among others Fine 1990, Annas 1981 at 196, Taylor 2008,
Lee 2010.
18 Kahn argues that many of Plato’s uses of ‘being’ are “overdetermined,” requiring that we
read more than one sense: for example, in Plato’s characterization of Forms at Phaedo 65d,
“Veridical idea, existence claim, and predicative syntax are all taken up” in the word ousia
(1981, 109); cf. Brown (1994).
19 On Vlastos’ view, Plato’s notion of being in the powers, Line, and Cave arguments in the
Republic amounts to reality, and the real is “that which is cognitively dependable, undeceiving”
(1981, 7). More specifically, Plato starts with a notion of epistêmê as a priori “logical certainty”
(ibid., 17), and assigns being to things to the degree that they admit of that kind of knowledge.
On this view, the Form of Beauty is such that we can know logical truths about it, and that is
what Plato means in giving it a superior ontological status; beautiful perceptibles are not sus-
ceptible to such knowledge, and that is what Plato means in giving them an inferior onto
logical status.
Which Sense Of ‘ being ’ ? 95
I want to show that the problem is at once harder and easier than it
seems. On the one hand, when correlating epistêmê with being Plato uses
‘being’ in such a variety of formulations that the task of finding a unified
syntactic account looks almost hopeless (see the formulations mentioned in
Contrasts 1–4 below). On the other hand—and crucially for my project—
we do not in fact need to determine the sense of ‘being’ in each passage in
order to fix on the epistemically relevant feature of Plato’s notion of being.
Our account will be compatible with many different interpretations of
‘being,’ as well as with the possibility that the word has different meanings
in different contexts, or that Plato’s usage is simply indeterminate. To under-
stand the basic concept of epistêmê as cognition of being, we need only
insist on one feature of Plato’s notion of being, one which is compatible with
a whole range of other details.
Here is my proposal. When Plato labels certain objects (e.g. the Forms) as
being, or says that they are, and goes on to say that they are the objects of
epistêmê, ‘being’ (or ‘to be’) has one constant function. This is the function
of assigning its referents a privileged status among all the things that exist.
Plato is saying or implying that these things most are, or perhaps really are.20
He is marking them as ontologically superior to other things, things which
do exist but lack this special status: things that are in a derivative sense, or
dependently, or to a lesser degree—the ontologically inferior. The notion of
ontological superiority is, I submit, one that Plato takes for granted without
explicitly theorizing. I say a bit about how to understand it in Section 4.
As to the meaning of ‘being’ in these various contexts, I leave open two
possibilities, on the grounds that I cannot see a way to decide between
them, and suspect that Plato did not determinately distinguish them. First,
perhaps ‘being’ has one of the meanings generally recognized—and perhaps
even different meanings in different contexts. (For example, perhaps it has
predicative-cum-existential import in Republic V’s powers arguments, while
at Euthyphro 11a (quoted in Chapter 9) it instead means essence.) But in
these contexts, Plato uses ‘being’ to express not just one of these familiar
senses, but a stronger claim, best captured by adding a ‘really’: he is saying
not simply for example that the Form is beautiful (the predicative reading),
or exists (the existential reading), but that it really is beautiful, really exists,
and so on—where this ‘really’ entails that it is ontologically superior.
20 This is not to be confused with what I have called the genuine sense, on which “the Form
of Beauty is” means that the Form of Beauty is genuinely beautiful, by contrast with particulars
which are sham imitations (Vlastos 1981; cf. Moline 1981).
Second, perhaps in at least some of these contexts ‘being’ simply means

ontological superiority: we should translate simply “the Form Is,” full stop.
In either case, the claim Plato takes as a basic principle is that epistêmê is
of the ontologically superior.21 Let us mark this use of ‘being’ and variants
with capitalization: Plato is saying that epistêmê is of Being, what Is.
3. Being as the ontologically superior22
It is not my purpose here to defend a detailed interpretation of Plato’s meta-

physics, neither across the corpus nor in any one dialogue. I want instead to
use a bird’s-eye overview to point out a general pattern throughout the Two
Worlds dialogues (and indeed, I will argue in Chapters 9 and 10, in the earl
ier and later dialogues as well). The pattern I have in mind should be fairly
obvious, and its existence fairly uncontroversial. It is this: throughout the
dialogues, and in a wide variety of contexts, Plato draws contrasts between
superior and inferior items, and uses variants on ‘being’ to label the superior
items. Moreover, wherever he does this, he correlates Being with epistêmê,
and the inferior items with doxa. In other words, Plato’s works show
variations on the following argument:
• Some things are special, and superior to the rest, in virtue of having
some feature x: for example stability and permanence, or genuineness, or
being an essence. (Plato gives different values for x in different contexts.)
• The x-having things are beings or what is (or some similar formula); the
x-lacking things have some lower ontological status.
• The x-having things are the objects of epistêmê, the x-lacking things the
objects of doxa.
The pattern is particularly explicit in the Two Worlds dialogues’ contrast

between Forms and perceptibles. Here Plato uses ousia and einai in a variety
of formulations to characterize Forms. The Form of something is a being
(ousia, Phaedo 77a), or the being itself (autê hê ousia, Phaedo 78d), or what
is (ho esti, Republic 507b), or that itself which each thing is (auto to ho
21 This raises the question of whether Socrates is fair after all in taking interlocutors like the
sight-lovers or Theaetetus to agree with the claim that epistêmê is of what is, i.e. in taking his
characterization of epistêmê to capture an uncontroversial, widely shared view; I address this
question below, in Section 7.
22 My discussion of the connection between epistêmê and being in this section and the next
has some important similarities to Szaif 1996.
Being As The Ontologically Superior 97
estin,23 Phaedo 78d; cf. 75b, 75d), or that which good or just is (Phaedo 75d),
or what a bed is (Republic 597a). Or Forms are in a superlative way: the
Form is what really is (ontôs esti, Republic 597d, Phaedrus 247c, Sophist
248a) or what fully or utterly or purely is (pantelôs on, elikrinôs on, Republic
477a; teleôs on, Republic 597a), or what is in nature (en têi phusei ousa,
Republic 597b), or what is as much as can be (einai hôs hoion te malista,
Phaedo 77a).
As for perceptibles, while Plato often casually refers to them as things
that are,24 he also often explicitly contrasts them with Forms by denying
them that label and assigning them instead some other status. That is, he
not only says that the Forms are in some special way, but in so doing contrasts
them with ordinary things, in a variety of ways. Looking at these contrasts
will give us more information about what makes Forms count as Beings:
(Contrast 1) What fully or purely is versus what is between being and not-
being (Republic 479c–e)25
(Contrast 2) Being versus becoming (genesis) (Timaeus 27d–28a, Republic
534a, Sophist 248a, Philebus 59a)26
(Contrast 3) What is F (or what the F is) versus “the many F things” (Phaedo
74d–76d, Republic 507b
(Contrast 4) What is versus its likenesses or images (Phaedrus 250a–250b,
Republic 597a)27
23 For discussion of this difficult phrase see Gallop’s notes ad loc (1975, 229–30).
24 Consider for example the contrast at Republic 507b between “the many beautifuls and
many goods . . . which we say are [εἶναί φαμέν]” and “the beautiful itself and the good
itself . . . which we label what each thing is [ὃ ἔστιν ἕκαστον προσαγορεύομεν];” cf. Phaedrus
249c, where “the things we now say are [ἃ νῦν εἶναί φαμεν]” are contrasted with “that which
really is [τὸ ὂν ὄντως]”.
25 The argument begins with a distinction between what fully (παντελῶς) is and what is
between purely being and in no way being (477a), and ends by identifying the Forms as the
former, perceptibles as the latter (479c–e).
26 “With the body we share in becoming, but through reasoning and with the soul we relate
to what really is [τὴν ὄντως οὐσίαν]” (Sophist 248a). Timaeus 27d–28a contrasts “what is always
being and has no becoming” with “what is always becoming and never is….becoming and get-
ting destroyed, but never really being [τὸ γιγνόμενον μὲν ἀεί . . . ὄντως δὲ οὐδέποτε ὄν]. Philebus
59a contrasts “things that always are” with “things that become and will become and have
become.” Throughout Republic VII Plato refers to the perceptible realm symbolized by the cave
as Becoming, and the intelligible realm of Forms as Being: see for example 518c and 534a.
27 A beautiful perceptible thing is a likeness (ὁμοίωμα) or image (εἰκών) of the Being (viz.,
the Form) (Phaedrus 250a–250b); the carpenter “does not make the Form, which we say is
what Is a bed [ὃ ἐστι κλίνη]…if he does not make what Is, he does not make something that Is,
but something such as to be like something that Is, but is not [εἰ μὴ ὃ ἐστιν ποιεῖ, οὐκ ἂν τὸ ὂν
ποιοῖ, ἀλλά τι τοιοῦτον οἷον τὸ ὄν, ὂν δὲ οὔ]. For if someone said that the carpenter’s bed is
something that completely [τελέως] Is . . . they would not be speaking truly” (Republic 597a).
These contrasts look to be different from one another, and they pull us
toward different explanations of the superiority of the Forms. Indeed, it is
no surprise that students of Plato’s metaphysics have come up with quite
different interpretations of the nature of the Forms. I will say something
about what ‘being’ might mean in these different contexts; then I will argue
that we need not in fact settle the question of its meaning in order to see the
crucial thing these contrasts have in common.
Contrast 1 is argued for on the basis of the claim that each of the many F
things will also have, or appear to have, the opposite quality not-F, while the
Form is always the same as itself (Republic 479a–b; for similar contrasts see
Phaedo 78d–e, Symposium 211a). It seems that here feature x—the feature in
virtue of which Forms alone are entitled to be called beings—is purity, by
contrast with admixture of opposites. If we confined our attention to pas-
sages like this, we would be inclined to say that ‘The Forms are’ simply
means that the Forms are pure instances of the corresponding qualities—
hence the popularity of the predicative interpretation of ‘being.’
Contrast 2 is slightly different: perceptibles are what merely become
rather than what is because they change; evidently feature x here is stability.
One could of course try to reconcile these two contrasts by making the first
basic: to be is to be pure; stability is necessary for purity because change
brings admixture of opposites. But the label ‘becoming’ very notably
emphasizes change as the mark of the ontologically inferior, rather than
impurity. If we confined our attention to the passages drawing Contrast 2
alone, we would be inclined to say that ‘The Forms are’ means that the
Forms are stable, eternal, unchanging: to use Kahn’s phrase, the verb ‘to be’
here seems to have “static value” (1981, 111).
Contrast 3 is often drawn on the basis of Contrast 1: in the very contexts
where Plato distinguishes between the Form which is always F and the
many Fs which are in some way not-F, he refers to the first as “the what is
F,” “the what the F is,” “the F itself,” or even “the nature of the F itself.”28 Yet
this way of putting it seems to make a claim not about purity (as in Contrast 1)
but instead about something else: the implication is that the Form is the
essence of F-ness, or is identical with the property F-ness, or has that prop-
erty in some primary, essential way, while the many Fs merely share in it
28 For this last see Republic 476b: lovers of sights and sounds love the beautiful sounds and
colors and shapes and all the things fashioned out of such things, but their thought is unable to
see and love the nature of the beautiful itself (αὐτοῦ δὲ τοῦ καλοῦ . . . τὴν φύσιν). For the other
formulations see among others Republic 507b, Phaedo 76b.
Being As The Ontologically Superior 99
derivatively. Here feature x is being an essence, and it is tempting—albeit

semantically awkward—to translate ‘being’ in these various formulations as
‘essence.’
Contrast 4 looks to make a different point altogether. Here feature x is
originality or genuineness: being the real thing by contrast with a copy.
Coupled with the pervasiveness of model-copy imagery in Plato’s depiction
of Forms even where he does not explicitly contrast what is with what
appears—in the Cave allegory, for example—these passages incline us
toward a reading on which ‘The Forms are’ means that the Forms are the
genuine articles.
Thus Plato seems to make quite disparate claims in different contexts
about the nature of Forms and their superiority to perceptibles, and there-
fore to imply quite different meanings for ‘being. This leaves students of
Plato’s metaphysics with the task of trying to reconcile these various mean-
ings by singling out one as basic. (For example, if ‘being’ ultimately means
being an essence, then Forms’ purity, stability, and role as paradigms all fol-
low from their being essences.29 Or if ‘being’ ultimately means being genu-
ine (by contrast with copies), all the rest follows from this: paradigms are
more stable, more pure, and better guides to the essence than their copies.)30
That is not my task, nor do I wish to take a stance on whether or not it is
a feasible one—on whether Plato does in fact have a single unifying view
about Forms, or instead has a plurality of independent views. I want simply
to point out that in all these contexts, despite the questions about its mean-
ing, the label ‘being’ plays the same role: it serves to mark out one class of
things as superior to the rest. ‘Being’ does not always have this function:
sometimes Plato uses it simply to say that something exists, or has some
property, or is the case, without further implication. In Contrasts 1–4, how-
ever, it is used to contrast its referents with other things, and to elevate them
above the rest. When Plato puts the claim that some things are more pure,
or stable, or original, than others by saying that these things alone are, he is
ascribing these things a privileged status in the order of reality. One promis-
ing although vague way to put it: he is saying that these are somehow what
really is. Before we consider what this might mean (Section 4), let me say a
bit more about how I take the implication to work.
29 For a version of this approach see Silverman 2002, 19–20: Being is a primitive relation
between a thing and its essence; the other ways that things are (such as identity, and existence)
flow from this.
30 Moline 1981.
As we saw above, sometimes Plato makes the privileging explicit by

qualifying ‘being’ with some intensifier like ‘fully’ (pantelôs, Republic 477a),
‘purely’ (eilikrinôs) (Republic 477a), ‘completely’ (teleôs, Republic 597a), or,
most often, with its own adverbial form, ‘beingly’ (ontôs, Timaeus 28a,
Phaedrus 247c–e, 249c, Sophist 248a, Republic 597d, among many others).
This last is often translated as “really,” and this is apt so long as we under-
stand ‘really’ in the special metaphysical way I will detail below, on which it
denotes ontological superiority. For example:
Is there something that is always being and has no becoming, and some-
thing that is always becoming and never is . . . . becoming and getting
destroyed, but never really Being [ontôs . . . on]? (Timaeus 27d–28a)
Here Plato moves from a simple statement of Contrast 2—Forms are stable
while perceptibles change—to the claim that the Form alone really is. My
proposal is that even where there is no such explicit move, we should under-
stand an implicit ‘really’ wherever we see ‘being’ as a contrast term. For
example, in Contrast 3, we should understand a contrast between the many
Fs and what really is F, or what the F really is.
Questions will remain about the sense of ‘being’ at issue in these different
contexts and—to repeat—I am not trying to settle them. I am instead saying
that whatever ‘being’ means in these contexts, it is amplified by an implicit
really.31
For example, in the powers argument of Republic V, when Plato moves
from claiming (1) that every one of the many beautiful things is sometimes
ugly (479a–b), to claiming (2) that the many beautifuls are “between being
and not-being” (479c–e), there are two possibilities. First, perhaps ‘being’ in
claim (2) has one of the senses others have proposed, but intensified by an
implicit really. Perhaps for example it has predicative sense, but the claim is
that the perceptibles are “between being really beautiful and not being beau-
tiful.” Plato is thus not simply repeating claim (1), but taking it to entail that
the perceptibles are not really beautiful. Or perhaps being in claim (2) has
31 I am as throughout bracketing Fine’s propositional reading, since I think Forms rather
than propositions are at issue in all these contexts. There is in fact an easy way to extend the
view to the propositional reading (although it is not at all what Fine has in mind): true proposi-
tions exclusively concerning Forms are really true; propositions involving perceptibles are at
best derivatively or loosely true. Indeed I think Plato would accept just this claim: see my dis-
cussion of truth in Chapter 4.1 and Chapter 7.6.
What Is Ontological Superiority? 101
existential sense, but again we must supply a really. Plato is thus not making
the bizarre claim that perceptibles only partially exist, but instead the para-
digmatically metaphysical claim (more on this in a moment) that they are
not what really exists. Likewise for other proposed interpretations: we can
accept one of the extant interpretations, but add a really. Second, perhaps
‘being’ in claim (2) simply means ontological superiority: the claim is that
perceptibles are not what really is, full stop—that is, the claim is that they
are ontologically inferior.
(Note that on either of these proposals there will remain a mystery about
how this argument at Republic V works. On any of the accounts I am offer-
ing, Plato owes us an argument connecting claim (1), about the impurity or
instability of the many beautifuls, to claim (2), about their ontological
inferiority. Why must the ontologically superior Beings be pure? That is a
question about Plato’s metaphysics that I will not undertake to answer here,
although I suspect that it is in part a matter of metaphysical prejudice
inherited from Parmenides, and perhaps in part due to an epistemologically-
driven argument implicit in one attributed to him by Aristotle:32 given that
epistêmê is stable, that cognition is of like-by-like, and that epistêmê is of
Being, Being must be stable, and hence pure.)
4. What is ontological superiority?
But what does it mean for something really to be—to be ontologically

superior? One might protest that this notion is empty until we decide what
‘being’ means. Is the idea that the ontologically superior exists more than
the ontologically inferior? That it is more F for some salient predicate? That
it is more stable, more genuine, more of an essence? In fact, however, the
notion of ontological superiority is intelligible in its own right—or at the
least, it is the sort of thing that Plato might have taken to be intelligible in its
own right, and indeed as not needing explanation.
Certainly it is an idea we find in his student. Aristotle argues that while
“being [to on] is said in many ways,” one kind of thing “is first” (prôton on);
note that like Plato he reserves ‘Being’ (ousia) in a strict sense for this kind
32 See Metaphysics 1078b12–17, quoted at the end of Chapter 2.

of thing (Metaphysics1028a10–15).33 (I say more about Aristotle’s notion of

primary being below). It is an idea that survives in modern metaphysics
too; indeed, it is arguably a basic metaphysical notion. Thus for example
K. Fine: “[T]here is a primitive metaphysical concept of reality, one that
cannot be understood in fundamentally different terms” (2001, 1).34
Is there anything more we can say, to give the notion substance? I hesitate
to attribute to Plato any very developed view: I see him as working with an
under-theorized notion which he takes to a large extent for granted.
Moreover, we might agree with Fine that the notion is primitive, not explic
able in other terms.
Nonetheless, Plato may well at least sometimes have in mind a view that
Aristotle attributes to him, and himself endorses in his theory of primary
being: the real is the fundamental—that on which other things depend. As
Aristotle puts it, “If the primary beings were not, it would be impossible for
anything else to be” (Categories 2b5–6). There is plenty of debate about how
to understand this notion, some of it echoing the dispute about Plato: is
being in this context existence? Essence?35 There is also a question about
whether Aristotle thinks that these dependence relations actually constitute
what it is to be a primary being, or are merely entailed by primacy. Without
settling these questions, however, we can recognize the crucial point that
Aristotle clearly had a notion he expressed in terms of ontological superior-
ity (primacy) on which the ontologically superior items are those on which
others depend. Moreover he thinks he finds exactly this notion in Plato:
Some things are said to be prior . . . in nature and being [protera . . . kata

phusin kai ousian], as many as admit of being without other things,
while those do not admit of being without them; Plato made use of
this distinction. (Metaphysics 1019a1–4)
33 Cf. Metaphysics 1019a4–6. Translations often give “substance” for Aristotle’s ousia, but it is
the same term we have seen translated in Plato as “being”—the verbal noun from εἶναι, “to be.”
34 Or: “When I was introduced to metaphysics as an undergraduate, I was given the follow-
ing definition: metaphysics is the study of ultimate reality. This still seems to me the best defin
ition of metaphysics I have seen” (van Inwagen 2015, 1, 4). For more on Fine’s and van
Inwagen’s views see below. For some discussion of the history of the view that metaphysics is
about reality, and defense of a version on which the notion of reality is specified as existence,
see McDaniel 2017.
35 See for example Peramatzis 2011, defending an essential reading: for one thing to be prior
for another is for the first to be what it is independently of the other but not vice versa.
What Is Ontological Superiority? 103
There is no consensus that any of Plato’s writings testify to precisely this

claim, nor even about what aspect of Plato’s theory Aristotle here has in
mind. It is however widely recognized that Plato construes perceptibles as
depending on the Forms, which are by contrast independent of percepti-
bles: consider most explicitly the Phaedo’s argument that Forms are the
causes, aitiai, of their participants’ properties (Phaedo 100c), or the perva-
sive idea represented in our Contrast 4 that Forms are the models of which
their participants are copies. Plausibly Plato, just like Aristotle, took this to
be part of the idea of ontological superiority: for the Beings to be real is in
part a matter of, or entails, them being the foundations of all else. (Note that
on this view the Form of the Good, on which other Forms depend for their
Being (Republic 509b), will be the most real thing, the supreme Being, while
the other Forms will be ontologically inferior to it. This is certainly one way
to make sense of the notoriously difficult claim that the Good is the source
of other things’ Being, and is itself “beyond being” (epekeina, 509b).36)
If this is Plato’s conception of ontological superiority, we can see him as
inheriting it from the Presocratics. In the Sophist, he refers to Presocratic
attempts to “define the things that are [ta onta], how many and of what sort
they are” (Sophist 242c).37 He goes on to say that we need to investigate the
notion of being, opening up important questions about its nature, but as
these lines show, the Sophist’s whole discussion presupposes a general notion
of being that Plato takes Parmenides and others to share. The Presocratics
were disputing over the nature and number of the primary or fundamental or
most real things, what Aristotle in discussing their theories will often refer to
as the archai, foundations or starting-points or first principles (Metaphysics
983b19), and sometimes as ousia, Being (Metaphysics 1028b3–6). Indeed,
this looks to be a basic metaphysical notion, from the pre-Socratics’ time
36 Thus Adam (1902, ad loc.): “As the cause of γένεσις, we may, in fact, regard the Sun as the
only true γένεσις, for all γιγνόμενα are derived from him. Similarly, the Good is not οὐσία in
the sense in which the Ideas are οὐσίαι; but in a higher sense it is the only true οὐσία, for
all οὐσίαι are only specific determinations of the Good.”
37 τὰ ὄντα διορίσασθαι πόσα τε καὶ ποῖά ἐστιν. He goes on to speak of some who report that
there are three beings (242c), or two or one (242d; the last are the Eleatics), or both one and
many (242e). Cornford translates τὸ ὄν in this part of the Sophist as “the real” or “reality,” say-
ing “This sense of the word has emerged from the contrast between the sort of existence
belonging to an eidolon [image, copy], and the real existence of the ὄντως ὄν” (1935, 216 note 1,
comment ad 242b–244b). Ι think this is on the right track, but too narrow: even if Plato him-
self sometimes wants to analyze what Is as what is real by contrast with images, he is unlikely to
be attributing this view to Parmenides and the other pre-Socratics. Instead, he is pointing to a
more general notion he shares with them: that of the ontologically superior.
down to ours, appearing variously in the guise of archai, substances, or fun-

damental realities, undergirding theories of degrees of reality and of
grounding. Certain things are more real than others, and the less real things
are somehow dependent on or derivative of the more real.
I suggest then that in the Two Worlds dialogues Plato is working with an
inchoate theory of ontological superiority as fundamentality: the things that
really are, the Beings, are the grounds of the other things that exist. I do not
however propose that this is definitively what ontological superiority
amounts to, for Plato—that is, I do not propose that we read this notion into
every contrast he draws between Being and something inferior. Rather, this
is one way he began to develop a more general, more neutral, and perhaps
even primitive notion of the real. Perhaps when he appeals to ontological
superiority in the Socratic dialogues or in the Theaetetus (as I will argue that
he does, in Chapters 9 and 10), he does not have fundamentality in mind at
all. Perhaps he does not always have it in mind even when privileging the
Forms as Beings.
There is another widespread way to flesh out the notion of ontological
superiority: the idea is that what is real is contrasted with what seems, or
appears. Some take this to be the core contrast that defines the notion of the
real.38 I think this too specific to capture the general notion we find in Plato.
For him, it will turn out to be one of several ways of developing the contrast
between what Is and something inferior, alongside the ones we have seen
above: I discuss it as Contrast 5, in Chapter 6. It will be a very important
contrast, and absolutely central to his epistemology, for—I will argue—it is
the contrast that underlies his notion of doxa. Nonetheless, I see no reason
to think that this is what Being means for Plato, nor that every time Plato
contrasts what Is with something inferior, this is what he has in mind. The
contrast with what seems captures an epistemologically salient aspect of
Plato’s notion of the real; it does not exhaust that notion.
Indeed, the safest characterization of reality, or ontological superiority, is
a very neutral one: just as in calling some things good one privileges them
above others on the scale of value, so in calling some things Beings one
privileges them above others on the scale of being. Uninformative though
this characterization may be, it is not empty. The idea that some things are
38 For example, K. Fine claims that the “metaphysical conception of reality” serves to “dis-
tinguish, within the sphere of what is the case, between what is really the case and what is only
apparently the case” (2001, 3); van Inwagen fleshes out the claim that metaphysics is about
ultimate reality as: “Metaphysics, then, attempts to get behind appearances and to tell the
ultimate truth about things” (2015, 4).
Epistêmê and the Ontologically Superior 105
evaluatively superior to others is a substantive one—some will deny it; thus

even if you do not have a developed theory of value, you are still making a
substantive move in privileging some things above others as the good ones.
Likewise, the idea that some things are ontologically superior to others is a
substantive one; many will deny it, or even reject it as incoherent. Thus even
if you do not have a developed theory of being, you are still making a sub-
stantive move in privileging some things above others as the Beings.
5. Epistêmê and the ontologically superior
Now to the crucial point for our understanding of Plato’s epistemology:

throughout the Two Worlds dialogues (and outside of them too, as I will
argue in Chapters 9 and 10), wherever Plato draws contrasts between Being
and the ontologically inferior he correlates epistêmê with Being, and doxa
with its ontologically inferior counterpart.
We have already seen this in the case of Contrast 1 (the Forms fully are,
while perceptibles are between being and not-being). This contrast is drawn
in the context of Republic V’s powers argument: Plato introduces “what fully
is” as epistêmê’s object and “what is between being and not-being” as doxa’s
(477a–b, 478d); philosophers, the ones with epistêmê, attend to what purely
Is; those who confine themselves to what is between being and not-being
are philodoxoi, lovers of doxa (479d–480a).
When Plato draws Contrast 2—the Forms are, while perceptibles merely
become—epistêmê is again correlated with Being, while doxa is of the infer
ior class, becoming. For example, in a passage we have seen already from
the Timaeus:
Is there something that is always Being, but has no becoming, and some-
thing that is always becoming, but never Being? The former is grasped by
noêsis with reason [logos], always being in the same ways, the latter in turn
is the object of [doxaston] through doxa with unreasoned perception,
becoming and passing away, but never really Being [ontôs de oudepote on].
(Timaeus 27d–28a)
We see the same correlation throughout the Sun, Line and Cave images in
the Republic. In the Sun analogy, when the soul attends to what Is (to on), it
has gnôsis and nous, while when it attends to what becomes and passes away
(to gignomenon kai apollumenon), it has doxa instead (508d). In the Cave
allegory, the educational process that takes a student from doxa to epistêmê
“turns the soul away from becoming toward what Is” (521d). When Socrates
summarizes the entire discussion of the Line and Cave, he makes it most
explicit:
Doxa is about [peri] becoming, while noêsis is about Being. (Republic 534a)39
In the Philebus, those arts which study things that come to be must use doxa
(59a); epistêmê and nous “possessed of the highest truth” are of “what is
really and is always the same” (to on kai to ontôs kai to kata tauton aei)
(Philebus 58a; cf. 59c–d).40
When Plato draws Contrast 3—Forms are what each F Is, by contrast
with the many F things—again the Beings are the objects of epistêmê. We
see this in the Republic’s contrast between sight-lovers and philosophers:
those who contemplate Beauty itself have epistêmê, while those who attend
to the many beautifuls have only doxa (479d–e). In the Phaedo, the philoso-
pher attains the goal of wisdom when his soul makes contact with “what Is
[tou ontos]” and “the Being that each thing really Is” (tês ousias ho tunchanei
hekaston on) (65c–e).41
When he draws Contrast 4—Forms are what Is, while perceptibles are
likenesses or images of them—once again it is the Beings that are objects of
epistêmê. In the Phaedo and Phaedrus, it is only through direct, disembod-
ied apprehension of the Forms, rather than their perceptible likenesses, that
we can have epistêmê (Phaedo 65a–66a; Phaedrus 247c–248b).42
39 Noêsis has just been said to include both dianoia and epistêmê (533e–534a).
40 See Chapter 2.3 for a discussion of the Philebus’ treatment of the doxa-using crafts as
lowly forms of epistêmê. For our purposes here, the crucial point is that there is a correlation
between ontological and epistemic superiority: the highest or truest epistêmê is of what
really Is.
41 “It is through reasoning that anything of the things that are [τι τῶν ὄντων] become clear
to it. And [the soul] reasons most finely when [undisturbed by the body] . . . it reaches out for
what Is,” because never with the body can one touch the Just itself, the Beautiful, “or in a word
what the Being of anything is.. [and the one who ignores the body] because whenever it is in
communion with the soul it disturbs the soul and doesn’t allow it to obtain truth and phronêsis,
isn’t this the one . . . who hits on what Is?” (Phaedo 65c–66a).
42 Fine 2016 argues that the Phaedo countenances epistêmê of perceptibles, in two ways.
First, the argument of 65c–66a (quoted in the preceding note) identifies the cognition of Forms
as phronêsis, leaving it open that epistêmê may be of perceptibles. To this I reply that the argu-
ment concerns the best epistemic condition, the one possessed by philosophers, which Plato
calls by various names, but which he consistently restricts to Being (see Chapter 1.2). Second,
she argues that, just as on her reading of the Republic, Plato shows not that the highest kind of
Epistêmê is of what is, revisited 107
Throughout the Two Worlds dialogues, then, epistêmê is correlated with

the ontologically superior. In Chapters 9 and 10 I will argue that we also see
a version of this pattern in the earlier dialogues, and in the later dialogue
which gave us evidence above for the truistic status of the claim that
epistêmê is of being, the Theaetetus. Even without the full-blown distinction
between Forms and perceptibles, Plato consistently privileges certain items
with the label ‘being,’ and makes these the province of epistêmê by contrast
with doxa. At any rate, for now we have seen enough to conclude that when
the Republic’s powers argument casts epistêmê’s defining object as “what is,”
Plato has in mind Being—the ontologically superior.
6. Epistêmê is of what is, revisited
I have argued that despite important differences in Plato’s metaphysical

claims in different contexts, and significant questions about how to inter-
pret his various uses of ‘being,’ we can identify a consistent, unified, general
account of the epistemically relevant notion of being. The defining object of
epistêmê is the ontologically superior: ultimate reality. When Plato identifies
being as the object of epistêmê, he is operating with a hierarchical ontology
in which some items get dignified with the term ‘being,’ and others deni-
grated with a contrast term. Epistêmê is of the ontologically superior.
I am not saying that the differences in the varying features he identifies as
conferring Being are unimportant for our purposes. They will matter quite a
bit to the details of Plato’s account of epistêmê, as we will see in the next
chapter. For example, where essence is the crucial feature x that makes
something a Being, epistêmê will amount to a grasp of essences, entailing
the ability to give definitions. A full, comprehensive account of Plato’s
epistêmê would thus have to take on the metaphysical questions that I have
bracketed here. Does Plato have a single account of Being that unifies the
various contrasts we have seen by making one primary (for example, to Be
is to be an essence, and all the other features flow from this)? If so, he will
also have a unified account of epistêmê across the dialogues (for example, it
cognition is only of Forms, but that it requires cognition of Forms. I see no sign that Plato is
trying to leave open this possibility, however, and I suspect Fine’s argument is motivated by the
wish to avoid the alleged pitfalls of a Distinct Objects interpretation. If my arguments in
Chapters 1 and 2 are successful, they undermine that motivation.
is first and foremost to be understood as the cognition of essences, and all

its other features flow from this.) Or does he have a hodge-podge of inde-
pendent views about what makes something a Being, in which case the
details of his account of epistêmê will differ from context to context—some-
times it is a grasp of essence, sometimes of stable entities, sometimes of
genuine entities, with the different features those differences in object will
entail? I will not offer any answer to these questions, not because I think
they are unimportant, but because they are downstream from the project I
am undertaking in this book. That is the project of identifying Plato’s basic
epistemological concepts and aims. I hope in this chapter to have made pro-
gress on that project by identifying the defining object of epistêmê, at a suit-
ably general level: Being, i.e., the ontologically superior.
A final note: if Plato’s claim that epistêmê is of what is turns out to mean
that epistêmê is of the ontologically superior, then does this claim really
deserve the status of unargued premise that he gives it in the conversations
we began with, in Republic V and the Theaetetus? The sight-lovers explicitly
reject the existence of Forms; so too would Protagoras, whose view Socrates
claims to be developing in the Theaetetus. More generally, both figures
would reject a hierarchical ontology on which only certain items are funda-
mental, ultimate, real. It would seem then that Socrates is cheating badly
when he puts forth his claim as one that his interlocutors accept. Far from a
conceptual truth or endoxon (common view), “epistêmê is of what is” turns
out to be a heavily theoretical, deeply controversial doctrine.
Are we then forced to choose between rejecting the proposed interpret
ation of being and accusing Socrates of egregious deception? I think not.
Socrates is indeed talking past his interlocutors in these conversations, but
only in a subtle and very in-character way. He thinks that his interlocutors,
and indeed all people, do in fact regard epistêmê as cognition of the onto
logically superior, but that most people have a very weak, metaphysically
inadequate view of ontological superiority. He thinks that everyone will
agree to the claim that epistêmê is of what really is. On the correct onto
logical theory (he thinks), this will mean that epistêmê is of the ultimate
Beings, things which are ontologically superior to other things—things
which exist but have lesser ontological status and which (perhaps) are
dependent on what is. On other theories—and on implicit, untheorized
views—the claim will mean something far less ambitious.
I will address the Theaetetus in Chapter 10, arguing that Protagoras
hears “epistêmê is of what is” as “epistêmê is of what is really, not merely
Epistêmê is of what is, revisited 109
apparently, the case.” As to the Republic’s sight-lovers, we can see Socrates

talking them into a grasp of the notion of ontological superiority, albeit a
limited and vague one. Here again is the crucial interchange, now divided
into stages:
(a) Does the one who has gnôsis have it of something or nothing? . . . —
He has gnôsis of something.
(b) Something that is or that is not [poteron on ê ouk on]?—Something
that is. For how could what is not be the object of gnôsis [gnôstheiê]?
(c) Are we then sufficiently assured of this, even if we examine it in
many ways, that what completely is is completely an object of gnôsis
[to men pantelôs on pantelôs gnôston], and what is in no way is in
every way not an object of gnôsis?—Most sufficiently . . .
(d) Well then. If something is such as both to be and not to be, wouldn’t
it lie between what purely is and what in no way is?—Yes, between.
(476d–477a)
If pressed to precisify the claim to which they are agreeing in Section (a)—
that epistêmê is of something rather than nothing—perhaps the sight-lovers
would say that there must be some object in the world about which one has
epistêmê, or perhaps that there is some content to all epistêmê (one has
epistêmê that such and such is the case). Socrates does not so press them,
and I see no reason to think there is a determinate fact of the matter as to
which claim Plato intends. In any case, we are to imagine nothing beyond
what is commonsensical and metaphysically tame.
What about the claim in Section (b), that epistêmê is of something that is,
rather than something that is not? The sight-lovers presumably take this to
be nothing but a restatement of the same claim to which they have just
agreed. Perhaps we are to think that they have in mind a fused existential-
cum-predicative claim: epistêmê is always of something that exists, i.e.
something has some properties—not of a total blank. Or that they have in
mind a veridical claim (in Kahn’s state-of-affairs sense): epistêmê is always
of something that is the case—not of something that is not the case.
In Section (c) Socrates strengthens (b)’s claim, with the ‘completely’
(pantelôs). I have argued above that he thereby brings out what he has had
in mind all along: epistêmê is of what really is, i.e. the ontologically superior,
by contrast with the ontologically inferior—a contrast he will make explicit
in Section (d). What about the sight-lovers, though—how do they hear this
new claim? I submit that they hear it as trivially true. On their unreflective,
implicit, and wholly common-sense ontology, there are only two degrees of
being (whether that is existence, predication, being the case, or some
fusion): total being and total non-being. Anything that is at all is completely.
If we ask them about the things concerning which they take themselves to
have epistêmê, namely beautiful sights and sounds (or the beauty of these
things, or facts about these things or states of affairs including them—whatever
Plato takes them to have in mind), “So then do you think that these things
are completely?” or even “Do you think these things really are?,” they will
certainly say yes. What other option, they might ask scornfully, could there be?
It is only in Section (d) that Socrates introduces a challenge to their com-
monsense ontology. Now he suggests a status between completely (or, here,
“purely”) being and totally not-being. Not everything that exists (or has
some quality, or is the case) is completely, purely, really. Simply existing (or
et cetera) is not enough to merit that status. Now we are in the territory of
substantive metaphysics, with Socrates appealing to his distinction between
the ontologically superior and the ontologically inferior. Sharper-witted
sight-lovers might interrupt at this point, refusing to concede the point:
“Everything either is or is not, no in-between!”
Or perhaps—and I offer this only as speculation—the only coherent con-
trast they can imagine between what really is and something inferior is the
contrast between reality and appearance, so they take this to be what
Socrates has in mind. They follow the ensuing conversation thinking, “We
have epistêmê because we know what really is beautiful. Someone who only
pays attention to things that merely seems beautiful lacks epistêmê.” If this is
right, then Socrates only introduces a substantive disagreement with
them—or, perhaps, persuades them of a new view—when he argues that
beautiful sights and sounds occupy this in-between status (479c); up until
then, he has their genuine agreement.
For comparison, imagine the following dialogue:
(a*) Does someone who thirsts thirst for something or nothing?—

Something.
(b*) Something that is drink, or something that is not drink?—
Something that is drink, for how could something that is not drink
be the object of thirst?
(c*) Are we then sufficiently assured of this, that what is completely drink
is completely such as to be the object of thirst, and what is in no
way drink is in every way not the object of thirst?—Most sufficiently.
The Basic Conception of epistêmê 111
(d*) Well then. If something is such as both to be drink and not to be

drink, wouldn’t it lie between what purely is drink and what in no
way is drink?
The claims in (a*) and (b*) are commonsensical. An agreeable interlocu-

tor will hear the claim in (c*) as commonsensical too, indeed as simply a
reiteration of the claim in (b*). It is only in (d*) that things get odd. A sharp-
witted interlocutor will baulk; she will say, “Huh? Everything is either drink
or not drink; what is drink is completely drink, what is not drink is in no
way drink; there is no in-between. You must be working with a strange the-
ory on which some things among those we can drink, or think we can
drink, or commonly regard as drinkable, are privileged among the rest as
the real drinks. I have no idea what you mean! Or, well, maybe you are just
trying to say that some things seem drinkable but in fact aren’t. Sure, go
ahead, you may speak that way . . . . ”
I have remained neutral about what the sight-lovers think in this conver-
sation, both because I want to show that my interpretation is available to
most parties to the debate, and because Plato leaves it very underdeter-
mined. I hope to have shown in any case that we can see Socrates’ basic
principle as a claim that they genuinely accept, indeed as something com-
monsensical, although he soon leaves common sense behind.
7. The Basic Conception of epistêmê
What we have seen is enough to give us an account of Plato’s basic concep-

tion of epistêmê. If cognitive kinds are essentially defined by their objects,
then epistêmê’s relation to Being is its most fundamental feature. All of
epistêmê’s other qualities are somehow explained by this relation. The power
of epistêmê and its accomplishments are the way they are because they are
directed toward Being. If we are looking for a basic definition of epistêmê—
something that answers the question “What is Plato trying to give an
account of: knowledge, understanding, Science, or something else?”—this
provides an answer. Epistêmê is in its essence cognition of Being.
In sum:
• Plato took “epistêmê is of being” to be common ground between all

who think about epistêmê, arguably a conceptual truth, certainly a tru-
ism or platitude.
• He inflated this platitude with a substantive metaphysical view: Being

is the ontologically superior.
• He also inflated it with his epistemological view that cognitive kinds
are defined by their objects: he took epistêmê to be, essentially, the
kind of cognition suited to grasp Being.
If this last claim is correct, then we should expect Plato’s various charac-
terizations of epistêmê throughout the dialogues to be explained by
epistêmê’s relation to Being. In the next chapter, I argue that this is just what
we find. In Chapter 5, we will seek a better understanding of the notion of
cognition of Being, asking how it relates to more familiar cognitive concepts
like knowledge, science, and understanding; at the end of the book we will
consider why Plato would make it the centerpiece of his epistemology.
4
The Basic Conception of
Epistêmê at Work
On an objects-based epistemology, each cognitive kind is the way it is because

it is that by which we cognize a certain kind of object. All its features are
explained by this relation. As we saw in the Introduction, Plato attributes to
epistêmê a number of features that are difficult to unify and explain, including
some that make it look quite a bit like the modern notion of knowledge, such
as being always true, and others that make it look very unlike knowledge,
such as requiring the ability to give definitions or being restricted to intelligible
Forms. If epistêmê’s defining object is Being, then “cognition of Being” should
constitute Plato’s basic conception of epistêmê: we should be able to appeal to
this in order to see why he attributes to epistêmê all the various features that he
does. In this chapter I aim to show that this expectation is borne out.
An important disclaimer: I am not trying to argue here that for each fea-
ture of epistêmê, taken on it its own, our account does a better explanatory
job than any rivals. (For example, it certainly cannot claim a better explan
ation for the definitional requirement than the view on which epistêmê
simply is definitional knowledge!) That is more than I need to show. My aim
is to confirm a hypothesis about Plato’s basic conception of epistêmê, a
hypothesis for which we have already seen much independent evidence, by
showing that it can do the work of explaining all the features Plato attributes
to epistêmê. I do in fact think that in many cases our account’s explanations
are more compelling than its rivals’, but that is not what I am trying to estab-
lish here, and I will not assess each rival explanation on its own terms. We
saw in the Introduction that there is something anachronistic, ad hoc, or
inadequate about most proposals for epistêmê’s identity. I have argued that
Plato’s texts give non-anachronistic, non-ad hoc support for the identifica-
tion of epistêmê as cognition of ultimate Being; I want now to show that this
interpretation is also adequate—that it can do the work.1
1 Compare: if there is a lot of confusing and inconclusive evidence about whodunnit, but
then we acquire good independent reason to hypothesize that it was the butler (say, a reliable
DOI: 10.1093/oso/9780198867401.003.0005
114 The Basic Conception of Epistêmê at Work
I will devote this chapter to Plato’s account of epistêmê in the Two Worlds
dialogues; in Chapters 9 and 10, after I have laid out my interpretation of
doxa, I will briefly consider the epistemology of the earlier and later dialogues.
1. Truth
Plato sometimes says that epistêmê is always true.2 It is easy to hear this as a
claim we nowadays make about knowledge, namely that unlike belief it is
factive: if S knows that p, it follows that p is true. In the Two Worlds dia-
logues, however, Plato sometimes makes a stronger, and stranger, claim
about the relation between epistêmê and truth: only in epistêmê, never in
doxa, do we attain the truth, because the objects we encounter through
epistêmê, and only these, are themselves true. This comes out especially in
the idea that the philosopher attains her goal of truth only when she makes
contact with the Forms, because these alone are true. For example, in the
Republic, philosophers alone are “lovers of the sight of truth” (by contrast
with doxa-having sight-lovers) (Republic 475e);3 in the Phaedo, the philoso-
pher spends time with the Forms, things which are “true and divine and not
the objects of doxa [adoxaston]” (Phaedo 84a); in the Phaedrus, true
epistêmê is attained only by those who in the afterlife see the Forms,
described as “the true things” or “the truth” (tôn alêthôn, 248c; tên alêtheian,
249b). Philosophers attain epistêmê because they grasp certain special
objects, the true ones.
The Basic Conception of epistêmê easily explains this exclusive connec-
tion between epistêmê and truth, granted one crucial premise: truth tracks
being. There is an enormous amount to be said about the relation between
truth and being in Plato.4 That the two are closely entwined is however
eyewitness reports it), then all we need do is to show that the other evidence is consistent with
this hypothesis, and we will have a strong case. We need to show for example that the butler
could have left the footprints and also broken down the door, not that the footprints uniquely fit
the butler’s shoe size nor even that the broken door shows more of the butler’s typical style than
of anyone else’s.
2 For example: there is true and false pistis, but only true epistêmê (Gorgias 454d); epistêmê
is un-false, apseudes (Theaetetus 152c).
3 Sight-lovers are “similar to philosophers” but are not “the true ones [τοὺς δὲ ἀληθινούς],”
because true philosophers do not love beautiful sights (likenesses of beauty) but instead “are
lovers of the sight of the truth [τοὺς τῆς ἀληθείας . . . φιλοθεάμονας],” i.e. the Form of Beauty.
Twice we get the contrast between the true, genuine articles and a likeness: while sham philo
sophers love sham beauty, true philosophers love the true thing.
4 For detailed treatment see Szaif 1996, summarized in Szaif 2018.
The explanatory requirement 115
abundantly clear, and even without probing the depths of that connection,
we get what we need for our purposes here. When Plato uses ‘being’ in the
special sense of the ontologically superior (Being), he uses ‘truth’ in a
similar way, as an exclusive property and indeed defining feature of Being.
Indeed, as many have observed, in such contexts ‘ousia’ and ‘alêtheia’ are
almost synonyms, and it is common to translate both Greek words
interchangeably as “reality.”5 The Forms not only are but are also “the true
and the divine” (Phaedo 84a); the Forms are the things on which “truth and
being shine” (Republic 508d); the objects higher up the Line are more true
(Republic 511e); the Form of Beauty by contrast with the perceptibles is
“the truth” (Republic 475e); in contemplating the Forms the soul is “seeing
what Is and beholding true things” (idousa to on . . . theôrousa talêthê,
Phaedrus 247d).6 Thus in his more metaphysical moods Plato will claim
that epistêmê alone gets at truth, because he holds that epistêmê alone is of
Being. (In Chapter 7.1 we will consider what to make of his claims, in other
moods, that doxa can be true.)
2. The explanatory requirement
In the Republic, Socrates characterizes the highest epistemic state as that in

which we can demonstrate how the first principle of everything (the Good)
is the cause (aitia) of all other beings.7 In the Phaedo, he argues that Forms,
contact with which brings wisdom, are the causes (aitiai) of all perceptible
phenomena (100c–e). In the Timaeus, Timaeus argues that one who loves
epistêmê and nous must study the intelligent cause of worldly phenomena
(46d–e). Epistêmê requires a grasp of causes, i.e. of explanations.8
As we saw in the Introduction, such claims—and more often a similar
one in the Meno (98a) which we will consider in Chapter 9—have led many
to conclude that Plato conceives of epistêmê not as ordinary knowledge, but
instead as what philosophers now call understanding: roughly, a systematic
grasp of explanatory relations.9 He thinks that to have epistêmê about beauty,
5 See for example Vlastos 1981. 6 See also Republic 519b, 525c, 527b.
7 See especially Republic 517c: one who has grasped the Good “must deduce [συλλογιστέα]
that this is the cause [αἰτία] of everything correct and beautiful for all things.”
8 On the Greek notion of aitia as cause and explanation, see for example Vlastos 1969.
9 See Introduction. On the Two Worlds dialogues especially Annas 1981, and Schwab 2016;
Burnyeat 1980 finds the view in the Theaetetus, Nehamas 1985 and Schwab 2015 find it in the
Meno. (Moline 1981 uses ‘understanding’ to translate ‘epistêmê,’ but is working with a different
conception of understanding on which it is less tied to explanation.)
for example, is not merely to know that certain things are beautiful, but to
understand why they are beautiful—because of the Form of Beauty, or
ultimately because of the Form of the Good.
Plato clearly does think of epistêmê as involving understanding. But this
does not entail that this is his bedrock conception of epistêmê—that he
thinks that grasping explanations is the essential, defining feature of
epistêmê. Our interpretation provides a different account of why epistêmê
has this feature. Epistêmê is cognition of ultimate Beings, and these, on
Plato’s ontology, are the causes of everything else. To grasp Being is there-
fore ipso facto to grasp explanations.10 Thus the explanatory requirement on
epistêmê falls out of the Basic Conception of epistêmê, together with a meta-
physical view about the nature of Being: that it has explanatory structure.
Does this entail that there is epistêmê of perceptibles after all, insofar as
they are explained by Forms? It does not. We can reconcile an account of
epistêmê as explanation-grasping with all the evidence we have seen—and
will see in Section 4—that it does not apply to perceptibles by following
Schwab 2016: Plato thinks that genuine explanatory connections only hold
between Forms themselves; the perceptible realm is too messy to admit of
genuine explanation. (Nonetheless, grasping Forms will enhance our cogni-
tion of perceptibles; I turn to this point in Section 5.)
3. Clarity, stability, and precision
In the Two Worlds dialogues, as well as in others, Plato attributes to epistêmê

certain features which he clearly thinks give it value, and in particular
render it superior to doxa. These are qualities we encountered briefly in
Chapter 2 in our discussion of Plato’s like-by-like view of cognition, where
I grouped them together under the name of cleanliness. Epistêmê is clear
(saphes: Republic 511e, Philebus 58c–59d), it is precise (akribes: Philebus
58c–59d), and it is stable (bebaion: Philebus 59b, monimon: Meno 98a; it
keeps us from wandering: Protagoras 356d–e, Republic 484b and 508d).
Plato is not as explicit as one would wish about the meaning of these
terms as applied to cognitive kinds. Detailed studies have been made of
10 It is widely recognized that the epistêmê-relevant causes are essences, one candidate for
Beings. For example see Gerson on Meno 98a: “the explanation for the truth of a true belief is
to be found in the nature or essence [ousia] owing to which something is correctly said to be or
to possess an instance of that essence” (2009, 30); for a more general treatment see Vlastos 1969.
Clarity, stability, and precision 117
various of them, with promising proposals about how they should be

understood.11 About stability he is fairly explicit: while the person with
doxa can be led by persuasion or passions or appearances to change her
mind, epistêmê makes its possessor stand her cognitive ground (Meno 98a;
cf. Protagoras 356d–e and the pseudo-Platonic Definitions 413c and 414b;
Socrates catalogues the ways doxa can be lost at Republic 413a–c). Precision
has something to do with accuracy: the Philebus attributes it to the use of
measurement by contrast with guesswork (Philebus 56b), so perhaps here
the idea is that epistêmê always gets the answer exactly right. Clarity has
something to do with freedom from confusion and contradiction (Republic
524c); Lesher (2009, 171) argues that it amounts to “full, accurate, and sure
awareness.”
The question we need to consider here however is not what these features
are, but whether the proposed Basic Conception of epistêmê can account for
them, and in fact we can show that it can without offering detailed accounts
of the features themselves. For Plato gives an almost explicit argument
deriving these features from epistêmê’s relation to Being. It goes like this:
(i) Epistêmê is of Being [Basic Conception]

(ii) Being is stable, clear, and precise
(iii) Cognitive grasps inherit the features of their objects [Like-by-like]
(iv) Therefore, epistêmê too is stable, clear, and precise.12
We have seen extensive evidence for (i) in Chapter 3. In Chapter 2.3, we

saw evidence for (ii): in the Two Worlds dialogues, Plato characterizes
Being by contrast with perceptible Becoming as clear, precise, and stable
(Republic 479c, 509d, 529d, Philebus 59b). Plato clearly intends some con-
nection with his frequent claim that Forms are unchanging, eternal, and
free from contradictory properties (see especially Republic 479a–b, 524a,
529d–530b, Phaedo 74b–c, Symposium 210e–211b, Timaeus 27d–28a). As
with their epistemic counterparts, my aim is not to offer detailed accounts
of these features. I want simply to emphasize that Plato consistently attributes
them to Forms: they are the distinctive marks of metaphysical superiority.
11 See for example Lesher 2009 on clarity, Perin 2012 on stability, and Carpenter 2015 on
precision.
12 This is the metaphysically explanatory counterpart of the evidential argument discussed
in Chapter 2.5, where Plato draws (ii) as a conclusion from the other claims: we can be justified
in accepting (ii) on the basis of the other claims, but it is (I argued there) explanatorily prior to
claim (iv).
Chapter 2.3 also presented extensive evidence for (iii): throughout the
Two Worlds dialogues, Plato holds that epistêmê inherits its cleanliness
from its clean objects. If you attend to things which are stable, clear, and
precise, you will have a stable, clear, precise cognitive condition, because
the objects are of that kind. (Recall especially Phaedo 79c–d: when inves-
tigating what is unchanging, the soul “has ceased its wandering and
remains always in the same state, because [hate] it is having contact with
things of that sort”; compare Republic 508d and Philebus 59a–b, quoted
in Chapter 2.3.)
Thus there is strong evidence that in the Two Worlds dialogues, Plato
thinks that epistêmê’s cleanliness is entailed by and inherited from its
objects. Moreover, the idea has some intuitive appeal, independently of the
general like-by-like doctrine of cognition. You cannot fully illustrate a
precise mathematical formula in a sketchy graph; you cannot adequately
represent a sharply delineated object with a blurry photograph or through
cloudy vision; you cannot capture a steady situation with a wavering glance.
Clear, precise, stable objects are adequately captured only by grasps that
share these qualities. If Beings are as clear, precise and stable as Plato’s
Forms, then cognition of them—that is, epistêmê—will itself be clear, pre-
cise, and stable as well. (We will come back to this idea in Chapter 7, where
I will argue that it plays a role in the view that there is no doxa of Forms.)
There are other explanations for Plato’s characterization of epistêmê as
clean. One might argue that he is simply following tradition (certainly his
predecessors and contemporaries consistently characterize epistêmê as
clear (saphes)).13 Some take the lesson to be that he is beginning from a
conception of epistêmê as a priori or “logically certain” knowledge: out of
a penchant for the rationally demonstrable, he recognizes only this kind
of knowledge as worthy of attention, ignoring empirical knowledge.14
I have tried to show that the Basic Conception can do the work instead.
Psychologically, Plato may have begun from a conviction that epistêmê is
clean. Philosophically, he derives that characterization as a conclusion:
beginning with the Basic Conception of epistêmê as cognition, and with
his metaphysical views about Being, he concludes that epistêmê must
be clean.
13 For extensive citations see Lesher 2009. 14 Vlastos 1981, cf. Vlastos 1985.
Restriction to Forms 119
4. Restriction to Forms
As we have seen, in the Two Worlds dialogues Plato seems to restrict

epistêmê wholly to the Forms, denying epistêmê of perceptibles. I want now to
show that this feature of epistêmê is also explained by the Basic Conception.
(I will also argue, against the Content Overlap view, that this means that
epistêmê is never about perceptibles.) The argument is very simple:
(a) Epistêmê is of Being [Basic Conception]

(b) The Forms are the only Beings; perceptibles merely become.
Conclusion: Therefore there is no epistêmê of perceptibles.
I will not here repeat the evidence for (a), since I discussed it at length in
Chapter 3. As for (b), we have seen evidence in the contrasts Plato draws
between the Forms, as Beings, and perceptibles as ontologically inferior. (To
take just one example, the things we can access through perception are not
Beings but instead “come into being and are destroyed without ever really
being” (Timaeus 28a); see further all the texts cited in Chapter 1.2 as evi-
dence for the Distinct Objects view, and the texts cited in Chapter 3.3 con-
trasting Forms and perceptibles.). We also see evidence for (b) in Plato’s
claims that the perceptible realm contains less, or no truth—where ‘truth’ is
used as almost interchangeable with ‘being.’ (This is what is sometimes
called “ontological truth”—not a property of representations, but of their
objects; not correspondence to reality, but reality itself.)15 The Forms alone
are that on which “truth and what Is shine” (Republic 508d); perceptible
things by contrast with intelligibles “partake less in truth and Being”16
(Republic 585d); the perceptible world is less true than the intelligible
(Republic 511e); just as there is no sunlight shining in the cave, so there is no
truth shining on the perceptible world (Republic 508d); hence those who are
stuck there are without experience of truth (alêtheias apeirous, Republic 519b);
through the senses we can access nothing true (Phaedo 83b).17
As to the argument from (a) and (b) to the conclusion, we have seen it
already in all the Contrast passages we saw which correlate epistêmê with Being
while downgrading perceptibles to a lower ontological status (Chapter 3.3).
15 See Szaif 2018. 16 ἧττον ἀληθείας τε καὶ οὐσίας μετέχει.

17 Philosophy encourages the soul not to suppose true anything it finds through the senses,
for such things are “perceptible and visible” (Phaedo 83b). A soul duped by the body will think
visible things most true (83c; cf. 83d), but in reality it is only by ignoring perceptibles and using
intellect on its own that one can grasp “the true and divine” (84a).
Epistêmê is of what Is; perceptibles have some lesser status; (therefore) the
perceptible realm is not the object of epistêmê (but is instead the object
of doxa).
We get a more elaborate version of this argument—one that provides a
rationale for (b)—in the Republic’s two extended arguments for the restric-
tion of epistêmê to intelligibles, the Book V argument that epistêmê is only
of Forms, and a less-discussed argument of Book VII (529b–30b).
It will help to consider the Book VII passage first. Socrates is discussing
the philosophical education that aims at epistêmê. He argues that astronomy
will be useful only as an aid to the study of imperceptible things rather than
as a direct source of epistêmê. Indeed we cannot acquire epistêmê of the stars
and planets in their own right, precisely because they are perceptible:
. . . if anyone tries to learn something about perceptibles [tôn aisthêtôn], by

gaping up or blinking down, I deny that he would ever learn – for such
things admit of no epistêmê. (Republic 529b–c)
The next lines explain why perceptibles cannot be the objects of epistêmê:
they are too messy to contain truth, where ‘truth’ is here used in the onto-
logical sense. While the heavenly bodies are the “most beautiful and most
precise” of perceptible things (529c), they are not beautiful or precise
enough to be true: their motions “fall far short of the true ones . . . which can
be grasped by logos and thought [dianoia] but not by sight” (529d). Just as
even the most beautiful mathematical diagrams are so deficient that a real
geometer “would suppose it ridiculous to investigate them seriously in the
expectation of grasping the truth in them [tên alêtheian en autois lêpsome-
non],” so too the motions of the stars are so messy that a real astronomer
“would regard someone as absurd who supposed that these were always the
same and never in any way deviate . . . and sought in every way to grasp the
truth [tên alêtheian labein] about these things” (529e–30b).18 Where there is
instability and imprecision there is no truth, and therefore no basis for
epistêmê. We can represent the argument as follows:
(1) Epistêmê is of Being (here, Truth). [Basic Conception] [= (a)]

(2) Being is clean.
18 Another philosopher might mean by these claims that there is truth there, but not truth
we can grasp; Plato, who as we have seen repeatedly denies that there is truth in the perceptible
world, means instead that there is no truth there to grasp.
Restriction to Forms 121
(3) Perceptibles are messy.

(4) Therefore, the Forms are the only Beings; perceptibles merely
become. [= (b)]
Conclusion: Therefore there is no epistêmê of perceptibles.
Turning now to Republic V’s powers argument, it is clear that we can

interpret it along the same lines. Epistêmê is of what fully Is ((1)–477a).
Perceptibles, meanwhile, are too messy to be what fully Is ((2)–(4)): this
explains the move from “every beautiful [perceptible] thing is in some way
ugly” to “every beautiful thing is between Being and not-being” (479a–c).
Only clean Forms have real Being. The epistemological thesis follows:
epistêmê is only of Forms, not of perceptibles ((4)–479e).19
Thus our Basic Conception of epistêmê makes sense not only of the
Republic’s restriction of epistêmê to intelligible Forms, but also of the actual
arguments it offers for that restriction. If we accept the Basic Conception,
we should happily join Aristotle, Alexander, Alcinous, and all the others up
to the 1960s who found it unproblematic to assign to Plato the view that
there is no epistêmê of perceptibles. Because epistêmê is of Being, the per-
ceptible world simply offers no place for it to get a hold. Just as you cannot
see sounds because there is nothing visible there, no color, you cannot have
epistêmê of perceptible things because there is none of its object there:
no Being.
Finally, I think it clear that these arguments rule out the Content Overlap
reading of epistêmê as well as less subtle Overlap readings. On the Content
Overlap view (discussed in Chapter 1.1; see especially Smith 2000 and 2012),
epistêmê can be about perceptibles even though it must be acquired through
contact with Forms. Thus a philosopher who has acquired epistêmê of
Justice by studying the Form can then apply that epistêmê in making
judgments about particular actions, and thereby have epistêmê about those
particular actions: she could for instance have epistêmê that starting a war
with Sparta right now would be just. What we have seen however seems to
rule out such readings. This is simply not the kind of subject matter about
which one can have epistêmê; the whole topic of when to start wars is
too unpredictable and messy to be the subject of epistêmê, because it is too
19 “Those who contemplate the many beautifuls, not seeing the Beautiful itself . . . and every-
thing like this, we say that they have doxa of everything, and do not have gnôsis of anything
which they have doxa of . . . What about those who contemplate the things which themselves
that are always the same? Won’t we say that they have gnôsis, and do not have doxa?”
(Republic 479e).
unpredictable and messy to contain Truth or Being onto which epistêmê can
latch. Thus, contra Smith, epistêmê can never be exercised in a judgment of
the form “starting a war right now is just;” that is the wrong kind of content.
Plato restricts epistêmê to Forms without qualification. Forms are not only
epistêmê’s sole input, but also its sole subject matter.
It may help to note that this is not such a strange position to hold. Recall
Isocrates’ claim that there is no epistêmê of practical matters because the
kairos (right occasion) “eludes our epistêmai” (Antidosis 184). There is no
stable, universal, precise grasp of the kairos because the areas in which the
kairos obtains are messy: areas in which individual cases vary widely, and so
there are no precise or universal rules. Compare the kind of thing we say
nowadays: “There is no science of human nature,” or “Humor is an art, not a
science,” where the claim is that there are no precise, universal laws because
the subject matter itself is too messy.20 These are claims about subject
matter: a clean grasp simply cannot latch on to a messy domain. Plato’s
epistêmê is far cleaner than Isocrates’, or than science as nowadays
conceived; if these latter are plausibly restricted in their subject matter, so
too is Plato’s epistêmê.
5. Objection: philosopher-rulers’ epistêmê
This brings us to an objection that has fueled much of the recent resistance
to the Distinct Objects reading of the Republic. The entire political theory of
the Republic is predicated on the idea that philosophers’ epistêmê qualifies
them to rule. If there is no epistêmê of the perceptible realm, however—no
epistêmê of when it is just to start a war, for example—how can epistêmê
be useful? Surely (the objection runs), Plato is thinking that good rulers
will have epistêmê about which human institutions and actions are just,
20 For example, in a popular science article about the debate over whether medicine is an art
or a science we learn that “Doctors on the far ‘art’ side of the spectrum maintain that every
patient and physician is different. Therefore, they believe there is no one right way to treat a
patient” (R. Pearl, “Medicine is an art, not a science: Medical myth or reality?” Forbes, June 12,
2014): to deny that medicine is a science is to deny that there are universal rules governing it.
Or we may deny that something is a science on the grounds that it is too imprecise and
unstable: “Physics can send a satellite to orbit Jupiter, tell you exactly when it will arrive and the
altitude it will orbit at. Economics can barely describe what happened yesterday—and without
any particular precision” (B. Ritholtz, “Ten reasons why economics is an art, not a science,” The
Washington Post, August 9, 2013).
Objection: Philosopher-Rulers ’ epistêmê 123
beneficial, or fine; therefore epistêmê about such particular, perceptible

things must be possible. (For this line of argument see especially Fine 1990
and Smith 2000.)
The first response to the objection is that, whether we like it or not, there is
clear evidence that even the philosopher-rulers lack the kind of completely
stable, clear, precise grasp of the perceptible world of the kind that could
qualify as epistêmê. On top of all the evidence we have seen above that such
a grasp is impossible, there is also direct evidence about the philosopher-
rulers’ cognitive condition, in Plato’s account of how the kallipolis first
degenerates into timocracy.21 The rulers miscalculate the proper breeding
times, for eventually, “although they are wise [sophoi]. . . they will not hit on
[the appropriate times for] fertility and sterility through calculation with
perception,” and so will conceive children when they ought not (546a–b);
these inferior offspring will drag down the city. One can imagine Isocrates
agreeing: childbearing is one of those domains in which everything hangs
on the kairos, opportune moment, but such matters “elude our epistêmai”
(Antidosis 184). Or as Arruzza puts it, “It is the irreducibility of sensible
reality to the perfect grasp of reason—its resistance to perfect knowledge—
that introduces corruption into the ideal city” (2018, 115). Epistêmê is
infallible, but the philosopher- rulers’ cognition of the perceptible realm is
subject to error; hence this cognition is not epistêmê.
The second response is that the problem that motivates these Overlap
readings—if epistêmê is exclusively of Forms, how can it be necessary for
practical expertise in the perceptible world?—has been much exaggerated.22
Plato himself evidently thinks it no problem at all, and indeed offers a near-
explicit and quite attractive solution. Broadly, the idea is that theory informs
practice. More specifically: the perceptible world is an image of the Forms,
and therefore epistêmê of the Forms guides, although does not strictly apply to,
our dealings in the particular world.
That this is Plato’s thought emerges from the discussion that follows
closely on Republic V’s powers argument. The argument that the best city
must be ruled by philosophers is often treated as if it ends with Book V, but
21 I owe this point to Cinzia Arruzza.

22 Although others have recognized that Plato gives us the resources for a solution. The view
I advocate here is similar to what Gonzalez 1996 defends in detail, and compatible with
Sedley 2007. It is also strongly suggested by a brief comment from Adam (1902, comment ad
476c): the philosopher-ruler needs epistêmê because he “cannot possibly frame political insti-
tutions on the model of Ideas which he does not know.”
it does not.23 Directly after concluding that only those who contemplate
Forms have epistêmê (479e–480a, the last pages of Book V), Socrates goes
on to show how it follows that philosophers should rule. The rationale he
gives is not that epistêmê applies directly to the perceptible world, but rather
that people with epistêmê of the Forms can use Forms as models (paradeig-
mata) on which to base the city, as an image:
[We cannot trust the rule of the city to] those who are deprived of any real
gnôsis of each thing that Is, and have no clear model [paradeigma] in their
soul, nor are able, just like painters, looking toward the truest thing and
always referring things to that and observing it as precisely as possible, in
this way also to establish worldly conventions [nomima] about fine things
and just things and good things, if they need to be established, and guard
and preserve the ones that have been laid down . . . . [So we should choose
as rulers] those who on the one hand have come to have gnôsis of each
thing that Is, and on the other hand are in no way deficient to the others in
experience24. . . (Republic 484c–d)
The perceptible world is an image and approximation of the intelligible

world (an idea we see most elaborately in the Cave allegory and in the
Timaeus’ cosmology; see Chapter 7.5 for discussion). Therefore an under-
standing of the intelligible world gives us an ideal model on which to base
our practice in the perceptible world, just as painters use their models to
guide them in producing images. Like “painters who use the divine model”
(500e), philosophers will use their epistêmê of Justice, Beauty, and the other
Forms to make the city and its inhabitants into good images of these—as
good as images can be:
They will take the city and the characters of people as a tablet . . . and as
they work they will look frequently in both directions, toward what is by
nature just and fine and moderate and all such things [i.e. the Forms], and
also toward the [just, fine, moderate, etc.] which they would instill in
humans . . . . And they will erase one part and re-paint another, until as
much as possible they have made the human characters insofar as is
possible dear to the gods. (Republic 501a–c)
23 This supports the alternate division of the Republic on which the argument that begins in
present-day Book V was included in a single book that ran up through the Cave allegory: see
Sedley 2013.
24 τοὺς ἐγνωκότας μὲν ἕκαστον τὸ ὄν, ἐμπειρίᾳ δὲ μηδὲν ἐκείνων ἐλλείποντας.
Moreover, the philosophers’ epistêmê of the Forms lets them use these not
only as models on which to mold perceptible images, but also as standards
by which to assess them. When they return to the city to rule, like freed
prisoners returning from the outside to the cave, the philosophers:
will see much better than the people there, and will have gnôsis of
[gnôsesthe] each image, what it is and what it is of, on account of having
seen the true things [t’alêthê] about the fine and just and good.
(Republic 520c)
This passage is sometimes put forth as decisive evidence that Plato recognizes
epistêmê of perceptibles (see especially Fine 1990), and as we saw above
the Book V powers argument does use words cognate with gignôskein
interchangeably with epistêmê. The verb can also mean to recognize or
discriminate, however, and in context that is all that it need mean. When the
philosopher first returned to the cave, “his vision was still dim, before his eyes
had adjusted” (516e–517a), and he was ridiculed for being bad at the activities
the prisoners honor: “sharply observing [kathorônti] the things that pass by,
and remembering best which ones are accustomed to go by before, after, or at
the same time, and from these things being most capable of predicting what is
to come” (516c–d). Once his eyes have adjusted, however, he will “see much
better” than the prisoners (520c): he will be better able to recognize or dis-
criminate the shadows. The claim Plato needs at 520c then is not that the
returned philosophers have epistêmê of the shadows; instead, he is saying that
because the rulers have epistêmê of the originals (the Forms), they are very
good at recognizing and distinguishing perceptible images of them. Epistêmê of
the Forms informs our cognition of the perceptible realm, whether in mold-
ing it or assessing it, because the perceptible is an image of the Forms. But this
does not mean that there is epistêmê of the perceptible realm itself.
Consider again Plato’s claim about the attitude geometers have toward
diagrams, or that true astronomers have about the motions of the visible
stars: they would think it absurd if someone “sought to grasp the truth [tên
alêtheian labein] about these things, which have body and are visible”
(Republic 530b). The truth is to be found only in the imperceptible models,
not in the visible images. But, as we saw in the previous section, this is just
to say that there is epistêmê only of the models, not of the images—precisely
the claim Plato makes about the visible stars (529b–c, quoted above: “such
things admit of no epistêmê”). Nonetheless, although Plato does not empha-
size the point in this particular context, experts will also be the best judges
of how well the images approximate the truth. Moreover—with some

practice in the requisite technical skills—they will also be the best at
producing images. Expert geometers will be the best judges of diagrams,
and with some practice will make the best diagrams. The same is true of
philosopher-rulers. They cannot have epistêmê of perceptibles, since these
do not contain sufficient truth or Being. They can however use their
epistêmê of the Forms to judge how well various perceptibles approximate
the Forms. Moreover—with some practice in the requisite technical skills,
i.e. with political experience (empeiria) (484d, quoted above)—they will be the
best at fashioning perceptibles into good approximations of the real things.
If this is right, then Plato has a complex view of the philosopher-rulers’
expertise which he does not make fully explicit. Here is an attempt to spell it
out. Ruling well—and indeed also living well25—is the business of a special
expertise. (He sometimes refers to this as a technê (346a), sometimes as
excellence in deliberation (euboulia, 428b), sometimes as an epistêmê in
what must be a loose sense equivalent to technê, because in these contexts
carpentry and medicine also qualify (428b–c), and sometimes as sophia,
wisdom (428b–c);26 I take this to show that he has developed no systematic
technical vocabulary here, and I choose ‘expertise’ as sufficiently general.)
This expertise has two components. First, a purely theoretical component:
epistêmê of Forms. Second, a practical component: excellence in molding,
discriminating, and evaluating images of the Forms. This second compo-
nent is not epistêmê, but it cannot be had without epistêmê: facility with
images presupposes familiarity with their originals. Plato gives no name to
this second component, nor does he say whether it is automatically entailed
by epistêmê or involves some additional process, although he does suggest
that it requires experience (empeiria).27
25 Philosophers, whether or not they are rulers, will use their epistêmê of Forms to guide
their own behavior in the perceptible world, eschewing unjust actions, pursuing just ones,
and so on.
26 See Claudia Yau, work in progress, for an argument that it is best conceived of as sophia.
27 Recall Plato’s comment after describing how the philosopher-rulers will use the Forms
as models: we should choose as rulers “those who on the one hand have got to have gnôsis of
each thing that Is, and on the other hand are in no way deficient to the others in experience”
(484d). And indeed, Plato requires the philosopher-ruler to spend fifteen years in the Cave of
political activity, gaining experience, before she becomes a ruler (539e–540a). For good dis-
cussion see Vasiliou 2015. Compare Cooper: “To be sure, merely knowing the good will not
by itself suffice – at any rate not for knowing how best to run a city. This is why Plato insists that
before coming to know the good his rulers must spend fifteen years getting experience in city
management (539e2–5). Still, the only part of what [the ruler, or the wise individual]. . . knows
that counts, on a strict view as knowledge (the sort that encompasses understanding, of why
things are the way they are and how they ought to be) is knowledge of the good-itself,” i.e. of
the Form (Cooper 1977, 156).
Thus theoretical epistêmê of Forms and practical facility with their images
are part of the same expertise. This follows a general rule which Plato has
proclaimed earlier in the Republic: the ability to know or recognize images
“belongs to the same craft and practice [technês kai meletês]” as the ability to
know or recognize the originals.28 If we take these to be two distinct com-
ponents of a single expertise, we have room to acknowledge both that
epistêmê is restricted to Forms, and that it is necessary for the expertise of
ruling well.
We can find a rough parallel in Aristotle’s account of craft (technê).
(I here present a contentious but compelling reading of his account, in line
with what we find for example in Devereux 1986).29 Aristotle distinguishes
two components of a craft like medicine: theory (logos) of the universals,
and experience-based judgments about particulars:
Regarding action, experience [empeiria] seems not to differ at all from

craft [technê]. Indeed those who are experienced succeed more than those
who have the account [logos] without experience. For experience is gnôsis
of particulars, while craft is [gnôsis] of the universal, and actions and pro-
ductions are all concerned with the particular: for one does not heal
Human, but Callicles . . .If then someone has the account [logos] without
the experience, and has gnôsis of [gnôrizêi] the universal but is ignorant of
the particular contained in this, he will often make errors. . .
(Metaphysics 981a12–23)
One aspect of medical expertise is the logos of the universals: someone who
“has the logos” can give an account of Human, for example, for they know
the universal essence. There is no corresponding account of the particulars:
these are the domain of experience instead.30 Moreover, although Aristotle
28 “And surely if there are images of letters appearing in water or in mirrors or anywhere, we
will not recognize [γνωσόμεθα] these before we recognize [γνῶμεν] the letters themselves, but
this belongs to the same art and practice . . . [and likewise, when we are well educated in music
we will be able to] recognize [γνωρίζωμεν] the forms of moderation and courage . . . both them-
selves and their images . . . thinking them to belong to the same craft and practice [τῆς αὐτῆς
οἰώμεθα τέχνης εἶναι καὶ μελέτης]” (Republic 402b–c).
29 Devereux argues that Aristotelian medical technê strictly speaking is of universals alone,
and hence not efficacious in action; what an effective doctor has is a two-part expertise which
combines technê of universals with empeiria of particulars. I am not convinced that Aristotle
consistently uses ‘technê’ to refer only to the universal component, but I am convinced by the
claim that the full, practically efficacious expertise involves both theory of universals and prac-
tical facility with particulars.
30 See also the contrast between technê and empeiria earlier in the discussion: it belongs to
empeiria to make particular claims about what cured Socrates or Callias, while technê instead
makes universal claims about what cures “all people of this sort, divided off in one class [εἶδος]”
(Metaphysics 981a10).
here says that experience without technê—that is, without the universal
logos—is more effective in action than the reverse, we can charitably infer
that he thinks the combination of the two the most effective of all. Theory is
only of universals, and so to be practically useful it must be supplemented
by experience of particulars. Experience of particulars on its own is not
wholly reliable, however: someone who also has the logos of the universals
will have a better grasp of particulars, because they can recognize the par-
ticulars as instances of the universal. An excellent craftsperson will have
both components: theoretical logos of the universals, practical facility with
particulars.
Analogously, on the view I am attributing to Plato, the rulers’ expertise
consists of theoretical epistêmê of Forms, and practical facility with
perceptibles.31 Moreover—a point on which Plato will insist much more
than empirically-minded Aristotle—the second cannot be had, in its best
form, without the first.
In spirit, this interpretation is not so different from some that are put
forth by Overlappers: epistêmê is primarily of Forms, but someone with
epistêmê of Forms can use that to have excellent cognition of perceptibles
(see for example Smith 2012 and Fine 1990).32 I agree that possessing
epistêmê of Forms allows one to be good at recognizing their images; my
disagreement is only with the claim that such recognition is itself an
instance of epistêmê. Why bother insisting on the point? First, because it
allows a literal reading of all the evidence we have seen that there is no
epistêmê about perceptibles (see especially Republic 529b–c on the impossi-
bility of epistêmê of the stars, Philebus 59a–b on the impossibility of stable
cognition of unstable things, and all the evidence we saw for like-by-like
cognition). Second—and more weightily, since of course Overlappers will
find other ways to read these passages—because once we get a fuller grip on
the nature of epistêmê, which is the task of the next chapter, we will see that
there are perfectly good reasons to restrict this kind of thing to Forms. For
now, that must be a promissory note.
31 The analogy is just an analogy: the theoretical logos that Aristotelean doctors have is not
at the epistemic level of Plato’s epistêmê.
32 “The power of knowledge [epistêmê] will provide the basis of the most reliable judgments
in regard to particulars . . . in applying the power of knowledge, philosophers would make an
appropriate connection to one or more Forms . . . which would result in their having the clear-
est possible concept of justice, which they could then apply in the judgment of instances.
Seeing the ‘resemblance’ of Justice in the rule of philosophers [something that belongs to the
realm of Becoming], the knower would judge such a rule to be an instance of justice”
(Smith 2012, 70).
If the philosophers’ excellent grasp of perceptibles is not epistêmê,

owever, what is it? A natural thought, which some have embraced, is that it
h
is doxa. On this interpretation, Plato agrees in a way with Isocrates
(Antidosis 184 and 271; Against the Sophists 8): the best we can have in
practical matters is good doxa. After all, perceptibles lack clarity and Being,
and thus—by the like-by-like doctrine—the best we can have of them is
doxa. This does not entail that the rulers’ cognition of perceptibles is on a
level with that of non-philosophers. Philosophers will have doxai that are as
good as doxai can be; as Schwab puts it, they will have “expert opinions,”
which, though their subject matter condemns them to fall short of epistêmê,
are far better than ordinary doxa.33 Certainly it is plausible that doxa of
images, when paired with epistêmê of their originals, is superior to ordinary
doxa: more stable, more true, more clear. Indeed Plato strongly suggests
this claim in the Timaeus, in a famous passage which we saw in part in
Chapter 1:
This world is a likeness [eikôn] of something [viz., a divine, unchanging

paradigm]. . . Accounts are akin to the things they expound. An account of
that which is fixed and stable and discoverable with the aid of intellect will
itself be fixed and unchangeable, so far as it is possible . . . while accounts of
what is made in the image of that other but is only a likeness, will them-
selves be only likely [eikotas], standing to accounts of the former kind in a
proportion: as Being is to becoming, so is truth [alêtheia] to conviction
[pistis].34 If then, Socrates, we do not prove able to render accounts which
in every way agree with each other and are precise . . . do not be astonished.
But if we are able to offer accounts which are no less likely than anyone
else’s, we ought to be well pleased . . . (Timaeus 29b–c)
Accounts of the perceptible world—and, as strongly implied, the cognitive

conditions they reflect and can produce—inevitably fall far short of the
cleanliness that characterizes epistêmê. Indeed, as Plato here implies and
elsewhere in the Timaeus explicitly states, the best we can have about the
33 “[W]e do not think that the doctor can prove that the course of treatment she prescribes
is best,” and so we regard her views only as opinions, but we respect them as expert opinions.
And “just as we take doctors’ medical opinions to be expert because they are informed by their
understanding of health, so too Socrates takes philosophers’ opinions concerning concrete
perceptible matters to be expert because they are informed by their epistêmê of Forms”
(Schwab 2016, 77–8).
34 For interpretations of this difficult analogy see Burnyeat 2005, Bryan 2012 and
Mourelatos 2014.
perceptible realm is pistis, or doxa (cf. Timaeus 27d–28a, 37b–c and 52a,
quoted in Chapter 1.2). Nonetheless, as commentators rightly emphasize
(see especially Broadie 2011, 33–38), because the perceptible world is an
image of the divine, we can have accounts and views of it that meet an
important standard: they are likely (eikos).35 Someone who recognizes the
perceptible realm’s status as image of the divine—and presumably even
more so someone who has expertise about the paradigm, whether or not
Timaeus is claiming that status for himself—can give a better account of
the perceptible realm than ordinary cosmologists, although he can never
have the highest kind of cognition about it. Likewise, one might argue,
philosopher-rulers with epistêmê of Forms can have doxai about the
perceptible realm that are far superior to ordinary people’s.
On the other hand, the Republic so downgrades doxa that it is not
surprising that we never find Plato explicitly attributing it to philosopher-
rulers, or to wise private individuals.36 Moreover, once I give my account of
doxa in Chapters 6–8, we will see some reason to doubt that this is really
what he has in mind. Unlike the cave-dwellers and sight-lovers who are
Plato’s exemplars of doxa, philosopher-rulers recognize perceptibles as mere
images of Forms; in this way, they are much more like people who have
something higher than doxa, namely dianoia, “thought” (see 510d–e on
mathematicians, discussed in Chapter 7.6).
Perhaps then if he had given the matter more attention he would have
said that the philosopher-rulers are exercising some cognitive power not yet
named: one that, like dianoia, looks back and forth between perceptibles
and intelligibles (see again the painter analogy, at 500e–501b), but unlike
dianoia is focused on evaluating and improving the perceptibles, rather
than on discovering truths about the intelligibles. Indeed, perhaps he would
wind up positing a distinction that Aristotle develops, between practical
and theoretical reasoning.37 Practical reasoning, or deliberation, shares with
doxa a concern with perceptibles, but is distinguished from doxa both by its
35 For discussion of ‘eikos’ as a term of praise, see the articles cited in the previous note,
along with Broadie 2011.
36 With one perhaps important exception: the proper youthful education will instill in the
rulers a stable, true doxa or dogma that they must protect the city (412e–414a). This is to be
secured by a myth (the Noble Lie). Perhaps we are to take it that—unlike the auxiliaries—the
rulers will later be able to replace this doxa with something better, but perhaps not.
37 The Republic certainly seems to lack this distinction; for discussion, and contrast with
Aristotle, see Devereux 1986 and Silverman 2010.
exclusively practical orientation, and also by its possibility of perfection,

phronêsis.38
Plato will return to and develop the notion of political expertise in the
Statesman explicitly and in detail, although in quite a different way, without
any appeal to Forms, and with extensive emphasis on the ability to deal with
messy, changing affairs (see especially 284a–e and 294a–296e). The Republic’s
account is brief, largely implicit, and leaves underdetermined the nature of
the practical component: perhaps it is doxa, perhaps something unnamed.
The account is nonetheless a coherent one. Moreover, it is one which recon-
ciles a Distinct Objects epistemology with a practical role for epistêmê:
epistêmê is required for expert cognition of perceptibles, but is itself only
of Forms.
38 See for example Nicomachean Ethics 1140b27.

5
What Is Epistêmê?
In answer to the question at the start of this book, I have argued that
Plato identifies epistêmê first and foremost as the cognition of a certain
kind of object: Being. We do not get at what epistêmê essentially is—at
Plato’s fundamental conception—by starting with epistêmê’s resem-
blances to recognized cognitive kinds like Science, knowledge, under-
standing, or others; nor do we get there by starting with the intrinsic
features in virtue of which it resembles these, such as its clarity, its stabil-
ity, or its explanatory nature. Instead, I have argued, all these features and
resemblances are grounded in the fact that epistêmê is of Being. When we
add to this Plato’s metaphysics—his view of the ultimate realities as
purely intelligible, as stable, precise, and clear, as essence-like, and as the
ultimate causes of all that can be explained—and his like-by-like doctrine
of cognition, we get a complete explanation of his characterization of
epistêmê. Epistêmê is a cognitive grasp suited to the ultimate realities:
that is why it is true, explanatory, stable, clear, and precise, and restricted
to the intelligible realm of Forms. Plato may have had independent rea-
son to attribute these features to epistêmê: perhaps he was preserving
contemporary views, or privileging philosophical reasoning over percep-
tion and experience. But these features are nonetheless philosophically
explained by the Basic Conception of epistêmê. Essentially, epistêmê is
cognition of Being.
This answer may however leave us nonplussed. What sense can we make
of this notion, cognition of Being? Is it a radical departure from other inter-
pretations of epistêmê? Is there any intuitive way to characterize it? Does it
correspond to anything else in the history of epistemology? Does it corres
pond to anything we recognize or care about now?
In this chapter I will offer some answers to these questions. At the end of
the book, I will return to offer an account of why Plato would make this
notion the centerpiece of his epistemology.
DOI: 10.1093/oso/9780198867401.003.0006
Extant interpretations 133
1. Extant interpretations
To begin with, my interpretation is new but not radical. While there has
been no clear recognition of “cognition of Being” as the fundamental defin
ition of Plato’s epistêmê, many insightful treatments come close to or fit very
well with this account.
Certainly older interpretations come close to it in spirit. As I mentioned
in Chapter 1, the standard English translation a century ago was ‘Science’
(see for example Bosanquet 1895, Jowett 1908, Taylor 1926, Shorey 1930).
This translation is clearly intended to pick up on the Latin Scientia, which in
turn stems from Aristotle’s epistêmê. (LSJ’s entry for epistêmê in the
nineteenth century cites the Republic’s powers argument, along with the
Posterior Analytics, as loci for the meaning “scientific knowledge, science”
(s.v. epistêmê, A.II.2).) As I will argue briefly below (Section 2), a central
feature of these notions is the connection to ultimate reality: Science is or
involves the cognition of the essences or other first principles.
More recently, as we have seen, some have argued that epistêmê is to be
identified with understanding. This too comes close in important ways—
especially on the version we find in Schwab 2016, on which understanding
is always grounded in essences. Understanding, on this interpretation,
requires a grasp of the deep realities that underlie everything else, and its
explanatory nature mirrors that of its objects.
Closer yet to my view, some identify Plato’s epistêmê as knowledge of
essences. Among others, Gonzalez characterizes epistêmê in the Republic as
“non-propositional knowledge of whatness (e.g. of what justice is…)” (1996,
27); Silverman construes Forms as essences and restricts epistêmê to Forms
(2002); Gerson characterizes epistêmê as “infallible mental seeing of Forms
or essences” (2009, 41).1 This type of view has a lot of explanatory power,
and also makes for compelling continuity with later epistemology (starting
with Aristotle’s nous). Even if we accept the metaphysical thesis that Plato
construes Forms as essences, however (and it is not my business in this
book to defend this, nor any thesis about the identity of the Forms beyond
their status as ontologically superior) I would want to revise or clarify the
thesis about epistêmê. It is not that Plato starts from an understanding of
1 Rowett 2018 construes epistêmê as knowledge of “types qua types,” arguably a relation of

the essence view.
134 What Is Epistêmê?
epistêmê as cognition of essences; instead, he reaches that view as a

substantive conclusion, derived from the Basic Conception of epistêmê
together with the metaphysical view that the Beings are essences.
My account is perhaps closest to what we find in Moline: epistêmê is “an
understanding of real objects” (1981, 119)—or more picturesquely in
Crombie: it is “a devouring of something real” (1962, vol 2., 42). I have
argued that the epistemically relevant sense of ‘being’ is something like
ultimate reality, and I think Crombie and Moline have got this part just
right. Their interpretations wind up however being too narrow, on my view:
they define the real as the contrast to copies or likenesses, while I think
Plato works in different contexts with different notions of the real—that is,
with different contrasts between the ontologically superior and inferior. So
like the essence view, this view captures part of the truth about epistêmê, but
a part that needs to be grounded in the more general and fundamental fact
that epistêmê is cognition of Being.
2. Counterparts
The notion of cognition of Being is not sui generis: it has historical

counterparts—precedents and descendants—which can help us to under-
stand it. This becomes clearest when we remind ourselves of the basic prem-
ises of this notion: first, that there is some privileged level of reality; second,
that an adequate grasp of this level will have certain features that suit it to
the role, and thus that it will be appropriately impressive in its own right.
Once we put it like this, we can recognize various counterparts.
One is the notion of a special power or state of my mind by which we
penetrate through superficial appearances to get at the underlying reality.
(The relevance of this notion will be confirmed by our account of doxa as
cognition of what seems or appears.) Scholars find such an idea at work in
two of Plato’s most influential predecessors, Parmenides and Heraclitus:
nous is “a direct of view of reality,” while doxa is stuck with what seems.2
Another historical counterpart—logically independent of the first,
although sometimes joined with it—is the notion of a special, deep under-
standing of the foundational ontological items that underlie and explain all
the rest. Arguably this notion was already at work in the Ancient
2 See among many others Lesher 1994 and von Fritz 1994, 26, 33, 40–1.
Counterparts 135
conception of what we would call a science, like medicine or geometry—for

which they used the word ‘epistêmê.’ The person with epistêmê about a
given domain grasps the fundamental truths or beings in that domain; they
go beyond superficial phenomena to grasp underlying causes and ultimate
principles.3 Certainly that is how Aristotle understands the sciences
(epistêmai): the scientist has a grasp of the ultimate realities in the domain,
the essences that ground all the rest. (See for example Posterior Analytics I.2,
I.14, II.3.)
Indeed, we can identify a very close cousin or descendant of Plato’s
epistêmê in Aristotle’s nous. For Aristotle too has the idea that there are spe-
cial ontologically privileged items which most of all are—the primary
beings simpliciter, or the primary beings in a given domain like biology or
geometry—and corresponding to that, the idea that there is a special
epistemically privileged power, the power of grasping these (see Posterior
Analytics II.19).
We see this same idea developed in the Neoplatonists’ Science of Being:
nous is direct apprehension of Being, where Being again is the ontologically
superior (see e.g. Plotinus Ennead V.1.4, or Proclus, In Parmenidem I.617).
We also see it in the medieval theory of Scientia, and from there in the early
moderns’ notions of Science or Knowledge as cognition of fundamental
entities—essences.4 On all these views there is a special kind of knowledge
or cognition very different from ordinary knowledge and very much worth
having; what makes it special and valuable is that it is cognition of the fun-
damental realities.
As for modern counterparts of epistêmê, there are many possible candi-
dates. Physics, theology, and foundational metaphysics could each make a
claim to be epistêmê’s heir: the ultimate science of the ultimate reality. For
while we might say that the expert physicist and the expert numismatist
have the same cognitive condition (knowledge, or understanding) with
regard to different objects, we might instead say that the physicist has
something special, a deep understanding, or a specially systematic and
3 See discussion of “rationalist” doctors, in for example Frede 1990.

4 For example, in Spinoza the highest form of cognition involves grasping a thing “through its
essence alone” (Treatise on the Emendation of the Intellect 19); for Leibniz the highest form involves
apprehension of necessary, eternal truths which are grounded in essences, and ultimately in God
(Monadology 29–45). Carriero 2013 argues at length that Descartes, Spinoza, and Leibniz all con-
strued the highest cognition as cognition of essences. For a similar treatment of some medieval
philosophers with explicit comparison to Plato, see Wolterstorff 1996 (quoted in the Introduction):
“the medievals were at one with Plato in their understanding of epistêmê,” because for them too
epistêmê is a grasp of “what is fully real . . . the necessary, eternal, immutable” (1996, 220–1).
explanatorily powerful one, and that this is due to the fact that what she
studies is nothing less than the foundations of reality. Or we might think
that the metaphysician has this, if we think of metaphysics as “an attempt to
know reality as against mere appearance, or the study of first principles or
ultimate truths” (Bradley 1908, 1), aiming at “deep truth, profound under-
standing—fundamental principles” (Horwich 2012, 24). If those descrip-
tions have some appeal to you, you are working with a notion directly
descended from Plato’s epistêmê.
3. A deep grasp of ultimate reality
Considering these counterparts of epistêmê helps us get an intuitive charac-

terization of the notion of cognition of Being. Although it is far from being
the centerpiece of modern epistemology (see next section), neither is it a
technical or archaic notion to which we can find no connection. We can
think of epistêmê as a deep grasp of ultimate reality. Whether or not you
accept the existence and importance of such cognition will of course depend
on your attitude to the notion of ultimate reality. In any case, however, we
can understand the idea—and recognize it, as I have just argued, in the self-
conception of some students of special disciplines like physics or metaphys-
ics or theology. It is the idea of penetrating through superficial phenomena
to get at the underlying nature of things, and in the process of reaching a
level of clarity and insight far beyond that of ordinary thought.
One might reject the notion of ultimate reality altogether, maintaining
that everything that exists is on the same ontological footing. Or one might
accept the metaphysical idea but deny that a grasp of ultimate reality is pos-
sible, or that it is valuable. Or one might argue that such a grasp is simply
knowledge, differing from other kinds of knowledge in its subject matter
but not constituting a distinct, special epistemic kind. All these would be
reasons to reject Plato’s notion of epistêmê. I am not making the case that we
should reincorporate this notion into our epistemology, let alone make it
the centerpiece. I am arguing that we can recognize what he is talking about.
Epistêmê, understood as cognition of Being, is not a narrowly technical
notion, nor one that has life only within Plato’s specific metaphysical sys-
tem. Instead it is his version of an idea that has had appeal throughout his-
tory and has its appeal to plenty of thinkers today: a deep grasp of ultimate
reality.
Epistêmê and knowledge 137
4. Epistêmê and knowledge
I began in the Introduction with the observation that many philosophers

today interpret Plato’s epistêmê as knowledge, either defending this inter-
pretation explicitly or assuming it without argument. I have tried to show
that this interpretation is at best unhelpful: if we begin from our modern
intuitions and theories about knowledge, and try to read these into Plato’s
epistêmê, we not only create problems for ourselves in trying to make sense
of many of his characterizations, but also wind up missing out on his real
starting point: the objects-based account of epistêmê as cognition of Being.
Now that we have fleshed out that account, however, it is worth going back
to the question of epistêmê’s relation to knowledge. Is the upshot that epis
temologists today have completely changed the subject, so that they are no
more part of a conversation with Plato than—one might argue—ethicists
who focus on the question of right action rather than the question of the
good life? Or can we see knowledge, as it is understood and discussed today,
as a descendant or cousin of Plato’s epistêmê, such that contemporary theor
ies of knowledge and Plato’s theory of epistêmê might be mutually
illuminating?
A full answer to this question would be a large project, requiring a
detailed study of contemporary epistemology’s notion of knowledge, which
I will not here undertake.5 But some brief remarks may help indicate how
one might try to connect our contemporary notion of knowledge with
Plato’s notion of epistêmê.
I argued in Chapter 3 that Socrates takes it as a basic principle about
epistêmê, common ground between himself and even his most anti-Platonic
of opponents, that epistêmê is of what is. He thinks that the right way to
understand this principle is in terms of an account of “what is”—Being, the
ontologically superior—which his opponents do not share. We can however
take them to be working with a deflated cousin of that notion: reality, how
things in fact are (see Chapter 3.6). When the Republic’s sight-lovers agree
that epistêmê is of what is, they are plausibly thinking that someone with
epistêmê is in touch with all the relevant facts in the domain, or all the
things that are beautiful—the things that are there to be known. The
5 The following draws on Moss 2019, which poses and answers the question in a bit more
detail. Fine (forthcoming) has an interesting treatment of the issue: using a distinction from
Rawls, she argues that Plato shares our concept of knowledge and works with a different
conception.
Theaetetus’ Protagoras is thinking that someone with epistêmê is in touch

with how things really, rather than merely apparently are.
When contemporary epistemologists take as a basic principle about
knowledge that it is factive, we can very plausibly take the notion of the
facts, or the truth, or true propositions, as ways of developing this general
notion of reality. Plato would of course think these the wrong ways: too
topic-neutral, too independent of the metaphysical distinctions he thinks
central to epistemology. Nonetheless, we can recognize these as a different
version of his own idea.
Thus we can think of modern knowledge and Plato’s epistêmê as two
developments of a common, basic idea, the idea of an especially good cogni-
tive relation to reality. On the modern version, there is no commitment to
metaphysically privileged objects, but only an appeal to the notion of facts,
or true propositions. Somehow a shift occurred: philosophers ceased to
center epistemology around the idea of a relation to special objects, and
instead began to think of it as a relation to facts. The result is that the notion
of knowledge becomes in principle divorced from metaphysics—or, more
precisely, independent of it. This is obviously related to the idea that most
philosophers now think of knowledge as an attitude toward propositions or
facts, whereas—I am arguing, in line with many others—Plato thought of
epistêmê primarily as a relation to objects. As I argued briefly above
(Chapter 2.5), this is thoroughly consistent with him holding that epistêmê
is a propositional attitude, which indeed I am inclined to think that he does.
(Here I am sympathetic to Fine 2004.) The point is that on Plato’s view, to
understand epistêmê we need to know not in the first instance that it is of
true propositions, but rather that it is about a certain kind of object.
(Likewise, one might agree that mathematical knowledge is propositional,
while holding that any account of what mathematical knowledge is must
make essential reference to its subject matter.)
Why did epistemology shift its focus from objects to propositions, from
the special subject matter of Being to the topic-neutral notion of proposi-
tions, from the deep nature of reality to the flat notion of facts? That is a
larger question than I can take on in this book, but considering why Plato
focuses on objects will provide an important part of the explanation.
5. Why epistêmê?
Understanding the notion of cognition of Being leads us to a very import

ant question: why take this as the privileged kind in one’s epistemology?
Why Epistêmê? 139
Why do epistemology this way at all, rather than starting from the kinds we
recognize nowadays, or from intrinsic, purely epistemic features like clarity
or stability? Why base one’s conception of the privileged cognitive kind on
its objects, and specifically why choose as the relevant objects Beings?
The answer to this question will involve an answer to the wider question
about Plato’s epistemological project, and we must thus postpone it until we
have given an account of Plato’s inferior cognitive kind, doxa. Here however
is a brief preview of what I will argue in the Conclusion. Plato’s epistemol-
ogy, like all his philosophy, is driven by ethical concerns, and in particular
by an overarching ethical question we see at work throughout the dialogues:
how should one live? His answer to this ethical question depends on a meta
physical conviction that there are ultimate realities—Beings—the appropriate
kind of contact with which constitutes living well. Moreover, he thinks that
such contact is cognitive. When he turns his attention to cognitive conditions,
therefore—to what we call epistemology—his ideal, the privileged kind he
wants to explain and recommend, the kind he dignifies with the title
epistêmê, is the cognitive grasp of Being.
6
Doxa Is of What Seems
What kind of thing is Plato’s doxa? This is a question nowadays rarely

asked.1 It is widely assumed that Plato had in mind something perfectly
familiar to us from commonsense contemporary epistemology: belief.
More precisely, the widespread view—reflected in standard translations
and background assumptions as well as sometimes in explicit discussions—
is that Plato uses ‘doxa’ to refer to two related cognitive phenomena, each of
which philosophers now refer to with the English word ‘belief.’2 First, what
we might call opinion: a deficient cognitive attitude, one that falls short of
knowledge due to inadequate grounding or subjective uncertainty. When
Plato denigrates doxa and sharply contrasts it with epistêmê, as for example
in the central books of the Republic, he has in mind opinion. Second,
generic believing, or “judgment”: taking something to be true. When Plato
describes doxa in neutral, non-pejorative terms, for example as the internal
analogue of assertoric speech (Theaetetus 189e–190a, Sophist 263e–264a,
Philebus 38b–39a), and when he construes epistêmê as not excluding doxa,
but instead as consisting in true doxa “with a logos” (Theaetetus 201c–d), he
has in mind judgment.3
If my arguments in Chapters 1 and 2 are correct, however, these interpret
ations are deeply problematic. For doxa, just like epistêmê, must be
understood fundamentally as the cognition of a certain kind of object: there
are certain things which are qualitatively similar to doxa, and doxa is in its
essence the power to cognize these. Neither judgment nor opinion as
1 A notable recent exception is Vogt 2012. If we go back fifty years and more we find much
more interest in the question, and a variety of compelling theories on which I will be drawing:
for example Cornford 1941, Crombie 1962, Gulley 1962, Sprute 1962, Havelock 1963.
2 I leave aside the question of whether ‘doxa’ on this interpretation—or for that matter the
English word ‘belief ’—is simply ambiguous between these two senses, or always means the
latter but sometimes connotes the former through implicature.
3 For these two senses of doxa see for example Sprute 1962 and Vogt 2012. Fine finds both
senses in the Republic’s powers argument: “In arguing that knowledge but not belief implies
truth, Plato is distinguishing knowledge as such from belief as such, where this leaves open the
possibility that knowledge is a species of belief. However, at some points in the argument he
uses ‘belief ’ [doxa] for mere belief: for belief that necessarily falls short of knowledge”
(2010, 325).
DOI: 10.1093/oso/9780198867401.003.0007
Doxa is of What Seems 141
owadays understood, however, are individuated or defined by their objects

n
at all. They are object-neutral attitudes. One can have beliefs, in either sense,
about anything whatsoever; there is no special quality that makes something
the kind of thing we can have beliefs about. Moreover, there is no likeness
or kinship between belief and its objects: one can have shaky, uncertain,
vague beliefs about the most stable and clear truths; one can have firm,
unwavering beliefs about the most fleeting affairs. (Recall that this is not
simply an issue about whether or not doxa is a propositional attitude: if my
arguments are correct, then if doxa is a propositional attitude, it is distin-
guished from other propositional attitudes by being about objects of a cer-
tain sort.)
The arguments of Chapters 1 and 2 thus suggest that doxa is radically
different from belief as understood today—different enough that we do not
do well to develop an account of doxa by relying on our views about belief.
Instead, we should begin from the idea that doxa is the cognition of a cer-
tain kind of object. My aim in the current chapter is to develop an account
that meets this criterion; later in the book I will go on to argue that this
account can make sense of Plato’s characterizations of doxa throughout the
dialogues.
I will argue that Plato’s doxa is indeed essentially to be understood as the
cognition of a special kind of object, one which can be identified in two
fundamental ways. First, what will sound to our ears like a mainly epis
temological characterization: doxa is of what seems. Second, an obviously
metaphysical characterization: doxa is of the ontologically inferior counter-
part of epistêmê’s object, whatever way the contrast is drawn. I will show
that these two characterizations are deeply unified, for it is one of Plato’s
central views—most radically developed in the Two Worlds dialogues, but
present from the start—that Being does not seem, while what seems is not
what Is. The ontologically superior is hidden, obscure, inaccessible without
difficult intellectual work—as most dramatically symbolized by the journey
out of the Cave. Meanwhile, what is readily manifest and accessible to the
human mind, what strikes us as clear and true and real when we are not
engaged in any radical questioning, turns out to be the ontologically
inferior—as most dramatically symbolized by the shadows that appear so real
to the prisoners in the Cave. The object of epistêmê is what Is but does not
readily appear; the object of doxa is what readily appears, but is not what Is.
Together with the like-by-like doctrine of cognition, this yields a Basic
Conception of doxa as cognition of what seems. With the metaphysical
developments of the Two Worlds dialogues, doxa becomes a cognitive
142 Doxa Is of What Seems
condition specially suited to grasp the murky, messy, perceptible realm of

Becoming. This account thus explains doxa’s restriction to perceptibles in
the Two Words dialogues, and also the other central features Plato attri
butes to doxa (Chapters 7 and 9).
Identifying doxa’s defining object as what seems will thus yield an answer
to our starting question: what is Plato’s basic conception of doxa? I consider
the resulting picture in Chapter 8. We will find that doxa does not corres
pond precisely to anything in our epistemology, although it is something
like empirical cognition, and something like delusion. Although it has
something in common with judgment, and a good deal in common with
opinion, there are significant differences too, and we only understand the
notion adequately when we begin, not from the assumption that doxa is like
belief, but instead from the characterization of doxa as cognition of the
ontologically inferior things that seem. In the Conclusion, I consider why
Plato would make this the salient inferior kind in his epistemology.
One important note is in order before I begin to develop my account.
Some might think this whole project obviously misguided, on the grounds
that doxa simply has to be belief. Belief is such a central epistemological
concept that anyone who does epistemology must be concerned with it, and
the only candidate for it in Plato’s system is doxa; ill fits with our usage are
thus to be explained by quirks in Plato’s theories of belief, not by a radical
difference in topic.
I agree that Plato recognized and often referred to something very like
our modern concept of belief. Along with his predecessors and contempor
aries, he often uses the verbs ‘hêgêsthai’, ‘oiesthai,’ ‘nomizein’ and ‘hupolam-
banein’ in ways we would translate with ‘believe’ or ‘think.’ They sometimes
use ‘doxa’ and its cognate verb, ‘dokein,’ in this way too, but even a brief
survey suffices to show that they often use them with much narrower con-
notations: connotations of mere seeming (Section 3 below). I will argue that
when Plato uses ‘doxa’ to pick out the salient inferior counterpart to
epistêmê, he has these connotations in mind.
In Chapter 8 I will consider the connections between Plato’s doxa and
our modern notion of opinion. As for the more general notion of belief as
judgement, it seems that throughout most of the dialogues Plato had little
interest in theorizing it at all, let alone under the rubric of doxa.4 In later
4 In the Republic Plato does recognize doxa and epistêmê as belonging to a common genus,
just as we class opinion and knowledge together as kinds of belief, but he has little to say about
the genus, and names it only with very broad terms that cover many non-cognitive states
Doxa and seeming 143
dialogues, the Theaetetus, Philebus, and Sophist, we do arguably find him

developing this notion, and using ‘doxa’ to name it.5 I will say something
about this later development in Chapter 10, but my main focus throughout
is on the notion that is, I shall argue, central to the epistemology of the
Socratic and Two Worlds dialogues: inferior, deficient cognition, which
Plato understands, I will argue, as cognition of what seems.
1. Doxa and seeming
In Chapter 3 I argued that epistêmê’s defining object is Being, where Being is

a contrast notion, dependent on the distinction between a superior and an
inferior level of reality. As we saw there in Contrasts 1–4, Plato has various
ways of describing the inferior domain: for example, what Becomes, or the
many F things, or or images. I will show in this chapter that there is one
characterization which unifies the rest; it is moreover one that Plato would
have thought too obvious to need special emphasis as the unifying one,
because he would have plausibly thought of it as part of the very concept of
doxa, or at least as a basic principle in need of no argument. The unifying
idea is that doxa is of what seems, by contrast with what really is.
Doxa is the noun formed from the verb dokein, standardly translated ‘to
seem.’ On one prominent use, a thing’s doxa is the way it seems to others: a
person’s doxa for example is her reputation. Dokein with the dative means
“to seem to…,” and so a person’s doxa in a second sense—the sense con-
trasted with epistêmê—is how things seem to her. Traditionally the former is
called the objective and the latter the subjective sense of doxa (see for
ex
ample Bosanquet 1895, 209 and Havelock 1963, 250–51). To avoid
unnecessarily confusing connotations6 I will characterize the former sense
as the object-sense, or as referring to “what seems,” and the latter as the
subject-sense; for discussion see the end of this section. (For convenience,
(dunameis, 477b–d; pathêmata, 511d). See Moss and Schwab 2019 for a fuller argument for
this claim.
5 As many have argued: see for example Gulley 1962, Sprute 1962, Vogt 2012, Moss and
Schwab 2019.
6 Since “objective seeming” may sound paradoxical, and “subjective seeming” tautologous.
when the contrast is not salient, I will continue to use ‘doxa’ without
ualification to pick out the subject-sense—as I have done up to now.)
q
Given this etymology, it is thus built into the meaning of doxa in the
subject-sense that it is a response to what seems. If we are looking for the
object of doxa, that which stands to doxa as color does to sight and Being to
epistêmê, we might therefore naturally turn to Plato’s notion of what seems.
It is certainly possible that we would find nothing of significance here:
perhaps Plato uses doxa without intending to emphasize this connotation,
or perhaps he construes seeming so neutrally that by “how things seem to
S” he understands nothing more than “what S thinks is the case” or “what S
holds true,” so that doxa is simply generic belief.
Indeed, the words do sometimes work this way. I will argue that when in
the later dialogues Sophist and Theaetetus Plato uses ‘doxa’ as a label for
“silent speech,” asserting something to oneself, he may well be appealing to
a neutral sense of seeming: one’s doxa is how things seem to one, i.e. generic
belief (see Chapter 10.3). This, however, turns out to be an exception to the
norm—a secondary sense of seeming and of doxa. Throughout the dia-
logues, and particularly in the Two Worlds dialogues, Plato makes use of a
metaphysically loaded notion of what merely seems, by contrast with the
true or real, and he has a correspondingly loaded notion of doxa not as
generic belief, but as a response to such appearances.
In making this argument I am closely following the interpretation of
doxa advanced by F.M. Cornford in a series of works from the 1930s and
1940s, and occasionally suggested by others.7 As Cornford puts it in his
commentary on the powers argument in Republic V:
Doxa and its cognates denote our apprehension of anything that ‘seems’:
(1) what seems to exist, sensible appearances, phenomena (2) what seems
true, opinions, beliefs, whether really true or false; (3) what seems right,
legal and deliberative decisions, and the ‘many conventional notions’ of
current morality (479d…) (Cornford 1941, 176)
7 The idea appears in Cornford 1933, 1935, 1937, 1941. Others too have sometimes noted

the significance of doxa’s etymology. Vogt makes note of the point early in her discussion of
Plato’s doxa, using it to suggest a connection that I will defend at length in Section 3: “It is help-
ful to recall the closeness of the Greek doxa with notions of seeming, appearances, and reputa-
tion. The verb dokein means ‘to appear,’, and doxa thus often denotes something in the domain
of appearances . . . If the Republic is read in Greek rather than translation, Book I is strikingly
continuous with the metaphysics of the middle books: doxa is associated with the domain of
appearance and perception, and contrasted with the domain of being” (Vogt 2012, 11–12).
Compare Bosanquet’s comment on Republic 476d: doxa is “ ‘Opinion’ or ‘seeming;’ the substan-
tive corresponding to the everyday Greek phrase ‘it seems to me,’ and conveying a suggestion
of difference from truth or reality” (1895, 209).
Doxa and seeming 145
Since Cornford tends to assert his interpretation more than to argue for it,
however, and since it has not had sufficient influence, I find it worth
resurrecting, defending, and developing the view in some detail here.
A clarification is in order before I turn to the evidence. It may seem a
mistake to speak of an ontological category of seemings that could serve as
the object of doxa. In other words, it may seem a mistake to take the
straightforward semantic distinction between object and subject senses of
‘doxa’ to track a distinction between metaphysical seemings out in the
world and epistemic seemings in the mind. For one might argue that all
seemings are mental: claims of the form “x seems F” always reduce to claims
of the form “S has an impression as of x being F.” I will nonetheless speak of
object-doxa and subject-doxa as distinct categories, and this for several
reasons.
First, clarity: Greek does distinguish between object and subject senses of
‘doxa,’ and my arguments below turn on the close connection in which
Greek thinkers use these two. The simplest way to put this is that they con-
strue subject-doxa (being seemed-to) as a response to object-doxa (what
seems): a person has doxa when she attends not to things as they are, but to
something else, what seems. So that is how I will talk. For much of the time,
the reader is welcome to take this as only a way of talking: to take all talk of
object-doxa as shorthand for talk of something subjective. If this is as far as
we go, we can take the idea that doxa is of what seems to mean that doxai
are superficial judgments based on how things affect or strike us, rather
than judgments that correct for perspective, question appearances, or
otherwise penetrate to the nature of things. For the most part this will suit
my purposes very well.
There is also however a more substantive reason for embracing talk of
object-doxa: Plato holds that there is an ontological realm of seemings or
appearances, things which are distinct from what is ultimately real or true,
but which nonetheless have some mind-independent existence. His canon
ical examples of seemings (which we will see in the next section) exemplify
this status: façades, images, and reputations are distinct from the real things
they imitate or mask, but they belong on the object side of the object-subject
divide; they are not merely a matter of how things strike a subject.8 We will
8 Not merely: perhaps the best way to construe Plato’s notion of object-seeming is as a
response-dependent one: for x to seem F is for x to have the tendency to produce impressions
of its being F in certain subjects. This tendency or disposition would however be rooted in
response-independent features of the object. For example: two sticks tend to appear to us to be
both equal and not equal because they are, in themselves, changeable and imperfect (Phaedo
74b). For a compelling defense of the notion of objective seeming, see Gabriel Lear (work in
see the same in his characterizations of the perceptible realm as something

that is not the real thing (what Is), but also not an illusion—instead, more
like a shadow or image, things which again have some mind-independent
existence (see Chapter 7.5). All this suggests that there are facts about what
seems rooted in something outside our minds, and it is thus not only intel-
ligible but even non-metaphorical to speak of an ontological realm of seem-
ing, something that can serve as the object of doxa just as color serves as
the object of sight or Being as the object of epistêmê. Although I suspect that
most of what I have to say about doxa can be secured on the reductivist,
metaphorical interpretation of object-doxa, it is worth noting that Plato
shows many signs of embracing the literal one.
2. What seems
Seeming in its object-sense is a contrast notion: it is one half of an antithesis,

defined in opposition to being or, sometimes, truth.9 Dokein and einai are
frequently presented in Greek texts as a contrasting pair, like logos/ergon or
nomos/phusis. To take two representative examples: “He wished not to seem
the best, but to be so”10 (Aeschylus, Seven Against Thebes 592–4, referred to
at Republic 361b–362a); “You who both are and seem to be the first among
the Greeks”11 (Gorgias, Palamedes 33). Although as the latter passage shows,
seeming and being can converge—things often seem just as they are—there
is nonetheless a conceptual distinction between the two. We also find an
antithesis between doxa and truth, alêtheia—which is not surprising, given
the frequent near-equivalence between truth and being (see Chapter 4.1).
Consider the division of Parmenides’ poem into two sections, Alêtheia and
Doxa, or a more mundane example from Antiphon:
For the doxa of the things done is advantageous for those who are able to
speak, but the truth is advantageous for those who do just and pious
things. (Antiphon, Second Tetralogy 3.2.2)
progress): for Plato, appearing is “a matter of a thing’s showing or displaying itself and the
appearance constituted by this activity is there for minds to register….”
9 In Chapter 3.4 I argued that the contrast with seeming is not part of the very concept of
Being, for Plato. This asymmetry is fitting, given the derivative status of seemings.
10 οὐ δοκεῖν ἄριστος ἀλλ’ εἶναι θέλει. 11 τοῖς πρώτοις οὖσι τῶν Ἑλλήνων καὶ δοκοῦσιν.
What seems 147
In all these contexts, the claim that x seems has two implications: (1) it is
easy to form an impression that x is a certain way—x strikes us that way
vividly or manifestly; (2) this does not entail anything about how things
really are.12 (Note that there are other words that function this way in Greek:
phainesthai, “appear,” and eoikein, “be like,” with adjectival and nominal
cognates. At various points in the corpus Plato uses these words as he often
uses doxa and dokein.13 I focus here on doxa because it is Plato’s most
frequently used term for the seeming that is opposed to being—and his
most frequently used term for that seeming’s subjective correlate, as we will
see below.)
In some contexts, the contrast between seeming and being is sharpened
into an opposition: how things seem is a very deceptive guide to how they
are; a thing’s seeming (its doxa in the object-sense) is a façade that is not
only ontologically distinct but also qualitatively very different from its
being. For some everyday examples, see LSJ s.v. dokeô I.4 “seem, pretend.”
We also find significant instances of this use in the works of some of Plato’s
influential predecessors. Parmenides contrasts “the unshaken heart of
persuasive truth” with “the things that seem [ta dokounta]” (B1.28–32).
Xenophanes concurs:
Now what is clear, no man has known/seen [iden] nor will any know it,
concerning the gods and about all the other things that I am saying. For
even if he should completely succeed in describing things as they come to
pass, nonetheless he himself does not know; seeming [dokos] is wrought
over all. (Xenophanes B34)
12 Compare Shakespeare’s Hamlet: “Why seems it so particular with thee? – Seems, madam!
nay it is; I know not ‘seems’ ” (Hamlet Act I scene 2).
13 The eikos, in place of to dokoun, is contrasted with truth and presented as the object of
doxa in Xenophanes B35 (“Let there be doxa of these things [δεδοξάσθω] as resembling the true
things [ἐοικότα τοῖς ἐτύμοισι]” and in Parmenides B8.50–62: “Here I end my trustworthy
account and thought about truth. Learn now from this point mortal doxai, hearing the decep-
tive ordering of my words….I declare to you this arrangement to be wholly resembling the
truth (ἐοικότα), so that no judgment of mortals will ever surpass you.” Plato picks up on this
use in the Timaeus (29b–d, discussed in Chapters 2.3 and 4.5) and in the Phaedrus (orators see
that eikota are more honored than truths, 267a). The eikos in these contexts, and to phaine-
menon or phantasma throughout the dialogues, function in the object sense (to characterize
things that seem) with connotations (1) and (2). They also have their subjective counterparts:
eikasia in the Republic (511d) and phantasia in the Sophist (264a). There is also pithanon and
its subjective correlate, pistis, which sometimes shows up in place of doxa (Gorgias 454d) or
alongside it (Timaeus 37b). On the eikos in Plato and his predecessors see Bryan 2012 and
Mourelatos 2014.
Heraclitus agrees too: “The most seemingly good [reputable] of them

recognizes and guards [only] what seems”14 (B28)—by contrast, it is

implied, with truth. Most dramatically, the poet Simonides says that
seeming “fights against truth”15 (fr. 76).
Plato himself makes extensive use of this notion of seeming throughout
the dialogues. The Republic is particularly rich with instances.16 The con-
trast between really being just and merely seeming so pervades the first part
of the dialogue: it is central to both Glaucon’s challenge and Socrates’
response (see specially 358a, 361a–362c, and 367b–d). In developing his
brother’s challenge Adeimantus even quotes the line we just saw from
Simonides:
[Someone raised on popular culture will think that] my being just, if I do

not also seem so [dokô], is no benefit . . . but a godlike life is said to belong
to the unjust person who has prepared a seeming [doxa] of justice [viz., a
reputation]. Surely, if “seeming,” as the wise show me, “even fights against
truth,” and is lord of happiness . . . I must draw a shadow-painting of virtue
around myself in a circle as a front and a façade [prothura kai schêma]. . .
(Republic 365b–c)
The first part of Republic also emphasizes this contrast between being and
seeming in other domains: the arguments turn on contrasts between real
and seeming friends (334c–335a), between what seems to be to the ruler’s
advantage and what really is (340c), between seeming to persuade people
and really persuading them (357a–b), and between seeming good and being
so (361b). The contrast recurs later in the dialogue: to take one notable
example among several, Socrates says that while everyone is content with
what seems just or fine even if it is not really so,
Obtaining seeming [ta dokounta] goods satisfies no one, but they seek the
real ones [ta onta], and here everyone disdains the seeming [doxa].
(Republic 505d)
In all these cases, Plato appeals to variations on doxa and dokein to express
that something can seem to be some way whether or not it really is: not all
14 δοκέοντα γὰρ ὁ δοκιμώτατος γινώσκει, φυλάσσει.

15 τὸ δοκεῖν καὶ τὰν ἀλάθειαν βιᾶται. 16 As emphasized by Vogt (2012).
Being-seemed-to 149
seeming friends are real friends, not all seemingly just people are really just.
Moreover, Plato’s pessimism in the Republic often implies the stronger thesis
we see in Simonides and others: things tend to seem quite different from
how they really are.
The contrast between seeming and being is pervasive in other dialogues
too. There are far too many instances to list here, but to take a few repre-
sentative examples, consider the Apology’s discussion of those who seem or
are reputed to be wise, but are not (21c), the Gorgias’ contrast between
the seeming good condition of body or soul and the real thing (464a), the
Theaetetus’ distinction between seeming and real advantage (172b), or the
Sophist’s claim that false images are a matter of “appearing and seeming but
not being”17 (236e).
Plato thus joins his predecessors and contemporaries in making promin
ent use of a robust notion of seeming on which what seems is sharply con-
trasted with what really is.
3. Being-seemed-to
Plato also joins his predecessors and contemporaries in making prominent

use of a corresponding notion of doxa as the subjective correlate of what
seems: a use on which doxai are reactions to how things seem by contrast
with how they are. If your attention is confined to what seems—if you are
attending to reputations, façades, images, or appearances—you will as a
result have doxa rather than epistêmê. To put it another way: subject-doxa is
contrasted with epistêmê just as object-doxa is contrasted with being or
truth, and the two contrasts run in parallel.
This correlation is manifest in Gorgias’ Helen, where orators make “what
is incredible and unclear appear [phainesthai] to the eyes of doxa” (Helen
13, cf. 10): appearances in the world are received on the subject’s end by
doxa. In his Palamedes, knowledge is of truth, while subject-doxa in the
subjective sense is of doxa in the object-sense, what seems:
You have trust in doxa . . . not knowing the truth….But you should not
trust those who have doxa but instead those who know, nor should you
17 τὸ γὰρ φαίνεσθαι τοῦτο καὶ τὸ δοκεῖν, εἶναι δὲ μή.

consider doxa more trustworthy than truth, but the opposite – the truth
more than doxa. (Palamedes 24)18
(In the contrast between doxa and knowing, the verb rendered as “have
doxa” is doxazein, usually translated “to opine”; here the sense is clearly
subjective. In the last line, where doxa is contrasted with truth, Gorgias
more plausibly has in mind its object sense: how things seem rather than
how they truly are.)
We see something similar in Xenophanes’ exhortation to “have doxa of
these things as resembling the true things” (Xenophanes B35). The “true things”
are the object of a higher cognitive condition (see the variants on the verb
eidenai in B34, quoted above); the things that merely resemble them are the
object instead of doxa. The implication again is that subject-doxa is of what
seems, by contrast with reality.19
The correlation persists after Plato’s time. In the commentary on
Aristotle’s de Anima attributed to Simplicius, the author refers to a use of
doxa to mean “cognition in accordance with the superficial and seeming”
(ps-Simplic. in De Anima 309.35–6).20 The correlation is very explicit in a
Stoic text:
We say that the wise person’s not having doxa [mê doxazein] is accompanied
by such characteristics as, first of all, nothing seeming [dokein] to them
so-and-so; for such ‘seeming’ [dokêsis] is non-kataleptic doxa . . . (Anonymous
Stoic treatise, (Herculaneum papyrus 1020, col. 4, col. 1= SVF 2.131, part;
translation based on Long and Sedley 1987).
Someone with no doxai is not subject to how things seem. The Stoics of
course do not mean that things in no way strike the wise person; the claim
is that things do not seem in the special way that constitutes doxa’s objective
correlate.
18 δόξῃ πιστεύσας . . . τὴν ἀλήθειαν οὐκ εἰδώς . . . ἀλλ’ οὔτε τοῖς δοξάζουσι δεῖ πιστεύειν ἀλλὰ
τοῖς εἰδόσιν, οὔτε τὴν δόξαν τῆς ἀληθείας πιστοτέραν νομίζειν, ἀλλὰ τἀναντία τὴν ἀλήθειαν τῆς
δόξης. I am translating the variants on eidenai here as “knowledge”; it is a worthwhile project,
but one I leave for another occasion, to ask how they relate to Plato’s conception of epistêmê.
19 In the other passage we saw from Xenophanes (B34), it is not clear whether he intends
dokos to have a subject sense or an object one, or perhaps to leave it indeterminate; in any case,
truth (object) and knowledge (cognitive condition) stand on one side of a divide, while seem-
ing stands on the other.
20 τὴν κατὰ τὸ ἐπιπόλαῖον καὶ δοκοῦν γνῶσιν. The author contrasts this with another use,
taking this to be the one intended by Aristotle in the passage to which this is a commentary
(De Anima 434a–10–12), which denies doxa to non-rational animals.
Being-seemed-to 151
Turning back to Plato, we find ample evidence of the correlation between

subject- and object-doxa.21 It comes out particularly clearly in his discus-
sion of rhetoric and sophistry, arts that he characterizes as both (a) con-
cerned with what seems rather than what is, and (b) producing doxa rather
than epistêmê.22 Consider especially two passages which move without
argument between the two characterizations:
[Phaedrus]: It’s necessary for the person who intends to be an orator to

learn (a) not what is really [tôi onti] just but what seems so [ta doxant’] to
the many who will pass judgment, nor what is really [ontôs] good or fine
but which things will seem so [doxei]. For persuasion is [produced] from
these things, and not from the truth. . . – [Socrates]: The orator, not know-
ing good and bad, undertakes to persuade a city in the same condi-
tion . . . about the bad, that it is good, and (b) having studied the doxai of
the many, persuades them to do bad things instead of good ones. . .
(Phaedrus 260a–c)23
[The sophist can] bewitch people who are young and standing still further
from the truth of things, showing them spoken images about everything,
(a) so as to make the things spoken seem true, and the speaker [seem] to
be the wisest of all about everything…[But when later the students] grasp
clearly the things that are, they will (b) change their former doxai.
(Sophist 234c–d)
Plato takes the claim that (a) the orator or sophist presents what seems just
(or good or fine) to entail the claim that (b) the orator is concerned with the
audience’s doxai. He takes the claim that (a) the sophist makes things seem
true and makes himself seem wise to entail the claim that (b) he has in so
doing produced doxai in the audience. Evidently Plato does not sharply dis-
tinguish between studying how things seem to people and studying their
21 Cornford argues that Plato is following Parmenides in this, citing his use of dokein and
also phainesthai in the first section of the Theaetetus and in Republic V (1933, 101; cf.
Palmer 1999).
22 In addition to the passages quoted below, for claim (a) see among others Gorgias 464a–b
(rhetoric makes the soul seem to be in good condition rather than to be so; Sophist 233c,
Republic 602b, and Gorgias 459b–c: “there is no need for the orator to know the things them-
selves, how they are [αὐτὰ μὲν τὰ πράγματα οὐδὲν δεῖ αὐτὴν εἰδέναι ὅπως ἔχει], for he has dis-
covered a mechanism of persuasion which makes him appear [φαίνεσθαι] to non-knowers to
more than the knowers.” For claim (b), see also Theaetetus 201a, discussed in Chapter 10.
23 Phaedrus speaks the first part of this passage; Socrates disagrees with the advice to
orators, but agrees that this is how things are done at present.
doxai; he does not sharply distinguish between influencing how things seem
to people and giving them new doxai. I suggest that he is taking it as obvi-
ous that a subject-doxa is the result of accepting how things seem, where
that is sharply contrasted with how things are. Consider also two comments
he makes in the continuation of the discussion in the Phaedrus:
In political speaking [the orator] will make the same things seem [dokein]
to the city at some times good and at other times in turn the opposite.
(Phaedrus 261d)
[The orator is one who] does not know the truth, but hunts down doxai.
(Phaedrus 262c)
In the first passage should we translate “he makes the same things seem
good and bad to the city,” or “he makes the city have the [subject-] doxa that
the same things good and bad”? In the second should we translate “hunts
down how things seem,” or “hunts down [subject-] doxai”? I submit that
there is no fact of the matter here. The claims are fundamentally ambiguous,
for to make x seem F to S is at the same time to give S the subject-doxa
that x is F.
We find a similar move with the same implication in a very different con-
text. In Republic VII, in describing the philosophical investigation that will
lead to epistêmê of the Good, Plato strongly implies that this method yields
epistêmê because it attends to being, while alternatives leave us with doxa
because they attend instead to seeming:
Whoever is not able to define the Form of the Good by logos. . . (a) being
zealous to examine it not in accordance with doxa but in accordance with
Being 24 . . . you will say that the person like that does not know [eidenai] the
Good itself nor any other good. But if he should in a way lay hold of some
image, you will say (b) he lays hold of it by doxa, not epistêmê, and spends this
present life dreaming and sleeping. (Republic 534b–c, emphases mine)
In Chapter 7 I will discuss how to understand the idea of investigating

Forms in accordance with being as opposed to seeming, but for now we can
note the pattern. In its first occurrence (in (a)) doxa is contrasted with being
(ousia), and so we should take it in the object sense, how things seem: we
24 μὴ κατὰ δόξαν ἀλλὰ κατ’ οὐσίαν.

The Basic Conception of doxa 153
can follow Adam’s translation of the italicized portion as “striving to test his
view not by that which seems, but by that which is.” In its second occurrence
(in (b)) the contrast is instead with epistêmê, so we now have doxa in the
subject sense, being-seemed-to. Investigating things in accordance with
what seems rather than with being gives one doxa rather than epistêmê; a
focus on seemings in the world produces being-seemed-to in the mind.
These unargued moves from claims of type (a) to claims of type (b) sug-
gest that Plato is working with an implicit theory of doxa by contrast with
epistêmê as a response to what seems by contrast with reality. I want to draw
out two implications of this.
First, we can add another contrast to Chapter 3’s list, one that again cor-
relates the ontologically superior with epistêmê and the ontologically infer
ior with doxa:
(Contrast 5): What Is versus what seems.
(In the passage from the Republic, Plato is contrasting Forms and percepti-
bles; in the passages on rhetoric, he is instead contrasting perceptibles as they
are and as they appear. Nonetheless both contexts distinguish something
ontologically superior from something ontologically inferior—although
in the second context the superiority is only relative. See Chapter 7.1 for
related discussion.)
The other implication is that while Plato correlates doxa with various
items, or with things under various descriptions, as we saw in Contrasts 1– 4
(where epistêmê is correlated with Being and doxa with the in-between, or
Becoming, or the many Fs, or images), the correlation with doxa in this par-
ticular contrast is special. The etymological connection, and the unargued
moves between claims about object- and subject-doxa, all suggest that Plato
takes “Doxa is of what seems” to be as obvious and uncontroversial as
“Epistêmê is of what is.”
4. The Basic Conception of doxa
This finding, together with the evidence that Plato’s epistemology is objects-
based, suggests a hypothesis: just as the defining object of epistêmê is Being,
the defining object of doxa is seeming. Plato inflates the linguistic association
between doxa and dokein into a substantive theory on which doxa is, in its
essence, cognition of what seems.
In the next chapter, I show that this hypothesis is borne out. The
interpretation can do the work we want from a Basic Conception: it can
explain Plato’s characterizations of doxa. Moreover, although Plato often
characterizes doxa’s object in other ways—as “what is between being and
not-being,” in the powers argument, or as perceptibles, or Becoming—all
these turn out to be unified under the category of seeming. In Chapter 8
I will turn to the question of how we should understand doxa on this
interpretation.
7
The Basic Conception
of Doxa at Work
On an objects-based epistemology, each cognitive kind is the way it is

because it is that by which we cognize a certain kind of object; all its features
are explained by this relation. If doxa’s defining object is what seems, we
should be able to appeal to that fact in order to see why Plato attributes to
doxa all the features that he does. In this chapter I aim to show that this
expectation is borne out.
As with epistêmê, my focus is on Plato’s characterization of doxa in the
Two Worlds dialogues, and I devote the most attention to the most mysteri-
ous facet of doxa by modern lights: the Two Worlds dialogues’ implication
that there is no doxa of Forms. I consider doxa in the earlier and later
dialogues in Chapters 9 and 10.
1. Truth and falsity
As we saw in Chapter 4.1, at times Plato says that only the objects we
encounter through epistêmê are true, implying that in doxa we never attain
the truth. Philosophers have epistêmê because they behold the truth; the
people with doxa behold not the truth about Beauty, but only something
inferior, sights and sounds (Republic 475e, quoted in 4.1); epistêmê belongs
to those who behold things “true and divine and not the objects of doxa
(adoxaston)” (Phaedo 84a).
At other points, however, Plato says that doxa can be true. The Two
Worlds dialogues pay less attention to the distinction between true and false
doxa than the Meno or Theaetetus, but even here it is important: false doxai
about important things are dangerous and must be eradicated; in some
contexts at least, true doxai are valuable and should be pursued. (See for
example Plato’s characterization of the educational program as instilling
and preserving true doxai in the young (Republic 412e–413a).)
DOI: 10.1093/oso/9780198867401.003.0008
156 The Basic Conception of Doxa at Work
Notably, this ambiguity about doxa’s relation to truth tracks an ambiguity

about doxa’s relation to being. We have seen plenty of evidence that doxa is
of the ontologically inferior by contrast with Being; when Plato speaks of
true doxa, however, he also speaks of doxa as having access to being. We see
this in his definition of true doxa in the Sophist: true doxa, as the inner
analogue of true logos, “states the things that are as they are” (legei . . . ta onta
hôs estin, Sophist 263b with 263e–264a). A similar claim makes a brief
appearance in the Republic, the very dialogue which makes the most explicit
claims to the effect that doxa is not of Being: to get truth (alêtheuein) is to
have doxa of the things that are (ta onta doxazein, 413a; see also Symposium
202a, where true doxa hits on what is (tou ontos tunchanon)).
We can resolve the ambiguity if we assume that in the contexts where he
speaks of true doxa, or of doxa of what is, Plato is not using ‘truth’ and
‘being’ with their special function of marking ontological superiority, but
instead only in one or another of the ordinary senses canvassed in
Chapter 3: what exists, what is the case, what is F for some property.1 In
these contexts, Plato will say that some doxai are of what is, namely the
doxai that accurately capture ordinary facts. Given the conceptual link
between being and truth (see Chapter 4.1), he will therefore also grant
these doxai the title ‘true.’
Our Basic Conception dictates a clear explanation for this fluctuation.
There are better and worse seemings, and correspondingly better and worse
doxai. A just person may veridically seem to be just or deceptively seem to
be unjust (Republic 365b); some images accurately reproduce the propor-
tions of the originals, while others distort them (Sophist 235d–236c).
Nonetheless, the line between seemings as a whole and Being is much
brighter than that between different kinds of seeming. Seeming is never
identical with Being, and so never touches the unqualified truth.
The Basic Conception predicts that when Plato is most focused on the
inferiority of seeming, he will be least inclined to speak of true doxa, and
that prediction is borne out. In the Two Worlds dialogues, as I will argue
below, doxa is of Becoming, and there is no truth or Being in the realm of
Becoming; it is no surprise then that the Republic’s elaborate epistemo
logical classifications in the Line and Cave discussions never mention the
difference between true and false doxa, but instead oppose doxa without
qualification to higher conditions. Indeed, it is not difficult to see how
1 For example, after defining true doxa as being of what is, the Theaetetus identifies “the
things that are” as “the things that come through the senses” (Theaetetus 194d).
Inferiority 157
someone could interpret Plato as entirely rejecting the notion of true doxa,
as indeed some of his Ancient interpreters did:
Plato held that the whole criterion of truth and truth itself was removed
from opinions [viz., doxai] and from the senses, and belonged to thought
itself and the mind. (Cicero Academica II.142, emphasis added)2
The Republic does introduce what looks to be a different way of distinguish-

ing between better and worse doxai: it identifies two species of doxa, pistis
(trust or conviction) and eikasia (imagination, conjecture). Corresponding
to these are two levels of ontology which are, I will argue below, both things
that seem. One is further from Being and the other closer, but both only
seem, and thus both are the objects not of epistêmê but of doxa.3
2. Inferiority
Unlike contemporary epistemology’s belief, in the sense of generic judg-

ment, doxa in the Two Worlds dialogues—and indeed throughout most of
Plato’s works—is an inferior condition.4 This is one of the features that dis-
tinguishes it from contemporary belief: belief is a generic phenomenon with
inferior kinds, like opinion, and superior kinds, like knowledge; thus to
acquire a superior grasp of something is not to lose one’s beliefs, but instead
to improve them. According to the Two Worlds dialogues, on the other
hand, to acquire epistêmê is to leave doxa behind.5 (In this way doxa is akin
to the contemporary folk notion of opinion. For connections between doxa
and opinion see Chapter 8.4.) The Basic Conception explains this neatly:
once someone grasps how things are, they have no need to rest content with
how things seem. This is a truism, independent of substantive metaphysical
2 omne iudicium veritatis veritatemque ipsam abductam ab opinionibus et a sensibus cogita-

tionis ipsius et mentis esse voluit. ‘Opinio’ is the standard Latin translation of ‘doxa.’
3 Plato does say that the objects of pistis are more true than those of eikasia, and we can
infer that a doxa based on acquaintance with a just person (for example) is truer than a doxa
based on acquaintance with a poetic representation of a just person; nonetheless, even the for-
mer does not get at the real truth. I return to this subject below.
4 Compare Vogt 2012.
5 As symbolized by the distinction between the lower sections of the Line as species of doxa
and the higher sections as something else, or by the inside of the Cave as the realm of doxa
(doxaston) and the outside as something else.
views about being and seeming; below I will show how Plato inflates this
truism to yield the Distinct Objects view.
At certain points in some later dialogues—especially the Theaetetus—
Plato does seem to treat doxa as generic belief, with epistêmê as a special
kind of doxa. Arguably here Plato’s use of ‘doxa’ simply changes. I discuss
this development in Chapter 10, arguing that even here we can see the Basic
Conception at work. In the rest of this book, however, I confine my atten-
tion to doxa as we see it in the Two Worlds dialogues (and, I will argue the
earlier ones too): an inferior cognitive condition, opposed to epistêmê.
3. Instability
A main feature of doxa emphasized throughout the dialogues is its instabil-

ity. Someone with a true doxa may well lose that doxa in the face of tempta-
tion or fear (most explicit at Meno 97e–98a, but strongly implied at Republic
430a, where special measures must be taken to ensure that correct doxa will
not be “washed out” by temptation or fear, and at Protagoras 356d–357a,
where epistêmê is what ensures that we will not change our minds in the
face of deceptive appearances), or through a process of persuasion (strongly
implied at Gorgias 459a–c, where those who lack epistêmê are subject to
persuasion; see also the Platonic Definitions, where doxa is defined as “open
to persuasion [metapeistos] by logos,” by contrast with epistêmê (414b–c).
This feature is easily explained by the Basic Conception, for Plato clearly
holds that appearances are shifty. They change depending on the circum-
stance, the perspective, the subject’s position in space relative to an object,
or in time relative to an event (Republic 598a, 602c, Phaedo 74b, Protagoras
356d–e). Thoughts based on how things seem will thus predictably
change—by sharp contrast with thoughts based on how things are. This is
obviously an instance of the doctrine of like-by-like, joined with the Basic
Conception: cognition of unstable objects (seemings) is itself unstable. In
Section 5 we will see more evidence for this explanation of doxa’s instability,
along with its other messy features: its unclarity and imprecision.
4. Persuasion
In several dialogues, Plato characterizes doxa as the product of rhetorical

persuasion, by contrast with teaching:
Persuasion 159
Nous and true doxa . . . must be said to be two different kinds, because they
come to be separately and are unlike. The former one arises in us through
teaching, the latter through persuasion [peithous] . . . And the former is
unmovable by persuasion, the later we can be persuaded to change
[metapeiston]. (Timaeus 51d–e; cf. Gorgias 454e)
Orators and lawyers. . . persuade, using their craft, not teaching but mak-
ing people have doxa of [doxazein] whatever they wish them to.
(Theaetetus 201a)
Indeed Plato seems even to define persuasion as the producing of doxa:
Don’t you mean by ‘persuading someone’ making them form a doxa?6
(Theaetetus 201b)
The connection between teaching and epistêmê looks quasi-definitional,

here and at other points in the corpus: a process only counts as teaching if it
results in epistêmê. But why should persuasion produce only doxa?
Plato’s texts suggest several simple pragmatic explanations. There is no
time to do better: imparting epistêmê is a lengthy procedure, which the ora-
tor under the clock cannot undertake (Theaetetus 201a–b, Gorgias 455a).7
There is no need to do better: the orator is concerned to influence people
rather than to promote truth (Phaedrus 260a). But these are contingent
restrictions: an orator with sufficient time, and noble motivations, could
overcome them. Plato seems to view the tie between rhetoric and doxa as
something stronger—as if there were something in the nature of rhetorical
persuasion, or something in the nature of doxa, that explains it. Persuading
just is producing doxa (Theaetetus 201a); doxa is the kind of thing that
arises from persuasion (Timaeus 51d).
Our Basic Conception of doxa provides a simple explanation of this tie,
and one that has ample support in Plato’s texts. Doxa is of what seems;
rhetoric trades in seemings rather than reality; therefore rhetoric pro-
duces doxa.
6 τὸ πεῖσαι δ’ οὐχὶ δοξάσαι λέγεις ποιῆσαι.

7 In the lines immediately following those quoted above: “Or do you think that there are any
teachers so clever that they could within the short time allowed by the clock adequately teach
people who were not present [not eye-witnesses] the truth . . . - Definitely not, but they could
persuade them” (Theaetetus 201a–b). Compare Gorgias 455a: “surely he wouldn’t be able to
teach so many people in such a short time such important affairs”.
We have already seen evidence that Plato presents rhetoric as exploiting

how things seem and as a result producing doxa in its audience, in Chapter 6.
In the Phaedrus Plato says both that (a) the orator presents what seems
good, just, healthy, wise, with indifference to the question of whether or not
these things really are such as they seem, and that (b) the audience comes
away with doxa rather than epistêmê about such things (Phaedrus 260a–c).
In the Sophist, he makes the same two claims about an art that he elsewhere
describes as almost indistinguishable from rhetoric (Gorgias 465c), namely
sophistry (Sophist 234c–d). The sophist produces and exploits mere seem-
ings by contrast with realities: he is able to “produce a seeming [doxan
paraskeuazein]8 as of being the wisest of anyone about everything,” although
he is not (233b); he makes his words seem (dokein) true whether or not they
are (234c). Indeed, the sophist’s false appearances are sharply contrasted
with the things that are, or realities (ta onta):
[But when later the students] approach close to the things that are [tois te
ousi] and are forced by their experiences to grasp clearly the things that
are, they will change their former doxai, so that . . . the phantoms in
[the sophists’] speeches are totally overturned . . . (Sophist 234d–e).9
Thus the sophist produces seemings rather than beings. And the result in
the audience is doxa: this is explicit at 234d (the false appearances instill
doxai in the audience, which the realities will dispel), and strongly implied
throughout, since it is clear both that the sophist influences his audience’s
thoughts and that he does not give them epistêmê.
Thus the Basic Conception explains why rhetoric produces only doxa:
rhetoric’s whole power consists in presenting what seems by contrast with
what is, and attending to what seems yields doxa.
8 Here we could interpret ‘doxa’ in either the object or subject sense.

9 The realm of what seems is here sharply separated from the realm of Beings. This is even
clearer after the dialogue’s long discussion of Being and not-being: “The sophist runs off into
the darkness of that which is not, which he’s had practice dealing with [viz, because he deals in
false seemings] . . .but the philosopher always uses reasoning to stay near the Form of Being. He
isn’t at all easy to see because that area is so bright and the eyes of most people’s souls can’t bear
to look at what’s divine” (Sophist 254a–b, trans. White). The sophist operates in the realm of
what seems but is not; the realm of Being is the domain instead of the philosopher—that is, it is
the separate realm of Forms.
Restriction to perceptibles 161
5. Restriction to perceptibles
In the Two Worlds dialogues, as we have seen at length in Chapters 1 and 3,

Plato consistently correlates doxa with the ontologically inferior realm—the
realm which he contrasts with Being, and which he describes variously as
Becoming, as what is between Being and not-being, and as the perceptible.
Doxa is set over what is between Being and not-being (Republic 478e); it is
about (peri) becoming (Republic 534a), or becoming is its object (doxaston,
Timaeus 28a); doxa grasps the perceptible realm (aisthêton, Timaeus 37b,
52a; cf. the Republic’s casual equation of the doxaston, object of doxa, with
the horaton, visible (see 509d with 510a). What Is, meanwhile is grasped by
epistêmê, by sharp contrast with doxa (in all the passages just listed); indeed
Being is adoxaston (Phaedo 84a).
Moreover, as I argued in Chapter 2, this correlation is by no means con-
tingent: Plato’s like-by-like account of cognition renders it necessary. Doxa
is a messy form of cognition: unstable, unclear, imprecise.10 It is thereby
suited to grasp only messy things. On the metaphysics of the Two Worlds
dialogues, this restricts it to the ontologically inferior realm of perceptible
objects and their images, for these are messy, and Being clean.11 Doxa is
thus by its very nature restricted to the ontologically inferior realm.
A good account of doxa must explain these features. Why is it specially
associated with the ontologically inferior? Why is it specially associated
with perception? And why is it not merely associated with the inferior realm
but confined to it, unable to grasp Being?
These questions pose particularly sharp problems for the view that doxa
is belief, in the sense either of generic judgment or ordinary opinion. These
are in no way specially connected with perception and the perceptible, nor
with anything ontologically inferior. Indeed, they can be about anything.
This makes the Distinct Objects view particularly mysterious when it comes
to doxa. Epistêmê is plausibly restricted in its objects precisely because it is
such a demanding mental state; belief, by contrast, is cheap! If you have
even just only heard of quarks, or Persians, or the factors of 8,000, you can
have beliefs about them—ill-founded, to be sure, and perhaps mostly false,
10 Meno 98a, Protagoras 356d–e, Republic 484b, 511a, Philebus 57b–d, 59d.
11 Like the many beautiful things, all things that come to be and pass away are unstable,
unclear, changing, contradictory; what Is, meanwhile, is stable, clear, precise, always the same
in every way (Republic 479a–b, Phaedo 74b–c, 78d–e, Symposium 210e–211b, Philebus 59a–b,
Timaeus 52a).
but beliefs nonetheless. How then could Plato insist that doxa is about the
ontologically inferior realm alone; how could he deny doxa of Forms? I
think this explains the insistence, even among some who accept that
epistêmê is restricted to Forms, that doxa cannot be restricted to percep
tibles (Vogt 2012, Schwab 2016).12
On the Basic Conception of doxa as cognition of what seems, there is an
excellent rationale for the restriction. It is a tenet of Two Worlds metaphys-
ics, I will argue, that what seems is the ontologically inferior perceptible
realm of Becoming, and that alone. Becoming seems, and so we have doxa
of it; Being does not seem, and so we have no doxa of it.
These claims are in a way simple and unsurprising. I am arguing that
Plato construes the perceptible realm as a realm of mere appearance, and
the realm of Forms as hidden, non-apparent. Plato is widely taken to hold
this kind of view; to regard the perceptible realm as a “veil of appearance”
obscuring the truth, as Nietzsche puts it.13 I take the time to defend the view
carefully because although it is often attributed to Plato it is rarely explicit in
his texts (I can find only two outright claims that perceptible realm is a
seeming or appearance: it is “the region that appears [phainomenên]
through sight” at Republic 517b; it is a phantasma, deceptive appearance, at
Timaeus 52c). Moreover, spelling out and defending the view will require
some precision. That is my project in the remainder of the chapter. In carry-
ing it out, I appeal to the theory of seeming that I argued is implicit in Plato
and his predecessors, defended in Chapter 6.2. On this theory, to say that x
seems means that (1) x strikes us vividly and manifestly as being certain
ways, but (2) this does not entail anything about how things really are. In
what follows I will argue that the realm of Becoming, as Plato characterizes
it in the Two Worlds dialogues, meets both criteria very well, while the
realm of Being fits neither.
5a. Becoming is not what Is
When Plato distinguishes a realm of Being from something inferior, the

inferior realm is a fortiori not the realm of Being. It is “between Being and
12 See the quotations in Chapter 1.1. from Annas 1981, 194, Fine 1990, 85, Vogt 2012, 51,
and Harte 2018, 142. For a more detailed argument that the interpretation of doxa as belief
motivates resistance to the Distinct Objects reading, see Moss 2020.
13 Daybreak, Section 474. Compare among others Cornford 1941 at 180: the perceptible
realm is a “world of mere appearances.”
not-being” (Republic 478e), it is not completely Being (teleôs de einai

on, Republic 597a), it becomes rather than Is (Republic 534a, Timaeus
27d–28a). Correspondingly (given the link between Being and truth which
we saw in Chapter 4.1), it is devoid of truth: the soul should “suppose noth-
ing true that it investigates through [the senses]” (Phaedo 83b), which is to
say that there is no truth in the perceptible realm; it is on the Forms by
contrast with perceptibles that “truth and what Is shine” (Republic 508d);
perceptible things “partake less in truth and being” than intelligibles
(Republic 585d). In every context in which Plato uses ‘Being’ or ‘truth’ to
mark the superiority of the one realm, he refuses those labels to the other.
The realm of Becoming—the ordinary, perceptible realm of the many
beautifuls, bigs, smalls, and so on—is not what Is; it is not the realm of
ultimate reality.
That Plato relegates the perceptible realm of Becoming to an ontologic
ally inferior status should be uncontroversial; there is however plenty of
controversy about how to interpret the view. This tracks the debates about
how to interpret Being reviewed in Chapter 3.3: in saying that perceptibles
fall short of Being, is Plato saying that they lack essences? That they have
their properties only in some deficient way, e.g. temporarily or impurely?
That they are copies rather than originals?14 Following my strategy in
Chapter 3, I will remain neutral between these interpretations, for again the
point I wish to emphasize is very general. I want simply to emphasize
that where Plato contrasts a superior intelligible realm with an inferior
perceptible one, he uses ‘Being’ to pick out the intelligible realm, and by
contrast relegates the perceptible realm to an ontologically inferior status.
The perceptible realm is not what Is—not what is ultimately real.
This gets expressed in the central metaphor of Plato’s Two Worlds
metaphysics: the entire perceptible realm of Becoming is a mere image of
the Forms. Not only the shadows on the wall but also the puppets that cast
them—that is, all perceptible things—are images of the things outside the
Cave (eidôla, 516a). Beautiful particulars are images of Beauty (Phaedrus
250d); they are likenesses of the Form as dream-images are likeness of their
originals (Rep 476c); the whole perceptible realm is related to the intelligible
realm “as the likeness is to what it is like” (Republic 510a); Being is the
14 For example: Plato means that beautiful perceptibles cannot be necessarily, but only con-
tingently beautiful Vlastos 1981), or cannot be essentially, but only accidentally, beautiful
(Nehamas 1975), or that they are in some sense are beautiful, but not in the primary sense, just
as an image of X is in some sense X but not in the primary sense in which X alone is X
(Allen 1960).
model (paradeigma, Timaeus 28a), Becoming the copy (eikôn, Timaeus 29b)
or imitation (mimêma, Timaeus 50c); even in an ideal city the characters
and customs are mere images of the Forms, as if painted on their model
(Republic 500e–501c). If the perceptible world is an image of the real world,
then the perceptible realm is not itself real: images are contrasted with
“what really is” (to ontôs on, Timaeus 52c, Sophist 240b); perceptible copies
of Forms are mere phantoms of them (phantasmata, Timaeus 52c).
Plato is indeed also interested in a distinction between original and
image within the perceptible realm—between for example a tree and its
shadow, or the genuine justice of a good person and the apparent justice of a
scoundrel with a good reputation. This is the distinction between the two
parts of the lower section of the Line, symbolized by the distinction between
puppets and their shadows in the Cave, and revisited in Republic X’s distinc-
tions between the carpenter’s bed and the painter’s, or the lawmaker’s virtue
and the poet’s (Republic 598a, 599d–600e).15 These distinctions are import
ant to Plato. Nonetheless, as the Line and Cave make particularly clear, he
regards even perceptible originals as, ultimately, mere images of something
higher. One section of Becoming is a more direct image of the Forms than
its shadows and copies, but it is an image nonetheless, not the reality. One
section of Becoming is closer to Being, but it is still not what Is.
One remark on this point: the famous summoners passage (Republic
523a–524d), which we will discuss below, is sometimes thought to show
that there is some ontological perfection in the perceptible realm, alongside
the imperfection—some being and truth. Here Plato argues that while per-
ception gives us confused reports about qualities like the Big and Small,
reporting on such matters deficiently (523e) it reports adequately that
something is a finger (523b).16 Does this passage say that perception gives
us the truth about fingers? Fine argues that it does—and therefore that we
have epistêmê of perceptibles (1990); others may infer at least that there is
truth available to perception, and thereby to doxa. But the passage does not
say that perception reveals the truth or being of a finger. Its point instead is
that perceiving a finger as a finger is unproblematic; such perception does
15 The carpenter makes a bed that is a copy of the Form, the painter a bed that is a copy of
that copy. Likewise the (good) lawmaker produces a character in the citizens that is a copy of
the Form of Virtue, while the poet’s characters exhibit a copy of that copy.
16 “Some things in perception do not summon the intellect to investigate, since they are
discerned adequately by the sense [ὡς ἱκανῶς ὑπὸ τῆς αἰσθήσεως κρινόμενα], while others in
every way call on intellect to investigate, since the sense produces nothing sound [ὑγιές]”
(523a–b).
not produce confusions that lead us to “summon reasoning and thought.”

Perception of a finger as big or small, by contrast, can lead us to do such
summoning; when this happens, “it first occurs to us to ask what at all the
Big is, and the Small” (524c). Hence the claim about perception is not that it
puts us in touch with the Being of fingers, but instead that it does not raise
questions about that Being. Perhaps Plato thinks that there is no Being of
fingers, no deep truth to be found by looking beyond perception: a finger is
just a bundle of seeming bigness, seeming smallness, seeming hardness, and
so on. Alternately, perhaps he thinks there is a deep truth about fingers, but
if so then this, just like the truth about beauty and bigness and the rest,
would reside in an imperceptible Form, not in a perceptible image. (If there
is a Form of Finger, it is not the kind of Form whose instances are particu-
larly contradictory, and therefore good at summoning thought, and that is
why he does not dwell on it here.) I do not know which of these views to
attribute to him. In either case, however, the passage in no way conflicts
with what we see so clearly throughout the Two Worlds dialogues: the per
ceptible world is devoid of ultimate Being and Truth.
5b. Becoming seems to be
Although the realm of Becoming is not what really Is, neither is it nothing
at all. It is something else: seeming.
We can find a semi-technical argument for this view in Plato’s texts. The
Republic’s powers argument identifies the object of doxa as what is between
Being and not-being, by contrast with what in every way is (478d–e). This
odd characterization is echoed in the Sophist’s contrast between “what really
is [to ontôs on],” and deceptive images which as “are in some way [pôs], but
not truly” (240b). An image is not what it purports to be, but it is some-
thing. What is it then? In various places Plato’s answer is clear: a seeming, an
appearance. The imitative artist makes “things that appear, but are not in
truth”17 (Republic 596e); a deceptive sophist is in the business of “appearing
and seeming but not being”18 (Sophist 236e). When we attend to an image—
such as the entire perceptible realm—we do not access what really is, but
neither are we in touch with nothing. Instead, we are in touch with what
17 φαινόμενα, οὐ μέντοι ὄντα γέ που τῇ ἀληθείᾳ.

18 τὸ γὰρ φαίνεσθαι τοῦτο καὶ τὸ δοκεῖν, εἶναι δὲ μή.
Republic V calls “what is between being and not-being,” and what these
other passages portray as what seems to be.
We can also find a more intuitive although less explicit argument that the
perceptible realm is a realm of seeming. It strikes us as being what there is,
all there is; without the active intellectual labor of questioning and resisting
and transcending how things seem, we are doomed to accept it as ultimate
reality. To use Plato’s analogies, it is like a dream-image that seems to be the
original (Republic 476c), or a shadow that seems to be an ordinary material
object (Republic 515a–c), or a cave that seems to be the whole world. We are
easily deceived into taking it for more than what it is. Indeed, that is our
default stance, one we must be educated into transcending, like prisoners
being led out of the Cave.
One might stop here to wonder just how to interpret this notion of
“seeming real.” Does the perceptible realm seem to be the very thing that is
in fact real, namely the Forms, the way that a dream-image of one’s mother
seems to be the very person of which it is an image, one’s mother? This
would give Plato a very implausible view of folk ontology, and indeed one
that contradicts his explicit claims. The sight-lovers do not acknowledge
that there is any such thing as the Form of Beauty (476b–c, 479a): in mis-
taking beautiful sights and sounds for “that thing itself which they resem-
ble” (476c), therefore, they cannot be thinking “These are the Form.” Instead
they must be thinking something like: this is all there is, this is the ultimate,
this is what there is to know about beauty. This interpretation gives us a
plausible understanding of the Cave prisoners’ world-view, too. It is not that
they appreciate the difference between shadows and reality while mis-
identifying the shadows as realities, as someone might be tricked into
taking a reflection for reality in a hall of mirrors. Instead, they simply think
“This is what there is.” (As strongly implied by Plato’s description: “Do you
suppose they would have seen anything other than the shadows . . . ? . . . .
Therefore in every way such people would consider [nomizoien] the truth to
be nothing other than the shadows of the puppets” (515a–c).) Or perhaps
they have no ontological thoughts at all, so that the shadows’ seeming real
to them is just a matter of their failing to notice that there are any questions
to be asked about their reality. Although the letter of Plato’s words at 515a–c
may suggest that the prisoners have a substantive ontological theory, one of
these quietist interpretations is more charitable.
In any case, we can ask why Becoming strikes us this way—why it is so
easy to take Becoming for reality, and so hard to recognize it as a mere
image. Consider the shadows in the Cave, which so thoroughly strike the
prisoners as real that even when they turn to see the puppets they think the
shadows “more true” (alêthestera, 515d). Plato does not say why the shadows
have this effect on the prisoners, but presumably it is because they are sim-
ply given: familiar, readily present, manifest to the prisoners without any
effort on their part. The shadows foist themselves on the prisoners’ atten-
tion, inviting their trust; in the absence of any motive for doubt, the prison-
ers take them at face value, not as derivative copies of something higher but
simply as what there is. Plato adds that the prisoners are fastened into place
by “chains of kinship with becoming,” attached to them by “feasts and that
sort of pleasures and luxuries” (519a–b). What keeps them from looking
away from the shadows—from doubting their reality and searching for
something better—is thus not only the inescapable presence of the shadows,
but also some gratification they provide.19
Likewise, the entire realm of Becoming foists itself on our attention and
invites our trust, because it is what we are given, and because it gratifies us.
In particular, it presents itself to us through perception. In the Republic the
realm of Becoming is described interchangeably as the object of doxa
(doxaston, 510a) and as the visible (horaton, 509d); in the Timaeus
Becoming “is the object of doxa [doxaston] through doxa with unreasoned
perception” (Timaeus 28a; cf. 52a); in the Phaedo, whenever the soul inves-
tigates “either through seeing or through hearing or through some other
sense . . . it is dragged by the body toward the things that are never the
same”—that is, toward Becoming (Phaedo 79c). The realm of Becoming is
what we access through our senses. And Plato clearly thinks that our default
human tendency is to accept as real what we perceive. All the many beauti-
ful things, big things, and so on are given to us through perception the way
that shadows are to the Cave prisoners: familiar, readily present, manifest to
us without effort on our part. Perceptibles lay claim to our attention and
invite our trust; in the absence of any motive for doubt, we take them at face
value as what there is.
19 This passage strongly suggests that the gratification is appetitive. Later it turns out that
there is thumoetic gratification too: the prisoners get honors and praises for skill at predicting
which shadow will appear next (Republic 516c–d). Plato does not explicitly say that this further
binds the prisoners to the Cave, but that is a plausible inference: one has added incentive not to
doubt the reality of a realm in which one is honored and admired.
This theme is particularly strong in the Phaedo’s characterizations of

philosophy:
The lovers of learning know that when philosophy gets hold of their soul,
it is imprisoned in and glued to the body . . . . and that the most terrible
thing about this prison is that it is due to appetites . . . Philosophy, taking
hold of the soul in this state, gently encourages it and attempts to free it,
showing that investigation through the eyes is full of deception, and inves-
tigation through the ears and the other senses too, and persuading it to
hold back from these, insofar as it is not necessary to use them . . . and to
trust [pisteuein] in nothing other than itself, because it can itself by itself
think [noêsêi] the things that by themselves Are; and to suppose
[hêgeisthai] nothing true that it investigates through other means, being
different in different situations. For that kind of thing is perceptible and
visible, while what [the soul] itself sees is intelligible and eternal . . .
(Phaedo 82d–83b)
It is part of the embodied human condition to think that what we perceive

is what is real; only philosophical investigation can break that trust.
Moreover, as with the Cave prisoners’ trust in shadows, this tendency is
exacerbated by our appetites and passions. We should abstain from bodily
pleasures and pains, the passage continues, because:
The soul of every person when it is strongly pleased or pained by some-

thing is forced also to think that the thing that causes the experience is
most clear and most true; but this is not so. And these are most of all vis
ible things . . . Every pleasure and pain just like a nail fastens the soul to the
body and rivets it and makes it corporeal, having doxa [doxazousan] that
whatever things the body says are true are true. For from this having doxa
in agreement with the body [homodoxein], and enjoying the same things,
it becomes like the body in manner and nurture. . . (Phaedo 83c–d)20
Thus the realm of Becoming strikes us as what there is, readily and mani-
festly, even though it is not. And on the theory of seeming I have defended
20 How do pleasure and pain make perceptibles seem more real? In the Cave allegory Plato’s
thought seems to be that the pleasure we take in shadows motivates us not to doubt their real-
ity. Here the thought is somewhat different: through pleasure and pain perceptibles affect us
strongly, manifestly, and vividly; they demand our attention.
(Chapter 6.2), this amounts to saying that, like a false friend or a false
reputation for justice, Becoming is something that seems rather than Is.
I have argued that Becoming seems real because it is readily present and
manifest to us, particularly through perception. The argument so far risks
simplification in two ways, however; addressing these is not only important
in clarifying what it means for Becoming to seem, but will also help us see
why Being does not seem.
5c. Seeming and Perception
First, the characterization of Becoming as perceptible must be interpreted

with some caution. If we assume too narrow a notion of perception, we will
exclude from the realm of Becoming many things that Plato clearly includes.
Certainly when he identifies Becoming as the visible realm (Republic 509d),
he has in mind more than the proper object of sight, color: he takes the
realm of Becoming to include ordinary material objects, things that come
into being and pass away and change (in Aristotle’s terminology, incidental
objects of perception as well as proper or common ones). More puzzling for
the modern reader, some of the things Plato puts in this realm we would not
count as visible or otherwise perceptible at all. Cave prisoners “contend
about the shadows of justice, or about the statues which cast them . . . never
having seen Justice itself ” (517d–e): since the Cave symbolizes the realm of
Becoming, we can conclude that this realm includes particular instances
of justice or injustice, along with images of them (such as the portrayals of
justice in poetry). Arguably we cannot literally see or hear the justice of a
particular action or character, but Plato counts these as paradigm examples
of Becoming.
How then should we understand the characterization of Becoming as
perceptible—and consequent on that, the relation between perception and
what seems, and consequent on that, the relation between perception and
doxa? I will not attempt to answer these questions in any detail here: a full
answer would require a detailed study of Plato’s theory of perception; more-
over I think that the Republic, by contrast with the later dialogues, leave
these questions seriously underdetermined.21 I will instead briefly sketch
two possible accounts, both consistent with my interpretation of seeming.
21 For some discussion of perception in the Republic see Burnyeat 1976 and Storey 2014.
The Timaeus gives us ample resources to define perception narrowly, and while it does not offer
First, perhaps Plato construes perception so broadly that he counts even

the justice of a just action as strictly perceptible. In this case, we may be
hard pressed to find any difference between what it is for something to be
perceptible and what it is for something to seem: perhaps we should under-
stand being perceptible in the ways I have characterized seeming so far—
being vivid, manifest, readily present—with narrowly sensory perceptibles
as paradigm cases. If so, then we will also be hard pressed to find any differ-
ence between the subjective correlates of perceptibles and of what seems,
namely perception and doxa. As in the theory Plato attributes to Protagoras
in the first part of the Theaetetus, they will be equivalent.22 This may in fact
be a plausible reading of the Republic, by contrast with later dialogues
including the Timaeus, which do very clearly distinguish perception from
doxa (see especially 77b). Perhaps in the Republic Plato thinks of perception
as a broad category equivalent to seeming, with sensory perception as only
one particular kind.
Alternately, perhaps Plato construes perception fairly narrowly, even in
the Republic, but nonetheless counts things like the justice of a just action as
part of the perceptible realm. Here is a possible rationale: he construes the
perceptible realm as the realm in which we are stuck so long as we trust
perception to put us in touch with the real.23 What I mean by this will
become clearer below when we consider the summoners passage (Republic
523a–524d), but in brief: on this reading, we should take the perceptible
realm to include all the strictly (i.e. sensorily) perceptible things, but also all
the other things that one can acknowledge and access without thinking that
there is anything unreal or derivative about those strictly perceptible things.
On this second reading, not everything that seems is literally perceptible,
but the perceptible realm coincides with the realm of seeming; therefore,
even though doxa is distinct from perception, it is confined to the percep
tible realm. Perhaps perception is one species of doxa, or perhaps doxa
much of a theory of doxa, it does show that it must be go beyond perception, since the appeti-
tive part has perception but not doxa (77b). In the Theaetetus, doxa and perception are expli
citly distinguished on the grounds that doxa but not perception is an activity of the soul
without the body, and has different objects (e.g. being and goodness, rather than proper per-
ceptibles like white and sweet) (184b–186c). In the Philebus, doxa arises from perception
(along with memory and certain “affections [pathêmata]” (39a). In the Sophist, one kind of
doxa arises from perception this is given a special name, phantasia (264a). (For discussion of
the relation between doxa and perception in the later dialogues see especially Frede 1999.)
22 See Chapter 10.1.

23 As Crombie puts it in discussing the Timaeus, doxa “is confined to the level (or levels) at
which sense-perception is decisive” (1962, vol. 2, 36).
operates “with perception” (Timaeus 28a), or perhaps doxa arises from

perception in combination with other factors, as at Philebus 39a.24
The Two Worlds dialogues may not offer a determinate or consistent
theory of the relation between doxa and perception. What they clearly do,
however, is to confine them both to the same realm, the ontologically infer
ior realm of Becoming.
5d. Seeming and reasoning
My argument that Becoming is something that seems because it is readily

present and manifest risks simplification in another way too: it may suggest
that all being-seemed-to is passive and unreflective. This is not Plato’s view:
he in fact categorizes certain reflective, critical attitudes as doxai, even in the
Two Worlds dialogues.25
This is most obvious in Republic X’s discussion of conflicting doxai in
cases of perceptual illusions (602e–603a). When a distant object looks small
but you calculate its true size, you find yourself with one doxa in accordance
with the appearances, another in accordance with calculation.26 This second
doxa does not take perceptual appearances at face value, but instead goes
beyond them. Although it is a doxa, it is not based on the unreflective
acceptance of how things seem.
I have argued elsewhere that this is not an isolated instance in the
Republic’s epistemology, but instead fits with a distinction established
earlier.27 In the Line and Cave passages Plato distinguishes between two
species of doxa, a complication we have so far mostly ignored: eikasia
24 In the Phaedo, Socrates says that in his youth he held a view on which doxa comes from
memory and perception (96b), but this view applies to epistêmê as well as doxa, and so is not a
serious candidate for Plato’s considered view in the Two Worlds’ dialogues.
25 It is particularly obvious that he does so in later dialogues, in passages which characterize
doxa as the conclusion of an internal dialogue: see Theaetetus 189e–190a, Sophist 263e–264a,
Philebus 38b–39a, and my discussion in Moss 2014. As I will argue in Chapter 10, however, this
may be a very different view of doxa from what we get in the Two Worlds dialogues.
26 Socrates contrasts “that in the soul which has doxa [δοξάζον] against the measurements”
(that is, in accordance with the appearances) with “that which does so in accordance with the
measurements” (Republic 603a).
27 See Moss 2014. Grönroos 2013 finds the distinction between being-based and appear-
ance-based beliefs in the Sophist’s distinction between regular doxa and phantasia (Sophist
263e–264b). He argues briefly that this distinction is vaguely anticipated in the Republic (both
in the Line passage and the optical illusion passage) as the distinction between “reasoned and
unreasoned belief ” (Grönroos 2013, 17); I am arguing that the distinction between appear-
ance-based and (lower-case ‘b’) being-based belief is there fully present.
(imaging, imagination, conjecture) is set over shadows, reflections, and

phantoms; pistis (trust, conviction, faith) is set over the ordinary physical
objects of which such things are images (509d–510a, 511e). We might at
first wonder what sense to make of this distinction: why should one’s cogni-
tive state change when one shifts one’s eyes from a shadow to the thing that
casts it? Most charitably construed, this is the same distinction explained in
different terms in Republic X’s optical illusion passage: the distinction
between beliefs based solely on perceptual appearances, and beliefs that
move beyond such appearances to consider how perceptible objects are in
themselves. Eikasia is utterly unreflective acceptance of appearances: Plato
characterizes it as set over shadows and images rather than material objects
themselves because it recognizes no distinction between objects “as they
appear” and “as they are.”28 Pistis is the product of some reflection and
reasoning (logismos), in which we recognize that distinction and try to get
to the truth about how things are; that is why Plato characterizes it as set
over perceptible objects themselves by contrast with their images. (Plausibly
Plato has in mind critical, reflective thoughts not only about the shapes and
sizes of perceptible objects, but also about which people and actions are vir-
tuous, which conventions just, and so on: for example, the thoughts we have
when we question poetic images of virtue, and reason out for ourselves
whom to admire.)
Thus much of what Plato calls doxa involves some reflection and reason
ing. Does this mean that it is a response not to what seems but to reality?
It does not. As both the Line and Cave images show, the contrast between
a material object and its image is an analogy for the contrast between the
whole perceptible realm of Becoming and the whole intelligible realm of
Being.29 Recognizing this second contrast makes one realize that the first is
insignificant. Consider the casual indifference to the distinction we see in
the Cave escapee’s description of the prisoners’ activity as dispute about “the
shadows of justice, or about the statues which cast them” (517d). From the
28 Plato uses these phrases in Republic X (598a) in arguing that the painter copies the
appearance of the bed rather than the material bed itself. The point easily extends to the illu-
sion case at 602c: just as the bed appears to be different shapes from different angles while itself
remaining the same, so would it appear to be smaller in the distance while itself remaining
the same.
29 Plato draws this analogy briefly at 510a: “the doxaston is to the gnôston as the likeness is
to thing to which it has been made like” (e.g. as the shadow is to the tree). The Cave allegory is
an extended expression of this idea: it uses the relation between real trees (outside the cave)
and copies of them (puppets and shadows in the caves) to symbolize the relation between intel-
ligible Forms and their perceptible participants.
perspective of someone who has seen the real thing, the difference between
the two levels of image is unimportant—they are both mere images.
The move from attending to the lowest level of images to attending to
their perceptible originals—the move from eikasia to pistis—is thus not a
move away from seemings to reality, although it can be a first step in that
direction. Someone who thinks that they have already reached reality at this
stage is deluded. They are still in a realm of seeming—that is, of things that
are readily present (although not quite so readily as the lowest level of
images), and of things that invite their trust. When someone calculates that
a distant object is bigger than it looks, even though she is questioning
appearances she is still looking for truth only in the perceptible realm. She
still takes the senses as authoritative: any doxa she has can in principle be
confirmed by them. (“This object would look large if I saw it close up; that
stick would look straight if I saw it out of the water.”) To move away from
trust in perception entirely—to move away from seemings—requires a
much more radical kind of questioning and calculation, a kind we see in the
Republic’s “summoners” passage, which I discuss in the next section. This
second kind will question not individual features of our perceptions and
their objects, but the whole enterprise of searching for truth in the percep
tible world.30 This is a difficult, radical step to take. Sextus and Descartes
demonstrate that one can go from noticing small inconsistencies in our per-
ceptions to doubting that the senses can ever be trusted to deliver the truth,
but most people never take that second step. That step is difficult, requiring
special methodical practices of doubting, precisely because it requires going
against our natural tendencies. The perceptible realm tends to present itself
to us as being what there is: even when we question some of its details, it
30 Compare two different ways in which someone might question her religion. There is a
radical way: she might find pervasive contradictions in her religion, and so come to reject its
authority altogether and look for truth elsewhere, thus leaving her religion behind. There is
also a more moderate kind of questioning and reflection, one that stays within the system: she
might notice occasional contradictions and tensions between what her priest says and what her
bible says, and so work to come up with some reconciliation. (Perhaps she realizes that the
priest’s sexism makes him predictably unreliable in certain situations, for instance, and learns
how to correct for that bias.) If she does this, we can say that she is in a limited way reflective
about her religion, but she still takes it as authoritative: any moral belief she has can in prin
ciple be confirmed by her religion’s views. (“The priest would approve of my becoming a mis-
sionary if I were male, and that shows that it’s the right choice.”) This is like realizing that sight
reports distant objects as small and submerged objects as bent, and so must be corrected, but
still taking the senses as ultimately authoritative. (“That object would look big if I were close to
it, and that shows that it really is big.”)
retains its grip on our trust. In other words, the whole perceptible realm,
not just the lowest level of shadows and images, seems real.
Thus Plato can allow a distinction between fully passive acceptance of
perceptual appearances, and reflective beliefs about perceptible objects as
they are rather than as they appear, while still holding that both are
responses to how things seem. To get beyond seemings to Being requires a
more radical move. In Section 5e, I show how Plato characterizes that move.
In the process, I show that our Basic Conception of doxa explains why there
is doxa only of Becoming—why there is no doxa of Being.
5e. Being does not seem
We saw in Chapter 6 that Plato takes seeming and being as contrast notions,
two sides of an antithesis. I want now to show that the Two Worlds dia-
logues inflate this distinction, so that just as what seems is not what Is, what
Is does not seem. In other words, following Parmenides, Plato draws a sharp
line between an ontologically inferior realm that is manifest to us through
perception, and an ontologically superior realm that is hidden, obscure,
accessible only through intellect. Being lies beyond the readily accessible
realm; it can be reached only through a radical questioning and criticizing
of everything that invites our trust. Given the above account of seeming,
this means that Being does not seem. Therefore, so long as we remain with
how things seem to us—that is, at the level of doxa—we cannot even be
thinking about what Is.
We see this view in Plato’s repeated insistence that Being cannot be
accessed through perception.31 Surely he means in part that Forms have no
color, sound, or other narrowly perceptible properties. He may however
have in mind a broader claim too, the counterpart of what we saw above
about Becoming: Forms are not obvious, given, manifest. It does not simply
seem to us that there is, for example, a Being of Justice, the same in all just
acts. That is instead something we discover only through difficult intellec-
tual labor—labor that begins with a rejection of the senses as guides to the
truth. So long as we take the senses as authoritative, and take the world in
31 For example, in the Republic’s characterization of Being as intelligible rather than visible
(509d), in the Timaeus’ claim that we grasp Being through intellect, while we grasp Becoming
through “doxa with perception” (28a), or in the Phaedo’s argument that we can come into
contact with the Forms only when we have left behind our senses along with our bodies
(65a–66e.)
which they are our guides as the ultimate reality—that is, as I just argued, so
long as we remain at the level of seemings—we will not have thoughts about
what underlies and explains this realm, which is to say that we will not have
thoughts about Being.
We can detect Plato making just this argument in a passage that, while
much-discussed, has been underappreciated for what it shows about the
contrast between doxa and epistêmê, and the transition between the two.
I have in mind the summoners passage of Republic VII.
Socrates is discussing the education that will “draw the soul away from
what Becomes toward what Is” (521d). Given that he has earlier identified
the former as the object of doxa (doxaston) and the latter as the object of
epistêmê (noêton), this should mean that the education lifts the soul out of
doxa toward epistêmê, and indeed that is how he will describe it at the end
of the discussion (533e–534a).32 Our account of doxa dictates that a move
away from doxa should be a move away from seeming. That is, in turning
the soul away from Becoming and towards Being, the education is turning it
away from things that seem (and hence are objects of doxa) toward things
that do not seem (and hence are not objects of doxa). The description of
how the soul-turning works illuminates what this could mean.
The education that effects the transition will begin by exploiting the
contradictory nature of certain perceptions. To illustrate, Plato gives the
example of sight reporting that the same finger is big (compared to a smaller
one) and small (compared to a bigger one), or touch reporting that the same
thing is hard and soft (compared to different things). When such
things happen,
The soul is puzzled about what at all the sense means by ‘the hard’ . . . The
messages to the soul are strange and in need of investigation. So it is likely
that it is in these kind of cases that the soul first attempts, summoning
reasoning and thought [logismon te kai noêsin], to investigate whether
each of the things announced [e.g. hard and soft, or big and small] is one
or two . . . and if each is one . . . then to get clarity about this, thought is
32 More precisely, in summing up the stages of ascent at the end of the discussion of educa-
tion, Plato describes it as an ascent from the lowest form of the doxa, eikasia, to the highest
form of epistêmê, noesis (533e–534a). He also describes the education as “a turning-around of
the soul from a kind of night-day to true day” (521c), which recalls the descriptions of people
with doxa as living in a dreaming state, and people with epistêmê as being awake (476c). (I take
this as clear license to treat the Line and Cave epistemologies as parallel, or near enough paral-
lel for our purposes: the lower half of each represents the perceptible realm of doxa, the upper
half of each the intelligible realm of epistêmê.)
compelled to observe the big and small not mixed together [as sight did]
but distinguished . . . And is it not from this [experience] that it first occurs
to us to ask what at all the Big is [ti oun pot’ esti], and the Small?
(Republic 524a–c, emphasis added)
If turning our attention from Becoming to Being means moving from doxa
toward epistêmê, then doxa must be the state that we are in before we notice
contradictions in our perceptions, and that we begin to leave behind when
we do notice these. Before the transition, we trust the reports of our senses,
not thinking to question them, and therefore not thinking to use any other
means of investigation. This is the condition illustrated by Cave prisoners
and sight-lovers, of going with how things seem: a perceptible object strikes
us as having a certain quality (a finger as big, a spectacle as beautiful, a
decree as just), and so we simply accept that it does. We go with how things
seem; we have doxa.
What the summoners passage shows is that so long as we remain at this
level, we are not even having thoughts about Being. For it is only when we
stop trusting the reports of the senses that we start even to ask questions
about Being: to ask “what at all the Big is.” It is not that at the earlier stage
we had false or inadequate views about what the Big is, it is that we had no
views at all: we had not even formulated the question. Our cognitions were
not “set over” the Being of bigness at all—they were about the particular
sizes of particular perceptible things. Thinking about Being thus requires
moving past what seems; therefore, doxa is never of Being.
Compare the condition of the sight-lovers, those exemplars of doxa who
think that what they can see and hear gives the final word about beauty and
so “do not acknowledge Beauty itself ” (Republic 476c)—do not acknow
ledge the Being of beauty. In virtue of this mistake, they are like dreamers,
unwittingly confining their attention to mere images. If a sight-lover begins
having thoughts about the Being of beauty—if she stops asking “Is this
spectacle more beautiful than that one?,” for instance, and instead starts
asking “what at all Beauty is”—she will ipso facto be waking from her dream.
She will be moving past the manifest, readily given, trust-inviting images
that seem beautiful, and directing her mind to the obscure, hidden, under
lying reality, what beauty Is. By definition then, since they are not about
what seems, such thoughts will not be doxai.
Perhaps the best way to bring out the idea is by way of a contrast between
the radical questioning of perception in the summoning cases, and the
more moderate questioning that can take one from eikasia to pistis. I argued
above that the latter kind of questioning fails to take us away from seemings:
when we ask whether a particular object is the size it appears to be from a
distance, we are still taking the perceptible world to be the whole world, and
taking perception to be our ultimate authority. (We are assuming that if we
were to see the object under better conditions, we could see how it really is.)
Notably, such questions make no mention of Being: we are not asking what
it is to be big or small, but only whether or not a particular object is big or
small. The questioning involved in the summoners passage, by contrast,
doubts whether the truth about size can be found at all in the perceptible
world (“Are bigness and smallness really always mixed up together as my
senses report?”), thereby leaving behind seemings; in the process, it brings
the mind to think about Being (“Is there instead one thing, the Big Itself, in
every way different from the Small? What is that thing?”). Thus this kind of
calculation, but not the first, counts as moving beyond how things seem,
and as turning one’s mind toward Being. These two steps go together: so
long as we remain at the level of seemings we are not thinking about Being.
Being does not seem.
I suspect this argument will be greeted with skepticism: surely Plato
thinks that Forms can seem. When Socrates refuses to relate what seems
(dokei) to him to be the case about the Good (Republic 506e), or when he
says that the Form of Equal never appears (phainetai) to be unequal (Phaedo
74c), he implies that the Forms seem to us in various ways. (I am not yet
addressing the closely related claim that there is textual evidence for doxa
of Forms: I turn to that in Section 6.)
Does Plato really say that Forms seem? We can attribute to him a consist-
ent view if we take his idea to be that Forms cannot seem directly, but only
through items ontologically distinct from themselves, namely perceptibles.
Plato very strongly implies such a view when he introduces the distinction
between Forms and perceptibles, in the lead-up to Republic V’s powers
argument:
The Beautiful and the Ugly . . . are each one . . . And about the Just and the
Unjust, and the Good and the Bad, and all the Forms, the same account
holds: each one is one, but through association with actions and bodies
and one another each one, appearing [phantazomena] everywhere, appears
[phainesthai] to be many.33 (Republic 475e–6a)
33 . . . αὐτὸ μὲν ἓν ἕκαστον εἶναι, τῇ δὲ τῶν πράξεων καὶ σωμάτων καὶ ἀλλήλων κοινωνίᾳ
πανταχοῦ φανταζόμενα πολλὰ φαίνεσθαι ἕκαστον.
Forms do in a broad sense seem or appear. But they do so only through their
images: for Beauty to appear, for example, is for some perceptible thing to
participate in it.34 Beauty itself is non-apparent: as we saw above, even to
direct one’s thoughts toward it requires the effort of questioning all that is
readily present and trust-inviting. To say then that Beauty seems to us to be
such-and-such is a loose way of saying that its images seem to us that way.
(Analogously, to say that your mother is right there on the television screen
is to say that an image of her is there. Mothers themselves cannot get onto
screens; they can only do so indirectly, through their images.)
Indeed, this appearing is so indirect that in other moods Plato says that
only the images appear, while the Form instead Is. We see this in the
Symposium: Beauty itself “will not appear [phantasthêsetai] like a hand or face,”
but is instead “itself by itself, with itself, always Being [on]” (Symposium
211a–b). We also see this in a passage on Forms and appearance in the
Phaedrus. Plato here uses visual language to describe our noetic apprehension
of the Forms during our disembodied existence: the Forms are apparitions
(phasmata, from the same root as phainesthai and phantazesthai) which we
are able to see (epopteuontes) in pure light (Phaedrus 250c); they are a blessed
sight and vision (opsin, thean, 250b). These are of course mere metaphors: it
is a crucial part of Plato’s argument here, as in the Phaedo (75b–c) that
when disembodied we have no senses, and that this is a major epistemic
benefit, for now we can grasp the Forms, through intellect. Thus we should
take the appearance language as metaphorical too: it is not the kind of
seeming that perceptibles do, but instead something analogous to that but
far better, a purely intelligible way of being manifest. Now consider what
Plato says about the Form of Beauty in particular: in our embodied condi-
tion Beauty still “glitters most vividly through our most vivid sense,” i.e.
sight (250d). Does that mean that the Form is itself apparent—a seeming
Form after all? No: the claim is that it shines only through its visible images
or likeness (eikônes, homoiôma, eidôla, 250a–d):
34 What about “through association with one another?” Many interpreters have proposed
excising or amending the text here, or interpreting the phrase as a reference to the co-instanti-
ation of Forms within perceptibles (for citations and discussion see Adam 1902, vol. 2, 363–4).
Others, including Adam (op. cit.) take it instead as a reference to the kind of “association of
Forms” Plato will address in the Sophist (251a ff): each Form shares in Being, Difference,
Sameness, and so on, and so appears to be many (it is, it is different, it is the same, and so on).
This would be surprising: nowhere else in the Republic does Plato acknowledge participation
of Forms in one another, and the Sophist’s countenancing of such associations is usually taken
as a revision of the Republic’s theory.
Wisdom [phronêsis] would produce terrible love, if such a clear image of it

coming through sight were available . . . but to Beauty alone is this allotted.
(Phaedrus 250d)
It is the images of Beauty that glitter so vividly, not Beauty itself. Forms are
metaphorically apparent to us in our unembodied conidition, but in our
embodied condition they do not directly appear to us; all the appearing is
done by their images.
One passage in the Phaedo is sometimes thought to show that Forms do
literally appear to us in our embodied condition: sticks and stones, while
being (onta) the same, appear (phainetai) equal to some and unequal to
others (74b), while the Equals themselves (the Form) never appeared

(ephanê) unequal (74c). But Plato does not here make the positive claim
that Forms appear; he instead makes only the negative claim, that they have
never appeared deceptively. Sticks and stones can be one way and appear
another, precisely because they are things capable of appearing; the Form,
he strongly suggests, is not.
What about when Plato uses dokein to denote how something which is in
fact a Form seems to non-philosophers—for example, how justice seems
(dokei) to Polemarchus (Republic 334b)35 or Thrasymachus (Republic
349a)?36 We should take the thought behind such claims to be the same:
Justice seems to them only through the various particular, perceptible
images of justice. When they say “Justice is helping one’s friends and harm-
ing one’s enemies,” or “Justice is the advantage of the stronger,” their use of
the abstract noun does not show that they recognize a single Being, Justice
itself, behind the many phenomena. Instead, they are talking about just par-
ticulars. Like the sight-lovers who make proclamations about Beauty, they
can make generalizations about particulars without rising above these to
have thoughts about Forms.
Alternately (but less plausibly), we might say that they are in fact budding
although confused philosophers. They really are trying to identify the Being
of Justice; like prisoners who have begun to ascend, they are thinking about
35 “This still seems to me, that justice benefits friends and harms enemies [τοῦτο μέντοι
ἔμοιγε δοκεῖ ἔτι, ὠφελεῖν μὲν τοὺς φίλους ἡ δικαιοσύνη . . .].” This certainly looks to be a seeming
about justice.
36 Socrates says that in discussing justice and injustice Thrasymachus has been stating “the
things that seem [to you] about the truth [τὰ δοκοῦντα περὶ τῆς ἀληθείας],” for example that
injustice is a virtue. Shortly afterwards he exhorts him not to answer questions about justice
“against your own doxa” (350e).
“what at all Justice is” (ti pote esti, 524c). In this case we should acknowledge
that their thought is directed toward the Form, but we should deny that
what they have is doxa. Below I will argue that Plato has the resources to
account for such thought, in his notion of a kind of cognition better than
doxa although worse than full epistêmê: dianoia.
To sum up this Section (5a–e): I have argued that the Basic Conception
of doxa makes sense of the Two Worlds dialogues’ restriction of doxa to the
ontologically inferior realm of Becoming. Becoming seems, and so we have
doxa of it; Being does not seem, and so we have no doxa of it. Thus there is
doxa only of perceptibles; there is no doxa of Forms.
Despite these arguments, however, and despite all the evidence we saw in
Chapters 1 and 2 that Plato carves up cognitive kinds by their objects, many
will protest that Plato simply must, and indeed plainly does, acknowledge
doxa of Forms. I turn to those objections now.
6. Objections: Doxa of Forms?
Some of the widespread insistence on doxa of Forms is motivated by the

assumption that doxa is ordinary belief. It is clearly possible to have beliefs—
in the sense of opinions or of generic judgments—about Forms: one might for
example believe that they do not exist, or that they are very hard to under-
stand. If doxa is belief, then, Plato should surely acknowledge doxa of Forms.
I hope to have undermined that motivation by showing that doxa is instead
something much more specific, a response to what seems. Nonetheless, the
spirit of the objection may still stand. Here is how it might be phrased:
Plato clearly recognizes thoughts about Forms that fall short of epistêmê – as
well he should, for how could we progress to epistêmê of something without
starting from some thoughts about it? Hence they must be doxa: what else
could they be? Therefore Plato must acknowledge doxa of Forms.37
Moreover, we have clear evidence that he does acknowledge such doxa:
Socrates outright avows doxa about the Form of the Good, at Republic
506c–e.38
37 Vogt argues that Plato regards doxai, construed as beliefs in the generic sense of taking-
to-be-true, as the starting points of inquiry, and thus must acknowledge doxai about Forms
(Vogt 2012, 10 and 68).
38 For arguments that this passage countenances doxa of Forms see among others Fine
1990, Taylor 2008, Vogt 2012. Fine 2016 also finds evidence in the Phaedo, which I dis-
cuss below.
Objections: Doxa of Forms? 181
If the objector were feeling charitable, she might even go on to suggest a

way that my proposed account of doxa could accommodate doxa of Forms:
You can maintain your view that doxa is of what seems, and even your
view that Forms do not directly seem, and indeed even many components
of a Distinct Objects interpretation of Plato’s epistemology. All you need
do is acknowledge that doxa of perceptible things is in an important sense
about the underlying Form. In other words, embrace a version of the
Content Overlap view (Chapter 1.1, 2.4). Distinguish what doxa is set over
(what seems) from what it is about, and you can address all my
objections.39
I will first address the alleged evidence that Plato is philosophically and
textually committed to doxa of Forms, arguing that it is far from decisive.
Plato does sometimes say things naturally interpreted as committing him to
doxa of Forms, but he has no need in his theory for such doxa, and we can
attribute to him an overall consistent theory on which it is not possible
(Sections 6a–b). This undermines the motivation for the Content Overlap
solution, but Content Overlap may seem independently plausible enough
that it is worth considering; I address it in Section 6c.
6a. Between doxa and epistêmê: dianoia
The main claim driving the objection is very compelling: Plato certainly
presents people who lack epistêmê as nonetheless having thoughts about
Forms. Whether or not we allow examples like Thrasymachus’ views of just
ice (Republic 349a), there are very clear cases throughout the Two Worlds
dialogues: Socrates and his interlocutors discuss the Forms extensively,
while disavowing epistêmê of them. Consider among many other instances
the agreement that each of the Forms is one rather than many, at Republic
476a or 507b, or—an example central to this debate, and to which we will
return below—Socrates’ thoughts about the Form of the Good (Republic
506a–e). These certainly seem to be thoughts about Forms. They are clearly
not supposed to be instances of epistêmê: Socrates explicitly denies that he
or his interlocutors have epistêmê of the Forms (see Republic 506c, 533a).
39 For extended defense of this view see Smith 2000, 2012, 2019; compare Szaif 2007. For
briefer remarks to this effect see Lafrance 1981 at 141–2, and Murphy 1951, 123–4.
Must we therefore accept that such thoughts are doxai, on the grounds that
there is nothing else that they could be?40
One response is to deny that these thoughts really are about Forms: they
are instead confused thoughts about images of Forms. Only when we have
fully, correctly grasped the Forms do we count as having thoughts about the
Forms themselves at all; someone who makes mere hypotheses about the
Good is no more thinking about the Form than someone who touches your
shadow is touching you. Some have taken this approach.41
I find this general line of argument compelling, but I will argue that there
is an important distinction to be drawn. Although Plato does indeed think
that doxai are too cognitively deficient to count as being about the Forms,
he wants to make room for imperfect cognition of Forms, of another kind.
In other words, I propose a different response to the objection that Plato
must acknowledge doxa of Forms: the objection rests on a false dichotomy.
Plato does in fact recognize a category between epistêmê and doxa. There is
something he describes as “between doxa and nous” (511d), or “clearer than
doxa, darker than epistêmê” (533d).42 This is something that differs sharply
from doxa in that that it counts as being set over (epi) intelligibles rather
than perceptibles (511d–e), but is somehow inferior to the highest epistemic
level, the full grasp of Forms. It is what Plato calls dianoia, “thought”: the
cognition exemplified by mathematicians, which uses perceptible things as
40 The analogy between the Good and the sun at Republic 506ff. “is undoubtedly supposed
to reveal something about the relation of the Good to the other Forms and it is not presented as
knowledge that Socrates has about the Form. The most natural thing, it seems to me, is to sup-
pose that it reveals something of the content of Socrates’ beliefs about the nature of the good”
(Baltzly 1997, 266); Baltzly takes it that such beliefs must be doxai (1997, 266). Compare
Vogt 2012, 53.
41 Gerson (2009, 41) argues that in doxa, by contrast with epistêmê, we are in touch with
propositions, while Forms can only be grasped in an unmediated, “non-representational” way.
Woolf (2013) reaches the conclusion from a very different route. Starting from Socrates’ argu-
ment in the Theaetetus that you cannot even have a doxa about something unless you already
know its distinguishing mark (209a–b), he attributes to Plato the view that you cannot have
something as an object of your thought unless you have a clear, accurate grasp of it. A putative
deficient grasp of a Form, then, is thus not really of the Form at all.
42 Plato refers to the very highest epistemic kind as nous in the Line passage, implying that
this is one species of epistêmê, along with the second-highest level, dianoia; when he returns to
sum up the epistemology described in both Line and Cave passages he reserves ‘epistêmê’ for
the very highest level, and uses “noêsis” for the whole upper half (533e, quoted below). Let us
follow Socrates’ advice not to worry too much about terminology (533d–e), and simply use
‘full grasp of Forms’ for the very highest level. With that in mind, we can see that all his classi-
fications agree in their treatment of dianoia: it is between doxa and the full grasp of Forms we
find at the highest level, whatever its name.
images of intelligibles, and relies on hypotheses (510b–e).43 Dianoia has

been largely neglected in the Two Worlds debate, although not entirely. It
plays a major role in the Line and Cave epistemologies, however, and is,
I shall argue, well suited to play precisely the role that my objectors’ argu-
ments attribute to doxa.
In making this argument I am following the lead not only of a few recent
scholars (see Gallop 1965, Gonzalez 1996, 273, note 50, and Gerson 2003 at
182–9), but also the Neoplatonists. These philosophers elevated dianoia to a
major role on which it fills several gaps left between doxa and epistêmê. In
particular, they assigned it the task we are concerned with here: accounting
for thoughts which are directed toward Forms without yet fully grasping
them. For example, Olympiodorus, commenting on Phaedo 65e, says that
the person who comes close but not all the way to knowing (eidenai) the
truth is “using the activity of dianoia” (dianoêtikê), while the person who
achieves the truth is using the activity of noêsis (noêra energeia) (In
Phaedonem, 5.4); see the next section for further citations.
To be clear, my aim is not to insist that Plato intended dianoia to play this
role, nor that he had a developed account of how it could do so. I am not
confident that he noticed the problem of how to account for beliefs about
Forms that fall short of knowledge; neither am I confident that he conceived
of dianoia as having precisely the right features to be able to solve the prob-
lem. Nonetheless, the fact that he included dianoia in his epistemology
shows at the least that he has the resources to solve the problem without
appealing to doxa. For dianoia is in the right genus: it is, I shall show, some-
thing superior to doxa, and about Being, but inferior to a full grasp of Forms.
I will start with a broad-strokes account that will be fairly uncontrover-
sial, and that will show dianoia to be in the right general category to
account for deficient thoughts about Forms. Then I will outline, without
decisively endorsing, a more detailed account of dianoia on which it plays
precisely this role, while also conforming to the constraints of objects-based
epistemology.
Republic V’s simple distinction between doxa and epistêmê gets refined in
the Line image: now there are two cognitive kinds over (epi) different
43 Confusingly enough Plato also occasionally uses “dianoia” to refer to something very
general, something like “cognition”: we call the dianoia of the philosophers gnômê (i.e.
epistêmê) and that of the sight-lovers doxa (Republic 476d); imitative poetry distorts the hear-
er’s dianoia, unless he knows what sort of thing it really is (Republic 595b). I will bracket this
use of “dianoia” in what follows, focusing on what is clearly a narrower, technical sense intro-
duced in the Line passage.
sections of the visible realm and two cognitive kinds over the intelligible
realm (noêton) (511d–e).44 Nous is the highest kind, in which we fully grasp
and can give an account of all the Forms rooted in an account of the Form
of the Good (511d). Dianoia is the penultimate level, exemplified by geom-
eters and other mathematicians. It is like nous in being set over (epi) intelli-
gibles (511d; cf. 534a).45 It is like doxa in that it makes use of perceptibles,
but it differs from doxa in how it uses them—namely, not as objects worthy
of attention in their own right, but instead as images of intelligibles:
[While geometers] use visible shapes, and make their statements about
these, they are not thinking about these, but about those things which
these are like, making their statements for the sake of [heneka] the Square
itself and the Diagonal itself, but not the one which they draw . . . seeking
to see those things which one cannot see except with thought.
(Republic 510d–511a; cf. 510b)
Sight-lovers do not acknowledge (nomizein) the Beautiful Itself, but con-

sider perceptible beautifuls the last word (476c); geometers by contrast are
aware that there is such a thing as a purely intelligible Square itself, and they
use perceptible squares as mere images of it. Plato illustrates this clearly
when he returns to dianoia in the context of Book VII’s discussion of the
education that brings one out of the Cave. In a passage we saw above
(Chapter 4.4), he shows that geometers treat diagrams not as an object of
study in which truth can be found, but instead as a tool for studying the
imperceptible truth; likewise, astronomers recognize that the motions of
visible stars “fall far short of the true ones . . . which can be grasped by logos
and thought [dianoia] but not by sight” (Republic 529dc–d).
Thus people with dianoia, unlike people with doxa, recognize the exist-
ence and importance of intelligibles, and as a corollary recognize the
44 Over the visible realm are eikasia and pistis, which in recapping the divisions Plato
c haracterizes as two forms of doxa (533e–534a). Over the intelligible are dianoia and nous
(or noêsis) (511d–e). In recapping, he switches terminology, calling the last level epistêmê, and
saying that it and dianoia are two forms of noêsis (533e–534a). For simplicity’s sake I will stick
with his first designation and refer to the highest level as nous.
45 “Are you satisfied as before to call the first portion epistêmê, the second dianoia, the third
pistis and the fourth eikasia? And [to call] these last two together doxa, and the first two
together noêsis? And [to say that] doxa is about Becoming, noêsis about Being? And that as
Being is to Becoming, so noêsis is to doxa, and that as noêsis is to doxa so epistêmê is to pistis
and dianoia to eikasia? But as to the relation in what they are set over [ἐφ’ οἷς], and the division
into two of each of these, the object of doxa [doxaston] and the intelligible [noêton], let us let that
go, Glaucon, lest it involve us in discussions many times longer than the one that came
before . . . ” (Republic 533e–534a).
limitations of perceptibles. In the language of the summoners passage, they

have got past a focus on the shapes of particular squares, and can ask the
question “what at all the square Is.” That is why Plato counts dianoia but not
doxa as being over (epi) intelligibles (511d–e).
Although dianoia is cognition of intelligibles, however, its grasp of them
is deficient. In dianoia:
the soul is forced to investigate from hypotheses . . . In [nous] however, it

goes away from hypothesis up to an unhypothetical first principle [viz.,
the Good] . . . proceeding by and through the Forms themselves . . . Those
who study geometry and calculation and such things hypothesize the odd
and the even and the shapes . . . .making these things hypotheses [poiêsame-
noi hupotheseis auta], not deeming it worthy to give an account of these to
themselves or to others. (Republic 510b–c)
Plato explains further in his discussion of the highest stage, now called dia-
lectic, in Book VII:
No one will [say] . . . that there is any other inquiry that attempts to grasp
about each thing what it Is. . . As to the others, which we say grasp Being to
some extent [tou ontos ti epilambanesthai], geometry and those [studies]
that follow it, we see that while they dream about what Is [peri to on], it is
impossible for them to see it waking, so long as they use hypotheses that
they leave untouched, not being able to give an account of them.
(Republic 533b–c)
Someone with dianoia is aware of intelligibles, can discuss them, and use
them to deduce true claims—for example, mathematical theorems. Indeed,
she is thinking about them, albeit only in a dream-like way. Nonetheless,
she does not fully grasp them. She is like someone who, although having
escaped the Cave, looks at shadows of the things outside rather than at the
things that cast them (532b–c). This is because she cannot “give an account
[logos]” of them—cannot define them, using a grasp of the Form of the
Good to explain their Being, as dialecticians do. That is why she does not
count as being at the top level:
And do you call a dialectician the person who grasps the account of the
Being of each thing? And the person who is not able to do this, insofar as
she is not able to give an account to herself or to another, to that extent
you will deny that she has nous about this thing? (Republic 534b)
We can thus sum up dianoia’s relation to intelligibles as follows. Someone

with dianoia:
(a) acknowledges the existence of intelligibles, and tries to think about

them, but
(b) relies on the use of perceptible images as illustrations of intelligibles,
rather than grasping them directly, and therefore
(c) only hypothesizes the existence and nature of intelligibles, rather
than being able to give accounts of them.
Now I will show how closely all this parallels Socrates’ deficient thoughts
about Forms. As geometers stand to the Square, so Socrates and his inter-
locutors stand to the Forms they discuss, on all three counts.
We see (a) and (c) explicitly in the Phaedo, where Socrates refers to his
views about Forms as hypotheses (100a–b). As Smith 2019 emphasizes, we
see (a) and (b) extensively in the Republic’s use of the Sun, Line and Cave
images to illustrate views about Forms, for the sun, a line, and the various
objects in the Cave allegory are all perceptibles. Indeed, here Socrates even
explicitly emphasizes that he is relying on images, and doing so because he
does not have a full grasp of the Forms. When Glaucon asks what goes on at
the highest epistemic level, Socrates replies,
You won’t be able to follow me further . . . for you would no longer be see-

ing an image [eikona] of what we are speaking of but the truth itself.
(Republic 533a)
He goes on to argue that only dialectic can give one a grasp of the Being of
things (533a–c). The clear implication is that their discussion must rely on
images, and that it must do so precisely because they do not yet grasp the
Forms. In other words, they are at the level of dianoia.
Note how well this account fits with what Socrates says in the very pas-
sage where he is often taken to avow doxa of Forms. (Here I am in agree-
ment with Smith 2019). When asked to state his own dogmata (506b—from
dogma, a stronger variant of doxa) about the Good, Socrates refuses, and
offers to provide an analogy instead:
Haven’t you observed that doxai without epistêmê are all shameful? The
best of them are blind – or do those have true doxa [alêthes ti doxazontes]
without nous seem to you any different from blind people walking the
right road? . . . . Let us let go of the what at all the Good itself is for now, for
to reach what now seems to me [tou dokountos emoi] is too much for the
present attempt. But I am willing to say what appears to be the offspring of
the Good and most like it, if you would like. (Republic 506c–e)
Socrates goes on to argue that the Good is like the Sun: as the Sun is the
source of both Becoming and vision in the perceptible realm, so the Good is
the source of both Being and epistêmê in the intelligible realm. Commentators
often refer to these claims as expressing Socrates’ doxai about the Good (for
example, Fine 1990, Vogt 2012, 53). But the lines just quoted shows that this
is wrong: Socrates explicitly refuses to state any doxai about the Good. How
then should we characterize his thoughts? They are clearly thoughts about
the Good. They are not epistêmê, which he denies having, nor are they doxai
about the Good, which he refuses to offer. They must then be something in
between. This points us toward dianoia, and indeed they fit the description
very well. They (a) acknowledge the existence of a purely intelligible Good
itself. They (b) use a perceptible object as an image—the sun. And they (c)
express undemonstrated hypotheses about the Good: Socrates simply
asserts that it exists, and that it is the source of Being and epistêmê, without
offering an “account of the Being” of the Good, a definition.
Thus Socrates’ thoughts about the Form of the Good fit the characteriza-
tion of dianoia much better than they do the characterization of doxa. Plato
has in dianoia the resources he needs to explain deficient thoughts about
Forms without appealing to doxa—without contradicting all the evidence
we have seen that there is no doxa of Forms.46
To be clear, I have been arguing only that Plato’s notion of dianoia shows
that he left room in his epistemology for something between doxa and a full
grasp of Forms, something which is in the right general category to explain
deficient thoughts about Forms. I am not arguing that Plato clearly and
determinately conceived of dianoia as able to play this role. Indeed, there
are several worries about how it could do so, which I will address
briefly here.
First, as others have noted, there is an important difference in the math-
ematicians’ attitude toward their hypotheses and Socrates’ toward his.
Mathematicians are content with their hypotheses, putting them forth “as if
they were clear,” not even aspiring to give accounts of them (510c); Socrates
46 In the next section I address the putative textual evidence that he nonetheless does char-
acterize such thoughts as doxa.
by contrast recognizes and aspires toward a higher kind of cognition—that

is, dialectic.47 Arguably however this is compatible with Socrates’ thoughts
being dianoia: Plato may be showing that there is a good use of dianoia, as
well as the typical more limited one.
Second, the account I gave above glossed over important questions about
the identity of dianoia’s objects. Not only is there extensive controversy on
this point, but it is also particularly important given the larger project of
this book. I will not here attempt to address the various interpretations on
their own terms, nor to show that my own proposal fits the text better than
its rivals: those are enormous tasks, best left for another occasion. (For a
recent interpretation at odds with what I propose here, see Smith 2019.)
Instead I will sketch how one could fill out the details in order to construct
an interpretation of dianoia that conforms to the constraints of objects-
based epistemology, while also showing dianoia able to account for defi-
cient thoughts about Forms.48
All the evidence we saw in Chapters 1 and 2 that Plato embraces a Distinct
Objects epistemology suggests that dianoia must have its own objects, espe-
cially when coupled with Plato’s claim that dianoia is “clearer than doxa but
darker than epistêmê” (533d).49 If cognitive kinds inherit their properties
from their objects, and dianoia is clearer than doxa and darker than the
highest kind of cognition, it must have its own objects, which are ontologic
ally superior to perceptibles but inferior to Forms.50 This is supported by
Glaucon’s response in the Line passage to Socrates’ contrasts between the
operations of nous and dianoia:
You wish to distinguish that [part] of what Is and is intelligible [to tou ontos
te kai noêtou] that is studied [theôroumenon] by dialectic as clearer than
that which is studied by the so-called crafts [viz., geometry and other
examples of dianoia]. (Republic 511c)
47 On this point compare Irwin 1995, 279.

48 I am particularly grateful to Damien Storey for discussion of this issue.
49 Here, as noted above, ‘epistêmê’ names the highest of four levels of cognition, what the
Line passage called ‘nous.’
50 Unfortunately Plato’s description of the Line seems to contradict his explicit claim that
dianoia is clearer than doxa: the lengths of the segments correspond to their clarity, and—as
Plato does not say explicitly, but as his mathematical instructions about how to construct the
line clearly entail—the two middle segments are equal, i.e. dianoia’s segment is equal to that of
pistis, the superior kind of doxa. I here join many interpreters who conclude that Plato simply
did not embrace the obvious inference that dianoia and pistis are equal in clarity (see for
example Smith 1996).
The claim is that there is one section of the intelligible realm that is studied
by dialectic, and another section studied by dianoia. This is also clearly
symbolized in the Cave allegory: at the penultimate stage, one attends to
shadows and reflections of the real objects in the outside world, by contrast
with the real objects that cast them.
This suggests that dianoia is set over images of Forms, but images that
are superior to perceptibles. Although dianoia makes use of perceptibles,
the images it is “over” are imperceptible—outside the Cave, part of the intel-
ligible realm. Let us then call dianoia’s objects “abstract images.” I propose
that Plato has in mind precisely the kind of thing we saw above as the direct
focus of dianoetic thought: hypothesized intelligibles.
This fits well with interpretations on which Socrates’ thoughts about
Forms are instances of dianoia. The idea is that dianoia’s objects include the
kind of images and analogies that Socrates employs: the city he constructs
in words to serve as a “perfect intelligible model” of the Form of Justice
(Sedley 2007, 267); this, the Sun image, and other logoi (discussions,
accounts) which are “images of Forms” (Gallop 1965, 120 and throughout).
What about the objects which Plato explicitly associates with dianoia: the
geometers’ hypothesized squares, angles, and so on? We can understand
these too as images of Forms. The Square itself is a Form, but since the
geometer cannot give an account of the Being of the square she does not
grasp this Form; instead her thought is set over an abstract image of it,
square-qua-hypothesis. (Thus this interpretation can, in an oblique way,
accommodate the interpretation going back to Aristotle on which dianoia’s
objects are “mathematicals”: purely intelligible objects, “different from per
ceptible things by their being eternal and unchanging, and different from
Forms by their being many of the same kind” (viz., many squares, many
triangles, and so on) (Metaphysics 987b14–18; see also 1028b19–21). On the
proposed version, mathematicals are hypothesized objects, images of math-
ematical Forms. (There are many hypothesized squares, all images of one
Form of the Square.) One might also develop the suggestion in a different
way: perhaps mathematicals are images of Forms because all the Forms are
essentially mathematical—unities, proportions, harmonies.)51
Thus I am arguing that to understand dianoia we need to distinguish
three levels of object: (i) the perceptible objects which dianoia uses as
images (the geometer’s diagrams, the astronomer’s stars, Socrates’ physical
51 For one version of this view of the Form of the Good and other Forms, see Sedley 2007.
sun or cave); (ii) the hypothesized intelligibles, which are abstract images of
Forms, which dianoia is over (the geometer’s hypothesized square, which is
not identical with but represented by the visible diagram; Socrates’ hypothe
sized Good-as-like-sun, which is represented by the physical sun); and (iii)
the Forms themselves, which dianoia fails to grasp, but “dreams about”
(Republic 533c, quoted above).52
A last comment on dianoia: the distinction between (ii) and (iii) seems to
conflict with stark Distinct Objects epistemology I have defended in this
book, for it opens a gap between what a cognitive kind is over and what it is
about. Is this a problem for my accounts of epistêmê and doxa?53 I think it is
not, for two reasons. If Socrates’ thoughts about the Good are really about
the Good, despite being “over” an abstract image of the Good, this shows
what is distinctive about dianoia. Socrates recognizes that he is working
with mere images: not only does he recognize that the sun is not the Good,
he also recognizes that his analogy between the sun and the Good is limited,
a mere hypothesis; he recognizes that he does not grasp the Good itself.
This ability to recognize an image as an image allows for a gap between the
different kinds of objects: between what cognition is over (ii), and what it is
about (iii), not to mention what it uses (i). Such a gap is precluded in doxa,
whose signature failure is the failure to recognize that its objects are only
images (see 6c below, and Chapter 8). It is also precluded in epistêmê, which
fully grasps its objects and has no need for attention to anything else.
6b. Textual evidence for doxa of Forms?
Readers may protest that this part of the Republic nonetheless provides
strong evidence for doxa of Forms. Even if the Sun analogy expresses
Socrates’ dianoia or something like it, in declining to state shameful, blind
doxai about the Form of the Good Socrates clearly implies that he has such
doxai. What are we to make of this?
We could render the passage thoroughly consistent with our theory if we
interpreted Socrates’ unspoken doxai as generalizations about perceptible
52 These correspond to the top three levels of the Line. Perhaps we should include the lowest
level in category (i) as well: although geometers and astronomers surely prefer to make use of
visible objects rather than copies of them, Socrates finds ample use for shadows and reflections
in the Line and Cave images.
53 For arguments that these are about precisely what they are over, see especially Chapter 2.1.
images of the Good—as I have argued we should understand the sight-lovers’

doxai about Beauty, or Thrasymachus’ doxai about Justice (Section 5e
above). This is however belied by Socrates’ reference to “what seems to me
about what at all the Good itself is”54 (506d–e, emphasis added). This “what
at all the x is” is the same language he will use in the summoners passage to
refer to Being, by contrast with its images. The strong implication is that he
has a doxa about the Being of the Good—the Form.
I conclude that in this passage Plato is loosening his terminology. This is
however nothing radical for him. The Republic is full of references to
epistêmê that do not fit the theory laid out in the central books: Plato refers
to various expertises about perceptibles, possessed by non-philosophers—
house building, flute playing and the like (428b–c, 438d, 601e–602a)—as
epistêmai. Notably these passages are rarely cited as evidence against the
Distinct Objects interpretation: evidently it is obvious even to committed
Overlappers that they are not relevant to the debate. Perhaps Plato is here
using ‘epistêmê’ in accordance with convention, as a synonym for ‘technê;’
perhaps he thinks it clear that what counts as epistêmê in this context would
not in the stricter contexts which consider the existence of Forms and of
philosophers.55 I suggest that we read 506d–e as employing similarly loose
terminology: Plato is relaxing his epistemology and going with convention
to use ‘doxa’ to refer to any cognition inferior to the best.
A different approach is offered by some Neoplatonist interpretations of
further apparent evidence for doxa of Forms, in the Phaedo.56 Here Socrates
uses ‘doxa’ and ‘doxazein’ (66b, 67b) to refer to the philosophers’ view that
they will attain the truth only after death, when they escape the prison of
bodily senses and contemplate “the things themselves” with their souls
alone (66d–e). This does indeed look to be a thought about Forms. It is cer-
tainly not the kind of thought available to a Cave prisoner, for it recognizes
that reality and epistemic success lie in a realm beyond the perceptible. It is
the kind of thought that, according to my arguments above, Plato should
characterize as dianoia: like Socrates discussing the Form of the Good, these
philosophers recognize the existence of Forms and are thinking “for the
sake of them” without yet fully grasping them. Neoplatonist commentaries
54 αὐτὸ μὲν τί ποτ’ ἐστὶ τἀγαθὸν . . . τοῦ γε δοκοῦντος ἐμοὶ.

55 For this contextualist interpretation, see Wedgwood 2018.
56 Fine notes this textual evidence in her 2016, along with the Neoplatonic responses, taking
them to support her Overlap view.
on the Phaedo agree with the spirit of this interpretation, while offering an
inventive explanation for Plato’s use here of ‘doxa’:
This is not the doxa receiving [its content] from below, but that which is
accomplished through dianoia. For there are two kinds of doxa.57
(Olympiodorus, In Platonis Phaedonem 5.13)
What is this ‘doxa’ of the philosophers?58 It is not from perception . . . nor
is it about perceptibles . . . but is like a dogma from dianoia and epistêmê.59
(Damascius, In Phaedonem, 1.103)
On this interpretation, Plato has a theory on which there are two kinds of
doxa: one deriving from perception and “about perceptibles,” another
derived from dianoia (and epistêmê) and about the intelligible realm. In
other words, Plato’s word ‘doxa’ is ambiguous between two referents. This
preserves the spirit of the Distinct Objects reading: the claim would be that
in all the passages where he seems to endorse Distinct Objects, or to associ-
ate doxa strongly with perception and Becoming, Plato has in mind the for-
mer kind; in passages like Republic 506c–e, he has in mind the latter instead.
Perhaps these scholars are reading back into the Phaedo the ambiguity in
‘doxa’ which we arguably find—as we will see in Chapter 10—in the
Theaetetus. There Plato at first uses ‘doxa’ to denote something almost
identical with perception, and then uses the same word apparently to
denote something more like generic judgement, which results from
dianoia (dianooumenê, Theaetetus 190a; cf. dianoia, Sophist 263e).60 At any
rate, their interpretative strategy lets stand a use of ‘doxa’ to denote a cogni-
tive kind which is restricted by its nature to the perceptible realm.
In sum, there are a small number of passages which imply that there are
doxai of Forms; given all the evidence that doxa is restricted to Becoming,
however, we should bracket these passages as loose, awkward, or (with the
Neoplatonists) ambiguous. With that done, we can attribute to Plato a con-
sistent theory of doxa, as follows. All cognitions lower than full grasp of the
Forms must make some use of perceptibles. If one accepts the perceptibles
57 δόξαν οὐ τὴν κάτωθεν δεχομένην, ἀλλὰ τὴν διανοίας οὖσαν ἀποτελεύτησιν· διττὴ γὰρ ἡ δόξα.
58 The commentary here is a series of questions about particular words from Plato’s text,
indicating that their sense is mysterious. For example, in the previous lines, Damascius has
asked “How is the body ‘bad’? . . . Or is it bad in relation to something . . . ” (1.102).
59 οἷον δόγμα τῆς διανοίας καὶ τῆς ἐπιστήμης.
60 Although in these passages ‘dianoia’ too has a more generic sense, rational thought,
rather than what is it at issue in the Divided Line.
as the last word, then one’s mind is set over perceptibles rather than
intelligibles, and one has doxa. If one instead uses the perceptibles as mere
images of hypothesized intelligibles then one’s mind is set over intelligibles
rather than over perceptibles, and so has dianoia, or some un-named
kindred condition. There is no doxa of Forms.
6c. Is doxa of perceptibles about Forms?
Now I turn to a final objection to the claim that there is no doxa of Forms.
Some have argued that if beauty (for example) really is a Form, then in hav-
ing the doxa that something is beautiful, or that beauty is good, one is ipso
facto having doxa about the real referent of one’s thoughts, the Form—even
if one is not aware of this.61 This suggestion is arguably bolstered by the idea
that perceptibles are images of Forms: one might argue that just as a dream
featuring a dream-image of one’s mother is a dream about the real thing,
one’s mother, so the sight-lovers’ doxa based on perceptible images of
Beauty is by Plato’s lights about the real thing, the Form.
Would Plato accept such a claim? I am sure that if you put it to him that
there are all kinds of relations between people’s thoughts and objects in the
world, some of which do not require that the thinker have a grasp of or even
be aware of the object—for example, if you were to explain the idea that one
could be very liberal in one’s de re belief attributions—he could be brought
to recognize that there is some sense in which thoughts about perceptibles
are about Forms, just as someone today could be brought to recognize that
there is a sense in which Plato’s thoughts about water were about H2O. What
I deny is that Plato had any interest in appealing to this idea in his discus-
sion of doxa. The relation between thoughts and objects that interests him is
the kind that involves the subject’s awareness of the object. Consider the
sight-lovers who are dreaming rather than awake, the Cave prisoners who
think the shadows are reality, or the poets, politicians, and craftsmen in the
Apology who think themselves wise despite not knowing what is really
important. What Plato wants to emphasize when he attributes doxa to such
people is that they are preoccupied with the wrong objects and unaware of
the right ones. Therefore he needs a terminology on which thoughts are
identified by the objects of which the thinker is aware—and thus on which
61 See especially Smith 2000, 2012, 2019; for a version of this idea, focusing on the reference
of language, see Harte 2007.
it is most natural to say that sight-lovers’ doxai are about mere appearances
and not about Beauty itself, even though the appearances are images of the
Form. Someone who fails to recognize an image as an image does not
merely contingently fail to be aware of the real thing, but is precluded from
being aware of it as such—and so it is very natural to say that her beliefs
about the copy of it are beliefs only about that.62 Thus we do not find out
anything important about Plato’s conception of doxa by noting that doxai
are arguably in some sense about Forms—indeed, no more than we would
learn something interesting about sight by noting that visible objects are
copies of Forms and thus that sight is (in some very strained sense)
of Forms.
Does Plato then show any signs of interest in the idea that thoughts
responding to images could in an important sense be about the originals?
Indeed he does—but, as I argued above, only in his discussion of dianoia.
Consider again a passage we saw on dianoia, this time in fuller context:
(i) All the other crafts are either directed toward human doxai and appe-
tites, or toward becomings and composition, or toward the care and nour-
ishing of things that grow and things that are composed; (ii) as to the
others, which we say grasp Being to some extent, geometry and those
[studies] that follow it, we see that while they dream about what Is [peri to
on], it is impossible for them to see it waking, so long as they use hypoth-
eses that they leave untouched, not being able to give an account of them.
(Republic 533b–c)
The arts described in (i) which concern themselves with things that come to
be and pass away—that is, in which we exercise doxa—do not at all “lay
hold in some way of Being” or “dream about Being,” even though the objects
of their attention are indeed images of and participants in Being. It is only
62 I am here largely in agreement with Gonzalez 1996: since perceptibles are “images or
imperfect instantiations of forms . . . in recognizing beautiful bodies as beautiful I can be said to
have some awareness of the form of beauty they imitate, but this is an awareness that fails to
distinguish the form from the imitations . . . It is therefore a dream-like awareness that con-
founds reality with what only resembles it” (Gonzalez 1996, 272). He goes on to say that doxa is
in a limited sense about the Form; my contention is that this limited sense is not one to which
Plato wanted to draw our attention. (Confusingly, in the next paragraph Gonzalez offers
Socrates’ thoughts about the Good as an illustration, but then explains in a footnote—rightly,
to my mind—that they are instead at the level of dianoia (ibid. 273, note 50)). Harte 2006 puts
forth a view of mistaking images for originals that is in many ways similar to what I have just
proposed, but on which one must in some sense cognize and refer to the originals in order to
count as making a mistake.
when we reach the level of dianoia, described in (ii)63—when we recognize

that Being exists, that it is distinct from perceptibles, and that perceptibles
are mere images of it—that our thoughts are in any noteworthy way
about Being.
In sum, the accomplishment of a cognitive power that is over (epi) images
can indeed be about the originals in a substantive way, but only when the
cognizer recognizes the images as images—and therefore, never in doxa. To
return to my claim at the beginning of this section, interpreters are right
that Plato recognizes deficient cognitions that are in an interesting sense
about Forms, but wrong to think that he must or does categorize such
thoughts as doxa.
Thus our account of doxa is borne out. Doxa is a response to what seems,
and to that alone. Forms do not directly seem; they seem only through their
perceptible images. Thus there is no doxa of Forms.
This leaves us with the question of why Plato chooses to present a theory
of a kind of cognition that is limited in this way to what seems. What sort of
thing is doxa, such that it should be limited to what seems—and why should
this kind of cognition be so salient to Plato as the main counterpart to
epistêmê? I turn to those questions in the next chapter.
63 The mention of geometry, and the talk of relying on hypotheses and not being able to give
an account, make it clear that Socrates is referring to what he called dianoia at 510c–511d; just
below he uses the term explicitly (533d).
8
What Is Doxa?
I have argued that Plato’s Basic Conception of doxa is cognition of what

seems. We do not get at what doxa essentially is by starting with familiar
notions like judgment or opinion, nor by starting with various features in
virtue of which it partly resembles them. Instead, all of doxa’s features are
grounded in the fact that its object is what seems, i.e. what is ontologically
inferior but apparent. When we add to this Plato’s metaphysical view of the
ontologically inferior as a messy image of Being, along with his like-by-like
doctrine of cognition, we get a complete explanation of his characterization
of doxa. Doxa is a cognitive grasp suited to what seems: thus it is deficient in
truth, it is unstable, unclear, imprecise, transmitted by shallow persuasion,
and restricted to the perceptible realm. Plato may have had independent
reason to attribute many of these features to doxa: perhaps he was preserving
contemporary endoxa. But these features are nonetheless explained by
the essential characterization of doxa. Essentially, doxa is cognition of
what seems.
This may however strike us a very odd thing to choose as a central cat
egory in one’s epistemology. It sounds like a technical notion, or an ad hoc
one, and certainly something marginal to our epistemic lives. Can we get
any intuitive grasp on this notion? Does it correspond to anything recog-
nized anywhere else in epistemology? Can we understand why Plato would
choose to devote so much attention to it?
In this chapter I will argue that, properly understood, the notion of
cognition of what seems is in fact central to Plato’s thought. Throughout
the dialogues Plato is particularly concerned to point out and warn
against two cognitive deficiencies: (1) the tendency to mistake images for
reality, and (2) the failure to look beyond particular phenomena to
underlying unifying explanations. Given his epistemology and metaphys-
ics, Plato holds that these two deficiencies go together, and thus he
is concerned with a particular way of thinking that exemplifies both.
This way of thinking, moreover, is naturally characterized as cognition of
what seems.
DOI: 10.1093/oso/9780198867401.003.0009
Doxa as dreaming 197
I will address the two cognitive deficiencies in turn, showing that Plato is
fixated on them, that he views them as cognition of what seems, and that he
explicitly treats them under the rubric of doxa; I will also explain how they
are connected, and will note where others have agreed on this interpretation
of doxa. In Chapter 9, I show why Plato makes doxa thus understood so
central to epistemology.
1. Doxa as dreaming
Consider how Plato introduces doxa in the powers argument of Republic

V. This text is central to our understanding of both doxa and epistêmê in the
Republic: it lays the groundwork for the distinction between them, and
thereby for the entirety of the Republic’s epistemological discussions. We
might thus expect him to introduce them by way of some attempt at defin
ition, but he does not. Nor does he begin by citing the characteristics he
associates with them in what follows—for example, the infallibility of
epistêmê and the fallibility of doxa (477e). Instead, he introduces his cogni-
tive kinds via exemplars: doxa is the way of thinking exemplified by sight-
lovers, epistêmê the way of thinking exemplified by philosophers. To
describe the difference between these two types of cognition, what he gives
us is a metaphor:
The person who acknowledges [nomizôn] beautiful things, but neither

acknowledges Beauty Itself nor is able to follow when someone tries to
guide him to the gnôsis of it—do you think that he lives dreaming or wak-
ing? Consider: isn’t dreaming this, if someone whether in sleep or while
awake thinks that the likeness is not a likeness but rather the thing itself to
which it is like [the original]? – I would certainly say such a person is dream-
ing. – What about the person who is opposite to these, who thinks that there
is some Beauty Itself….? – Very much awake. – Surely then we would cor-
rectly say that the thought [dianoian] of the one, since he knows, is knowledge
[gnômê], and that of the other is doxa, since he has doxa [doxazontos]?
(Republic 476c–d)
Doxa is like dreaming—and that is the most informative thing to say about
it. There is no further argument to establish that the sight-lovers have doxa:
they have that status because they are like dreamers who think they are
198 What Is Doxa?
awake, attending to mere images of Beauty while ignoring the real thing,
and mistaking the images for reality.
Clearly Plato takes this to be an important metaphor for doxa. He uses it
again in describing his next illustrations of doxa.1 Cave-prisoners are living
in a dream: a city ruled by people who have left the cave and seen the truth
is the only one “governed by people who are waking rather than dreaming”
(520c), for the move from the cave to the real world is “a turning-around of
the soul from a kind of night-day to true day [alêthinên]” (521c). He uses
the metaphor again in distinguishing ordinary people from philosophers:
someone who examines things in accordance with seeming [kata doxan]
rather than Being only has doxa, and thus “spends this present life dream-
ing” (534c).
This last passage reminds us very explicitly how the idea of doxa as
dreaming fits with the Basic Conception of doxa. What seems are images
and appearances, while reality is non-apparent (Chapter 7.5); thus confining
one’s attention to what seems while ignoring the underlying realities means
mistaking images for reality—that is, dreaming.
That is one handle we can get on the notion of doxa, then. Doxa is, like
delusive dreaming, the mistaking of an image for what is real. This may
seem to us a condition of only minor interest or epistemological relevance:
today none but a radical skeptic would think we spend much time in this
condition. Plato, however, clearly considers it a pervasive condition: the
Cave prisoners are “like us” (515a); the sight-lovers and their fellow dream-
ers are the targets and products of the culturally pervasive poetry, sophistry,
and rhetoric that present mere images instead of reality (for example
Republic 600e, 602b). In the book’s final chapter I will argue that Plato also
considers this mistaking of images for reality to be the most ethically signifi-
cant cognitive deficiency, and for this reason assigns it a central place in his
epistemology.
An important clarification is in order: does Plato think that all doxa is
delusive, or does he recognize an equivalent of lucid dreaming—focusing
on images while at the same time recognizing them as such?2 We saw some
reason to think that the philosopher-rulers have doxa about perceptibles,
1 Cave prisoners are confined to the realm that symbolizes the visible realm of becoming
(517b), earlier called the realm of doxa (doxaston, 510a); Plato explicitly labels the prisoners as
having doxa (doxazein) at 516d.
2 In dianoia one attends to perceptible images but one’s thought is turned toward their
originals; I am asking whether Plato recognizes a different possibility: attending to perceptibles
in their own right, while in the background recognizing that they are images of Forms.
Doxa as atheoretical thought 199
despite recognizing these as images of Forms (Chapter 4.5). Certainly in the

Timaeus the world-soul has doxa of perceptibles, despite also having nous of
intelligibles (Timaeus 37b–c). Moreover, the Republic’s sight-lovers, those
paradigmatic dreamers, are extreme cases: they are described as not merely
having doxa, but as loving it—they are philodoxoi (480a). Perhaps it is their
attachment to their doxa, rather than the mere condition of having doxa,
that makes them unable to recognize perceptibles as mere images.
Here then is a speculation: Plato thinks that attention to what seems
tends overwhelmingly to lead to acceptance of what seems as real (delusive
dreaming), but that awareness of the original can allow one to navigate the
dream-world while recognizing it as such. (The philosopher-ruler for
example could have a fictionalist attitude: “According to the fiction that
there is justice in the perceptible world, I judge that this course of action is
the just one for the city to take.”) If so, then delusive dreaming is paradig-
matic of doxa but not essential to it. Doxa is cognition of what seems; what
seems tends to, but need not always, convince us that it is real.
2. Doxa as atheoretical thought
There is a second cognitive deficiency that Plato emphasizes throughout the

dialogues, that he also characterizes as doxa, and that he also would natur
ally construe as cognition of what seems. I have in mind what we might call
atheoretical thought: confining one’s attention to particular phenomena,
rather than going beyond these to seek hidden, underlying, unifying explan
ations. This is a category we can easily recognize, even though it is not the
focus of attention in contemporary epistemology. (If for example you have
ever felt frustrated with students who fixate on the particular details of a
thought-experiment rather than attempting to draw general conclusions,
you are familiar with the phenomenon!)
Plato repeatedly draws attention to this way of thinking, treating those
who are stuck in it as epistemically inferior, and sometimes explicitly label
ing their thinking as doxa. We have already seen one notable instance of
this, in the summoners passage (Republic 523a–524d): one leaves the realm
of doxa behind only when one stop asking if particular perceptible fingers
are big or small, and instead asks “what at all the Big is.” Doxa is focused on
particular, observable properties of particular objects or events; it has no
truck with non-observable things like unifying essences.
200 What Is Doxa?
We can see just this same mindset in the Republic’s exemplars of doxa, the
sight-lovers, who attend to beautiful things but cannot be brought to
acknowledge the Beautiful itself (Republic 476c). They confine their
thoughts to particular, observable phenomena; presumably they are able to
make generalizations about these; they are however unable or unwilling
to think beyond these to underlying universal truths. As Stokes puts it, they
do not go in for “intellectual abstraction” (Stokes 1992, 106).
We find a characterization of the same mindset in a passage with notable
echoes of the ascent from the Cave, the Theaetetus’ digression about the dif-
ference between the philosopher and the lawyer:
It escapes [the philosopher’s] notice if he is human or some other beast; but

what at all a human is [ti de pot’ estin] and what befits such a nature . . . this
he seeks and investigates . . . Whenever he’s forced to converse about the
things at his feet and the things before his eyes he’s laughable . . . Whenever
however he drags up someone upward,3 and the other is willing to step out
from “What injustice did I do to you or you to me?” toward the investi-
gation [skepsis] of justice itself, and injustice, what each of them is and how
it differs from all other thing . . . the tables are turned
(Theaetetus 174b–175d)
The philosopher is concerned with things “upward,” while the lawyer is con-
cerned with “things at his feet and before his eyes,” i.e. the perceptible realm.
Although Plato uses no explicit epistemic vocabulary here, the Two Worlds
dialogues would say that the one who looks up to the intelligible realm has
epistêmê, while the one who stays down in the perceptible realm has doxa.
Moreover, just as in the summoners passage, the epistemically inferior per-
son confines his attention to questions like whether a particular thing has
some property F (is human, just), while the person who ascends to epistêmê
does so by asking “what at all the F is.” Plato is again contrasting the philos-
opher’s interest in abstract, universal truths with a kind of thinking stuck at
the level of perceptible particulars—doxa.
We also see Plato very prominently drawing attention to this same mind-
set throughout the Socratic dialogues, where Socrates repeatedly attempts
to turn his interlocutors away from a certain kind of wrong answer to the
question “What is virtue,” or piety, or courage, and toward one that would
3 ἑλκύσῃ ἄνω, a clear echo of the Cave allegory (Republic 515e).

Doxa as atheoretical thought 201
manifest epistêmê. The right kind of answer is an account of the essence, the
unifying, universal cause of particular phenomena being as they are; the
wrong kind of answer is, once again, fixated on the particular individual
phenomena themselves. See for example Socrates’ complaint to Euthyphro:
“I did not bid you to teach me one or two of the many pious things, but that
form [eidos] itself, by which all the pious things are pious” (Euthyphro 6d);
compare his complaint that Meno has offered him a “swarm” of virtues,
rather than the one thing that makes them all virtues (Meno 72a–c). The
deficient mindset is illustrated with particular clarity by Hippias’ refusal to
recognize any such thing as “the beautiful,” above and beyond particular
beautiful things, in the Hippias Major:
Socrates: Are not all beautiful things beautiful by [by virtue of] the beauti-
ful? – Hippias: Yes, by the beautiful. – Soc: This being something?…Tell
me then, what is this, the beautiful [ti esti touto to kalon]? – Hip: Does the
one who asks this want to find out anything but what is beautiful [ti esti
kalon]? – Soc: I do not think so, but what the beautiful is, Hippias. – . . . Hip:
These do not differ . . . I will answer what the beautiful is . . . A beautiful
maiden is beautiful….[This] is what seems so to everybody [i.e. what
everybody thinks: ho pasin dokei] (Hippias Major 287c–288a)4
Hippias claims to have epistêmê about beauty, just as Meno claims to have
epistêmê about virtue, and Euthyphro about piety. In each case Socrates
argues that his interlocutor’s inability to give the right kind of answer to the
“what is x” question—their inability to move past atheoretical concern with
particulars—shows that they lack epistêmê; the obvious implication is that
they have doxa instead.
A final example of Plato’s interest in this mindset: consider his contrast in
the Gorgias between technê (craft, expertise), and empeiria (experience).
Empeiria is an epistemically inferior condition, possessed by many who
purport to have genuine expertise; in this way the distinction is parallel to
that between doxa and epistêmê. Moreover, empeiria’s signature deficiency
is its inability to give an account of the nature and the cause of the phenom-
ena with which it deals.5 It thus shares with doxa a limitation to particular
4 Stokes 1992 also draws attention to the comparison between Hippias and the Republic’s
sight-lovers; see also Baltzly 1997. Even if the Hippias Major is apocryphal, the author clearly
means to portray a mindset which is the same one we find in many of Socrates’ interlocutors.
5 Flattery “isn’t craft [technê], but mere empeiria, because it has no account of the nature of
whatever things it applies [or to what] it applies them, so that it’s unable to state the cause
202 What Is Doxa?
perceptible phenomena, by contrast with hidden, universal causes. I suggest

then that Plato views empeiria as a special species of doxa, or a capacity
constituted by doxai; his denigration of empeiria in the Gorgias is yet
another instance of his concern with atheoretical thought.
I hope to have shown that that Plato is interested in characterizing and
warning against the dangers of a mindset that we might call atheoretical
thought, that he consistently contrasts it with epistêmê as an inferior
epistemic condition, and that he sometimes explicitly labels it as doxa. It
should be obvious from my arguments in Chapter 7 how this characterization
of doxa fits with the Basic Conception. The realm of particular, perceptible
phenomena is the realm of what seems, by contrast with the realm of
unifying, hidden, intelligible Being. Confining one’s attention to particular
phenomena thus amounts to cognizing only what seems.6
To put it more metaphorically, and make explicit the connection with the
other characterization of doxa I drew out above, doxa is atheoretical because
doxa is dreaming. Doxa mistakes images for realities, where most gobally
this means mistaking the realm of perceptible particulars for the entirety of
what there is—which amounts, on Plato’s ontology, to ignoring essences and
causes, i.e. to being atheoretical. It is no coincidence that Plato fixates on
and unites under one label these two cognitive errors—being stuck with
images and being stuck with particular phenomena—for on his view they
are two descriptions of the same failing: being stuck with what seems.
3. Counterparts and extant interpretations
The idea that doxa is cognition of what seems is, as I mentioned above,
prominent in Vogt 2012, and is a centerpiece of Cornford’s work on Plato.7
Although Cornford does not put it this way, he holds that doxa is something
like dreaming: focusing on appearances rather than reality. (See also
[aitia] of each thing” (Gorgias 465a; cf. 501a). Flattering knacks like rhetoric have other fail-
ings—they are concerned with pleasure rather than the good—but this is their distinctively
epistemic failing, and as 465a shows it is this failing that constitutes them as mere empeiria.
6 Again, if the philosopher-rulers do have doxa, we can imagine them switching in and out
of this mindset, although never fully inhabiting it since they are of course aware of the under
lying Forms.
7 The view is mentioned in his commentaries on the Republic, Timaeus, and Theaetetus, as
well as in his discussion of Parmenides (Cornford 1933, 1935, 1937, 1941).
Counterparts and extant interpretations 203
Crombie 1962.)8 Smith 2019 emphasizes the connection between doxa and
dreaming, in arguing (as I did in the previous chapter) that thoughts like
Socrates’ thoughts about the Good are something better than doxa, namely
dianoia. Harte (2006) takes the difference between dreamers and wakers—
the mistaking of images for reality—to be key to understanding the differ-
ence between those who lack epistêmê and those who have it.
What about the idea that doxa is atheoretical thought? Although the view
has fallen out of favor—and indeed, to judge by current scholarship, largely
even out of memory—there is in fact a long tradition of interpreting Plato
this way.9 Ficino finds in the Theaetetus the view that doxa is “particular
reasoning” by contrast with universal reasoning.10 Proclus, working with a
similar interpretation of the Timaeus, emphasizes doxa’s lack of access to
causes:11
The doxastic [power] comprehends the logoi of perceptibles, and is the

thing that cognizes the beings of these,12 and contemplates the that, while
being ignorant of the cause . . . For that is also how correct doxa differs
from epistêmê, in that it cognizes the that only, while the latter is also
capable of contemplating the cause.13
(Proclus in Timaeum 2.248 11–22, comment ad Timaeus 28a;
cf. 2.343.10–11)
8 “Knowledge . . . is the devouring of something real. The mathematician who knows all
about triangles . . . has devoured triangularity; triangularity has become part of his mental
equipment. The school-boy on the other hand . . . has devoured not triangularity but a likeness
of it only; and his state is one not of knowledge but of belief. The difference between the two
states is that in knowledge what has been devoured is an objectivity entity, whereas in belief
what has been devoured is no more than a likeness of such an entity” (Crombie 1962, vol. 2, 42).
9 Others have defended a closely related view: doxa is perception-based belief (see
Didaskalikos 4.5; Gerson 2009 Chapter 3). The atheoretical-thought interpretation is in essence
a more liberal version of the perception-based-belief interpretation: it allows as doxa thoughts
which arise only indirectly from perception, or which are difficult to trace to any sensory per-
ception at all, while still confining doxa to the realm where perception is authoritative, by con-
trast with the realm of intelligible Beings.
10 Opinio “judges all these things [features of perceptibles] in its own proper and particular
way, at certain times and places and circumstances . . . For this reason there is need of another
power . . . We say that this [latter] reasoning or cogitation is universal, while we call opinion
itself particular reasoning and cogitation [opinione ipsam particulare ratione cogitationemque]”
(Summary of Theaetetus, Divini Platonis Opera Omnia, 92).
11 In this discussion and throughout he also distinguishes both from dianoia, which like
doxa cognizes perceptibles, but unlike doxa grasps their causes as well as what they are.
12 As Lautner puts it, this claim means not that doxa grasps the “hidden essences” of per-
ceptibles, but rather that unlike perception it grasps “that all the proper perceptibles inhere in
one object,” such as the apple underlying the red and sweet (Lautner 2002, 258–9); see my brief
discussion of Proclus’ In Timaeum 2.249.13–27 in Chapter 1.3.
13 τὸ μὲν ὅτι θεωροῦν, τὴν δὲ αἰτίαν ἀγνοοῦν. . . τὸ ὅτι γινώσκει μόνον, ἐκείνης καὶ τὴν αἰτίαν
θεωρεῖν δυναμένης.
204 What Is Doxa?
On Proclus’ version, doxa sounds very like Aristotle’s empeiria (experience)

(Metaphysics I.1, Posterior Analytics II.19, Nicomachean Ethics VI.7–8).
Empeiria is perception-based cognition that operates on particular things,
registering their similarities enough to make generalizations about them
and to manipulate them for practical purposes, but without ever recogniz-
ing the underlying universals that explain why they are as they are. Aristotle
contrasts it with higher cognitive kinds—phronêsis, technê, nous—which
grasp universals and causes.14 This notion of empeiria goes on to play a
central role in the debate between Empiricist doctors, who forswear the
attempt to discover hidden causes of perceptible phenomena, and
Rationalist doctors who engage in just this inquiry (see Galen, On the Sects
for Beginners, Chapter 5, and M. Frede 1990). It also shows up in the closely
related debate between Skeptics, who forswear inquiry into non-evident,
hidden natures and causes in general, and their dogmatist opponents who
build theories about such things (see Sextus, Outlines of Pyrrhonism
2.97–129). If I am right, these philosophers are all discussing close descend-
ants of Plato’s doxa.
This view of Plato’s doxa is echoed by many interpreters in more recent
history. It is strongly implied by Bosanquet’s description of doxa as the level
at which “commonplace minds are content to rest, while those possessed by
the true spirit of science—the philosophers—insist on criticizing all judg-
ments till in every group of phenomena they have detected a single and
central principle” (1895, 213, in summarizing Republic 479a ff.). Nettleship
draws a similar contrast between the mind of one with doxa and “the philo-
sophic mind,” where the latter “constantly looks for principles or laws or
unities of which the manifold of our experience is the phenomenon” (1906,
195). Hackforth describes doxa as “the state of mind that cannot rise above
perceptibles” (1945, 127); Havelock gives a lengthy account of doxa as “the
non-abstract state of mind” (1963, 248).15 Two books published in the same
year in different languages argue at length that, in the contexts where Plato
associates doxa with the perceptible realm by contrast with the intelligible,
he has in mind empirical cognition or empirical knowledge (Sprute 1962
14 For example: “We believe that knowing [εἰδέναι] and being proficient belongs more to
craft than to experience….This is because craftsmen know the cause while the [merely] experi-
enced do not. For the experienced know the that but don’t know the why, while craftsmen
know the why and the cause” (Metaphysics I.1 981a24–30).
15 On Havelock’s view Plato has in mind specifically the mindset inculcated by the tradition
of oral poetry. I think this is too narrow, but he is right to see connections; we should take that
mindset as paradigmatic but not exhaustive of doxa.
Doxa and belief 205
and Gulley 1962). Even the paper that helped to spark the Overlap
movement, Gosling 1968, recognizes this as one strain in Plato’s account of
doxa.16 In recent years the view has largely disappeared, with occasional
resurgences;17 a new book on Plato’s epistemology proposes an account that
we can see as a version of this view, on which doxa is always of particular
concrete tokens, while epistêmê is of abstract types (Rowett 2018).
On some of these accounts this is Plato’s core or sole notion of doxa; on
others, his use of ‘doxa’ is ambiguous between several notions, including
something like generic belief, but this is the notion he has in mind in the
contexts that invite the Distinct Objects reading.
Indeed this interpretation goes very much hand in hand with the Distinct
Objects reading, and owes much of its motivation to that reading. The claim
that doxa is restricted to perceptibles, so problematic when doxa is con-
strued as judgment or opinion, becomes not only reasonable but almost
analytic when we interpret doxa in this way instead. There is no mystery in
saying “There is no empirical cognition of Forms,” or “Those who occupy
the non-abstract state of mind are not thinking about Forms”; indeed these
claims are obviously true.
The challenge to this interpretation of doxa is then primarily to show that
it is in fact well-grounded in Plato’s texts rather than cooked up simply to
make sense of the Distinct Objects passages. Many of those who have
offered versions of the interpretation have done this very well, and I hope to
have added to this support. The main contribution I wish to make is to add
that doxa has this character precisely because it is cognition of what seems.
4. Doxa and belief
Now that we have a developed account of Plato’s doxa we can return to a

question raised at the start of the book: how does it relate to our contempor
ary notion of belief—either in the sense of generic judgment, or in the sense
of belief that falls short of knowledge, what we might call opinion? I have
16 Sight-lovers are inadequate because “they try to answer questions about beauty and the
rest by empirical observation. So the δόξα in question is tied to empirical observation, and
there is a tendency throughout the Republic to echo Parmenides’ division whereby δόξα is the
best you can hope for if you concentrate on observation” (Gosling 1968, 127).
17 “In Book 5 of the Republic, Plato follows Parmenides in his characterization of opinion as
the mental condition of ordinary people who take what their senses tell them for knowledge”
(Schofield 2011, 54, on the sight-lovers).
206 What Is Doxa?
already given extensive arguments that it is quite different from either of

these, so much so that we go wrong in trying to use our views about belief
to guide our interpretation of doxa. Doxa is defined by its objects, while
judgment and opinion are topic-neutral. (As with my discussion of know
ledge and epistêmê, I want to urge that this distinction is orthogonal to the
question of whether doxa is a propositional attitude: if it is, the proposi-
tions in question can only be about a certain kind of object.) Hence doxa is
confined to perceptible things, while one can make judgments or form
opinions about anything at all.
Nonetheless, as with epistêmê and knowledge, one could arguably trace a
lineage from Plato’s notion of doxa to our modern notions of belief. I will
discuss the case of judgment in the next chapter: arguably in the Theaetetus
and other later dialogues, Plato metaphysically deflates the Basic Conception
of doxa as cognition of what seems so that what is left is the notion of how
we take things to be. The connection to opinion is stronger. If we start with
the notion of a cognitive response to what seems, and then interpret seem-
ing not as an ontological notion at all but instead as a purely epistemic
one—how things strike the subject, how the subject tentatively or shakily
takes things to be—then we get the idea of a cognitive condition whose hall-
mark is subjectivity, instability, and a deficient grasp on truth; in other
words, we get something very like the modern notion of belief that falls
short of knowledge.
5. Why doxa?
I hope to have shown in this chapter that cognition of what seems is neither
an obscure nor an ad hoc notion. It is instead a cognitive condition with
which Plato is centrally concerned in his epistemology; it is also something
perfectly intelligible to us, although far from the center of epistemology
today. The obvious question then arises: why should this particular cogni-
tive condition play such a large role in his epistemology—why should it
be the salient inferior condition which he chooses to focus upon, and to
contrast with epistêmê? Answering that—along with the corresponding
question about epistêmê understood as cognition of Being—is the task of
the final chapter. First, I will turn to consider how the account of Plato’s
epistemology developed here makes sense of his views outside of the Two
Worlds dialogues.
9
Epistemology in the Earlier Dialogues
Does Plato have an objects-based epistemology throughout the dialogues?

We should expect the answer to be no. An epistemology on which cognitive
kinds are distinguished, defined, and qualitatively determined by their
objects depends on a detailed, robust metaphysical view of the objects of
cognition—and so we should not expect to see this kind of epistemological
view in dialogues without a developed metaphysics. Plato’s particular version
of objects-based epistemology, I have argued, depends on a metaphysical
view which inflates the distinction between ontologically privileged Beings
and ontologically inferior seemings—and so we should not expect to see
this kind of epistemological view in dialogues without a full-blown Two
Worlds metaphysics.
Nonetheless, as I have already suggested at several points, the epistemo-
logical views we find fully developed in the Two Worlds dialogues are
anticipated in the earlier dialogues, and survive in some form in the later
ones.1 I want now to show in a bit more detail that this is so, and to examine
the repercussions for our understanding of these dialogues. I will not argue
that the Basic Conceptions are fully developed in these dialogues, but I will
argue that they are at work—that what the Two Worlds dialogues present is
best understood as a highly developed version of a view Plato worked with
all along.
A warning: my discussion in this chapter and the next will be brief and
sketchy. I ignore many dialogues altogether; even where I give sustained
treatments, I ignore many important details, and engage very little with the
extensive secondary literature. My aim here is simply to show, in broad
strokes, how to read the earlier and later dialogues through the lens of the
epistemology we find in the Two Worlds dialogues. I leave more thorough
1 I use these chronological terms for convenience and without commitment, but the argu-
ments of this chapter do fit naturally with a developmentalist account of Plato’s works on which
views nascent in the Socratic dialogues are developed in the Two Worlds dialogues, and then
questioned in dialogues like the Theaetetus.
DOI: 10.1093/oso/9780198867401.003.0010
208 Epistemology in the Earlier Dialogues
treatments—including a defense of the resulting readings against many

rivals—for other occasions.
1. Epistêmê in the earlier dialogues
In Chapter 3 we saw that the Two Worlds dialogues consistently privilege

certain items over the rest, mark them with various phrases involving the
word ‘being,’ and make these things the province of epistêmê. Although the
earlier dialogues show no developed doctrine of Forms, they do show strik-
ing instances of this same pattern. That is, here too Plato is working with the
view that epistêmê is of Being.
To begin with a very clear example, Socrates rejects Euthyphro’s defin
ition of the pious as the god-loved on the grounds that it reveals only an
affection (pathos) of the pious rather than its being (ousia) (Euthyphro 11a).
Thus here we have another contrast between something labeled ‘being’ and
something else:
(Contrast 6) The Being of something versus its affections.
Here feature x—the feature which earns something the title ‘being’—is
e vidently something like being-an-essence. The same seems to be the case
in a second contrast prevalent in the Socratic dialogues:
(Contrast 7) What Fness Is versus the many F things (Euthyphro 6d, Laches
191d–e, Meno 72a–d, 74d)
Here too feature x is something like being an essence, something that com-
bines universality and explanatory power: the being of F is that one thing,
the same in all F things, in virtue of which all F things are F.
Both of these contrasts clearly anticipate ones we see in the Two Worlds
dialogues between Forms and perceptibles: in particular, contrast 7 looks
almost identical to Contrast 3 (Phaedo 74d–76d, Republic 507b), and in
drawing it Plato identifies “what Fness is” with the form (idea, eidos:
Euthyphro 6d–e, Meno 72c). Nonetheless, there is no evidence that he is yet
working with the full-blown theory of Forms. Nor indeed does he seem to
be working with any full-blown hierarchical ontology of the kind we get in
the Two Worlds dialogues: nowhere in the Socratic dialogues do we get the
idea that non-essential features of Fness, or the many F things, are shadowy
Epistêmê in the earlier dialogues 209
images of the essence. There is however a crucial parallel to the contrasts we

see in the Two Worlds dialogues: certain things are somehow special, and
the way to mark this is with the label ‘being.’ Moreover, even if the Socratic
dialogues do not subscribe to a fully hierarchical ontology, they do strongly
suggest that these special things—roughly, essences—are in some way fun-
damental, and the other things somehow derivative: in asking “what at all
piety is” (hoti pot’ eiê), for example, Socrates is seeking that by which other
things are pious, i.e. the cause or explanation or ground of other things
being as they are (Euthyphro 6d–e, cf. Meno 72c).2
Thus the contrasts we have seen mark a substantive distinction within the
realm of the things that exist: certain things, labeled with ‘being,’ are privil
eged above the rest in a way that at least prefigures the Two Worlds’ distinc-
tion between ontologically superior and inferior levels. With some caution,
then, I will use the capitalized ‘Being’ here too: even if he does not yet have a
developed metaphysical theory of degrees of being, and even if he does not
yet downgrade the other items in his ontology, Plato is using ‘being’ to mark
out things which are somehow ontologically special.
Moreover, here too the Beings are the province of epistêmê. (Below I con-
sider the important question of whether or not epistêmê in these dialogues
is only of Being; for now I want to show that Being’s status as epistêmê’s
object is at the least a privileged one.) The definition Socrates demands
from Euthyphro is meant to vindicate Euthyphro’s claim to epistêmê about
the pious (Euthyphro 4e–5a): if Euthyphro has this epistêmê, he must be
aware of the Being of the pious. The assumption again is that epistêmê has
Being as its special object. We can discern this same idea in Socrates’ fre-
quent insistence throughout the definitional dialogues that someone with
epistêmê of Fness must be able to give a special kind of statement (logos)
about Fness: one that says “what at all it is” (hoti pot’ estin, Euthyphro 11a,
Meno 71a; ti pote on, Laches 191e), “the being, what at all it is” (peri ousias,
hoti pot’estin, Meno 72b), “being an F” (tôi melittas einai, Meno 72b), or “that
which is Fness” (ho tunchanei ousa aretê, Meno 72c–d) (emphases added).
When Socrates characterizes the account necessary for epistêmê of something
as an account of what it is, he is again correlating epistêmê with the Beings.
(One might object that the emphasis is on the ‘what’ rather than the ‘is’: the
contrast is between what something is and what it is like (Meno 71a–b). But
Socrates equates “what it is” with the ousia (Meno 72b; cf. Euthyphro 11a),
2 τὸ εἶδος ῷ πάντα τὰ ὅσια ὅσιά ἐστιν . . . μιᾷ ἰδεᾷ τά τε ἀνόσια ἀνόσια εἶναι καὶ τὰ ὅσια ὅσια
(Euthyphro 6d–e; ἕν γέ τι εἶδος . . . δι’ ὃ εἰσὶν ἀρεταί (Meno 72c).
and this encourages us to hear, in these mentions of “what something is,” an

emphasis on the ‘is’ as well.) Just as in the Two Worlds dialogues, then, to
have epistêmê about any subject one must grasp a special kind of item, dis-
tinguished from ordinary things and marked off with variants of the word
‘being.’
I want now to show briefly that two other important features which these
dialogues attribute to epistêmê could in principle be explained by epistêmê’s
association with Being. In Section 3, I will consider whether this is in fact
Plato’s intention: whether he is in these dialogues already thinking of
epistêmê’s relation to Being as the bedrock principle which explains its other
features, or instead simply as one important feature among others.
First, the connection with Being can explain the early dialogues’ defin
itional requirement. Socrates’ insistence that anyone with epistêmê about
something must be able to give a definition of that thing has often been
taken to show that Plato has a very strange view of knowledge: that he was
confused about knowledge,3 that he departed sharply from contemporary
usage by simply stipulating that epistêmê is or includes knowledge of defin
itions, or that he started with a very specific and demanding conception of
the abilities involved in epistêmê which entails that it must include know
ledge of definitions.4 If we assume that Plato is working with the Basic
Conception, we have a simpler explanation. The definitional requirement
falls neatly out of the principle that epistêmê is of Being when coupled with
a Socratic metaphysical view that the ontologically privileged things are
essences. To have epistêmê about piety, for example, involves being in cogni-
tive contact with the essence of piety. Very plausibly, Plato thinks that to be
in such contact is to be in the kind of cognitive condition that can be
expressed in a definition.5
3 Thus Geach: “We know heaps of things without being able to define the terms in which we
express our knowledge” (1966, 371).
4 “The Priority of Definition principle . . . is not part and parcel of a ‘style of mistaken think-
ing’ about knowledge and definition [as widely thought], but a specific case of a general
Platonic principle of epistemology: the principle that epistêmê requires a logos” (Prior
1998, 112).
5 Do we need a further substantive thesis about epistêmê to explain the claim that one must
be able to give the definition? More economically, we can get this out of Plato’s other views. He
seems to think that the ability to state one’s thoughts calls for no special explanation: in the
Charmides, after saying that Charmides must have some opinion (doxa) about what temper-
ance is, Socrates adds: “And since you speak Greek you can surely also state that which appears
to you?” (κἂν εἴποις δήπου αὐτὸ ὅτι σοι φαίνεται) (Charmides 159a). Perhaps he is implicitly
working with a view that will become explicit in the later dialogues: thoughts are silent logoi in
the soul, which can be expressed out loud as uttered logoi (see Sophist 263e–4a and Theaetetus
190a, discussed in the next chapter: these dialogues identify silent logos as doxa, but in contexts
Epistêmê in the earlier dialogues 211
Second, the connection with Being can explain the Meno’s claim that
epistêmê results from a “reasoning out of the cause [aitia]” (98a)—a claim
that has reasonably led some interpreters to assimilate epistêmê here to
understanding.6 I have already discussed the evidence for a similar inter-
pretation of the Two Worlds dialogues (see Chapter 4.2), but it is worth
making explicit how the connection between epistêmê and Being could
explain the Meno’s claim as well. What is the cause that the slave must grasp
in order to turn his true doxa about the square into epistêmê? Socrates does
not make the answer explicit, but as we just saw above he identifies as
essences as explanatory. Thus, very plausibly he has in mind the essences of
geometrical entities—the nature of line, triangle, square, and so on.7 He is
looking for explanations such as: squares can be doubled by using the diag-
onal as base because the essence of square is such-and-such. (If so, he is
anticipating the idea Aristotle will make explicit in his own discussion of
geometrical epistêmê: we come to have epistêmê of geometrical theorems
when we can demonstrate how they are explained by the first principles of
geometry, namely the essences of the basic shapes.)8 What is the cause that
an inquirer into virtue must grasp in order to turn their true doxai about
virtue into epistêmê? Again Socrates does not make it explicit, but very
plausibly he has in mind precisely what he has shown Meno to be ignorant
of: the essence of virtue, the answer to the “what is it” question. Virtue can
be taught (if indeed it is teachable) because the essence of virtue is know
ledge of the good (if indeed that is its essence).
In the Two Worlds dialogues, the explanatory requirement on epistêmê
falls out of the Basic Conception of epistêmê, together with a metaphysical
view about the nature of Being: that Beings are causes (see Chapter 4.2). We
have now seen that the same could be true of the Meno, for here too, to
grasp epistêmê-conferring explanations is to grasp Being.
One note before we leave the topic of explanation: the Meno’s view seems
to be that someone who grasps a Being, and uses it to explain an ontologically
where, I argue below, Plato is considering a broad view of doxa on which epistêmê is a species).
Thus having epistêmê about x entails having a logos about the being of x in one’s soul, one that one
can state in conversation. Alternatively, perhaps Plato is already working with an implicit version
of the like-by-like view of cognition we see in the middle dialogues. The Being of x is something
deep, clear, precise; thus the grasp of it—namely, epistêmê—must be similarly deep, clear and pre-
cise, and on the cognitive side those qualities manifest as the ability to state clearly what one grasps.
6 See especially Nehamas 1985 and Schwab 2015; I follow their interpretations closely here.
7 For defense of this interpretation see Gerson 2009, 30, and Schwab 2015.
8 See for example Posterior Analytics II.2–3; for discussion see among others Bronstein 2016.
derivative fact, thereby comes to have epistêmê of that ontologically deriva-

tive fact. With an explanatory account grounded in the essence of square,
one who starts off with true doxa about how to double the area of a square
can convert it into epistêmê (reading 85c–d with 98a). Presumably, although
Socrates does not say so explicitly, the same holds in the case of virtue: with
an explanatory account grounded in the nature of virtue, one who starts off
with true doxa about virtue being teachable, or beneficial, can convert that
into epistêmê. If this is right, then epistêmê is not restricted to Beings.9
Arguably Plato’s thought is that epistêmê requires a grasp of Being, but can
then be extended to other things as well. (Someone with epistêmê of virtue’s
Being can use this to get epistêmê of whether or not virtue is teachable;
someone with epistêmê of piety’s Being can use this to get epistêmê about
whether or not particular actions are pious.10) If this is what Plato has in
mind, it would constitute an important difference between his earlier and
later epistemology, for in the Two Worlds dialogues, I have argued at length,
epistêmê is exclusively of Being.
Does this show that the Meno rejects Distinct Objects? And does it entail
that in the earlier dialogues Being is not the defining object of epistêmê—
that Plato sees an important connection between epistêmê and Being, but is
not yet working with the Basic Conception on which epistêmê is to be
understood, essentially, as cognition of Being? If so, then the explanatory
accounts I have proposed above are misleading: although Plato could in
principle derive epistêmê’s features from its connection with Being even in
the earlier dialogues, that is not what he is doing.
I postpone these questions until we have discussed the earlier dialogues’
view of doxa. For now, we have seen that these dialogues work with the
principle that epistêmê is of Being, and that they characterize epistêmê in
ways consistent with the idea that here too this principle constitutes the
Basic Conception.
9 Perhaps Plato would be sympathetic to a view like Aristotle’s in the Posterior Analytics, on
which there is a difference between cognition of the essences and of the derivative phenomena:
one who has a specially excellent grasp of the first principles, what Aristotle calls nous, and
who can demonstrate how these principles explain the derivative phenomena, thereby comes
to have demonstrative epistêmê of the derivative phenomena. But Plato makes no such distinc-
tions here, and it seems reasonable to attribute to him the view that epistêmê of Beings can
yield epistêmê of the rest.
10 This is implied by Socrates’ question “How can I know what a thing is like when I don’t
know at all what it is?” (Meno 71a); certainly Socrates is widely interpreted as holding that
defi nitional epistêmê is prior to other epistêmê, rather than that other epistêmê is impossible.
Doxa in the earlier dialogues 213
2. Doxa in the earlier dialogues
The Socratic dialogues do not distinguish a realm of perceptible things that

seem from a realm of intelligible Beings, nor do they systematically associate
doxa with cognition of what seems. They do sometimes exploit the verbal
connection between the subject and object senses of ‘doxa’: in the Crito, for
example, the Laws tell Socrates that escape would confirm the jurors’ doxa
of him, since one who destroys the laws would seem (doxeien) to destroy
the youth as well (Crito 53b–c). But there is no emphasis on the inferiority
of what seems, and there is no widespread use of ‘doxa’ to characterize an
inferior cognitive condition, nor does any other term take its place.11 The
Meno firmly establishes doxa as the inferior counterpart of epistêmê, but
without any special emphasis on the connection with seeming.
Nonetheless, these dialogues certainly present an inferior cognitive
condition which has much in common with doxa as we have seen it in the
Two Worlds dialogues, sometimes using vocabulary related to seeming. For
example, lacking epistêmê and being influenced by how things appear
(phainetai) or seem (dokei) means being cognitively unstable, “wandering”
in how things seem to one (Hippias Minor 372d–e)12 or even in one’s doxa
(Alcibiades1, 117a–b).13
Furthermore, the concerns that motivate the Two Worlds’ dialogues
conception of doxa are evident here too. We saw some evidence in Chapter 8:
the early dialogues draw an enormous amount of attention to an inferior
cognitive condition in which people are focused on particular perceptible
phenomena, unable to think beyond these to hidden unifying essences and
causes. Plato suggests that this is the cognitive condition of ordinary people
by showing that Socrates’ interlocutors tend not even to understand his
definitional questions: they proffer lists of the many F things, or varieties of
Fness, rather than a unifying essence. (See for example Euthyphro 6d, and
11 For example, the Apology draws attention to (a) the contrast between those who falsely
seem (δοκεῖ) wise, to themselves and others, and those who really are (Apology 21c), and to (b)
a corresponding contrast between those who suppose they are wise and those who actually
know (οἴεταί τι εἰδέναι οὐκ εἰδώς, Apology 21d), but it does not use dokein or doxa to mark the
second contrast, nor does it reserve the word it does choose, oiesthai, for the condition of being
taken in by what seems: at 29b, Socrates supposes (οἴομαι) correctly that he does not know.
12 “Sometimes indeed the opposite of these things seems to me, and I wander about these
things. It’s clear that this is because of not knowing [ἐνίοτε μέντοι καὶ τοὐναντίον δοκεῖ μοι
τούτων και πλανῶμαι περὶ ταῦτα. δῆλον ὅτι διὰ τὸ μὴ εἰδέναι].” (Hippias Minor 372d–e).
13 “Whenever someone does not know [εἰδῇ], necessarily their soul wanders [πλανᾶσθαι]
about this thing . . . And does your doxa wander about these things too [πλανᾶται σου ἡ δόξα]
[viz., how to ascend to heaven]?” (Alcibiades I, 117b).
especially Hippias Major 287c–288a, quoted in Chapter 8.2.) And although

Plato does not emphasize the point in these dialogues, we can certainly
construe this condition as the condition of cognizing things only as they
seem: in this mindset people focus on what is obvious and apparent (the
particular perceptible phenomena), unaware of the underlying hidden deep
truths. Moreover, I will argue in the Conclusion, these dialogues also treat
ordinary people as dreamers, in one specific but all-important arena: they
mistake things that merely seem good for things that really are good. These
dialogues also emphasize another aspect of dreaming: many of Socrates’
interlocutors think they have epistêmê when they do not.
3. Distinct Objects and objects-based

epistemology in the earlier dialogues
We have seen important resemblances between the epistemology of the

earlier dialogues and that of the Two Worlds dialogues. But are the Basic
Conceptions of doxa and epistêmê already present in the earlier dialogues,
or only anticipated?
One reason to doubt that Plato is already working with a full-blown
objects-based epistemology is that the earlier dialogues seem not to sub-
scribe to Distinct Objects. In the Meno, a person who lacks epistêmê can
have “true doxa about those things which he does not know [peri toutôn
hôn ouk oide]” (85c)—that is, about things which one could have epistêmê.
Indeed, because a true doxa can be turned into epistêmê (when “tied down
with a reasoning out of the cause” (98a)), it seems that the initial doxa and
the resulting epistêmê will have the same subject matter: both will be about
how to double the area of a square (85c–d), or about the road to Larissa
(97a–b). These observations have motivated some to resist the Distinct
Objects interpretation even of the Two Worlds dialogues, on the grounds
that it would impute a radical inconsistency between these and the Meno
(see especially Fine 1978 and 1990). I want here to show that the inconsist-
ency is in fact neither so large nor so significant. First, there are some signs
of a Distinct Objects view even in the earlier dialogues. Second, insofar as
these dialogues support Overlap, the explanation is metaphysical. The epis-
temology at a general level remains constant, with epistêmê associated with
Being and doxa with its inferior, readily apparent counterpart; the changes
are driven by changes in Plato’s metaphysics.
Objects and Objects-Based Epistemology 215
Let us review the views we have found in the earlier dialogues about the
objects of cognitive conditions. To begin with:
(1) There is an ontological distinction between privileged Beings and

derivative phenomena.
(2) Epistêmê has a special relation to the Beings: without grasping
Beings one cannot have epistêmê but will instead by consigned to an
inferior cognitive condition, which Plato often calls doxa.
Claims (1) and (2) are obviously shared with the Two Worlds dialogues.
There is however an important difference in the metaphysics, with reper-
cussions for the epistemology. In the earlier dialogues, Plato distinguishes
but does not separate Beings from other things, for the essences or forms
are in the ordinary things, making them be what they are. The nature of
piety (for example) is in pious actions, making them pious. Thus claim (2)
should be read as: epistêmê and doxa may have distinct objects, but not sep
arate ones. Epistêmê is of one aspect of piety—namely the essence—while
doxa is about another aspect: particular instances or accidents of piety. If Plato
does hold a Distinct Objects view here, it is one unlike what we see in the
Two Worlds dialogues, and closer to what we saw in Aristotle (Chapter 1.4):
epistêmê is of the unchanging essential features of things, doxa of their other
features.
Does Plato hold a Distinct Objects view in these dialogues though?
A third claim from the earlier dialogues, which we saw above, may suggest
that he does not:
(3) Epistêmê of Being improves one’s cognition of the derivative

phenomena.
In this general form, claim (3) is shared with the Two Worlds dialogues
too. As we saw in discussing the philosopher-rulers (Chapter 4.5), epistêmê
of Forms improves one’s cognition of perceptibles. On the Two Worlds’ ver-
sion, I argued, the claim is consistent with Distinct Objects: even philo
sophers cannot have epistêmê of perceptibles, but only some other kind of
enhanced cognition.14 The earlier dialogues however, as we saw in Section 1,
14 I argued that Plato might call this doxa, or might want to give it another name, such as
practical wisdom.
may subscribe to a stronger version of (3): arguably here epistêmê of Being

allows for epistêmê, nothing less, of the derivative phenomena. (One who
has epistêmê of what virtue is will also have epistêmê of what is like, e.g.
teachable; one who has epistêmê of the nature of squares will also have
epistêmê of how to double a square.)
In fact there is a possible Distinct Objects reading of the earlier dialogues
consistent with this stronger version of (3)—or more precisely, a possible
way to flesh out what Plato here leaves under-theorized. It turns on reading
strongly into the Meno the notion of doxa as atheoretical cognition. The
idea would be that the improved cognition, although it is epistêmê, is not in
fact about the very same objects as the original inferior cognition was. In
gaining epistêmê one shifts one’s attention from particular, perceptible
instances or qualities to the universal features that underwrite them—the
essences, and also whatever features these robustly explain. For example:
the Meno’s slave has only true doxa about squares, even at the end of the
mathematical discussion (85c–d), because his thoughts are about particular,
perceptible squares. Just like Meno at the start of the dialogue, who does not
yet grasp the idea that there is some Being of virtue, distinct from the many
virtues—and just like the Republic’s sight-lovers and cave-dwellers—he has
not yet turned his mind to Being. If he studies geometry further, he will
start to get the idea that there is a single nature of the square itself, common
to all squares; then, and only then, will he leave doxa behind. When he has
grasped the nature of the square, and can use it to demonstrate a theorem
about doubling the area of a square using the diagonal—that is, when he has
tied down his true doxa with a reasoning out of the cause (98a)—he will
replace his doxa with epistêmê.15 The resulting epistêmê will not be about
particular perceptible squares (although it will of course inform our judg-
ments about them). Instead it will be about the form of the square itself:
about how the essence of squareness explains the various other features of
squareness.
There is, then, a viable Distinct Objects reading of the Meno. But I wholly
grant that it is not an obvious one: here, unlike in the Two Worlds dialogues,
Plato makes no clear effort to restrict the objects of each cognitive kind.
15 That is, when one ties down a true doxa (Meno 98a) one ceases to have doxa on this topic,
for the resulting epistêmê is not a kind of doxa (just as in coming to have knowledge, one ceases
to have an opinion, for knowledge is not a kind of opinion). For arguments that this is the
Meno’s view, see Gerson 2009, 28–30; and Vogt 2012, 13–14.
Objects and Objects-Based Epistemology 217
This brings me to the crucial point: if in fact the earlier dialogues do not
embrace Distinct Objects, while the Two Worlds dialogues do, this is not a
puzzling inconsistency, but instead just what we should expect.
In the Socratic dialogues there is no metaphysical doctrine of separation
between forms or essences and the derivative items they explain. In the Two
Worlds dialogues, Being occupies a realm of its own, separate from the
ontologically inferior. As this ontological divide grows, so too does the epis-
temological one: access to Being no longer entails or involves access to the
inferior.
Moreover, in the earlier dialogues there is no suggestion that the deriva-
tive items are intrinsically unsuited to be objects of epistêmê. Socrates thinks
that kinds or instances of virtue (for example) are varied and different, like
bees (Meno 72b–c), but he never claims that they are so messy as to be
beyond the reach of epistêmê, as we have seen him do in the Two Worlds
dialogues (Chapter 4.4).
Here then is a hypothesis. Right from the start Plato took it as a basic
principle that epistêmê is primarily or specially of Being, and that there is a
widespread inferior cognitive condition which fails to get at Being. Only
however with the robust separation of Being from other things, things
which were thereby downgraded, did he come to hold that epistêmê is only
of Being: in the earlier dialogues, he held that Being is epistêmê’s primary
and special object, but not its only one. Thus the earlier dialogues subscribe
to a moderate Overlap view much like what some readers find in the
Republic.16
So much for Distinct Objects in the earlier dialogues. What about the
main subject of this book, objects-based epistemology? (Recall that a ver-
sion of this is compatible even with a moderate Overlap reading like the one
just proposed: cognitive kinds would be defined by their primary objects,
just as on the strict version they are defined by their exclusive objects
(Chapter 1.1)). Is Plato already working with full-blown objects-based Basic
Conceptions of epistêmê and doxa, on which epistêmê’s relation to Being,
and doxa’s to what seems, constitute the basic feature from which all others
are derived? Or in the earlier dialogues are these simply important features
of epistêmê and doxa, alongside others, without serving as fundamental uni-
fying features?
16 See Harte 2018 and Kamtekar 2009, discussed in Chapter 1.1.

I find no strong evidence for full-blown objects-based conceptions in the

earlier dialogues. We saw above that the Two Worlds dialogues explicitly
ground features of cognitive kinds in features of their objects: for example,
they ground the stability of epistêmê in the stability of Beings, and the
instability of doxa in the instability of Becoming (see Chapters 4.3 and 7.3).
In the Meno, by contrast, Socrates introduces the claim that epistêmê is
stable and doxa unstable without argument, as if it were something that
Meno will simply accept, and we are evidently meant to take this as a basic
feature that needs no special explanation.
I want to leave the question open, for what we have seen is compatible
with both options. Perhaps the Basic Conceptions are at work already, and
robustly explain Plato’s characterizations of epistêmê and doxa. Or, more
plausibly, these conceptions are inchoate in these earlier dialogues, waiting
for the full-blown metaphysics of the Two Worlds dialogues and with it the
robust doctrine of objects-individuated and objects-defined cognition.
In either case, we have seen that that the association between epistêmê
and Being, and between doxa and the ontologically inferior or what seems,
are present in and at least capable of doing explanatory work in the earlier
dialogues’ epistemology. Here then is my proposal: although in the earlier
dialogues Plato may not already be committed to the Basic Conceptions, his
views in these dialogues are such that when he does come to a more explicit
and developed epistemology, along with a more inflated metaphysics, these
are the natural conceptions for him to adopt.
10
Epistemology in the Theaetetus
A thorough study of Plato’s epistemology across the dialogues would require

detailed interpretations of the later dialogues, but given this book’s focus on
the Two Worlds dialogues, I will confine myself here to a brief study of the
Theaetetus, with some comments on the Sophist and Philebus. (I have also
discussed these dialogues at various points earlier in the book, noting that
the Philebus, despite its presumed late date, is very much a Two Worlds
dialogue).
The Theaetetus obviously merits special attention here because its focus is
epistêmê. On its surface, the epistemology of the dialogue appears to clash
radically with that of the Two Worlds dialogues. Not only does the
Theaetetus draw no association between epistêmê and Forms (which are
indeed never explicitly mentioned), but it devotes extensive attention to the
hypothesis that epistêmê is perception, and even after rejecting that hypoth
esis gives frequent examples of epistêmê of perceptibles. Moreover, there is
no attempt to distinguish epistêmê and doxa by their objects.1 Indeed the
two are distinguished only minimally: the first hypothesis of the dialogue
casually equates them (epistêmê is perception, where perception is equated
with appearance and doxa (see below)), and the second and third hypoth
eses construe epistêmê as a kind of doxa: true doxa (187b), or true doxa with
an account (201c–d). Nonetheless, I want to show, we can see the Basic
Conceptions very much at work in the Theaetetus.
1 Indeed, it may seem that the dialogue rejects an objects-based conception of epistêmê at
the outset, since Socrates rejects Theaetetus’ list-definition of knowledge (geometry, cobbling)
on the grounds that it says what things epistêmê is of (τίνων) rather than what it is (146c–e).
But Socrates’ principal objection to the proposed definition is that it enumerates the species of
epistêmê rather than offering a unified definition; the complaint about objects comes in because
the way to enumerate the species of epistêmê is to name various domains (e.g. “epistêmê of
shoe-making”). Socrates’ objection to this is fully compatible with his holding that the correct
definition is by way of the general object. (By analogy, someone might reject a definition of
appetite by enumeration of its species—“thirst and hunger—that is, appetite for drink, appetite
for food”—while holding that appetite is in fact to be defined by its generic object, as desire for
the pleasant.)
DOI: 10.1093/oso/9780198867401.003.0011
220 Epistemology in the Theaetetus
My discussion will be brief and sketchy, ignoring many interesting

aspects of the dialogue’s epistemology and failing to engage extensively with
others’ interpretations. I am not aiming to present a reading as being by all
measures the best reading of this dialogue. Instead I aim to show that it can
be read as consistent with the reading of Plato’s epistemology I have devel
oped. In fact I think the reading I propose an attractive one on its own
terms, but this is not the place to defend it against its many rivals. It is how
ever worth noting that, despite some similarities, the interpretation I pro
pose is not the one we find in Cornford 1935 and some Ancient interpreters
on which the Theaetetus offers an implicit defense of the theory of Forms.
My reading is consistent with this one, but also with the view that Plato has
retained the basic tenets of his epistemology—epistêmê is of what Is, while
doxa is of something distinct and inferior, what seems—while deflating his
metaphysics, so that Being and seeming are no longer to be identified as the
intelligible realm of Forms and the perceptible realm of Becoming.
It will be clearest to proceed by considering Plato’s treatments of epistêmê
and of doxa together, in each of the dialogue’s three sections in turn.
1. Theaetetus’ first definition
Despite its radically anti-Platonic premise, we can clearly find the main
elements of the Two Worlds’ dialogues’ epistemology in the first part of the
dialogue.
Theaetetus hypothesizes that epistêmê is perception. Socrates argues that
Theaetetus’ hypothesis is equivalent to Protagoras’ Measure Doctrine, in the
process equating perception with appearance and with doxa.2 He then
argues that the hypothesis also entails Heraclitus’ flux ontology. In other
words, Socrates takes thesis (1) to entail (2) and (3):3
(1) Epistêmê is no different from perception, i.e. from doxa. [Theaetetus]

(2) Being is no different from what seems. [Protagoras]
(3) Being is no different from Becoming. [Heraclitus]
Why should these relations between the theses hold? There is a good deal
of debate over how to understand Socrates’ surprising development of
2 I give evidence below. Compare Frede 1999.

3 For a more complex interpretation of the structure, see Burnyeat 1982.
Theaetetus ’ first definition 221
Theaetetus’ hypothesis. I will present an interpretation that makes sense of

the argument by showing it to presuppose the main tenets of Plato’s objects-
based epistemology: epistêmê is of Being, doxa is of what seems, and cogni
tive kinds are defined by their objects.
To put it briefly, I propose that the entailments between theses (1)–(3)
hold because epistêmê is of what Is, while doxa is of what seems, which
is identical to what becomes. If the cognitive conditions turn out to be
identical—thesis (1)—then so too must be their objects (theses (2) and (3)).
Now for some evidence, beginning with the connection between epistêmê
and Being. As we saw in Chapter 3, in this portion of the dialogue Socrates
treats the connection as obvious, common ground between all disputants.
In developing Theaetetus’ hypothesis he states as an explicit assumption
that “epistêmê is of what is” (152c) and in refuting the hypothesis he relies
on the claim that epistêmê grasps being (186e).4 What does he mean by
‘what is’ and ‘being’? There is an enormous amount of interpretative contro
versy over the meaning of these words in the refutation passage and its con
text (184b–186e), which I will address in the next section. Much less
attention has been paid to the question of what the words mean in the earl
ier part of the dialogue, but it is obviously most charitable to think that the
meaning is the same in both contexts. (See Chapter 3.1. The claim that
epistêmê is of what is, or being, functions as a premise in both the develop
ment and the refutation of Theaetetus’ hypothesis; if the meaning has
changed, then the refutation misses its mark). What I want to show here is
that whatever ‘what is’ and ‘being’ mean here, they play the same role that
we have seen elsewhere: privileging certain things above others as onto
logically superior.
Theaetetus states that epistêmê is perception; Socrates takes this to entail
that as things are perceived by or appear or seem to each person, so they are
for that person, i.e. are true.5 There is a clear conceptual contrast here
between what is, or truth, and what seems: Protagoras’ radical view that the
4 See the texts discussed in Chapter 3.1 as expressing what I there called ASSUMPTION:
“In matters of hot things and all things like that, as each person perceives things, so they are for
that person . . . Perception therefore is always of what is [tou ontos], and without falsehood
[apseudes], as befits epistêmê” (152c); “Perception has no share in grasping truth, since it has
no [share in grasping] being [ousia], and therefore neither [has it any share in] epistêmê. And
therefore perception is not the same as epistêmê” (186e).
5 See, in addition to the passages quoted just below: “As each thing appears to me, so it is for
me [οἷα μὲν ἕκαστα ἐμοὶ φαίνεται, τοιαῦτα μὲν ἔστιν ἐμοί]” (152a); “as each person perceives
things, such things also are for that person” (152c); “the things that seem are true for the per
son to whom they seem [τὰ ἀεὶ δοκοῦντα . . . τῷ δοκοῦντι εἶναι ἀληθῆ]” (158e).
two always converge indeed presupposes such a contrast. Indeed, we see

appeals to this contrast throughout the discussion. For example, Protagoreans
hold that:
whatever seems [dokoun] on each occasion to each person this also is for
the person to whom it seems;6 they wanted to hold this both in other cases
and not least concerning just things, saying that more than anything
whatever a city lays down as seeming [doxanta] [just] to it, these things
also are just. (177c–d emphases added; cf. 172b)
Moreover, Socrates presents as obstacles to the Protagorean view cases like

dreams, hallucinations, and misperceptions, in which (according to com
mon sense) “far from what appears [phainomena] to each person also being,
quite the opposite: nothing that appears is” (158a, emphases added).
Thus Socrates’ use of ‘being’ in this section of the dialogue is a contrast
use, and the relevant contrast is one central to the argument of this book,
the one we have seen to underlie Plato’s notion of seeming and his notion of
doxa (see Chapter 6.3):
(Contrast 5) What Is versus what seems (dokei).
With some caution, I will thus use the capitalized ‘Being’ to mark this
notion. I do not mean to say that Plato is here committing to a full-blown
Two Worlds metaphysics on which the realm of Being is distinct from the
realm of seeming: that would involve radically talking past Protagoras and
Theaetetus. I simply mean that here too Plato is privileging one kind of
thing by contrast with others, and marking it with the label ‘being.’
Furthermore, here too as with the other instances of this contrast we have
seen, Being is marked out as the object of epistêmê. Why does Socrates take
the hypothesis that epistêmê is perception to entail the Protagorean view
that things are just as they seem or appear? Because epistêmê is of Being,
while perception is of what seems, and therefore if epistêmê is perception,
what Is and what seems completely coincide—Protagorean relativism holds.
I will return in the next section to the interpretation of ‘being’ later in the
dialogue. For now, we have seen that in the first part we have the familiar
idea that epistêmê is of Being.
Now let us consider how this first part of the dialogue treats
doxa. We find Socrates drawing a casual but systematic—indeed almost
6 τὸ αἐὶ δοκοῦν ἑκάστῳ τοῦτο καὶ εἶναι τούτῳ ᾧ δοκεῖ.

Theaetetus ’ first definition 223
analytic—correlation between doxa and what seems, much as he does in

the Two Worlds dialogues. As we just saw, he variously states Protagoras’
Measure Doctrine as the thesis that things are for each person as that per
son perceives them to be, as they appear, or as they seem. He then moves
without any argument to the claim that on this view each person’s doxa is of
what is. We see this at 161c–d, where he moves from “what seems [to
dokoun] to each person also is” to “whatever someone forms a doxa of
[doxazêi] through perception is true,” and from there to “no one will be
more authoritative in investigating whether another person’s doxa is true or
false.”7 We also see it in the famous self-refutation argument, which equates
the thesis that “what seems [to dokoun] to each person also is for the person
to whom it seems” with the thesis that there is no false doxa (170a–c; cf.
179c). That doxa is of what seems or appears is treated as an obvious truth,
in need of no emphasis nor argument.
Finally, this part of the dialogue not only correlates epistêmê with Being
and doxa with what seems, but implies a robust objects-based epistemology
on which cognitive kinds are individuated and determined by their objects.
This emerges in Socrates’ argument that Theaetetus’ epistemology entails
Heraclitean metaphysics, on which all is such constant flux that “nothing
ever is, but everything becomes”8, and hence all the things that we say are
beings we should call becomings instead (152d–e). Why does Theaetetus’
view entail Heraclitus’? The argument makes sense if we take Socrates to
hold that, given a lack of distinction between epistêmê and perception or
doxa, and the ensuing lack of distinction between what Is and what seems,
there can be no real distinction between Being and Becoming. The epis
temological distinction stands and falls with the ontological one. This is an
argument that presupposes an objects-based, Distinct Objects epistemol
ogy. Indeed it is very similar to an argument we have seen Timaeus make
very compactly in drawing out a consequence of this kind of epistemology
in the Timaeus (with the caveat, to be addressed below, that here Plato is
talking about Forms):
If then nous and true doxa are two different kinds [genê], then these things
do by all means exist, themselves by themselves, the Forms that are
7 Compare 166e–167c, which equates the thesis that things are for each person as they
appear, i.e. as they seem, with the thesis that as each person forms doxa (doxazein) of things so
they are. “To the sick person the things he eats both appear and are bitter [φαίνεται . . . καὶ
ἔστι] . . . It is impossible to form doxa about things that are not [i.e., whatever one forms a doxa
about is] [οὔτε γὰρ τὰ μὴ ὄντα δυνατὸν δοξάσαι] . . . whatever seems just and fine to a city also is
[δοκῇ . . . καὶ εἶναι].”
8 ἔστι μὲν γὰρ οὐδέποτ’ οὐδέν, ἀεὶ δὲ γίγνεται (152e).
imperceptible to us and objects of nous only [nooumena]. If however, as it

appears to some, true doxa and nous differ not at all, then whatever we
perceive through the body must be assumed to be the most stable things
there are. (Timaeus 51d)
Here too, epistemological and ontological distinctions stand and fall

together: if, and only if, epistêmê and true doxa are distinct, so too are Being
and Becoming. If there is some cognitive condition superior to true doxa,
there must also be some ontological entity superior to mere Becoming; if
instead so-called epistêmê is nothing more than true doxa, then so-called
Being is nothing more than Becoming.9 Collapse the epistemological dis
tinction—as Theaetetus’ hypothesis does—and the ontological distinction
collapses too.
Why hold that the epistemological and ontological distinctions are mutu
ally entailing? We know why Plato holds this view in the Timaeus: because
epistêmê is exclusively of Being, and doxa exclusively of Becoming.10 Why
hold the specific thesis about stability, that if there is no epistêmê distinct
from true doxa then there is no stable Being? Again, we know why Plato
holds this view in the Two Worlds dialogues: because epistêmê inherits its
stability from its objects; the existence of a stable cognitive condition
depends on the existence of stable objects like Forms. And we can see the
same principles at work in Socrates’ argument from Theaetetus’ epistemology
to Heraclitus’ metaphysics. We might paraphrase his thinking as follows:
(P1) Epistêmê is of Being, obviously.

We normally think of epistêmê as stable, and of being stable because Being
is stable—since, as you know, I hold:
(P2) Cognitive kinds inherit their qualities from their objects.
But now you, Theaetetus, are proposing that epistêmê is unstable: conflicting,
changeable, varying with varying appearances. In short, you’re reducing
epistêmê to doxa. Well then:
9 That is the meaning of the last phrase: to say that perceptibles are the most stable things
there are is to say that there are no genuinely stable items, and hence, given Plato’s metaphysics,
no Beings.
10 Compare an argument one might make about the paradigm object-individuated powers
of Republic V: “If sight and hearing really are two different kinds, then colors do exist, distinct
from hearing. But if sight and hearing do not differ at all [i.e. if ‘sight’ is just a misleading name
for hearing, according to some strange proposal], sounds are all there is. Sight is different from
hearing if and only if color is different from sound.” Why would someone hold this view?
Because they think sight is by nature of color, and is individuated as a sense—counts as a sense
distinct from hearing and the others—for that very reason.
Refuting Theaetetus 225
(P3) If epistêmê is unstable, then by P1 and P2, Being must be unstable too.
But that’s just another way to say that:
(C) There really isn’t any Being as we ordinarily think of it, only Becoming:
you’re reducing Being to Becoming, like Heraclitus.
Finally, before leaving this section of the dialogue, a comment on the notion
of Being. I have been comparing the Theaetetus’ argument to the Timaeus’,
but I do not mean to suggest that Socrates here follows Timaeus in identify
ing the Beings as Forms. As we saw above, when Socrates introduces the
claim that epistêmê is of what is (152c), the claim is meant to be obvious and
acceptable to his interlocutors, Theaetetus and the imagined Protagoras. It
would be radically slippery of Socrates to be appealing here to Forms.
Nonetheless, he is still picking out the ontologically superior, i.e. something
privileged over and contrasted with other things and marked with the
label ‘being’: what Is by contrast with what appears (152a—Contrast 5); or
in the next lines by contrast with what becomes (152d—Contrast 2 from
Chapter 3).
Of course the reader may suspect that what Socrates has in mind are
those paramount non-seeming, non-becoming entities from the Two
Worlds dialogues, the Forms. But his interlocutors can hear the claim very
differently. On the surface, Socrates is simply appealing to the ordinary con
trasts between how things are and how they appear, and between things that
stably remain the same and things that change, contrasts that underlie the
commonsense notion of stable, mind-independent reality, a notion with
which Protagoras and Heraclitus are perfectly familiar. At 152c, Protagoras
can go along with the argument, thinking “Yes Socrates, of course I agree
with you that epistêmê is of what really is, not what merely seems to be.
Most people think that means that going with how things seem is not
enough for epistêmê. But I argue that how things seem is how they really
are. And therefore going with how things seem—that is, perceiving—simply
is epistêmê.” So we can read this part of the dialogue as surreptitiously
assuming a Two Worlds metaphysics, or as studiously remaining innocent
of it. In either case, I hope to have shown that it presupposes an objects-
based epistemology on which epistêmê is of what Is, and doxa of what seems.
2. Refuting Theaetetus
After refuting Heraclitus’ view and Protagoras’, Socrates turns to a direct

refutation of the thesis that epistêmê is perception, in an argument
I discussed briefly in Chapter 3. Perception grasps only proper perceptibles

like color and sound, and cannot grasp any of the “common” features of
objects, including being (184b–186b). Therefore:
Perception has no share in grasping truth, since it has no [share in grasping]

being [ousia], and therefore neither [has it any share in] epistêmê. And
therefore perception is not the same as epistêmê. (186e)
As before, being is the special province of epistêmê. But what does Plato
have in mind? There is an entire body of literature debating the meaning of
‘being’ in this part of the dialogue. Is Socrates referring to Forms? To
essences? To objective reality? Or is he using ‘being’ as the copula, so the
claim is that perception cannot make predications?11
Once again, we do not need to settle this question to show that ‘being’ has
the function of picking out the ontologically superior. The argument that
perception has no share in being rests on a metaphysical distinction, yet
another contrast between something labeled ‘being’ and something lesser:
(Contrast 8) The Being of Fness (or of F things?) versus perceptible Fness.

(Theaetetus 186b–c)12
Whatever ‘being’ means here, Plato is drawing a distinction between ordin

ary things accessible to ordinary cognition (the hardness that we can touch)
and their Being, accessible only through special intellectual effort (the soul
has to “go back and compare in relation to one another, and try to
discern. . . . [using] calculations [analogismata]”(186b–c))—just as he does
in the Socratic and Two Worlds dialogues. The hardness we perceive is not,
in the strict sense, what Is. So whatever precisely Socrates has in mind by
‘being’ here, he is once again privileging one kind of thing above others, and
marking it out with the label “being.” Moreover, he is once again arguing
that this special kind of thing is the object of epistêmê. In other words,
11 Some representative interpretations: Burnyeat 1976 (copula), Cornford 1935 (Forms),
Cooper 1970 (objective reality), Modrak 1981 (essences).
12 “Don’t we perceive the hardness of the hard [thing] through touch, and the softness of the
soft likewise?—Yes—But the being and that they are [τὴν δέ γε οὐσίαν καὶ ὅτι ἐστὸν], and the
opposition in relation to one another and the being in turn of this opposition, the soul itself
going back and comparing in relation to one another tries to discern for us.—Certainly.—Then
surely the affections that reach the soul through the body are present to humans and beasts as
soon as we are born, while the calculations about these things, regarding both being and bene
fit, scarcely arrive to those to whom they do arrive even after a long time, and through a lot of
effort and education” (Theaetetus 186b–c).
Theaetetus ’ second definition: two senses of doxa? 227
epistêmê is correlated with the ontologically superior—even if that is here a

less robust notion than what we saw in the Two Worlds dialogues. This I
think rules out the straightforward version of the copula reading, but it
allows all the others that have been proposed.13
3. Theaetetus’ second definition: two senses of doxa?
The upshot of the refutation is that epistêmê is not perception, because per
ception does not grasp Being. Given the first part of the dialogue’s casual
equivalences between doxa and perception, we would expect Socrates also
to conclude that epistêmê is not doxa, but here the dialogue takes a surpris
ing turn. Theaetetus, with Socrates’ approval, identifies the soul’s activity of
investigating Being as forming doxa (doxazein), and then offers a second
definition: epistêmê is true doxa.
What has happened? The Basic Conception of epistêmê is still in place:
epistêmê is of Being. But what about doxa? Unlike in the Two Worlds dia
logues, it is no longer an inferior kind of cognition consigned to an inferior
realm, but something that can itself attain truth and Being. And unlike in
the Two Worlds dialogues, it is not something contrasted with epistêmê, but
rather a genus of which epistêmê is a special species. It is the familiar con
ception of doxa as inferior cognition that seems to be at play in the first part
of the dialogue, as strongly suggested both by the close connection with
appearance and perception, and by the fact that the Protagorean equation of
doxa with epistêmê is presented as radical. Why have things changed now?
The most plausible interpretation is that Plato is not revising his old view
of doxa, but instead, while leaving that intact, pressing the word ‘doxa’ into
service to do another task as well: to refer to something for which he has no
other term, namely the generic condition of judging or taking-to-be-true—
what modern epistemologists call belief. That he really has changed the
topic is strongly suggested by the characterization he goes on to give of doxa:
The soul when it thinks [dianooumenê] is doing nothing other than dia
loguing, asking itself questions and answering them, and affirming and
13 If we assimilate the copula reading to the objective-reality reading—grasping that the
white is sweet (for example) means grasping that it really is the case that the white is sweet, i.e.
the white really, objectively is sweet—then this reading fits too. Of course that is a departure
from what Burnyeat and followers propose, but it is one that makes the copula reading a much
better fit with the use of ‘being’ in the first part of the dialogue, and is moreover plausible on its
own terms, I would argue: see Moss 2014.
denying. And whenever it has determined something, either gradually or

by leaping quickly, and affirms the same thing and does not disagree, we
put that down as its doxa. So I call forming a doxa ‘saying’ [to doxazein
legein], and I call doxa a logos spoken not to another nor with voice, but
silently to oneself. (Theaetetus 189e–190a)
Contrast the closest thing we get to a definition of doxa in the Republic:

dreaming, mistaking a likeness for the thing it is like (476c–d). It is very
hard to see these as two different descriptions of the same phenomenon.
What we have here in the Theaetetus looks like a description not of cogni
tively deficient dreaming, but instead of generic belief-formation. The
description is broad and neutral: it seems to apply not only to appearance-
based beliefs like the dreamings of the Republic, or to the doxai discussed
earlier in the Theaetetus, but also to the most reflective, precise beliefs about
hidden underlying truths: epistêmê.
I am thus inclined to agree with those who have argued that Plato’s use of
‘doxa’ at this point simply becomes ambiguous (see especially Vogt 2012,
84, and Sprute 1962). Earlier (in the Meno and Two Worlds dialogues), doxa
was negatively characterized as inferior—unclear, unstable—while there is
nothing negative in the present characterization. Earlier, doxa was con
trasted with epistêmê; now doxa is instead its genus. It is no accident that
while in other contexts Plato’s use invited the translation ‘opinion,’ here it
invites the translation ‘judgment.’14
Nor is this a one-off usage: the Sophist gives an account of doxa as silent
affirmation or denial (263e–264a), very like what we see at Theaetetus
189e–190a. The Philebus is a harder case. It offers something similar, but
with a restriction to thoughts arising from perception (38b–39a): one begins
from a perceptual appearance that x is F, and then after an inner dialogue
asserts that it is F or is not. This implies a view of doxa closer to what we
find in the Two Worlds dialogues—and thus fits well with the decidedly
Distinct Objects treatment of doxa and epistêmê at Philebus 58c–59d. On the
other hand, the Philebus does show a more positive assessment of doxa than
some dialogues: true doxa is grouped with epistêmê and nous in the class
that forms Socrates’ candidate for the good (21b; cf. 11b). Perhaps Plato
here holds a view of doxa which is somehow hybrid of the other two. Or
perhaps—a suggestion to which I will return below—the Philebus view is
14 See among many others the translation by Levett, with Burnyeat’s commentary (1990);
even Cornford opts for ‘judgment’ at this juncture (1935).
Theaetetus ’ second definition: two senses of doxa? 229
the standard for these later dialogues: even here in the Theaetetus Plato
means to restrict doxa to judgments that begin from appearances, which
signals that the attempt to define epistêmê as a kind of doxa will fail.
Suppose however that Plato’s use of ‘doxa’ is indeed ambiguous—that he
here introduces a new use of the word to denote generic belief. If we assume
that he is working throughout with the Basic Conception, the ambiguity is
not arbitrary, because the notion of what seems is similarly ambiguous. As
we have seen at length, Plato often sharply contrasts how things seem with
how things are. But ‘dokein,’ just like our ‘seem,’ can also be quite neutral.
Correspondingly, ‘dokei moi’ (literally “it seems to me”) can simply mean
“This is my impression, this is how things appear to me,” with emphasis not
on any contrast with reality, but instead on the subjectivity of the report: I
am telling you my take on the situation. This latter use fits very well with the
use of doxa to mean something like generic judgment or belief: one thinks
through a question, and the conclusion one draws—how things now seem
to one, on reflection—is one’s doxa.15
It is of course confusing to use the word in both ways, especially within
the same dialogue—and especially in dialogues like the Theaetetus and
Sophist, which in other contexts draw enormous attention to the contrast
between seeming and being. Arguably when in the Sophist Plato introduces
a new term, ‘phantasia,’ he is trying to disambiguate: now ‘phantasia’ takes
on the task of referring to cognition of what merely seems (Grönroos 2013),16
leaving ‘doxa’ to denote generic judgment alone. In other work, with
Whitney Schwab, I argue that Aristotle resolves the ambiguity in a different
direction, keeping ‘doxa’ for the inferior kind of cognition and introducing
a new term, ‘hupolêpsis,’ for generic belief (Moss and Schwab 2019).
At any rate, we can see the Basic Conception at work on both sides of the
alleged ambiguity: it underlies both the characterization of doxa in the first
part of the dialogue, as equivalent to perceiving or being-appeared-to, and
the new account of doxa in the remainder of the dialogue, as judgment or
generic belief. Plato retains the platitude that doxa is of how things seem,
developing it in two different ways.
15 There are other explanations for why Plato would choose ‘doxa’ as his label for this kind
of cognition. Perhaps he is picking up on the older root of dokein in dechesthai, to accept: one
considers a proposition, and then it becomes one’s doxa once one accepts or approves it. I am
grateful to Alex Mourelatos and Paul Woodruff for discussion of this point.
16 Although Grönroos argues that ‘doxa’ now comes to denote not generic judgment, but a
superior species thereof, namely a grasp of essences.
4. Theaetetus’ third definition
The final section of the Theaetetus rejects the identity of epistêmê and true
doxa, but not on the sort of grounds a reader of the Republic would expect.
There is no claim that they are incompatible with one another, let alone that
they have different provinces; the claim is instead that one needs to add
something to true doxa—namely an account (logos)—to get epistêmê. It is
no surprise that this discussion is so widely interpreted as developing a
Justified True Belief account of knowledge:17 Plato seems to be treating doxa
as generic belief or judgment, implying that epistêmê and doxa can have the
same subject matter, and distinguishing epistêmê from other true doxai by
some extra hold it has on the truth.
One natural conclusion to draw is that Plato is now, late in his career,
inventing a new kind of epistemology: he has finally hit on the topics of
knowledge and belief. That may be right. On the other hand, there are
plenty of signs—as many interpreters have noted—that even here he is
working with a very different conception of epistemology, one much closer
to that of the Two Worlds dialogues. I will make the case very briefly,
reviewing some points that others have made and adding some new ones.
First, the attempt to define epistêmê as true doxa with a logos fails:
Socrates and Theaetetus are unable to come up with an adequate account of
logos, and the dialogue ends in aporia. A natural interpretation (defended
recently in Sedley 2004) is that Plato is rejecting the entire enterprise of
defining epistêmê as a kind of true doxa: as in the Two Worlds dialogues, he
thinks epistêmê and doxa exclude one another; even accompanied by the
very best accounts, a true doxa will never be more than doxa.
Second, there are signs that the attempt fails precisely because Plato still
wants to distinguish epistêmê and doxa by their objects. On a traditional
interpretation, the Theaetetus is Plato’s reductio of epistemology without the
Forms: as the Anonymous Commentator puts this interpretation (which he
rejects), “having proposed to investigate epistêmê, in the Theaetetus [Plato]
shows what it is not about [peri], while in the Sophist he shows what it is
about”—viz., Forms.18 Ancient interpreters also interpreted the dialogue as
17 Indeed by Gettier himself in arguing against such an account (Gettier, 1963, 121 n.1).
18 2.34–9. Sedley 1996 attributes the view to the Didaskalikos; it was revived in the last cen
tury by Cornford: “The Platonist will draw the necessary inference. True knowledge has for its
objects things of a different order—not sensible things, but intelligible Forms and the truths
about them . . . The Theaetetus leads to this old conclusion by demonstrating the failure of all
attempts to extract knowledge from sensible objects” (1935, 162–3).
Theaetetus ’ third definition 231
confining doxa to perceptibles. Rather than taking the arguments

istinguishing doxa from perception (at 184b–187a) to show that doxa is
d
generic belief, they took them to show that doxa makes judgments about
perceptibles: judgments that go beyond what perception itself can do, but
that are still confined to perceptible objects.19
Even if we reject these particular readings, here is an argument—
admittedly a highly speculative one—that the argument refuting the second
hypothesis turns on the idea that epistêmê is of Being, and doxa of what
seems. Socrates begins as follows:
There is a whole craft showing you that [true doxa] is not epistêmê . . . the
profession of those greatest in wisdom, those called orators and lawyers.
For they persuade, using their craft, not teaching people but making them
have doxa [doxazein poiountes] of whatever they want . . . And don’t you
mean by ‘persuading someone’ making them form a doxa [doxasai]?
(Theaetetus 201a–b)
The idea that rhetoric persuades by producing doxa, rather than teaches by
producing epistêmê, is one we discussed in Chapters 6.3 and 7.4. There we
saw that Plato tends to explain this feature of rhetoric with another: orators
present what seems rather than what really Is (see Phaedrus 260a–c, 261c–d,
Sophist 234c–d). Could he have the same thought in mind here? Consider
how the argument continues:
So surely whenever the jurors have been justly persuaded about things
which can be known only by someone who sees them [idonti monon estin
eidenai], and not in any other way, and they judge from hearsay [ex akoês],
acquiring true doxa, they have judged without epistêmê, but are correctly
persuaded, if they juried well? (Theaetetus 201b–c, emphases added)
Why could only an eye-witness have epistêmê about a crime? This claim,
reminiscent of the Meno’s claim that the person who knows the road to
19 See Chapter 1.3. They found evidence in the distinction between proper perceptibles as
the province of perception, and commons—where commons are still features of perceptible
objects—as the province of doxa: Proclus defines doxa as cognition of the essences of percepti
bles. They also found evidence in the wax block model, where doxa results from matching a
memory of something formerly perceived with an occurrent perception (Theaetetus
191c–195a): the Didaskalikos, presumably drawing also on the Philebus, takes this to show that
“doxa is a combination of memory and perception” (Didaskalikos 4.5).
Larissa is the one who has walked it himself (97a–b), seems to suggests that
epistêmê requires direct experience of its objects. On a once-prevalent inter
pretation, Plato took epistêmê to be “knowledge by acquaintance,” and
ignored the possibility of “knowledge by description,” or got confused
between them (see especially Cross and Woozley, 1964, 171–78). But why
would Plato, who even in the Theaetetus so elevates the epistemic role of
reasoning, and so downgrades the epistemic role of perception, think per
ceptual contact with an object more epistemically valuable than an accurate
account of it?20 Perhaps instead the claim is something specially connected
to rhetoric: indeed, that courtroom testimony can yield only doxa, not
epistêmê, was a staple of contemporary rhetoric.21 Perhaps then Plato is
appealing to his characterization of rhetoric as persuading by presenting
what seems. The orator produces doxa because he can only ever present an
image or reputation of the crime, even if a veridical one, not the thing itself.
Witnessing the crime oneself is a metaphor for being in touch with things as
they really Are. Thus the argument distinguishing epistêmê from true doxa
appeals to a version of Distinct Objects: epistêmê is of the thing itself (e.g.
the crime itself, what it Is), while doxa is of the appearance or reputation
(the lawyer’s report, the crime as it seems to those not in touch with the
thing itself).
If so, then doxa is not after all generic belief, but something narrower:
belief based on how things appear. We should revise our interpretation of
doxa in the dialogue’s second and third hypotheses: the apparently new use
and definition were never meant to widen doxa’s scope beyond the realm of
seeming.
Moreover, if this reading of the jury passage is right, then the definition
of epistêmê as true doxa fails because epistêmê has a special object which is
beyond the reach of doxa: Being. This would suggest that the third defin
ition, true doxa with a logos, has the task of capturing precisely this feature,
and in fact we see some evidence that it does. Socrates explains the idea that
the relevant logos is a list of a thing’s parts (the second account of logos) as
follows:
The person who is able to recount the being [ousian] of the wagon by
going through those one hundred parts has by this addition added logos to
20 Others argue that the experience of walking the road is a metaphor for the intellectual
labor of working out the explanation for oneself: see for example Fine 2004.
21 See for example Antisthenes’ Ajax 1.1–1.8 and Isocrates’ Antidosis 52–4.
Epistemology beyond the Two Worlds Dialogues 233
their true doxa, and instead of having doxa has become an expert and has
epistêmê of the being of the wagon.22 (Theaetetus 207b–c)
When Socrates rejects this account, it is on the grounds that being able to
list the parts is not sufficient for having a good grasp on them; the suggestion
that epistêmê requires grasping the being is left untouched. Moreover, this
characterization of epistêmê-conferring logos has a good Platonic p edigree:
see especially Republic 534b, where only the dialectician has nous, because
only they “grasp the account of the Being of each thing” (ton logon . . . tês
ousias).23 The third account of logos, as naming a thing’s distinguishing
mark, seems to move in a different direction, but its failure could be
taken as a sign that the right answer has been touched on and then aban
doned: the person with epistêmê is the one who can give a logos of a
thing’s Being.
My discussion of the Theaetetus’ epistemology has been brief, and has left
out many important details. I hope however to have shown in broad out
lines that there is a compelling reading of the dialogue on which Plato
retains an objects-based epistemology (in developing Theaetetus’ first defin
ition), retains the Basic Conception of epistêmê, and either expands the
Basic Conception of doxa, or retains it throughout while also using ‘doxa’ to
introduce a new notion: generic belief.
5. Epistemology beyond the Two Worlds Dialogues
In the Two Worlds dialogues, Plato puts forth full-blown objects-based con
ceptions of his two main cognitive kinds. In this chapter and the previous, I
have argued that we can trace these conceptions to truisms he works with
and concerns he holds in the earlier dialogues, and that we can see them
very much at work in his late treatise on epistemology, the Theaetetus.
While—just as we should expect—Plato’s objects-based epistemology is at
its most developed in the dialogues which most develop the metaphysical
divide on which it is premised, it is anticipated earlier and survives later,
and it serves as a useful lens through which to interpret Plato’s epistemology
as a whole.
22 ἀντὶ δοξαστικοῦ τεχνικόν τε καὶ ἐπιστήμονα περὶ ἁμάξης οὐσίας γεγονέναι.
23 Compare Gorgias 501a, where crafts are epistemically superior to knacks because they
can give a logos of the nature (phusis) and the cause of their objects.
Conclusion
Plato’s Ethical-cum-Metaphysical Epistemology
What is Plato doing when he talks about epistêmê and doxa? I hope to have
shown conclusively that he is doing something very different from what
epistemologists do today.
First, his central epistemological categories are very different from ours.
His epistemology is centered around the contrast between a deep grasp of
ultimate reality and atheoretical thought that mistakes appearances for real-
ity. These are notions at best marginal to epistemology today. Perhaps we
can trace important continuities between these notions and our contempor
ary notions of knowledge and belief; conceivably these continuities are
strong enough that there is an interesting case to be made that Plato shares
our epistemological concepts, while presenting different substantive theor
ies about them.1 Even if this project is successful, however, the differences
between Plato’s notions and ours are profound.
Second, his entire approach to epistemology is radically different from
ours. His epistemology is objects-based, and hence dependent on meta-
physics; ours is focused instead on notions like justification, evidence,
rationality—purely epistemological notions, we might say—with little or no
attention to the question of cognition’s objects. Indeed, Plato would think
most of what goes on in epistemology today badly misguided, because it
ignores metaphysics. Most obviously, he would consider most of our con-
trasts between knowledge and belief as merely contrasts between different
kinds of doxa. Consider once again the claim in the Timaeus that a robust
distinction between nous and doxa entails a corresponding ontological dis-
tinction between Being and becoming:
If then nous and true doxa are two different kinds, then these things do by
all means exist, themselves by themselves, the Forms that are imperceptible
1 See the suggestions in Chapter 5.4 and 8.4.
DOI: 10.1093/oso/9780198867401.003.0012
Conclusion 235
to us and objects of nous only. If however, as it appears to some, true doxa

and nous differ not at all, then whatever we perceive through the body must
be assumed to be the most stable things there are. (Timaeus 51d)
Without a Two Worlds metaphysics, Plato would think, we are not entitled
to a two-kinds epistemology; we are just drawing fine distinctions between
fundamentally shoddy cognitive phenomena.
Others have seen that Plato’s epistemology is bound up with his meta-
physics. What has not been fully enough nor widely enough appreciated,
however, is just how significant this fact is for understanding what his
epistemological categories are—and how radically different they are from
ours. Knowledge and belief are object-neutral phenomena: one can have
them about any kind of thing whatsoever. This means that knowledge and
belief can be defined and understood independently of their objects and,
therefore, largely independent of ontology. Of course, one’s ontology will
influence one’s substantive epistemological views: for example, if you think
that there is no external world you will think that very many of our beliefs
are false. But even an ontological view as radical as this has no bearing on
one’s views about what knowledge or belief are. Indeed, the only way that
contemporary epistemology conceptually presupposes ontology is through
the widespread view that knowledge and belief are propositional attitudes.
If you share this view, and take it to be a substantive one, then you would
seem to be committed to the existence of propositions. You also think that
it is an essential feature of knowledge that it is an attitude only to true
propositions; perhaps you think it is an essential feature of belief that it is an
attitude towards all sorts of propositions. There is certainly some analogy
here to the epistemology I am attributing to Plato, but it is a fairly loose one;
moreover, the propositions one knows or believes can be about any objects
whatsoever. Plato’s epistemology, I have argued, is dependent on ontology
in a much more fundamental way. His very distinction between epistêmê
and doxa, at least at its most developed, entails an ontology which distin-
guishes Beings from ontologically inferior but apparent things. This means
that, despite plenty of interesting and important similarities, Plato’s epistêmê
and doxa are radically different from knowledge and belief as we under-
stand these today.
My main concern in the book has been to show that Plato’s epistemology
is fundamentally different from ours in this way, and to investigate—once
236 Conclusion
freed of the constraints imposed when we try to read him through our own
views—just how his epistemology works. Recognizing the differences will
however raise an important question: why would Plato do epistemology his
way? Knowledge and belief, understood more or less as philosophers under-
stand them now, are salient and important phenomena, obviously worthy of
study; why devote so much attention instead to obscure or possibly empty
notions like the deep grasp of ultimate reality, or atheoretical “dreaming”?
Likewise, the contemporary project of investigating epistemological phe-
nomena in their own right is one we can make sense of; why see them
instead as fundamentally defined by objects?
In this final chapter I will try to show that given Plato’s overarching philo
sophical concerns and projects, his way of doing epistemology makes excel-
lent sense—so much so that even without the detailed studies we have
made, we could predict that his central epistemological categories and
approach would be what I have argued that they in fact are. I am returning
to the big picture I sketched briefly in the Introduction, filling it out a bit
with textual evidence and indicating how the arguments of this book have
supported it.
It is widely recognized that Plato’s philosophy is largely driven by ethical
concerns, and in particular by an overarching ethical question we see at
work throughout the dialogues: How should one live? (For example: Republic
352d, Gorgias 500c, Apology 28b.) It is also widely recognized that his epis-
temology has a role in his answer to this question: the best life includes epi
stemic excellence as a crucial ingredient. (Consider the Socratic dialogues’
equation of epistemic excellence with virtue, which is in turn constitutive of
or at least absolutely necessary for happiness,2 or the Republic’s depiction of
the life of the philosopher as the happiest.)
I propose that this ethical role for epistemology is the key to understand-
ing it. Plato’s epistemology is not first and foremost driven by purely epis
temological concerns, such as cataloguing our truth-apt mental states, or
responding to skepticism, or analyzing current epistemological concepts. It
is driven instead by ethical concerns. That is, the cognitive kinds and qual
ities that are salient to Plato are precisely those that feature importantly in
the answer to the question of how one should live. It is no accident that
2 See for example Laches 199d, Charmides 174c, Euthydemus 280b–282b, and Meno
87d–89a. There is a debate about the exact relation of virtue to eudaimonia on Socrates’ view
(for one compelling interpretation see Vlastos 1984); all parties agree however that virtue, and
thereby epistêmê, plays a crucial role in living well.
Conclusion 237
epistêmê tends to enter the scene in clearly ethical contexts: in earlier

dialogues, as a candidate for being virtue; in the Republic, as a necessary
condition for ruling city or soul well, and hence for virtue and happiness
(428c–d); in the Philebus, as a candidate for the human good (11b); even in
the Theaetetus, widely regarded as a work of epistemology in the modern
vein, as a virtue (epistêmê is equated with wisdom (sophia) (145e), which is
a virtue (145b)). The questions that motivate Plato’s concern to identify and
define epistêmê, to distinguish it from doxa, and to exhort us to aspire to it,
are primarily questions about what the best life is and how to achieve it.
How does this explain the distinctive features of his epistemology? I will
paint an answer in broad brush, since the central claims are familiar and
fairly uncontroversial, and my main task is to show how they fit together
to make sense of the account I have offered in this book of Plato’s
epistemology.
Although Plato’s answer to the ethical question of how one should live
seems to vary over the course of the dialogues, we can identify some central
common themes.
First: there are certain things which are genuinely, objectively, really
good, and the goal of life is or involves being related in the right way to what
is really good.
Second: there is a salient danger of wasting one’s life and harming oneself
by pursuing what merely seems good instead.
Third: therefore the key to achieving the goal—at the very least, a neces-
sary condition—is being right about what is really good. The nature of desire
is such that our desires push us toward whatever it is that we think good.3
What determines whether or not we pursue what is really good is therefore
whether or not we are correct about what is good, i.e. whether or not we are
able to detect what is really good by contrast with what seems so (see espe-
cially Meno 77a–78a, Protagoras 358b–d, and Republic 505a–e.)4
3 There are vexed questions about how this works, but they do not bear on our question
here and so I set them aside. For example, what is the relation between Socrates’ claims that we
desire what is really good, and his claims that we desire what appears good? (For discussion
see. Barney 2010.) Is only rational desire for the good or seeming good, or are non-rational
desires like appetites and spirited desires also for what seems good to those parts of the soul?
(For defense of the latter answer see Moss 2008).
4 If we do not know (ἴσμεν, ἐπισταίμεθα) the Good itself, no other good can benefit us;
pleasure seems (δοκεῖ) good to the many, and wisdom to the refined; no-one is satisfied to
attain the seeming (δοκοῦντα) goods, but only the real things (ὄντα), so no-one values the doxa
(Republic 505a–e). (Perhaps in the Republic and other dialogues which countenance non-
rational desires there is also a second crucial factor: that the rational part of the soul be free to
pursue what it thinks good. But this condition seems to be inseparable from the epistemic one:
238 Conclusion
With this ethical picture in mind, when Plato turns to consider

epistemological phenomena the distinction most salient to him will be
between the cognitive condition that allows for a successful grasp of real
goods and the cognitive condition responsible for a misguided attachment
to seeming goods. Thus the cognitive condition he singles out as that to
which we should aspire is that of being in touch with what really is. The
cognitive condition he singles out as a danger is that of confining attention
to what seems.
We see this big picture in the Socratic dialogues. Socrates thinks most
people are wrong about what is good: they pursue wealth, honor, pleasure,
and power, wrongly thinking these will make them happy (see for example
Apology 29d–30b, Gorgias 467a). What is really good is what Socrates
exhorts us to pursue: wisdom and virtue.5 The biggest obstacle to pursuing
these goods, meanwhile, is the failure to recognize them as such. This pre-
vents one from recognizing that one does not already possess what is good,
and therefore leaves one content with the sham goods one thinks real. Most
people waste their time pursuing things that, if good at all, are only good for
those who also have virtue (Apology 30b, Meno 87d–89a), failing to recog-
nize the supreme value of virtue. Moreover, because they think they already
have or are pursuing what is good, they are not motivated to pursue virtue:
this is the condition Socrates warns against so emphatically in the Apology,
and elsewhere too, of failing to recognize one’s own ignorance. In other
words, the biggest obstacle to pursuing the good is the false seeming that
one has already achieved it:
This is what is so hard about ignorance: while not being [onta] fine and
good nor wise, seeming [dokein] adequate to oneself. For the one who
does not think themselves lacking will not desire that which they do not
think they need. (Symposium 204a)6
as the Republic’s discussion of degenerate souls shows, those whose rational parts are enslaved
tend to have false conceptions of the good.)
5 For example, “We must treat as most important not life, but the good life. . . and the good
life, the fine life and the just life are the same” (Crito 48b); “Aren’t you ashamed that you care
for money, that you may have as much as possible, and care for your reputation and honor, but
neither care for nor think of wisdom, truth, or your soul, that it may be as good as possible?”
(Apology 29d–e). See also the arguments that wisdom, i.e. virtue, is the unconditional good
which renders other things valuable (Meno 87d–89a, Euthydemus 280b–282b).
6 See also Socrates’ injunction to the jury: they should value his exhortations to wisdom and
virtue far above the gratifications provided by an Olympic victor, for “that person makes you
seem happy, while I make you be happy” (Apology 36d).
Conclusion 239
This entails that the cognitive condition of being right about what is good
is a necessary condition for happiness:
With error removed and correctness guiding, it is necessary that those in

such condition would do finely and well in every action, and those doing
well would be happy. (Charmides 171e–172a)
Indeed, Socrates sometimes suggests a stronger thesis: this cognitive

condition (which is identical to virtue) in fact constitutes happiness.7 The
goal of life is to be in the right relation to the genuine goods. Sometimes this
right relation is spelled out in terms of “using” the goods correctly, where
cognitively grasping them is not this correct use itself but only a necessary
condition of it (Meno 87d–89a, Laches 199d, Euthydemus 280b–282b).8 At
other points it looks as if epistêmê is itself the right relation: simply knowing
the good, which is being virtuous (in these contexts), is being happy
(perhaps implied by the claim that a good person cannot be harmed, at
Apology 41d).
In either case, the ethically salient distinction in epistemology, and thus
the natural one for Plato to focus on in his epistemological investigations,
will be the distinction between grasping what really is good and being stuck
at the superficial level of what merely seems good.
When we come to the Two Worlds dialogues, with their inflated meta-
physical distinction between what Is and what seems, the importance of this
cognitive distinction is correspondingly increased. Now Plato identifies as
what is really good the Forms, and first and foremost the Form of the Good.
Because the Good bestows Being (Republic 509b), what Is and what is good
are now fully coextensive: the genuine goods are the Beings. (Becomings
can be good only derivatively or indirectly, by participating in the Form of
the Good: the “light” of the Good does not shine directly on them.) Thus
the goal of being in touch with what really is good can now be reformulated
as being in touch with what really Is. In other words, the goal of life is now
to interact in the right way with the Forms.
This view of eudaimonia is most vividly illustrated in the Republic’s Cave
allegory, where the philosopher, the one who has escaped from the dark
7 For good discussion of the evidence and the controversy, see Bobonich 2011.
8 “Since we are all eager to be happy, and since we appear to become so by not only using
things but using them correctly, and since it is epistêmê that provides correctness and good
fortune, it seems that every person must prepare himself by every means so that he becomes as
wise as possible” (Euthydemus 282a).
240 Conclusion
cave of images to the light of Being and truth, rightly recognizes himself as
happy (eudaimonizein) and pities those who remain behind (Republic 516c,
cf. 518b). We see it clearly elsewhere too: to take just a few examples, in the
Phaedo the goal of life is to escape the prison of the body to be in contact
with the Forms; the Phaedrus has the same vision of the blessed afterlife.
The dialogues that characterize ultimate reality as consisting in Forms also
characterize eudaimonia as contact with those Forms, and characterize the
life that ignores or fails to reach these as misspent.9
This view has something in common with one that is familiar to us, often
as a response to skeptical scenarios. Plato uses the Cave allegory to repulse
us from the life of the deluded prisoners, and thus make us realize that we
value contact with reality; philosophers still make the same kind of argu-
ment two millennia on. Consider Nozick’s meditation on one upshot of his
Experience Machine thought-experiment:
We want to be importantly connected to reality, not to live in a d

elusion . . .
What we want and value is actual contact with reality. . . .
(Nozick 1989, 106)
Plato’s view goes beyond this, however: we should aspire to contact not
merely with reality in the sense of what is actually the case, but with certain
entities, the things that Are, for these are the very best things, and the best
condition for us is thus a condition of being connected with them.
Moreover, the Two Worlds dialogues make clear that the primary kind of
contact we can have with these Beings—the way to interact with them, to
“have” what is good—is cognitive contact. We come closest to the Forms by
contemplating them, being aware of them, getting to know them, just as the
escapee from the Cave beholds and contemplates the outer world and the
sun. As the Didaskalikos puts it, “the good for humans is in the epistêmê and
contemplation [theôria] of the primary good, which one might call God and
the primary nous” (27.1).
Plato does countenance other relations with Forms—we can and should
love them, imitate them, and thereby come more fully to “participate” in
9 For a version of this argument based on a reading of Diotima’s speech in the Symposium
see Wedgwood: “happiness consists in standing in the right relations to the intrinsic values”
(2009, 308), where the intrinsic values turn out to be Forms. Wedgwood argues that the best
relation to the Forms is instantiation, with cognition only indirectly serving this goal (because
wisdom makes one more excellent and thus more like the Form of the Beautiful (2009, 312)); I
will argue that cognition is itself the best relation.
Conclusion 241
them—but cognition is a necessary condition for these others.10 Moreover,

it is cognition that he describes as actually constituting eudaimonia. Here is
the Republic’s description of the person who achieves the goal of life, namely
the philosopher:
The real lover of learning strives by nature for what Is, and he would not
linger over each of the many things which people have doxa of [doxa-
zomenois] as being,11 but he will go on, and will not dull or let go of his
erôs, until he touches the nature of each thing that Is [autou ho estin hekas-
tou tês phuseôs]…by which, having approached and mixed with what
really is [tôi onti ontôs], and having begat nous and truth, he has gnôsis and
truly [alêthôs] lives, and is nourished. . . . (Republic 490a–b)
The philosopher’s goal—which is the correct goal, the one everyone would
aim at if they knew what was good—is contact with the Forms. This contact
alone yields real happiness—the philosopher alone “truly lives” (by contrast
with ordinary people who, as the Cave allegory puts it, are dreaming).
Moreover, this contact, described though it is here in sexual terms, is clearly
cognitive. The philosopher’s goal is to have excellent cognition of the Forms,
and this is living well.
We see the same idea in the Phaedrus’s divine myth. When the soul after
death is disembodied and unencumbered, it is finally able to gaze on the
Forms, which are what really are, and are the objects of real epistêmê;12 now,
finally, it is contented (agapâi) and feels good (eupathei) (Phaedrus 247c–d).
We see it also in the Symposium: the life worth living (biôton, 211d), the
life that is the goal (telos) of erotic education, is the life in which one has
gnôsis of what Is beautiful, and is “contemplating and being-with [the Form
of Beauty]…not laying hold of an image . . . but laying hold of the real thing
[tou alêthous]” (Symposium 212a).13
10 See for example Republic 500c: the philosophers, “looking at and beholding [horôntas kai
theômenous] [the Forms]…imitate them and make themselves as much as possible like them.”
Compare Timaeus 90a–d, where we become like the divine harmonies of the cosmos, and
thereby obtain the goal of life, through contemplating and understanding those harmonies.
More generally, Plato almost always describes philosophers’ contact with the Forms using vis-
ual metaphors, which are metaphors for cognition.
11 τοῖς δοξαζομένοις εἶναι πολλοῖς ἑκαστοις.
12 οὐσία ὄντως οὖσα . . . περὶ ἣν τὸ τῆς ἀληθοῦς ἐπιστήμης γένος (Phaedrus 247c).
13 Here the soul goes on to beget true virtue, and thereby become immortal. Does this show
that cognitive contact with reality is of merely instrumental value? One could interpret it that
way, but more plausibly such contact is constitutive of this begetting and immortality, wholly or
in part.
242 Conclusion
We see it very clearly in the Phaedo: the goal of life for the
philosopher—which is the true goal of life, i.e. true happiness, by contrast
with the things wrongly thought worth pursuing by lovers of the body—is
to grasp Being through reasoning with the soul itself apart from the body,
i.e. to cognize Being and thereby “obtain truth and wisdom [ktêsashtai
alêtheian te kai phronêsin]” (Phaedo 65c–66a). As in the Cave allegory, this
should be our goal in life: to escape confused cognition of the ontologically
inferior, and to gain cognition of what Is.
Thus the Two World’s dialogues answer to the question “How should one
live?” centrally involves cognitive contact with the ultimate Beings. Such
contact is at a minimum a crucial component of the goal of life, enabling
other goods like imitation of these Beings; in many places, Plato’s view
seems to be simply that such contact is the goal of life—is happiness itself.
Thus when Plato turns his mind in these dialogues to what we would call
epistemological questions, his target—his ideal, the privileged kind he
wants to explain and recommend—is precisely this eudaimonia-conferring
cognition of Being. That is the centerpiece of his epistemology, and there-
fore that is what he emphasizes and dignifies with labels like ‘noêsis,’
‘phronêsis, ‘epistêmê.’
Meanwhile, these dialogues clearly identify the chief obstacle to living
well as excessive attachment to what merely seems, the ontologically infer
ior. The Cave allegory again provides a good illustration: the prisoners are
pitiable because they are stuck in the Cave, i.e. in the realm of Becoming;
worse, they are ignorant of their own plight and thus have no motivation to
escape it. If this is the central ethical danger, then when Plato turns his mind
to epistemology the condition salient to him as inferior will be the one that
keeps us in the Cave: dream-like cognition that mistakes what seems for
reality. Given his ethical concerns, this is the kind of cognition he thinks we
must be on our guard against—that we must identify, learn to recognize,
and try to avoid. That is why he chooses doxa as the deficient counterpart of
epistêmê, the negative focus of his epistemological inquiries.
For Plato, the bright line in ethics is between blessed contact with Being
and pitiable immersion in seeming. Therefore the bright line in e pistemology—
the salient division between higher and lower—is the line between accepting
what seems as final, and going beyond this to find what Is.
Bibliography
J. Adam (1902), The Republic of Plato, 2 volumes, Cambridge.

R.E. Allen (1960), “Participation and Predication in Plato’s Middle Dialogues,”
Philosophical Review 69: 147–64.
J. Annas (1981), An Introduction to Plato’s Republic, Oxford.
D.M. Armstrong (1973), Belief, Truth and Knowledge, Cambridge.
C. Arruzza (2018), A Wolf in the City: Tyranny and the Tyrant in Plato’s
Republic, Oxford.
D. Baltzly (1997), “Knowledge and Belief in Republic V,” Archiv für Geschichte der
Philosophie 79: 239–72.
J. Barnes (1980), reply to M. Burnyeat, “Socrates and the Jury: Paradoxes in Plato’s
Distinction between Knowledge and True Belief,” Proceedings of the Aristotelian
Society, suppl. vol. 54: 193–206.
R. Barney (2010), “Plato on the Desire for the Good,” in S. Tenenbaum ed., Desire,
Practical Reason, and the Good, Oxford, 34–64.
H. Benson (2000), Socratic Wisdom: The Model of Knowledge in Plato’s Early
Dialogues, New York.
C. Bobonich (2011), “Socrates and Eudaimonia,” in D. Morrison ed., The Cambridge
Companion to Socrates, Cambridge, 293–332.
B. Bosanquet (1895), A Companion to Plato’s Republic for English Readers, London.
A. Boylu (2011), “The Powers Argument in Plato’s Republic,” History of Philosophy
Quarterly 28: 107–24.
G. Boys-Stones (1997), “Plutarch on the Probable Principle of Cold: Epistemology
and the De Primo Frigido,” Classical Quarterly 47: 227–38.
F.H. Bradley (1908), Appearance and Reality: A Metaphysical Essay, New York.
S. Broadie (2011), Nature and Divinity in Plato’s Timaeus, Cambridge.
D. Bronstein (2016), Aristotle on Knowledge and Learning: The Posterior
Analytics, Oxford.
L. Brown (1994), “The Verb ‘to Be’ in Greek Philosophy,” in Language, ed., S. Everson,
Cambridge, 212–36.
J. Bryan (2012), Likeness and Likelihood in the Presocratics and Plato, Cambridge.
M. Burnyeat (1976), “Plato on the Grammar of Perceiving,” Classical Quarterly
26: 29–51.
M. Burnyeat (1980), “Socrates and the Jury: Paradoxes in Plato’s Distinction between
Knowledge and True Belief,” Proceedings of the Aristotelian Society, suppl. vol.
54: 173–91.
M. Burnyeat (1982), “Idealism and Greek Philosophy: What Descartes Saw and
Berkeley Missed,” Philosophical Review 91: 3–40.
M. Burnyeat (1984), “Aristotle on Understanding Knowledge,” in E. Berti ed.,
Aristotle on Science: The Posterior Analytics, Proceedings of the Eighth Symposium
Aristotelicum, Padua, 97–139.
244 Bibliography
M. Burnyeat (1990), The Theaetetus of Plato, with trans. by M. Levett, Indianapolis, IN.
M. Burnyeat (2005), “ΕΙΚΩΣ ΜΥΘΟΣ,” Rhizai 2: 143–65.
T. Butler (2007), “On Today’s Two Worlds Interpretation: Knowledge and True
Belief in Plato,” Southern Journal of Philosophy 45: 31–56.
G.R. Carone (2001), “Akrasia in the Republic: Does Plato Change His Mind?” Oxford
Studies in Ancient Philosophy 20: 107–48.
A. Carpenter (2012–13), “Judging Strives to Be Knowing,” in L. Brisson and
N. Notomi eds, Plato’s Republic: Proceedings of the IXth Symposium Platonicum,
Sankt Augustin.
A. Carpenter (2015), “Ranking Knowledge in the Philebus,” Phronesis 60: 180–205.
J. Carriero (2013), “Epistemology Past and Present,” Proceedings of the Aristotelian
Society 113: 175–200.
J.M. Cooper (1970), “Plato on Sense-Perception and Knowledge: Theaetetus
184–186,” Phronesis 15: 123–46.
J.M. Cooper, (1977), “The Psychology of Justice in Plato,” American Philosophical
Quarterly 14: 151–157.
F.M. Cornford (1933), “Parmenides’ Two Ways,” Classical Quarterly 27: 97–111.
F.M. Cornford (1935), Plato’s Theory of Knowledge: The Theaetetus and the
Sophist, London.
F.M. Cornford (1937), Plato’s Cosmology: The Timaeus of Plato, London.
F.M. Cornford (1941), The Republic of Plato, Oxford.
I.M. Crombie (1962), An Examination of Plato’s Doctrines, 2 volumes, London.
R.C. Cross and A.D. Woozley (1964), Plato’s Republic: A Philosophical Commentary,
London.
D. Devereux (1986), “Particular and Universal in Aristotle’s Conception of Practical
Knowledge,” Review of Metaphysics 39: 483–504.
J. Dillon trans. and ed. (1993), Alcinous: The Handbook of Platonism, Oxford.
D. Ebrey (forthcoming), “The Sun, the Good, and the Value of Knowledge.”
S. Everson ed. (1990), Epistemology, Cambridge.
S. Everson (1997), Aristotle on Perception, Oxford.
G. Fine (1978), “Knowledge and Belief in Republic V, ” Archiv für Geschichte der
G. Fine (1990), “Knowledge and Belief in Republic V–VII,” in S. Everson ed.,
Epistemology, Cambridge, 85–115.
G. Fine (2004), “Knowledge and True Belief in the Meno,” Oxford Studies in Ancient
Philosophy 27: 41–81.
G. Fine (2010), “Aristotle’s Two Worlds: Knowledge and Belief in Posterior Analytics
1.33,” Proceedings of the Aristotelian Society 110: 323–46.
G. Fine (2016), “The ‘Two Worlds’ Theory in the Phaedo,” British Journal for the
History of Philosophy 24: 557–72.
G. Fine (forthcoming), “Epistêmê and Doxa: Knowledge and True Belief, in the
Phaedo,” Bulletin of the Institute of Classical Studies.
K. Fine (2001), “The Question of Realism,” Philosophers’ Imprint 1: 1–30.
M. Frede (1990), “An Empiricist View of Knowledge: Memorism,” in S. Everson ed.,
Bibliography 245
M. Frede (1999), “Observations on Perception in Plato’s Later Dialogues,” in G. Fine

ed., Plato I: Metaphysics and Epistemology, Oxford, 377–83.
D. Gallop (1965), “Image and Reality in Plato’s Republic,” Archiv für Geschichte der
D. Gallop (1975), Plato: Phaedo, Oxford.
P.T. Geach (1966), “Plato’s Euthyphro: An Analysis and Commentary,” The Monist
50: 369–82.
L.P. Gerson (2003), Knowing Persons: A Study in Plato, Oxford.
L.P. Gerson (2009), Ancient Epistemology, Cambridge.
E. Gettier (1963), “Is Justified True Belief Knowledge?” Analysis 23: 121–3.
F Gonzalez (1996), “Propositions or Objects? A Critique of Gail Fine on Knowledge
and Belief in Republic V,” Phronesis 51: 245–75.
J.C.B. Gosling (1968), “Doxa and Dunamis in Plato’s Republic,” Phronesis
13: 119–30.
G. Grönroos (2013), “Two Kinds of Belief in Plato,” Journal of the History of Philosophy
51: 1–19.
N. Gulley (1962), Plato’s Theory of Knowledge, London.
R. Hackforth (1945), Plato’s Examination of Pleasure: A Translation of the Philebus,
with Introduction and Commentary, Cambridge.
V. Harte (2006), “Beware of Imitations: Image Recognition in Plato,” in New Essays
on Plato: Language and Thought in Fourth Century Greek Philosophy, F. Herrmann
ed., Swansea, 21–42.
V. Harte (2007), “Language in the Cave,” in Maieusis: Essays on Ancient Philosophy in
Honour of Myles Burnyeat, D. Scott ed., Oxford, 195–215.
V. Harte (2018), “Knowing and Believing in Republic V,” in V. Harte and R. Woolf
eds, Rereading Ancient Philosophy: Old Chestnuts and Sacred Cows, Cambridge,
141–62.
E.A. Havelock (1963), Preface to Plato, Cambridge, MA.
R.D. Hicks (1907), Aristotle, De Anima, Cambridge.
P. Horwich (2012), Wittgenstein’s Metaphilosophy, Oxford.
T. Irwin (1995), Plato’s Ethics, Oxford.
H.H. Joachim (1951), Aristotle: The Nicomachean Ethics, Oxford.
T. Johansen (2012), The Powers of Aristotle’s Soul, Oxford.
B. Jowett (1908), The Republic of Plato, Oxford.
B. Jowett and L. Campbell (1894), Plato’s Republic, Oxford.
C. Kahn (1981), “Some Philosophical Uses of ‘To Be’ in Plato,” Phronesis 26: 105–34.
C. Kahn (1996), Plato and the Socratic Dialogue, Cambridge.
R. Kamtekar (2009), “The Powers of Plato’s Tripartite Psychology,” Proceedings of the
Boston Area Colloquium in Ancient Philosophy 24: 127–50.
Y. LaFrance (1981), La theorie Platonicienne de la doxa, Montreal.
P. Lautner (2002), “The Distinction between Phantasia and Doxa in Proclus’ In
Timaeum,” Classical Quarterly 52: 257–69.
G.R. Lear (work in progress), “Appearances of Forms and the Form of the Beautiful”.
D.C. Lee (2010), “Interpreting Plato’s Republic: Knowledge and Belief,” Philosophy
Compass 5: 854–64.
246 Bibliography
J. Lesher (1994), “The Emergence of Philosophical Interest in Cognition,” Oxford

Studies in Ancient Philosophy 12 (1994): 1–34.
J. Lesher (2009), “The Meaning of Saphêneia in Plato’s Divided Line,” in
M. McPherran ed., Plato’s Republic: A Critical Guide, Cambridge, 171–87.
H.G. Liddell and R. Scott (1940), A Greek–English Lexicon, revised H.S. Jones, Oxford.
A.A. Long and D.N. Sedley (1987), The Hellenistic Philosophers, 2 volumes,
Cambridge.
J. Lyons (1963), Structural Semantics, Oxford.
K. McDaniel (2017), The Fragmentation of Being, Oxford.
J. McDowell (1977), Plato: Theaetetus, Oxford.
D. Modrak (1981), “Perception and Judgment in the Theaetetus,” Phronesis
26: 35–54.
J. Moline (1981), Plato’s Theory of Understanding, Madison, WI.
J.E. Moravcsik (1979), “Understanding and Knowledge in Plato’s Philosophy,” Neue
Hefte fur Philosophie 15/16: 53–69.
W.E. Morris and C.R. Brown (2016), “David Hume,” in E.N. Zalta ed., The Stanford
Encyclopedia of Philosophy (Spring).
J. Moss (2007), “What Is Imitative Poetry and Why Is It Bad?,” in G.R.F. Ferrari ed.,
The Cambridge Companion to Plato’s Republic, Cambridge, 415–44.
J. Moss (2008), “Appearances and Calculations: Plato’s Division of the Soul,” Oxford
Studies in Ancient Philosophy 34: 35–68.
J. Moss (2014), “Plato’s Appearance-Assent Account of Belief,” Proceedings of the
Aristotelian Society 114: 213–38.
J. Moss (2019), “Is Plato’s Epistemology about Knowledge?” in S. Hetherington and
N. Smith eds, What the Ancients Offer to Contemporary Epistemology,
Abingdon, 68–85.
J. Moss (2020), “Plato’s Doxa,” Analytic Philosophy 61: 1–25.
J. Moss and W. Schwab (2019), “The Birth of Belief,” Journal of the History of
A. Mourelatos (2014), “The Conception of Eoikôs/Eikôs as Epistemic Standard in
Xenophanes, Parmenides, and in Plato’s Timaeus,” Ancient Philosophy 34: 169–91.
N.R. Murphy (1951), The Interpretation of Plato’s Republic, Oxford.
A. Nehamas (1975), “Plato on the Imperfection of the Sensible World,” American
Philosophical Quarterly 12: 105–17.
A. Nehamas (1985), “Meno’s Paradox and Socrates as Teacher,” Oxford Studies in
Ancient Philosophy 3: 1–30.
R.L. Nettleship (1906), Lectures on the Republic of Plato, London.
F. Nietzsche (1997), Daybreak: Thoughts on the Prejudices of Morality, trans.
R.J. Hollingdale, eds M. Clark and B. Leiter, Cambridge.
R. Nozick (1989), The Examined Life: Philosophical Meditations, New York.
J. Palmer (1999), Plato’s Reception of Parmenides, Oxford.
R. Pasnau (2013), “Epistemology Idealized,” Mind 488: 987–1021.
R. Pasnau (2017), After Certainty: A History of Our Epistemic Ideals and Illusions, Oxford.
M. Peramatzis (2011), Priority in Aristotle’s Metaphysics, Oxford.
C. Perin (2012), “Knowledge, Stability, and Virtue in the Meno,” Ancient Philosophy
32: 15–34.
Bibliography 247
T. Poulakos (2001), “Isocrates’ Use of Doxa,” Philosophy and Rhetoric 34: 61–78.
W.J. Prior (1998), “Plato and the ‘Socratic Fallacy’ ” Phronesis 43: 97–113.
C.J. Rowe (2012), Plato: The Republic, London.
C. Rowett (2018), Knowledge and Truth in Plato: Stepping Past the Shadow of
Socrates, Oxford.
M. Schofield (2011), “Plato in His Time and Place,” in G. Fine ed., The Oxford
Handbook of Plato, Oxford, 36–62.
W. Schwab (2015), “Explanation in Epistemology of the Meno,” Oxford Studies in
W. Schwab (2016), “Understanding Epistêmê in Plato’s Republic,” Oxford Studies in
D. Scott (1987), “Platonic Anamnesis Revisited,” Classical Quarterly 37: 346–66.
D. Sedley (1996), “Three Platonist Interpretations of the Theaetetus,” in C. Gill and
M.M. McCabe eds, Form and Argument in Late Plato, 79–103.
D. Sedley (2004), The Midwife of Platonism: Text and Subtext in Plato’s
Theaetetus, Oxford.
D. Sedley (2007), “Philosophy, the Forms, and the Art of Ruling,” in G.R.F. Ferrari
ed., The Cambridge Companion to Plato’s Republic, Cambridge, 256–83.
D. Sedley (2013), “Socratic Intellectualism in the Republic’s Central Digression,” in
G. Boys-Stones, D. El Murr and C. Gill eds, The Platonic Art of Philosophy,
Cambridge, 70–8.
P. Shorey (1930), Plato: The Republic, Cambridge.
S. Sierksma-Agteres (2015), “ ‘Say Goodbye to Opinions!’ Plutarch’s Philosophy of
Natural Phenomena and the Journey to Metaphysical Knowledge,” in M. Meeusen
and L. van der Stockt eds, Natural Spectaculars: Aspects of Plutarch’s Philosophy of
Nature, Leuven, 57–71.
A. Silverman (2002), The Dialectic of Essence: A Study of Plato’s Metaphysics,
Princeton, NJ.
A. Silverman (2010), “Philosopher-Kings and Craftsman Gods,” in R. Mohr
and B. Sattler eds, One Book, the Whole Universe: Plato’s Timaeus Today, Las
Vegas, NV.
N.D. Smith (1996), “Plato’s Divided Line,” Ancient Philosophy 16: 25–46.
N.D. Smith (2000), “Plato on Knowledge as a Power,” Journal of the History of
N.D. Smith (2012), “Plato on the Power of Ignorance,” Oxford Studies in Ancient
Philosophy, suppl. vol.: 51–73.
N.D. Smith (2019), Summoning Knowledge in Plato’s Republic, Oxford.
B. Snell (1924), Die Ausdrücke fur den Begriff des Wissens in der vorplatonischen
Philosophie, Berlin.
J. Sprute (1962), Der Begriff der Doxa in der Platonischen Philosophie, Gottingen.
M. Stokes (1992), “Plato and the Sightlovers of the Republic,” Apeiron 25: 103–32.
D. Storey (2014), “Appearance, Perception, and Non-Rational Belief: Republic
602c–603a,” Oxford Studies in Ancient Philosophy 47: 81–118.
J. Szaif (1996), Platons Begriff der Wahrheit, Freiberg.
J. Szaif (2007), “Doxa and Epistêmê as Modes of Acquaintance in Republic V,” Etudes
Platoniciennes 4: 253–72.
248 Bibliography
J. Szaif (2018), “Plato and Aristotle on Truth and Falsehood,” in M. Glanzberg ed.,
The Oxford Handbook of Truth, Oxford, 9–49.
A. Taylor (1926), Plato, The Man and His Work, London.
C.C.W. Taylor (2008), “Plato’s Epistemology,” in G. Fine ed., The Oxford Handbook of
Plato, Oxford, 165–90.
P. van Inwagen (2015), Metaphysics, Boulder, CO.
I. Vasiliou (2015), “Plato, Forms, and Moral Motivation,” Oxford Studies in Ancient
G. Vlastos (1969), “Reasons and Causes in the Phaedo,” Philosophical Review 78:
291–325.
G. Vlastos (1981), “Degrees of Reality in Plato,” in Platonic Studies, 2nd edition,
Princeton, NJ, 73–4.
G. Vlastos (1984), “Happiness and Virtue in Socrates’ Moral Theory,” Proceedings of
the Cambridge Philological Society 30: 181–213.
G. Vlastos (1985), “Socrates’ Disavowal of Knowledge,” Philosophical Quarterly
35: 1–31.
K. Vogt (2012), Belief and Truth: A Skeptic Reading of Plato, Oxford.
K. von Fritz (1994), “Nous, Noein, and their Derivatives in Pre-Socratic Philosophy,”
in A.P. Mourelatos ed., The Pre-Socratics: A Collection of Critical Essays, Princeton,
NJ, 23–85.
R. Wedgwood (2002), “The Aim of Belief,” Nous 36: 267–97.
R. Wedgwood (2009), “Diotima’s Eudaemonism: Intrinsic Value and Rational
Motivation in Plato’s Symposium,” Phronesis 54: 297–325.
R. Wedgwood (2018), Plato’s Theory of Knowledge,” in D. Brink, S. Sauvé Meyer,
and C. Shields eds, Virtue, Happiness, and Knowledge: Themes from the Work of
Gail Fine and Terence Irwin, Oxford, 33–56.
F.C. White (1972), “῾Ως ἐπιστήμη οὖσα’: A Passage of Some Elegance in the
‘Theaetetus,’ ” Phronesis 17: 219–26.
N. White (1976), Plato on Knowledge and Reality, Indianapolis, IN.
N. White (1979), A Companion to Plato’s Republic, Indianapolis, IN.
N. White (1992), “Plato’s Metaphysical Epistemology,” in R. Kraut ed., The Cambridge
Companion to Plato, Cambridge, 277–310.
N. Wolterstorff (1996), John Locke on the Ethics of Belief, Cambridge, 219–21.
P. Woodruff (1990), “Plato’s Early Theory of Knowledge,” in S. Everson ed.,
R. Woolf (2013), “Plato and the Norms of Thought,” Mind 122: 171–216.
E. Zeller (1920), Grundriss der Geschichte der griechischen Philosophie, Leipzig.
Index Locorum
For the benefit of digital users, indexed terms that span two pages (e.g., 52–53) may, on
occasion, appear on only one of those pages.
AESCHYLUS 1028a34–6 61–2
Seven Against Thebes 1028b19–21 189
529–4 146 1049b12–17 61–2
1078b12–17 35, 82
ALCINOUS
Nicomachean Ethics
Didaskalikos
1138b27 7n.13, 21n.16
4.2–8 36
1138b26–32 44n.70
4.5 203n.9, 230–1
1139a6–12 16, 43, 65–6
27.1 240
1139a8–11 67
ANTIPHON 1139b19–21 44
Second Tetralogy 1139b20–4 43n.65
3.2.2 146 1140a1–2 38n.55
1140b25–28 16, 43, 65n.24
ARISTOTLE Magna Moralia
Categories 1196b15–29 66–7
2b5–6 102
CICERO
Posterior Analytics
Academica
88b37 43n.66
I.30–32 36
88b30–89a3 43
II.142 157
89a3–4 43n.66
89a6–8 44 DAMASCIUS, In Phaedonem
89a23–37 80n.46 1.102–3 192
Topics DIONYSIUS OF HALICARNASSUS
114a23 44 De Compositione Verborum
De Anima 45.17 45
402b9–16 16, 61
EMPEDOCLES
404b16–27 69
B 109 68
409b26–7 68
415a16–22,16–2215 FICINO, Divini Platonis Opera Omnia
418a24–5 62 6 40n.61
434a10–12 150n.20 92 203
Metaphysics 93 40
981a10 127n.30 GALEN
981a12–23 127 On the Sects for Beginners 5 204
981a24–30 204n.14 GORGIAS
987b14–18 189 Helen
987a32–b1 35 10 149
1019a1–4 102 13 149
1019a4–6 102n.33 Palamedes
1021a33–b2 63 24 149–50
1028a10–15 101–2 33 146
250 Index Locorum
HERACLITUS 79c–80b 72
B28 148 82d–83d 168–9
ISOCRATES 83b 162–3
Against the Sophists 83b–84a 33–4, 119, 119n.17
8 129 83c–84a 72–3
Antidosis 84a 114–15, 155, 161
52–4 232n.21 84a–b 78
184 45, 122–3, 129 90c 76–7
271 45, 129 96b 171n.24
100c 103, 115, 186
OLYMPIODORUS Cratylus
In Platonis Phaedonem 439c–440d 35n.49
5.4 183 Theaetetus
5.13 192 145b–e 236–7
PARMENIDES 146c–e 219n.1
B1.28–32 147 152a 221n.5
B8.50–62 147n.13 152a–d 223–5
PLATO 152c 90–1, 114n.2, 221
Euthyphro 152d–e 223
4e–5a 209–10 152d–160c 35n.49
5a 4 158a 222
6d–e 200–1, 208–9, 213–14 158e 221n.5
11a 95–6, 208–10 161c–d 222–3
Apology 166e–167c 223n.7
21c 149, 213n.11 170a–c 222–3
21d 213n.11 172b 149, 222
28b 236 174b–175d 200
29b 213n.11 177c–d 222
29d–e 238n.5 179c 222–3
29d–30b 238 184e–185a 54
36d 238n.6 184b–186e 54–5, 91–2, 169n.21,
41d 239 225–7
Crito 186e 221
48b 238n.5 187b 4, 219
53b–c 213 189e–190a 140, 171n.25, 192,
Phaedo 210n.5, 227–9
65a–66a 106, 174n.31, 242 191c–195a 230–1
65c–e 106 193c 37
65d 94n.18 194d 156n.1
65e 183 195e–196a 37n.52
66b–67b 23, 23n.21, 27, 191–2 201a–c 6, 151n.22, 158–9, 231–2
74b 145n.8, 158 201a–210d 230–3
74b–c 21n.14, 71–2, 117, 161n.11, 177, 201c–d 5, 6n.9, 21–2, 140, 155
17974d–76d 97, 208–9 207b–c 232–3
75b–d 96–7, 178 208e 22n.19
76b 6, 98n.28 209a–b 182n.41
77a 96–7 Sophist
78d–e 33, 71–2, 96–8, 161n.11 233c 151n.22
79a–b 33, 71, 7879c–d 33–4, 35n.49, 234c–d 151, 160, 231
71–4, 79, 118, 167 234d–e 160
Index Locorum 251
235d–236c 156 270d–271e 59n.16
236e 149, 165–6 272c 152
240b 163–6 Alcibiades I
242c–e 103–4 117a–b 213
248a 96–7 Charmides
251a 178n.34 159a 210n.5
254a–b 160n.9 167e 58n.14
263b 156 167e–168d 6, 56–8, 62–4, 86n.1
263e–264a 140, 156, 171n.25, 171n.27, 168a 64
192, 210n.5, 228–9 168d 54
263e–264b 171n.27 165c–d 52–3
264a 147n.13, 169n.21 171e–172a 239
Statesman 174c 236n.2
284a–e 131 Laches
294a–296e 131 191e 208–10
Parmenides 199a 59–60
134a 86n.2 199d 236n.2, 239
Philebus Euthydemus
11b 228–9, 236–7 280b–282b 236n.2, 238n.5, 239
21b 228–9 Protagoras
34a 37 312c–e 59–60
35a 58n.14 356d–e 72, 116–17, 158, 161
38b 37 356d–357a 158
38b–39a 140, 171n.25, 228–9 358b–d 237
39a 169–71 Gorgias
56b 116–17 447c 59–60
57b–d 161n.10 449d 59–60
58a 34, 75–6 451d 59–60
58a–59b 75–6 454d 4, 114n.2, 147n.13
58c–59d 71–2, 106, 116, 228–9 454e 159
59a–b 35n.49, 79, 97, 117–18, 455a 159
128, 161n.11 459b–c 151, 158
59c–d 34 463b 59n.16
Symposium 464a 149, 151
202a 156 465a 201–2
204a 239 465c 160
210e–211b 35n.49, 71–2, 117, 161n.11 466e–468d 13–14, 57
211a–b 98, 178, 467a 238
212a 241 468c 63–4
Phaedrus 500c 236
247c–d 34, 96–7, 114–15, 241 501a 201–2, 233n.23
247c–248b 106 Meno
249c 97n.24 71a 209–10
250a–b 97 71a–b 6, 209–10
250a–d 178–9 72a–d 200–1, 208–10
250d 163–4 72b–c 208–9, 217
260a 151–2, 159–60, 231 77a–78a 57, 237
261c–d 152, 231 81c 6
267a 147n.13 81c–86b 6n.8
270d 28 85c–d 212, 214, 216
252 Index Locorum
PLATO (cont.) 477d–478e 6, 15–18

87d–89a 236n.2, 238–9 477e 4, 15, 16n.5, 17, 18n.9, 83, 197
97a–b 6, 214, 231–2 477e–478b 28
97d 4 478a 16nn.5–6, 17–18, 52–6, 86
98a 5–6, 6n.9, 21–2, 40n.62, 71–2, 83, 478a–c 64, 71–2, 74
115–17, 116n.10, 158, 161n.10, 211, 478e 161–3, 165–6
214, 216 479a–b 35n.49, 98, 100–1, 117,
98b 1, 83 161n.11, 166
Hippias Major 479c–e 6, 19, 29n.29, 97, 100–1,
287c–288a 201, 213–14 106, 117
Hippias Minor 480a 198–9
372d–e 213 484b 71–2, 116, 161n.10
373d–e 52–3 484c–d 19n.10, 124–6
375d–376a 53n.3 490a–b 71, 73–4, 240–1
Ion 500b–d 78
537c–538a 59–60 500c 241n.10
Republic 505d 148
334b 179 500e–501c 124, 130–1, 163–4
334c–335a 148 505a–e 237
340c 148 506a–e 21–2, 177, 181–2, 186–7,
346a 126 190–1507b 96–7, 97n.24, 98n.28,
349a–350e 179, 181–2 181–2, 208–9
352d 236 507c–e 54
353a 25–6 508d 29, 33–4, 73–5, 79, 105–6,
357a–b 148 114–16, 118–19, 162–3
358a 148 509b 103, 239
361b–362a 146, 148 509d–510a 34, 117, 161, 167, 169,
365b–c 146, 156 171–2, 174n.31
367b–d 148 509d–511e 75
402b–c 127n.28 510a 161, 163–4, 167, 198n.1
412e–414a 130n.36, 155 510b–511a 182–5
413a–c 116–17, 156 510c 187–8, 195n.63
414b–c 158 511a 161n.10
428b–d 75–6, 126, 191, 236–7 511c 188
430a 158 511d 1n.2, 51n.2, 64, 142n.4,
437d–438d 57–9, 62–3 147n.13, 182–3
438d 21n.16, 27, 75–6, 191 511d–e 77–8, 182–5
443e 21–2 511e 71–2, 114–15, 119
475e 114–15, 155 515a–c 166, 171–2
475e–476a 177–8 515d 166–7
476a 181–2 515e 200n.3
476b 98n.28, 166 516a 163–4
476c 163–4, 166, 175n.32, 176, 184, 516c 239–40
197–8, 200, 228 516c–517a 125
476d–e 4, 183n.43 516d 167n.19, 198n.1
476d–477a 109–11 517b 162, 198n.1
476e–477b 88–90 517c 115n.7
477a 86, 89n.5, 96–7, 100, 105 517d–e 169, 172–3
477b 26n.24 518b 239–40
477c–480a 18–29, 52–6, 88–90, 105, 518c 50, 73–4, 97n.26
121, 197–8 519b 115n.6, 119, 166–7
Index Locorum 253
520c 21–2, 125, 198 77b 169–70

521c–d 105–6, 175, 198 90a–d 77–8, 240–1
523a–524d 164–5, 175–7, 199 Definitions
523e 54n.9 414b–c 71–2, 116–17, 158
524a 35n.49, 54, 71–2, 77–8, 117 413c 116–17
524c 116–17, 179–80 PLOTINUS
525c 115n.6 Ennead V
527b 115n.6 1.4 135
529b–d 31, 128, 184 5.1 47
529b–530b 120–1 5.8 46
529d–530b 35n.49, 71–2, 117, 9.7 46
125–6 Ennead VI
532b–c 185 9.3 46
533a–c 155–86, 189–90, 194–5
PLUTARCH
533d 182–3, 188, 195n.63
Adversus Colotem
533d–e 182–3
13 37, 46
533e–534b 156–85, 177n.33
De Iside et Osiride
534a 6, 31, 106, 161, 183–4
383c–d 46
534b–c 6, 152–3, 185, 198, 233
539e–540a 126n.27 PROCLUS
585d 119, 162–3 In Timaeum
595b 183n.43 2.240–64 39n.60, 46
596e 165–6 2.243.18–22 39
597a–d 96–7, 100, 162–3 2.248 11–22 203
598a 158, 164, 172n.28 2.248.16–17 54n.9
599d–600e 164 2.249.2–3 39
600e 198 2.249.8–9 38
601c–d 21n.15 2.249.13–27 39, 203n.12
601e–602a 75–6, 191 2.254.24–27 38
602b 151n.22, 198 2.255.3–12 39
602e–603a 171, 172n.28 2.339.14–17 38
611e 73–4 2.343.10–11 203
Timaeus Theologica Platonica, I.15.17–20 38
27d–28a 6, 32, 38–9, 46, 97, 100, 105, In Parmenidem
117, 119, 129–30, 161–4, 167, 170–1, I.617 135
174n.31, 203 SEXTUS EMPIRICUS
29b–c 76, 129–30, 147n.13, 163–4 Against the Mathematicians
37b–c 32, 70, 71n.32, 129–30, 147n.13, 7.147–9 45
161, 198–9 Outlines of Pyrrhonism
46d–e 115 2.97–129 204
50c– 163–4
SIMONIDES
51d 83, 159, 223–4, 234–5
76 148
52a 6, 32, 34, 129–30, 161,
161n.11, 167 XENOPHANES
52c 162–4 B34 147
67b–c 54 B35 150
Index
For the benefit of digital users, indexed terms that span two pages (e.g., 52–53) may, on
occasion, appear on only one of those pages.
Adam, J. 18n.7, 21n.15, 22n.20, 41, and essence 94, 96, 98–9, 107–8,
55n.10, 78n.45, 103n.36, 123n.22, 133–4, 208–12
152–3, 178n.34 and Forms, see under Forms
Alcinous, see under Didaskalikos as the real or ontologically superior,
Allen, R.E. 163 see under ontological superiority
Annas, J. 7n.19, 18n.7, 21n.15, 25, 54–5, 94, belief
115n.9, 162n.12 and doxa, see under doxa
Antiphon 146 de re 193–4
Aristotle and propositions, see under propositions
cognition of like by like in 67–9 two kinds: opinion and judgment 140,
Distinct Objects epistemology in 19–45, 157–8, 206
65–7, 79–80 Benson, H. 7n.19, 59–60
doxa in, see under doxa Bobonich, C. 239n.7
empeiria in, see under empeiria Bosanquet, B. 7n.11, 22n.20, 41n.63, 94, 133,
epistêmê in, see under epistêmê 143–4, 144n.7, 204–5
interpretation of Plato 35–6, 69–70, 82–3, Boylu, A. 18n.7
101–3, 189 Boys–Stones, G. 37n.53, 46
ontological superiority (primary Bradley, F.H. 135–6
being) in 101–2 Broadie, S. 129–30
phronêsis (practical wisdom), Bronstein, D. 211n.8
see under phronêsis Brown, L. 18n.7, 94
powers, activities, and objects 15–17, Bryan, J. 129n.34, 147n.13
61–3, 65–7 Burnyeat, M. 7n.19, 90n.7, 91n.10, 94,
technê (craft) 127–8 115n.9, 129n.34, 169n.21, 220n.3,
Armstrong, D. 1n.3, 24n.23 226n.11, 227n.13
Arruzza, C. 123, Butler, T 18n.7
Baltzly, D. 18n.7, 20, 42, 182n.40, 201n.4 Carone, G. 63

Barnes, J. 13n.1 Carpenter, A. 21n.14, 117n.11
Barney, R. 237n.3 Carriero, J. 7, 47, 135
being Cave allegory 19–22, 31, 75, 84–5, 105–6,
contrasted with becoming 6, 31–2, 76–7, 124–5, 141, 156–89, 198, 240–2
97–8, 105–6, 162–4, 225, 234–5, 239 Cicero 36, 157
contrasted with seeming, see cognition, cognitive
under seeming as general terms covering doxa and
existential, predicative, veridical and other epistêmê 51
senses 94–6 like by like doctrine of in Aristotle 67–9
and epistêmê 86–93, 105–8, like by like doctrine of in Plato 69–79
see also under Forms Cooper, J.M. 94, 126n.27, 226n.11
Index 255
Cornford, F.M. 3n.5, 11–12, 17, 22n.20, 40–1, Ebrey, D. 71n.32, 72n.33
69–70, 90n.7, 94, 103n.37, 140n.1, 144–5, eikos (likely) 129–30, 147n.13
162, 202–3, 226, 228n.14, 230n.18 eikasia (imagination) 31n.33, 75, 157, 171–4
craft, see under technê Empedocles 68–70, 72–3
Crombie, I.M. 1n.2, 4n.6, 7n.16, 42, 54n.8, empeiria (experience) 125–7, 202, 204
134, 140n.1, 170n.23, 202–3 epistêmê
Cross, R.C. 20, 24, 231–2 and being, see under being
and definitions 222
Damascius 23n.21, 192 and essences 133–4, 209–10
Devereux, D. 127, 130–1 and eudaimonia 237–42
dianoia (“thought”) 182–90 and explanation 6, 115–16, 211–12
Didaskalikos 36–7, 203n.9, 230nn.18–19 and knowledge 2–8, 23, 42, 48, 84, 137–8
Dillon, J. 36 and metaphysics 135–6
Dionysius of Halicarnassus 45 and nous, see under nous
Distinct Objects epistemology and propositions, see under propositions
in Aristotle, see under Aristotle evidence and science 7, 41–2, 64–5, 122, 133–5
for in Plato 26–35, 52–61, 79–82 and Scientia 7, 36, 39–40, 47, 133, 135
in early and Middle Platonists 15–45 and understanding 7–8, 11–12, 115–16,
in Isocrates 14–45 133–6, 211
in medieval and modern philosophy 47 Basic Conception of 111–12, 136
in the Meno 214–17 counterparts in Parmenides and
in Neoplatonists 46, 155 Heraclitus 134
in Parmenides 37, 40, 42, 134, 174 historical interpretations of 7, 35–42, 133
in the Theaetetus 220–4, 230–3 in Aristotle 7, 44, 134–5, 211–12
doxa in Isocrates 14–45, 122
and atheoretical thought 199–202 in Plato’s time 13–14
and belief 2–8, 14–24, 42, 48, 84, 140–3, in the earlier dialogues 208–12
157–8, 205–6, 227–9 in the Theaetetus 90–2, 220–2,
and dreaming 197–9, 202 226–7, 232–3
and empeiria, see under empeiria nonstandard uses of the word 21n.16, 27,
and perception 54–5, 161, 169–71 51–75, 191
and propositions, see under propositions, recent interpretations of 2, 7–8, 48, 133–4
propositional attitudes synonyms for 1n.2, 27, 242
Basic Conception of 141–2, 196–7 epistemology, contemporary vs.
in Aristotle 44, 82n.48, 150, 229 Platonic 1–3, 234–5, see also under
historical interpretations of 7 belief and under knowledge
for doxa 35–42, 202–5 ethics
in Isocrates 45, 232n.21 Ancient and modern, as analogy to
in Parmenides 146, 151n.21, 205n.16 epistemology 8–10
in Platonists 7, 203–4 relation to Plato’s epistemology 236–42
in the earlier dialogues 213–14 Everson, S. 2n.4, 50n.1
in the Philebus and Sophist 228–9 explanation (aitia), see under epistêmê
in the Theaetetus 222–3, 227–9, 231–2
object vs. subject senses 145–6 Ficino, M. 39–40, 203
nonstandard uses of the word 21n.16, 27 Fine, G. 2n.4, 3n.5, 9n.23, 13n.2, 18n.7,
recent interpretations of 2, 7–8, 48, 202–3 20–1, 21n.16, 23, 25–6, 27n.25, 28, 30,
reflective vs. passive 171–4 34n.44, 44n.67, 53n.5, 78n.44, 88n.4,
true vs. false 155–7 93n.16, 94, 100n.31, 122–3, 125, 128,
two kinds 191–2, 227–9 137n.5, 138, 140n.3, 162n.12, 164–5,
and what seems 143–4, 149–53 180n.38, 187, 191n.56, 214, 232n.20
256 Index
Fine, K. 101–2 Lear, G.R. 145n.8

Forms Lee, D. 94n.17
and Being 19, 96–100, 103, 105–7, 225 Leibniz 7n.14, 47, 135n.4
and doxa 177–81, 190–5 Lesher, J. 116–17, 118n.13, 134n.2
and epistêmê 19–20, 27–34, 71–9, Locke 47
82–4, 117–22 Lyons, J. 27n.25
Frede, M. 135n.3, 169n.21, 204, 220n.2
McDaniel, K. 102n.34
Galen 204 metaphysics, relation to Plato’s
Gallop, D. 34n.44, 72n.35, 97n.23, 183, 189 epistemology 10, 82–4, 214–18
Geach, P.T. 210n.3 Modrak, D. 226n.11
Gerson, L. 11–12, 21n.14, 65, 116n.10, Moline, J. 2n.4, 11–12, 95n.20, 99n.30,
133–4, 182n.41, 183, 203n.9, 115n.9, 134
211n.7, 216n.15 Moravcsik, J. 7n.19
Gettier, E. 1n.3, 230n.17 Moss, J. 8n.20, 44n.71, 74, 137n.5, 142n.4,
Gonzalez, F. 18n.7, 21n.14, 29n.29, 123n.22, 162n.12, 171nn.25,27, 227n.13,
133–4, 183, 194n.62 229, 237n.3
Gorgias of Leontini 146, 149–50 Mourelatos, A. 129n.34, 147n.13
Gosling, J.C.B. 9n.23, 18n.7, 20, 42, 204–5 Murphy, N.R. 41, 181n.39
Grönroos, G. 171n.27, 229
Gulley, N. 7n.16, 11–12, 140n.1, Nehamas, A. 7n.19, 11–12, 115n.9,
143n.5, 204–5 163n.14, 211n.6
Nettleship, R.L. 204–5
Hackforth, R. 7n.11, 204–5 Nietzsche 162
Harte, V. 18n.7, 20, 25–6, 28n.28, 80, nomima (conventions) 19n.10, 124
162n.12, 193n.61, 194n.62, nous
202–3, 217n.16 in Plato 1n.2, 183–4
Havelock, E.A. 140n.1, 143–4, 204–5 in Aristotle 134–5, 212
Heraclitus 35, 68, 82, 134 in Parmenides and Heraclitus
Hicks, R.D. 62n.19, 69n.30, 71n.31 42, 134
Horwich, P. 135–6 in NeoPlatonists 39, 46
Hume 47 Nozick, R. 240
infallibility 4, 17, 123 Olympiodorus 23n.21, 183, 192

Irwin, T. 188n.47 ontological superiority 87–107, 134–6, 153,
Isocrates 44–5, 122–3, 232n.21 208–9, 223–7
in Aristotle, see under Aristotle
Joachim, H. 71n.31 and fundamentality 88–102
Johansen, T. 16n.4, 62 and reality vs. appearance 104, 110, and
Jowett, B. 7n.11, 22n.20, 41n.63, 133 see under seeming
and truth 114–15
Kahn, C. 21n.14, 94, 98, 109 opinion, see under belief
Kamtekar, R. 18n.7, 20, 25–6, 28n.28, Overlap readings (“One World”
63n.21, 80, 217n.16 epistemology) 14–26, 80–1
knowledge
by acquaintance vs. by description 231–2 Palmer, J. 42n.64, 151n.21
and epistêmê, see under epistêmê Parmenides 37, 40, 42, 46, 101, 103–4, 134,
and propositions, see under propositions 146–7, 151n.21, 174, 205n.16
Pasnau, R. 9n.22, 47n.77
LaFrance, Y. 181n.39 Peramatzis, M. 62n.18, 102n.35
Lautner, P. 39n.59, 46n.75, 203n.12 perception, see under doxa
Index 257
Perin, C. 117n.11 science, see under epistêmê

pistis (conviction) 27, 32, 129–30, 147n.13, Scientia, see under epistêmê
171–4, 176–7, 188n.50 Scott, D. 37, 46
Poulakos, T. 45n.72 Sedley 21n.14, 23n.21, 123n.22, 124n.23,
Prior, W. 210n.4 189, 189n.51, 230–1
phantasia (appearance) 147n.13, 169n.21, seeming
171n.27, 229 and doxa, see under doxa
phronêsis contrast with Being 146–9
as eqivalent to epistêmê 1n.2, 27, 33–4, ontological 145–6
72–3, 106n.42 and perceptibles 162, 165–74
in Aristotle 66, 130–1 Sextus Empiricus 45, 148, 204
and philosopher–rulers 130–1 Shorey, P. 7n.11, 19n.10, 41n.63, 133
philosopher–rulers 122–31 Simonides 148
pistis (conviction) 1n.2, 27, 31n.33, 32, 75–7, Sierksma–Agteres, S. 46
114n.2, 129–30, 147n.13, 157, 171–4, Silverman, A. 94, 99n.29, 130n.37,
176–7, 184n.44–45, 188n.50 133–4
Platonists: see under Alcinous, Damascius, Smith, N.D. 9n.23, 18n.7, 20, 21n.13, 25–6,
Ficino, Olympiodorus, Plotinus, 80, 121–3, 128, 181n.39, 186–8,
Proclus, Plutarch 193n.61, 202–3, 21nn.15–16
Plotinus 38, 46–7, 135 Snell, B. 1n.2
Plutarch 14–37, 46 sophia
power (dunamis) Spinoza 7n.14, 47, 135n.4
and objects and accomplishments Sprute, J. 7n.16, 11–12, 140, 140n.1, 143n.5,
52–80 204–5, 228
in Aristotle 15–16, 61–9 stability 71–2, 76, 98, 116–18, 120, 158,
powers argument (Republic 218, 223–4
476e–480a) 14–19, 28–9, 52–6, 60–1, Stokes, M. 18n.7, 94, 200, 201n.4
86, 88–90, 95–6, 100–1, 121, 133, Storey, D. 169n.21
144, 197–8 summoners argument (Republic
practical wisdom, see under phronêsis 523a–524d) 164–5, 175–7
precision (akribeia) 71–2, 75–6, 116–18, Szaif, J. 9n.23, 18n.7, 20, 21n.13, 50n.1,
120, 129 74n.39, 96n.22, 114n.4,
Proclus 7, 20, 135, 203 119n15, 181n.39
Distinct Objects epistemology 46
interpretation of Plato 38–9, 70 Taylor, A. 7, 133
propositions, as relata of epistêmê, doxa, Taylor, C.C.W. 18n.7, 94n.17, 181n.39
knowledge and belief 9n.23, 20–1, technê (craft) 52–3, 59–60, 75–6, 81, 126–8,
23n.22, 28–30, 43–4, 66n.26, 84, 93–4, 191, 201–2
138, 140–1, 205–6, 235 truth 113–15, 155–7
Two Worlds epistemology – see under
reality, see under ontological Distinct Objects epistemology
superiority Two Worlds dialogues 19–20
rhetoric 59–60, 151–2, 158–60, 231–2
Rowe, C.J. 31n.36, 133n.1 understanding, see under epistêmê
Rowett, C. 8n.21, 133n.1, 204–5
van Inwagen, P. 102n.34, 104n.38
Schofield, M. 205n.17 Vasiliou, I. 126n.27
Schwab, W. 11–12, 13n.1, 31n.36, Vlastos, G. 7n.17, 21n.14, 94,
44n.71, 7nn.19–20, 115n.9, 116, 95n.20, 115n.5, 116n.10,
129, 133, 161–2, 211n.6, 229, 118n.14, 163n.14,
142nn.4–5 236n.2
258 Index
Vogt, K. 8n20, 18n.7, 20, 22n.17, 23–6, 28n.28, White, N. 11–12
41, 140nn.1,3, 143n.5, 144n.7, 148n.16, Wolterstorff, N. 7n.15, 135n.4
157n.4, 161–2, 180nn.37,38, 182n.40, Woodruff, P. 7n.18
187, 202–3, 216n.15, 228 Woolf, R. 21n.14, 182n.41
von Fritz, K. 134n.2 Woozley, A.D., see under Cross
Wedgwood, R. 191n.55, Xenophanes 147, 150

240n.9
White, F.C. 91n.9 Zeller, E. 40

Moss - Plato's Epistemology. Being and Seeming

Uploaded by

Copyright:

Available Formats

Moss - Plato's Epistemology. Being and Seeming

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

Moss - Plato's Epistemology. Being and Seeming

Uploaded by

Copyright:

Available Formats

Plato’s Epistemology

That correct doxa is something different from epistêmê is

Throughout the dialogues Plato contrasts a superior kind of cognition with

One major presupposition of this answer, sometimes made explicit, is

theories of knowledge and belief to guide or constrain our interpretations of

despite describing it with varying terminology, and despite sometimes

1. Epistêmê and doxa versus knowledge and belief

• Epistêmê but not doxa is infallible or unerring, always true (Republic

• Epistêmê—according to some dialogues, although not others—is built

The fact is however that quite a number of Plato’s epistemological views

• Epistêmê requires reasoning out explanations (Meno 98a).

inherently deficient cognition,20 or cognition of particular tokens rather

2. The plan

Abandoning that identification in favor of any alternative, however, will

instead on the notion of happiness. A theory developed to address the

22 For this interpretation of Aristotle’s epistemology, see Pasnau 2013.

epistemological project which shows it to be not a muddled or nascent

• There is a crucial metaphysical distinction between two levels of

If one comes with this metaphysical and ethical background to epis­temo­

• Epistêmê and doxa are each essentially to be understood as cognition

bolster them by showing their place in an overall interpretation of Plato’s

(466e–468d). Whether he means literally to redefine certain words, or

also translate “abilities,” “capacities,” “faculties”), and presents a general

1. The Two Worlds debate

Let us begin with a brief overview of the Republic’s powers argument.

(1) Epistêmê and doxa are distinct powers (dunameis) (477d–e).9

So much is explicit. As to the interpretation, however, there is a raging

If this interpretation of the powers argument is correct, then Plato is

cognitive powers have distinct objects.11 I will therefore refer to the

prop­os­itions: epistêmê to the set of true propositions, doxa to the set of

We can refer to these and many related interpretations as Overlap readings,

paradoxical. Conversely, Socrates and his interlocutors evidently have many

17 See especially Vogt 2012.

of knowledge, sometimes called ‘opinion.’ Beliefs of both kinds can in

[I]f, without any preconceived theories, we consider knowledge and belief,

The thought is frequently echoed; for example:

even though other knives can be used to prune (Republic 353a).24 On a

2. The prima facie case for Distinct Objects

I think the answer is clear: it is the natural, simple, straightforward reading

First, there is the much-discussed passage from Republic V with which

[W]e clearly agree that epistêmê is different from doxa?—Different.—So each

There is much to be said about whether or not these interpretations are

Strategies for deflating the ‘over’ of Republic V as describing a mere norm or

Doxa is about [peri] becoming, noêsis34 is about being. (Republic 534a).

These look like outright denials that epistêmê of perceptibles is possible—

Later in the dialogue we get a distinction between Forms:

which it falls to noêsis to investigate, and the other kind . . . per­

These claims doubtless could be interpreted as describing mere norms, but

Cognition of intelligibles is by nature different from cognition of per­cep­

This looks to be a statement of a Distinct Objects view. Here the contrast is

40 τῷ τῆς διανοίας λογισμῷ.

Whenever it attends instead to shifty perceptibles, investigating things

3. A brief history of the Distinct Objects interpretation

Aristotle evidently thinks that Plato holds both an extreme metaphysical

Might Aristotle have intended a subtle claim compatible with an Overlap

Again there are no qualifications or subtleties. Perhaps ‘concerned with’

Or suppose you turned for illumination to the Neoplatonists. Far from

Moreover, each kind of cognition is similar in character to its cor­res­pond­

when it directs itself toward the contemplation of intelligibles uses both

What then is Scientia? The comprehension of divine things by certain rea-

If one comes with this metaphysical and ethical background to epistemo

propositions: epistêmê to the set of true propositions, doxa to the set of

which it falls to noêsis to investigate, and the other kind . . . per

Cognition of intelligibles is by nature different from cognition of percep

Moreover, each kind of cognition is similar in character to its correspond

On a popular interpretation, Parmenides holds a Distinct Objects epistem

principle of like-by-like.31 Moreover, the doctrine applies both to powers