This page intentionally left blank

Descartes’ Cogito
Saved from the Great Shipwreck

Perhaps the most famous proposition in the history of philosophy is

Descartes’ cogito, “I think, therefore I am.” Husain Sarkar claims in this
provocative new interpretation of Descartes that the ancient tradition
of reading the cogito as an argument is mistaken. It should, he says, be
read as an intuition. Through this new interpretative lens, the author
reconsiders key Cartesian topics: the ideal inquirer; the role of clear
and distinct ideas; the relation of these to the will; memory; the nature
of intuition and deduction; the nature, content, and elusiveness of
‘I’; and the tenability of the doctrine of the creation of eternal truths.
Finally, the book demonstrates how Descartes’ attempt to prove the
existence of God is foiled by a new Cartesian Circle.

Husain Sarkar is Professor of Philosophy at Louisiana State University.

 Descartes’ Cogito
Saved from the Great Shipwreck

HUSAIN SARKAR
Louisiana State University
  
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press

The Edinburgh Building, Cambridge  , United Kingdom
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521821667

© Husain Sarkar 2003

This book is in copyright. Subject to statutory exception and to the provision of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2003

-
isbn-13 978-0-511-07035-8 eBook (EBL)
-
isbn-10 0-511-07035-7 eBook (EBL)

-
isbn-13 978-0-521-82166-7 hardback
-
isbn-10 0-521-82166-5 hardback

Cambridge University Press has no responsibility for the persistence or accuracy of

s for external or third-party internet websites referred to in this book, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
 For
Ashifa
part ballerina, part lover of children, my daughter

Reason with them in the most courteous manner.

(The Koran, Chapter 16, “The Bee,” verse 125)
What are you saying? You know? Is this certain and beyond all
doubt? Is this the sole surviving timber from the great shipwreck
that is to be hung up as an offering in the temple of truth?
Pierre Bourdin, “Seventh Set of Objections,” in The
Philosophical Writings of Rene Descartes, Volume 2, 317.
 Contents

Preface page ix
Acknowledgments xiii
Abbreviations xvii

1 The Prolegomena to Any Future Epistemology 1

I The Making of an Ideal Seeker 3
II The Method: The Rationalist Thread 10
III The Tree of Philosophy 16
IV Method, Morals, and Bootstraps 22
2 The Problem of Epistemology 33
I Types of Problems 34
II Directive to Dismantle 39
III Two Models of Doubt 45
IV Doubt and Principles 50
3 The Solution: Cogito 58
I The Nature of the First Principle 60
II The Thought Experiment 65
III The Experiment Evaluated 80
IV The Eucharist Objection 83
V Doubt and the Cogito 86
VI The General Rule and Truth 91
4 A Skeptic against Reason 101
I Why Natural Reason? 102
II “Buy All or Nothing” 109
III Attempting to Step Out of the Circle 116

vii
viii Contents

IV No Escaping from the Circle 121

V Another Failed Attempt 126
VI How Not to Read the Meditations: A Skeptic’s Reply 130
5 The Five Ways 138
I The Five Ways. . . 139
II . . .Plus One 168
6 Cogito: Not an Argument 176
I The Preliminaries 177
II The Core of the Claim 181
III The Proof 186
IV Skepticism and the Theory of Deduction 195
7 The Content of the Cogito 209
I A Source of the Mistake 209
II The Content of the Cogito 214
III Ryle and the Elusive ‘I’ 226
8 Memory, Explanation, and Will 236
I The Role of Memory 236
II Discovery, Explanation, and the New Logic 240
III Will, Cogito, and the Purposes of God 249

Appendix A: Comments on Jeffrey Tlumak’s “Certainty

and Cartesian Method” 269
Appendix B: Comments on Robert Nozick’s “Fiction” 273
Appendix C: Cogito and the Port-Royal Logic 278
Appendix D: Bacon and Descartes 284
Appendix E: Comments on Anthony Kenny’s “Descartes
on the Will” 287

Bibliography 294
Name Index 299
Subject Index 302
 Preface

Rene Descartes offered an ultimate truth, famously known as the cogito.

But there have been virtually no takers. Some have thought that it
was merely an analytic statement, a statement empty of content; some
have thought that he should have begun with a less complex, a less
unwarranted statement (but then he could not have derived the cogito
from it); some have averred that it was an argument that was badly
in need of repair (and when repaired, one that could not possibly do
what Descartes had wanted it to do); and a distinguished philosopher
once argued that the truth of the cogito, if that is what it is, is at best
odd, “degenerate.” This has been the litany for 350 years.
Here is Descartes in Discourse on the Method:

But immediately I noticed that while I was trying thus to think everything false,
it was necessary that I, who was thinking this, was something. And observing
that this truth, “I am thinking, therefore I exist” was so firm and sure that all the
most extravagant suppositions of the skeptics were incapable of shaking it,
I decided that I could accept it without scruple as the first principle of the
philosophy I was seeking.

Then, in his “Replies to the Second Set of Objections”:

When we observe that we are thinking beings, this is a sort of primary notion,
which is not the conclusion of any syllogism; and, moreover, when somebody
says: I am thinking, therefore I am or exist, he is not using a syllogism to deduce
his existence from his thought, but recognizing this as something self-evident,
in a simple mental intuition.

ix
x Preface

In effect, Descartes is saying both that the cogito is an argument (read-

ing the ‘therefore’ as a conclusion indicator) and that it is not. There
is simply no interpretation of Descartes that will not fail to explain at
least some of the text. One might as well try to square a circle.
Consequently, in this book I have attempted to do the next best
thing, to provide an interpretation that will save as much as possible of
what is profound and interesting in Descartes. There are two sides to
the cogito debate: what it is and what it is not. There are novel and fairly
conclusive reasons why the cogito cannot be construed as an argument.
I offer these and thereby show what the cogito is not. A great deal of
what Descartes has said militates against construing the cogito as an
argument. It was with some of those reasons that this book was begun.
But if the cogito is not an argument, what then is it? It is an experi-
ment. When that experiment is conducted, the mind intuits – learns
through an insight – the truth that the cogito expresses. The exper-
iment will teach an individual thinker by example what he cannot
learn by relying on someone else’s experiment, or on that of a col-
lection of individuals, no matter how well conducted the experiment
and precise the reporting. However, it is an experiment that can be
performed only after the mind has been cleansed and prepared. Now,
these two claims are, of course, independent. Whether the view that
the cogito is an experiment, to be performed by an individual thinker,
is true has no implication for whether the cogito is an argument. In fact,
I think that the negative claim offered here, namely, that thecogito is
not an argument, is decisive. But that is not so with the positive thesis
as to what the cogito is.
The investigation into the cogito, this first principle, led me to un-
cover the basic structure of Descartes’ method. I wanted to focus not
on the details but on the essence of his method. For I was certain
that the cogito and the method were too intimately linked to be under-
stood one without the other. At any rate, getting the method right was
a prerequisite to being right about the cogito. The focus on method
and cogito inevitably led to other issues in Descartes, such as logic and
skepticism, will and memory, discovery and explanation, reason and
experience, eternal truths and the general rule. I have explored none
of these as fully as I might have. I have had a more limited aim: In
each instance, the scope of the inquiry was sharply restricted by what
it had to contribute to the cogito and the method. Thus, I have tried to
 Preface xi

bring in a sequence of distinct issues in Descartes and to balance and

counterbalance them. The result, I trust, is a fresh perspective.
Montaigne wrote an essay entitled, “That Our Happiness Must Not
Be Judged until After Our Death.” Borrowing the same theme from
Ovid, I propose a principle of reconstruction and call it the Sulmo prin-
ciple, namely, that a philosopher’s system must not be reconstructed
and judged until after his death. For what he once assumed, he may
later radically question (as did Descartes on the certainty of mathema-
tics); what he once sketched, he may in time expand in great detail (as
did Descartes on the unity of the sciences); what he once wrote, he may
subsequently abrogate (as did Descartes, virtually, on the immortality
of the soul); what he once wrote, he may remain silent about for the
rest of his life (as did Descartes on the taxonomy of problems). Only
after his death can one say that, if he had tied his views together at
the end into a single consistent system, he would have done so in this
way or that; only then can one judge the worth of the system. When
attempting to reconstruct, there are enough avenues to explore, in
harmony with the philosopher’s general view. This allows for much
guesswork. The whole, and only the whole, reconstructed at the end,
should be judged for historical plausibility and philosophical sound-
ness. At any rate, it is the principle of Sulmo that I have used in this
book.
The ideal author of a book on Descartes would be someone who
has mastered all of the French and Latin texts of Descartes as they cur-
rently exist in Charles Adams and Paul Tannery, Oeuvres de Descartes,
and in F. Alquié, Oeuvres philosophiques de Descartes. He would be
familiar with all of the responses of Descartes’ contemporaries, and
with the commentaries from the seventeenth century, such as Tepelli’s
Historia Philosophiae Cartesiane; with eighteenth-century works, such as
Rousseau’s The Profession of Faith of a Savoyard Priest; and with items from
the nineteenth century, such as Duboux’s La Physique de Descartes, down
to the commentaries of the twentieth century. Not only would he have
kept track of all these items, their moorings and interconnections,
and how each of them measures up to the original, he would also be
familiar with all of the recent work in metaphysics, philosophy of mind,
philosophy of religion, philosophy of science, history of physics, and
mathematics and logic, so that he could determine how far ahead of
Descartes we are in a given field, or how far behind. Then, and only
xii Preface

then, would he write a book on Descartes. I fall miserably short of that

ideal.
This book is not in the same line as the works that sweep over a
vast range of the philosophy of Descartes, such as those of Baker and
Curley; Gaukroger, Kenny, and Kemp Smith; Wilson and Williams. It
is a commentary on Descartes’ Meditations, which he called “my little
book dealing with First Philosophy.” It is, moreover, a commentary
on only the first few pages. Such, I think, is the power of his views.
I have strenuously aimed to be fair to the history of philosophy, and
struggled not to be silly in any anachronistic way: In short, I have tried
to argue with Descartes as if I were his contemporary, not he mine.
This, in part, explains the frequent citation of page and passage from
Descartes. My ultimate aim has been to reconstruct Descartes’ ideas in
a manner that demands that we see Descartes’ achievement for what it
is: He achieved in his field what Archimedes had only dreamed about
in his.

Baton Rouge, Louisiana

February 1, 2002
 Acknowledgments

In the fall of 1989 I went to England with the oldest version of this
book, commenced in November or December of the year before,
and returned in the spring of the following year with lots of notes
and marginalia for the making of a better one. The result of sub-
sequent attempts at doing just that is here. For that result, and the
trip across the Atlantic and back, I am grateful to two institutions:
Louisiana State University, for granting me sabbatical and other leave,
and the University of Cambridge, for electing me a Visiting Scholar
for 1989–90. But nearly as much is owed to the commons flanking
Victoria Avenue, with a narrow part of the River Cam providing the
margin, where I spent long hours, day and night, and did my best
thinking, such as it was, under the rarely clear English sky. The second
version was completed at the end of September 1992.
Then, for several years, I turned to work on other things (while in-
termittently thinking about and teaching Descartes). Nine years later,
in February 1998, I went to the University of Oxford as a Faculty Visitor
for the Hilary Term. I owe a debt not only to the University of Oxford
but also, once again, to Louisiana State University for making this pos-
sible by granting me sabbatical leave. This time I went with a somewhat
better version of the book, in hopes that good things might come from
my visit. They did. Anthony Kenny provided written comments; he was
encouraging in several details, found the second version better than
the skimpy first, but remained steadfast in his belief that the cogito is an
argument. I mulled a long time over his detailed written comments.

xiii
xiv Acknowledgments

I then rewrote the relevant material to show that his view is confronted
by some very serious difficulties. Of my readers, he was the only one
to clearly perceive that, among other things, I was trying to show that
a continuum can be seen, or forged, between the early and later work
of Descartes; he is not entirely convinced by it, although he was kind
enough to say that I have made a serious case for it.
Gordon Baker not only proved to be a very kind host but also,
through many long conversations on the manuscript, led me happily
to believe that perhaps there is something to the central thesis of this
book; nevertheless, I reconsidered several of my arguments in light
of his remarks and upon reading his own book on Descartes (which
he wrote with Katherine J. Morris). In the final analysis, I have al-
ways relied on his written, rather than his spoken, word. (Oxford’s
best gift to me, however, was the discovery of the novels of Jim
Crace.)
I am grateful to Katholieke Universiteit Leuven for the invitation to
participate in their Thursday Lecture series. In October 1998 I read a
paper there entitled “In Defense of Skepticism,” an early draft of what
now sits as sections II–IV of Chapter 4. I started work on the penul-
timate version in May 1999, furiously working for about five months,
doing minor editing work for several weeks thereafter, and finishing
in January 2000. The final version was started in late July 2001 and
ended in January 2002.
I am grateful to Richard A. Watson for his generous encourage-
ment. Alas, even what he approved I had to remove for lack of space.
Although he agreed substantially with the core thesis, John Compton’s
trenchant and detailed written criticisms – especially on the self and
self-identity – left little doubt about what he thought of Descartes’ pos-
itive thesis. Had I even fragmentary knowledge of the work of Edumnd
Husserl, I know he would have found my responses better informed
and my subsidiary theses less objectionable.
Jeffrey Tlumak commented in the margins of the manuscript, vir-
tually on every third page. I refrained from answering every one of
his objections, or the book would have been twice as long. I have an-
swered, at least to my own satisfaction, several of his objections; several
others have remained unanswered simply because I did not know how
to accomplish that within the confines of the allotted space. The force
of his other objections I have duly acknowledged. I owe him additional
 Acknowledgments xv

gratitude because he provided me with his painstaking comments at a

time when he was passing through an incredible hardship.
Of the anonymous referees for Cambridge University Press, I single
out three: for their old-fashioned, easily discernible, Kantian goodwill;
and for resonating with the central thesis of this book and saving it from
a thousand infelicities, thus making it a better book, both in form and
substance – I thank them for that favor, too. Whatever appeal this
book may possess by way of artwork and diagrams is owed entirely to
my son, Casim Ali. I am grateful as well to Russell Hahn for his superb
copyediting. Without the persistence – I almost said protection – of
Terence Moore, the editor, this book would have seen the light of day,
but under a very different seal.
Last and greatest is my debt to Catherine Wilson. She has been a
provider of innumerable corrections, a host of tough philosophical
questions, and acute analysis; she has also been a dispenser of encour-
agement. No doubt this manuscript is far less objectionable than it
would have been otherwise, had she not given it her time and exper-
tise. Indeed, the title of the book stands a bit modified, thanks to her.
For my ending I will adapt one of Machiavelli’s beginnings and
address my benefactors collectively thus: “With this I send you a gift. I
have endeavored to embody in it all that long experience and assiduous
research have taught me. But you may well complain of my lack of
talent when my arguments are poor, and of the fallacies of my judgment
on account of the errors into which I have doubtless fallen many times.
This being so, however, I know you have a right to complain that I
should have written without giving you cause to be satisfied, which, if
it bears no proportion to the extent of the obligations that I owe you,
is nevertheless the best I am able to offer you.”
 Abbreviations

AN Antoine Arnauld and Pierre Nicole, Logic or the Art of Think-

ing. Translated and edited by Jill Vance Buroker. Cambridge:
Cambridge University Press, 1996.
AT Charles Adam and Paul Tannery, editors, Oeuvres de Descartes.
Volumes 1–12, revised edition. Paris: Vrin/CNRS, 1964–76.
C Edwin M. Curley, Descartes against the Skeptics. Cambridge,
Mass.: Harvard University Press, 1978.
CSM The Philosophical Writings of Descartes, Volumes I and II. Trans-
lated by John Cottingham, Robert Stoothoff, and Dugald
Murdoch. Cambridge: Cambridge University Press, 1984.
CSMK The Philosophical Writings of Descartes, Volume III. The
Correspondence. Translated by John Cottingham, Robert
Stoothoff, Dugald Murdoch, and Anthony Kenny. Cambridge:
Cambridge University Press, 1991.
G Stephen Gaukroger, Descartes: An Intellectual Biography. New
York: Oxford University Press, 1995.
K Anthony Kenny, Descartes: A Study of his Philosophy. New York:
Random House, 1968.
M Michael de Montaigne, An Apology for Raymond Sebond. Trans-
lated and edited by M. A. Screech. New York: Penguin, 1987.
R Genevieve Rodis-Lewis, Descartes: His Life and Thought. Trans-
lated by Jane Marie Todd. Ithaca, N.Y.: Cornell University
Press, 1999.
V Jack Rochford Vrooman, Rene Descartes: A Biography. New York:
G. P. Putnam’s Sons, 1970.
W Bernard Williams, Descartes: The Project of Pure Inquiry. New
York: Penguin, 1978.

xvii
xviii Abbreviations

WC Bernard Williams, “The Certainty of the Cogito.” In Willis

Doney, editor, Descartes: A Collection of Critical Essays. London:
Macmillan, 1968, pp. 88–107.
WM Margaret D. Wilson, Descartes. New York: Routledge and Kegan
Paul, 1986.
 Descartes’ Cogito
Saved from the Great Shipwreck
 1

The Prolegomena to Any Future Epistemology

In 1628, Rene Descartes received an invitation to a meeting at the

home of Cardinal Bagni, papal nuncio. Descartes brought with him
Father Mersenne, a Minim friar, and M. de Ville-Bressieu, a physician
of Grenoble. This was no ordinary meeting. It consisted of well-known
honnets gens of Paris. They had met to hear a famous doctor-chemist
by the name of Chandoux. Chandoux was an expert on base metals
who three years later was to be executed for peddling fake currency.
Chandoux, charming and fluent, was denouncing the verbiage of
scholastic philosophy as it was usually taught in the Schools. There
was little new in what he said, for it was mostly in the vein of Francis
Bacon, Pierre Gassendi, and Thomas Hobbes. Yet he wanted his sys-
tem of philosophy to appear fresh and novel. Whatever Chandoux
said, everyone applauded. That is, everyone save Descartes.
The founder of the oratory, and perhaps the most powerful religious
thinker of the Counter-Reformation, Cardinal Berulle, observed this.
He asked Descartes what he thought of Chandoux’s speech that had
so thrilled the audience. Descartes demurred, saying “that he could
not speak in opposition to the feeling of the savants present.”1 But the
Cardinal did not relent. At last, Descartes spoke. He began by prais-
ing Chandoux’s denunciation of scholastic philosophy. But then he
argued against the speaker and “that great and learned company” for

1 Elizabeth S. Haldane, Descartes: His Life and Times, 108. The details of the references
are given in the bibliography.

1
2 The Prolegomena to Any Future Epistemology

taking probability as the central notion and not the notion of truth. If
one were satisfied with something merely probable, he argued, then
one could easily take false statements to be true and true statements to
be false. As evidence, he asked that someone in the audience propose
what he deemed to be an incontestable truth. Someone volunteered,
and Descartes proceeded to show in twelve arguments, relying on the
notion of probability, that the proposed statement was false.2 He then
asked that someone propose a statement that he took to be incon-
testably false. Once again, with reasoning by probability as his guide,
he showed the statement to be true. He thus demonstrated that our
minds can become victims of the notion of probability. The audience
was duly stunned, and some openly deserted Chandoux on the spot.
The savants begged to know if there was a method, “some infallible
means to avoid these difficulties.” Descartes replied that there was his
own method. “I made the whole company recognize what power the
art of right reasoning has over the minds of those who have no learning
beyond the ordinary, and how much better founded, and more true
and natural, my principles are than any of those which are currently
received in the learned world” (CSMK, 32; AT I, 213). Such a method
would be useful not only in metaphysics, but also in mechanics and
medicine. Cardinal Berulle, whom the young philosopher met with
privately shortly afterward, was impressed beyond words. With the full
weight of his ecclesiastical authority, he urged Descartes to write and
publish his views, on the ground that he, Descartes, “was responsible
to God for giving to mankind what had been delivered to him.”3 Thus
was born, some nine years later, Discourse on the Method – and with it,
the history of modern philosophy.

2 On October 5, 1637, Descartes wrote to Father Mersenne, complaining that Fermat

had misunderstood him: “He thought that when I said that something was easy to
believe, I meant that it was no more than probable; but in this he has altogether mis-
taken my meaning. I consider almost as false whatever is only a matter of probability;
and when I say that something is easy to believe I do not mean that it is only probable,
but that it is so clear and so evident that there is no need for me to stop to prove it.”
(CSMK III, 74; AT I, 450–451)
3 Elizabeth S. Haldane, Descartes: His Life and Times, 110. Alas, the private meeting with
Cardinal Berulle – Haldane undoubtedly got it from Adrien Baillet’s (1649–1706)
La Vie de Monsieur Descartes, the first biography of Descartes – has been contested by
Genevieve Rodis-Lewis in her marvelous book Descartes: His Life and Thought. (See
R, 67–69 and 240, note 21, for further details on this episode.)
 I The Making of an Ideal Seeker 3

Granting the possibility of knowledge, what kind of person can

pursue and possess knowledge? Descartes thinks that only a certain
kind of person can, or at any rate should, embark on the pursuit of
knowledge and come to possess it. Section I of this chapter delineates
the making of such an ideal knower, who should be armed with a
method in his pursuit, like a traveler who ought to carry a map on his
journey. Section II provides just such a rationalist method. Section III
presents Descartes’ famous tree of philosophy: This is Descartes’ view
of what the completed structure of science would look like. Finally,
section IV presents the moral code a pursuer of knowledge should
abide by, and I raise the question of whether Descartes is attempting,
in this endeavor, to raise himself by his own bootstraps.

I. The Making of an Ideal Seeker

It is our modern liberal view that anyone, man or woman, of any station
in life, can embark on studying any discipline, at any time, and at
any place, and that what he or she learns will depend on how hard
he or she works. There are no other restrictions. This view was not
always held. Descartes, for instance, did not hold it. He thought not
only that it was necessary for a person to possess certain intellectual
and emotional qualities, but also that he had to undergo an initial
period of preparation before he could finally embark on a strenuous
philosophical inquiry.
Descartes became aware only very slowly of the problem of the ideal
seeker. In Rules for the Direction of the Mind, composed around 1628
and published posthumously, Descartes was hardly aware of the prob-
lem, even though he had said, “Where knowledge of things is con-
cerned, only two factors need to be considered: ourselves, the knowing
subjects, and the things which are the objects of knowledge” (CSM I,
39; AT X, 411). Descartes had scarcely said anything in this work about
the knowing subject as an ideal inquirer. But in Discourse on the Method,
first published anonymously in 1637, he was quite interested in that
problem. That issue was shelved, or at best the solution presupposed,
when he came to write the Meditations on First Philosophy, published in
1641. It was once again in the limelight in the unfinished dialogue
The Search for Truth, composed, according to one authority, sometime
4 The Prolegomena to Any Future Epistemology

during the last seven years of his life.4 (There is an interesting parallel
in his treatment of mathematics. Descartes assumed the truthfulness of
mathematical statements without question in theRules for the Direction
of the Mind, but in subsequent works, such as the Meditations on First
Philosophy, he felt he could no longer make that assumption and tried,
as we know, to justify even those truths.)5
My aim in discussing this issue is threefold. First, I want to give
prominence to a historical issue that has been cast aside, if occasionally
noticed. Second, I want to focus afresh on the problem of the reliability
of reason. Third, and far more importantly, I hope to show that a
proper understanding of the nature of the ideal seeker in Descartes
will provide us with one powerful argument, among others, in defense
of the central thesis of this book.
From Sextus Empiricus to Michele de Montaigne, the problem of
the ideal seeker is hardly in the background. These philosophers had
concerned themselves with the problems and pitfalls facing an ordi-
nary seeker. Concerned as he was to respond to the skeptic, it is scarcely
surprising that Descartes should have said much that revolved around
this topic – although it is surprising that he never explicitly discussed
the issue, by this name or any other. In what follows, I am clearly offer-
ing a reconstruction, namely, a systematic reconstruction of an answer,
based on the Cartesian texts, to the question, “How is an ideal seeker
made?,” as if Descartes had explicitly chosen to ask and answer that
question.
An ideal seeker after truth has to pass through four stages.6 The first
stage consists of his “original state of ignorance” (CSM II, 413; AT X,

4 For other conjectures see R, 196–197, note 6.

5 See C, 35–38. That Curley overstates the case by underplaying the method presented
in Rules for the Direction of the Mind and overplaying the method presented in the
Discourse on the Method, and in subsequently published works, does not detract from
what he says about Descartes’ evolving view of both method and mathematics.
6 Since this is admittedly a reconstruction, my primary task is to invite the reader to
consider not only whether Descartes clearly delineated the four states, but also his
thinking that the ideal seeker passes through these states as if they were stages in a
progressive order. Descartes did not explicitly develop the notion of an ideal seeker
and put it to epistemic use, nor did he take a stand, for or against, on a progressive
order of such states, since he did not treat this issue explicitly. However, there is some
historical evidence to suggest that the proposed reconstruction is not entirely alien
to Descartes’ philosophy; indeed, it might be seen to play a vital role in it. See, for
example, the final chapter of this book, pages 266–267 and notes 31 and 32.
 I The Making of an Ideal Seeker 5

519).7 Initially, everyone belongs in this group. Out of this group are
sifted those desiring to be seekers after truth from the others who have
no such desire; given their dispositions, the nonseekers are unsuited
for the philosophical task. This constitutes the second stage. From
the group of those desiring to be seekers are distinguished, on the
basis of certain right qualities, potentially ideal seekers from those
who are not. This is the third stage. These potentially ideal seekers
have finally to undergo preparation – study and reflection – in the
fourth and last stage, as a way of making them ideal seekers before
actually commencing the philosophical task.
The first stage, the original state of ignorance: “[A]s regards reason or
sense,” says Descartes, “since it is the only thing that makes us men and
distinguishes us from the beasts, I am inclined to believe that it exists
whole and complete in each of us” (CSM I, 112; AT VI, 2).8 Then
it would appear that anyone, at the start, is fit for the task of philo-
sophical inquiry; but there are hindrances. Each normal person, at
birth, has the senses of taste, smell, touch, sight, and hearing fully and
dominantly functioning in him; reason, at this point, plays a small
and subservient role. Here commences the growth of “the first
obstacle” (CSM II, 406; AT X, 508). For the senses are essentially
imperfect: They often deliver false reports about the external world;
our inclinations are quite corrupt, our nurses foolish; our appetites
and teachers are opposed, our instincts blind. Thus, we are all in the
original state of ignorance, and the problem is how to emancipate
ourselves from it so that we may become fit truth seekers.
The second stage, the stage of sifting: There are two types of indivi-
duals – “types of minds” – who are clearly unsuited for philosophical
inquiry.

First, there are those who, believing themselves cleverer than they are, cannot
avoid precipitate judgements and never have the patience to direct all their

7 In the letter of February 27, 1637, to Mersenne, Descartes wrote, “I was afraid that
weak minds might avidly embrace the doubts and scruples which I would have had to
propound and afterwards be unable to follow as fully the arguments by which I would
have endeavoured to remove them. Thus I would have set them on a false path and
been unable to bring them back.” (CSMK, 53; AT I, 350) Scholars are not agreed on
the exact date of this letter.
8 “[F]ew,” wrote Descartes to Mersenne on October 16, 1639, “are capable of under-
standing metaphysics.” (CSMK, 65; AT II, 596) It must follow that the final group of
inquirers after truth would be inordinately small.
6 The Prolegomena to Any Future Epistemology

thoughts in an orderly manner; consequently, if they once took the liberty of

doubting the principles they accepted and of straying from the common path,
they could never stick to the track that must be taken as a short-cut, and they
would remain lost all their lives. Secondly, there are those who have enough
reason or modesty to recognize that they are less capable of distinguishing
the true from the false than certain others by whom they can be taught; such
people should be content to follow the opinions of these others rather than
seek better opinions themselves. (CSM I, 118; AT VI, 15)

In short, none of these men are “of a fairly robust intellect” (CSM II,
320; AT VII, 475).
Descartes’ fear of losing an individual in the morass of doubt was a
genuine one. For him, knowledge was a guide to action, and actions
were necessary to the making of a good person. Thus, ignorance and
confusion could easily produce poor or evil deeds. Even a good method
could produce, in someone incompetent, a bad person. This result
must be avoided at all cost. For learning is of secondary importance in
comparison to good deeds.

A good man is not required to have read every book or diligently mastered
everything taught in the Schools. It would, indeed, be a kind of defect in his
education if he had spent too much time on book-learning. Having many
other things to do in the course of his life, he must judiciously measure out
his time so as to reserve the better part of it for performing good actions – the
actions which his own reason would have to teach him if he learned everything
from it alone. (CSM II, 400; AT X, 495–496)

The moral risks are plainly too high for anyone who is incompetent to
embark on the kind of enterprise Descartes has in mind.
Who, then, is fit for the philosophical task? I am attempting to
search for minimal conditions or qualities that a person must possess,
in Descartes’ view, in order to perform that task; anyone who possesses
anything more is more than qualified. In short, I am looking for neces-
sary conditions, jointly adding up to a sufficient condition, that would
make a person an ideal seeker.
The third stage, the stage of determining the right qualities: The ideal
seeker must be someone of at least average intelligence, who has
reached the age of discretion, whose senses are in good condition,
who is blessed with a modicum of insight and has common sense; this
eliminates the necessity of having gone to School (and thus having
 I The Making of an Ideal Seeker 7

received training in grammar and logic). Peter Ramus, whose logical

system Descartes had studied, had defined such a person as a syllogistic
reasoner, and not just as a reasoner. So in Ramus’ view an ideal seeker
would be essentially equipped with syllogistic reasoning. Not so for
Descartes; he maintained that he had never presumed his own mind
“to be in any way more perfect than that of the ordinary man” (CSM I;
111, AT VI, 2).9 Descartes would have been quite pleased with John
Locke’s remark that “God has not been so sparing to men to make
them barely two-legged creatures, and left it to Aristotle to make them
rational, i.e., those few of them that he could get so to examine the
grounds of syllogisms.”10
The ideal seeker must have a quick wit, a sharp and distinct imagina-
tion, ample and prompt memory, and the strongest ability to reason;
he must be skilled at ordering his thoughts, troubled by no cares or
passions, and capable of seeing clearly into his own actions; he must
not be precipitate in his judgments, nor influenced by custom and
example; he must allow adequate time in planning his work, and pro-
ceed confidently in this life. Only such an ideal seeker will persevere
unswervingly in this task and eventually discover the truth, and having

9 This was no mere false modesty. It was typical of the newfound confidence in rea-
son and the belief that reason, whole and complete, was universal in man. Descartes
conducted himself accordingly. Thus, he taught his servant, Jean Gillot, and Dirk
Rembrandtsz, a cobbler, mathematics; the former became director of an engineering
school at Leiden. Noting his talents, Descartes hired Henry Schulter as his manser-
vant, so that Schulter might assist him in his experiments. The captain of a ship on
which Descartes had traveled was so impressed with Descartes’ vast knowledge of
meteorology that when they reached Stockholm, the captain boasted to Christina
that Descartes had taught him more in three weeks than he had learned in sixty
years at sea. Clearly, Descartes’ theory belied his practice: Ordinary people, without
any formal learning, can learn difficult and important things. Perhaps this was the
net result of his Jesuit education: “The equality the Jesuits established among [the
students],” he wrote, “hardly treating the highest born any differently from the most
humble, was an extremely good invention” (R, 11; see also vii, 184–186).
The provisional title of Discourse on the Method was Project for a universal science which
might raise our nature to its highest degree of perfection. Next the Dioptric, the Meteors, where
the most curious matters which the author could find to give proof of the universal science he
proposes are explained in such a manner that even those who have never studied can under-
stand them. He suggested that an ideal seeker should be at least twenty-four years old
(CSMK, 120; AT II, 347), because “the younger they are, the less liberty they have,”
due to the soft nature of their brains (CSMK, 190; AT III, 424), which makes them
unfit for learning.
10 John Locke, An Essay Concerning Human Understanding, Volume 2, p. 391.
8 The Prolegomena to Any Future Epistemology

discovered it, be able to persuade others of it. Such seekers will be

able to persuade “even if they speak only low Breton and have never
learned rhetoric” (CSM I, 114; AT VI, 7).
There might be a conflict of propositions here. In the “ Fourth Set of
Replies,” Descartes had warned that the Meditations on First Philosophy
should be studied only “by very intelligent and well-educated readers”
(CSM II, 172; AT VII, 247). One might conclude that Descartes had
not made up his mind whether he wanted his ideal seeker to be just
intelligent, like Polyander in The Search for Truth (in which Polyander
is to Eudoxus what the slave boy was to Socrates in Meno), or whether he
wanted an ideal seeker who was very intelligent. Again, did Descartes
want his ideal seeker to be initially without education, as Polyander
was? Or did he want the ideal seeker to be someone initially with a
solid education? And yet, says an excited, marveling Eudoxus, who
has taken Polyander through the cogito, “Would you have thought
that an uneducated man who had never bothered to study could rea-
son with such precision, and be so consistent in all his arguments?”11
(CSM II, 415; AT X, 522) Obviously, being well educated is not a nec-
essary condition for being an ideal seeker. Descartes is concerned, in
his “Fourth Set of Replies,” to fend off the objection that his method
of doubt will engender doubt in the believers, and turn many a per-
son away from the truths of faith. Descartes’ counter would have
been that such men, if they turned away from their faith, would be
precipitate in their judgment and hence would not qualify as ideal
seekers.
It is not clear whether the qualities that a person possesses, such
as the qualities of quick wit, prompt memory, and sharp imagination,
or the qualities of being precipitate in one’s judgments and having
modest reasoning abilities, are essential properties or accidental ones.
If merely accidental, then those eliminated at the stage of sifting can

11 Since this is of some importance later, I cite the historical root of this approach. In the
Prologus of Raymond Sebond’s Natural Theology, written in the 1420s or early 1430s,
Sebond wrote: “And there is no need that anyone should refrain from reading it or
learning it from lack of other learning: it presupposes no knowledge of Grammar,
Logic, nor any other deliberative art or science, nor of Physics nor of Metaphysics. . . .”
(Appendix II, in Michele de Montaigne, An Apology for Raymond Sebond, xli–xlii) Such
was the man Polyander; such was the ideal seeker who could be persuaded of what
Descartes was trying to persuade him.
 I The Making of an Ideal Seeker 9

return to the fold by appropriately training themselves, acquiring the

necessary prerequisites to be an ideal seeker. If essential, then the set
of ideal seekers constitutes a natural class; genuine knowledge seekers
would be born, not made. Inasmuch as Descartes maintains that reason
exists in each person whole and complete, he must maintain the more
realistic doctrine, as follows: All persons are capable of discovering the
truth, some more than others. Those who make poor seekers are those
in whom reason is clouded by a host of contingent factors over which
they have little control.
“Having thus prepared our understanding to make perfect judg-
ments about the truth, we must also learn to control our will by distin-
guishing good things from bad, and by observing the true difference
between virtues and vices” (CSM II, 405; AT X, 506). This is putting
the cart before the horse: One cannot prepare the understanding to
make perfect judgments without the will; if the will is not in control,
it will make poor affirmations or denials. I find it surprising how very
little Descartes says about the will in the earlier portions of either the
Meditations on First Philosophy or the Discourse on the Method, given its
central importance in his epistemology. For one thing, it is only the
will’s affirmation that introduces the question of truth or falsity into the
discussion. Without the will, such questions cannot arise, and so knowl-
edge seeking cannot proceed apace without the will. Descartes speaks
of the intellectual qualities of the seeker, of the morals he should adopt
while engaged in his philosophical quest, but there is virtually nothing
about the will or the goodness of the will, how it should be controlled
and trained, and so on, in order that it may act without error.12
The fourth stage, the stage of preparation: The potentially ideal seeker
does not jump into making philosophical inquiries, not yet. He has to
prepare himself. He travels and gathers experience of men and the
world; he moves in the company of gifted men. (He reads books; and, as
a daily routine, he engages in the study of mathematics. These clearly

12 Why not think, one might ask, that the will is trained through enacting the analytic
method of the Meditations? I have two reservations: First, there is no evidence that
Descartes intended that; second, if the will – of a mature individual – is to make
appropriate choices as it wades through the Meditations, would it not already have to
possess goodness, say, if it is not to run afoul and choose erroneously? As an antidote
to my reservations, see the splendid Chapter 2, “Descartes: Willful Thinking,” in
Michael Losonsky, Enlightenment and Action from Descartes to Kant: Passionate Thought.
10 The Prolegomena to Any Future Epistemology

go far beyond the necessary conditions for the making of an ideal

seeker. Descartes did these things, but he did not make Polyander, his
example of an ideal inquirer, do them.) Thus, the ideal seeker trains
his mind, deepens it, makes it more powerful, so that when he finally
embarks on his philosophical inquiries he will be a person whose mind
is properly balanced between intellectual and emotional matters, and
his will will be strong and clear. This, then, is the nature of the ideal
seeker, and this is how he is made.
Polyander says, “I am a man who has never engaged in study or
accustomed himself to turning his mind so far away from things that
are perceivable by the senses” (CSM II, 408; AT X, 512). Epistemon, a
bookish man, asserts, “I agree that it is very dangerous to proceed
too far in this line of thinking” (CSM II, 408; AT X, 512). Eudoxus
(playing the role of Descartes) counters thus: “I confess that it would
be dangerous for someone who does not know a ford to venture across
it without a guide, and many have lost their lives in doing so. But you
have nothing to fear if you follow me.” (CSM II, 408; AT X, 512) A
strong and bold explorer can lose himself without a guide; a man of
common sense and discretion can lose himself, too, without someone
to guide him in his search for knowledge. Thus, even the ideal seeker
needs a guide, a method.

II. The Method: The Rationalist Thread

If Descartes had been asked, “What is the aim of science?,” he no doubt
would have replied, quite simply, “The absolute truth.” He took truth
to be indefinable, but he might have granted the following distinction.
There is phenomenal truth, truth p , and there is rational truth,truthr .
When we combine truthp and truthr , we get absolute truth. What, then,
are these two species of truth? Descartes wanted our theories of the
world to at least match our experiences and experiments. The theories
should “enable us to explain all natural phenomena [i.e., the effects
that we perceive by means of our senses]” (CSM I, 248; AT VIIIA, 80).
Such theories are truep . What cannot explain the deliverances of our
sense experiences is, at a minimum, not phenomenally true, and hence
not absolutely true.
Now, it is entirely possible for two theories to be truep , that is,
phenomenally true, without their being truer , that is, rationally true.
 II The Rationalist Thread 11

Descartes gives a simple example to illustrate this (what philosophers

now call the empirical equivalence of theories).

However, although this method may enable us to understand how all the things
in nature could have arisen, it should not therefore be inferred that they were
in fact made in this way. Just as the same craftsman could make two clocks
which tell the time equally well and look completely alike from the outside
but have completely different assemblies of wheels inside, so the supreme
craftsman of the real world could have produced all that we see in several
different ways. (CSM I, 289; AT VIIIA, 327)

A theory is determined to be truer if it appears true when viewed

in the natural light of reason. Such, for example, are the truths of
logic, mathematics, and metaphysics. The truer theory will correctly
describe how the wheels are assembled within. The truep theory will
correctly describe how the two clocks look and how they tell time.
In other words, granting that the supreme Craftsman could have de-
vised various hidden mechanisms to produce the same observable
effects, then the truep theory will explain all of the observable ef-
fects, and the truer theory will capture the veiled internal mecha-
nism of the world that produces these effects, and that truth will
be unmistakably exhibited to the natural light of reason. Thus, an
absolutely true theory not only will get the phenomena right – it
will be truep –, but also will get the mechanism right – it will be
truer .
One might worry that the foregoing is a less-than-exhaustive way of
describing Descartes’ own problem-situation, because of the problem
posed by the micromechanical, that is, that which we do not directly
perceive by sense, but which is visualizable. There seems to be a ten-
sion in Descartes between the idea that underlying structures are fully
determinate and could be perceived with good microscopes, and the
idea that they can be approached only by “reason”, that is, by model
making and intramental model comparison and exclusion. It is still,
one might suppose, a very live question whether rational truth just
stands in provisionally for the micrographic, or whether subvisible
structures are still just phenomenal truth and ultimately have to be
supplanted by nonvisualizable rational truth.13

13 I owe this objection to Catherine Wilson.

12 The Prolegomena to Any Future Epistemology

I have no fully satisfactory answer – certainly none that is histori-

cally satisfactory – to allay this worry, save this: Drawing on a common
antirealist position, I would say that there does not appear to be a third
type of truth. Let us say that a model depicts a microphenomenon.
Then, the model is to be viewed either as a truthp or a truthr. If it is not
regarded as a truthp , it may for a while have an uncertain status, until
it is explained or “approached” by reason. If the model has some sort
of truth not characterized by either of these kinds of truth, I am not
able to determine what it might be that would cohere with Descartes’
method and metaphysics. Finally, if the underlying structures are fully
determinate, and we can “approach” them only vis-à-vis our model
making, never quite getting at the underlying reality, then this can
easily be shown to lead to a kind of skepticism from which Descartes
cannot be saved.
For Descartes, science was systematic knowledge: neither a patch-
work quilt, nor a mere network, of propositions. It was built on a
rational foundation and not on guesswork or conjecture; it was indu-
bitable and nothing less. Given the aim of science, there was a method
to match. “I formed a method whereby, it seems to me, I can increase
my knowledge gradually and raise it little by little to the highest point”
(CSM I, 112; AT VI, 3). What was this method?
But first, what exactly did Descartes mean by method ?

By a ‘method’ I mean reliable rules which are easy to apply, and such that if one
follows them exactly, one will never take what is false to be true or fruitlessly
expend one’s mental efforts, but will gradually and constantly increase one’s
knowledge till one arrives at a true understanding of everything within one’s
capacity. (CSM I, 16; AT X, 371–372)

Descartes was not offering a method whose set of rules could be

mechanically applied in order to churn out new and novel truths. If
that is what is meant by a method – an algorithm – then Descartes did
not offer a method. A truth table is an algorithm used to determine if
an argument in propositional logic is valid; a Venn diagram is an algo-
rithm used to determine if any of Aristotle’s 256 categorical syllogisms
is valid. Descartes is not offering anything remotely similar. This does
not prevent him, like most current philosophers of science, from being
optimistic and claiming that if his method is adopted, then the likeli-
hood of discovering more and more scientific truths, the discovery of
 II The Rationalist Thread 13

which lie within the province of human capacity, is far greater than the
likelihood of discovering such truths using any alternative method.
These, then, are the four major rules of Descartes’ method:

The first was never to accept anything as true if I did not have evident knowl-
edge of its truth: that is, carefully to avoid precipitate conclusions and precon-
ceptions, and to include nothing more in my judgments than what presented
itself to my mind so clearly and so distinctly that I had no occasion to doubt it.
The second, to divide each of the difficulties I examined into as many parts
as possible and as may be required in order to resolve them better.
The third, to direct my thoughts in an orderly manner, by beginning with
the simplest and most easily known objects in order to ascend little by little,
step by step, to knowledge of the most complex, and by supposing some order
even among objects that have no natural order of precedence.
And the last, throughout to make enumerations so complete, and reviews
so comprehensive, that I could be sure of leaving nothing out. (CSM I, 120;
AT VI, 18–19)

The first two rules are primarily rules of analysis. Essentially, they
maintain the following: (i) Keep doubting a proposition until you have
no occasion to doubt it (otherwise, by implication, reject it). Such a
method of doubt will lead one to reject not only hasty conclusions
but also vague, untested conceptions or preconceptions, such as the
notions of vacuum, gravity, and substantial forms. (ii) Divide the given
problem into several smaller, yet clearly more manageable, problems.
The requisite analysis should be carried to the limit, so that the solutions
of the smallest and simplest problems can better lead to the solution
of the larger problem with which the analysis began. (iii) Accept as
true only that which presents itself to the mind as clear and distinct.
The combination of (i) and (iii) will yield certain knowledge and not
merely probable knowledge.
Perhaps one can now more sympathetically appreciate Descartes’
reaction to Galileo. On October 11, 1638, Descartes wrote to Mersenne,
“It seems to me that he [Galileo] lacks a great deal in that he is
continually digressing and never stops to explain one topic completely,
which demonstrates that he has not examined them in an orderly fash-
ion and that, without having considered nature’s first causes, he has
sought only the reasons for a few particular effects, and thus he has
built without foundations” (V, 115). This was no case of sour grapes.
In terms of the foregoing, Descartes was complaining that Galileo had
14 The Prolegomena to Any Future Epistemology

proceeded in a disorderly fashion, that he had tried to explain a few

particular effects without taking his analysis down to the deepest level.
Thus, Galileo had failed to uncover the basic axioms of his science of
physics, which would have enabled him to explain not just the parti-
cular effects, but all of the facts in the given domain of knowledge.14
Moreover, it was only the truth and certainty of the axioms of physics
that would have given firm anchor to Galileo’s science. Galileo had
failed in that respect, and hence had “built without foundations”. This,
then, is the structure of discovery, not just in physics and geometry but
in any field of knowledge.
Next, consider rules 3 and 4 of the method. These are primarily
the rules of synthesis. Essentially, these rules maintain the following:
(i) Assume, even if perhaps contrary to appearance, some natural, not
man-made, order among the objects of your investigation. One might
ask, “Isn’t Descartes saying that one has to impose an order by deciding
what to take up first, even if there is no natural order?” Well, Descartes
wrote to Mersenne on May 10, 1632, while absorbed in the study of
astronomy, “For although [the stars] seem very irregularly distributed
in various places in the heavens, I do not doubt that there is a natural

14 Sir Karl Raimund Popper has objected that just such a structure of knowledge as
Descartes was proposing was obscurantist in its demand for ultimate explanations in
terms of essences; see his Conjectures and Refutations: The Growth of Scientific Knowledge,
in particular Chapter 3, section 3. Popper states, “Thus my criticism of essentialism
does not aim at establishing the non-existence of essences; it merely aims at showing
the obscurantist character of the role played by the idea of essences in the Galilean
philosophy of science” (105). Popper fears that if essences are postulated, as in
Galileo (or in Descartes), useful questions will not be asked (106), and this will pre-
maturely stop the flow of knowledge.
This need not be so. There is nothing inherently wrong in the doctrine of essences;
in any event, it is compatible with the doctrine of conjectures and refutations. Thus,
no one who postulates essences, described at one level, need claim that he has in fact
discovered those essences. He may simply regard himself as being at an earlier level,
awaiting further descent into deeper and deeper worlds described at the correspond-
ing levels. In Popper’s words, “the world of each of our theories may be explained,
in its turn, by further worlds which are described by further theories” (115). Such
a philosopher, like Descartes, is simply proposing what the structure of knowledge
should look like, when there is knowledge at hand. To claim that thereis knowledge
at hand, Descartes proposed a different theory. The former is an ontological claim,
the latter an epistemic one, and the two can rest side by side, at ease with each other.
Descartes, more than any other philosopher in the history of philosophy – Francis
Bacon included, in my view – was the one who determined, for a long time, what was
to be regarded as the structure of knowledge.
 II The Rationalist Thread 15

order among them which is regular and determinate” (CSMK, 38; AT I,

250). I suspect, then, that Descartes’ reply to the foregoing question
would be twofold: First, there is a natural order in the world, and
second, often we discover that natural order by starting our search
by imposing a conjectured order on the system we are investigating.
(ii) Begin with the simplest and most easily known objects that precede
the rest of the objects in the order. “[W]e term ‘simple’ only those
things which we know so clearly and distinctly that they cannot be
divided by the mind into others which are more distinctly known”
(CSM I, 44; AT X, 418). (iii) Ascend to the more complex objects in the
order in a slow step-by-step manner, via less complex objects, the steps
being dictated by the rules of the subject matter at hand. (iv) Omit
nothing by enumerating everything that lies within the domain of
your subject matter.15 Finally, (v) carry out comprehensive reviews and
enumeration.
This method of analysis presupposes that we can analyze a problem,
break it down into simpler units, like building blocks, without the
aid of a theory, a perspective. It presupposes that there is only one
unique way of breaking down the problem, that there is a “best order”
in which all items can be arranged. But this is quite an implausible
assumption, at least as a general statement. There are as many ways
of analyzing a problem as there are ways of viewing the objects in a
domain, and there are generally quite a few ways of viewing a given set
of objects. That this presents a difficulty for Descartes is not hard to
see. If the perspective is granted, then the analysis can be carried out.
But that leaves the correctness of the perspective in question. For if the

15 “I said also that the enumeration must be well-ordered. . . . if we arrange all the rele-
vant items in the best order, so that for the most part they fall under definite classes,
it will be sufficient if we look closely at one class, or at a member of each particular
class, or at some classes rather than others. If we do that, we shall at any rate never
pointlessly go over the same ground twice, and thanks to our well-devised order, we
shall often manage to review quickly and effortlessly a large number of items which
at first sight seemed formidably large.” (CSM I, 27; AT X, 390–391)
To quote in full rule 5 of Rules for the Direction of the Mind: “The whole method
consists entirely in the ordering and arranging of the objects on which we must
concentrate our mind’s eye if we are to discover some truth. We shall be following
this method exactly if we first reduce complicated and obscure propositions step by
step to simpler ones, and then, starting with the intuition of the simplest ones of
all, try to ascend through the same steps to a knowledge of all the rest.” (CSM I, 20;
AT X, 379)
16 The Prolegomena to Any Future Epistemology

perspective were false or poor, the analysis would be worthless. On the

other hand, if the perspective is not granted, then the analysis cannot
even begin. As the dogma has it, there cannot be a theory-neutral
observation (or analysis).

III. The Tree of Philosophy

After the cogito, these from Principles of Philosophy are perhaps the most
famous lines in Descartes: “ Thus the whole of philosophy is like a
tree. The roots are metaphysics, the trunk is physics, and the branches
emerging from the trunk are all the other sciences, which may be
reduced to three principal ones, namely medicine, mechanics and
morals.”16 (CSM I, 186; AT IXB, 14) Descartes’ tree of philosophy
would look like Figure 1.1. The arrows in the diagram indicate notions
such as dependence, reduction, support, grounded in, resting secure in, and
other such loosely similar ideas, since Descartes did not have a precise
conceptual notion of how the sciences at the top of the tree were
related to those at the lower levels, or to the one at the bottom. “ I will
also add,” said Descartes to Clerselier, in a letter written in June or
July 1646, “that one should not require the first principle to be such
that all other propositions can be reduced to it and proved by it. It is
enough if it is useful for the discovery of many, and if there is no other
proposition on which it depends, and none which is easier to discover.”
(CSMK, 290; AT IV, 444–445; my emphasis) To cover this multitude
of concepts, I shall use the colorless term dependent on. When he was
younger, Descartes had thought that all of the sciences, if their links or
derivations were properly established, could be held in the mind’s eye
quite easily. “ The sciences are at present masked, but if the masks were
taken off, they would be revealed in all their beauty. If we could see
how the sciences are linked together, we would find them no harder to
retain in our minds than the series of numbers.” (CSM I, 3; AT X, 215)
Descartes’ tree of philosophy could aid in that cause.

16 The tree analogy is also presented in Francis Bacon’s Advancement of Learning, II, v.1,
v. 3; De Augumentis, III, i–ii, quoted in Edwin A. Abbott’s Francis Bacon: An Account of
His Life and Works, 354–355. On November 10, 1619, Descartes had some remarkable
dreams, in one of which he dreamed of a dictionary representing all the sciences
gathered together; see V, 54–59.
 III The Tree of Philosophy 17

.... Theorems . . . .

Axiomsi Axiomsj Axiomsk .... Axiomss .... Axiomsz

The Other Sciences

Theorems Theorems Theorems

Axioms of Medicine Axioms of Mechanics Axioms of Morals

Theorems

Axioms of Physics

Theorems

Axiom of Metaphysics

Descartes’ Tree of Philosophy

figure 1.1.

To illustrate: Suppose pathology to be one of the sciences defined,

say, by axiomsi ; this field of science can be reasonably regarded
as dependent on the science of medicine. Given a theory of how
the heart functions – for example, William Harvey’s theory of the
18 The Prolegomena to Any Future Epistemology

heart and the circulation of the blood, a theory that Descartes rightly
considered important enough to argue with – Descartes approached it
in a Hobbes-like manner, but came away less satisfied than Hobbes was
with Euclidean geometry.17 He disagreed with Harvey’s explanation,
and showed how Harvey’s empirical results could eventually be shown
to be more satisfactorily dependent on the axioms of medicine,
and not on Harvey’s outmoded Aristotelianism. Furthermore, the
axioms of medicine could be made to be firmly dependent on the
theorems, and hence on the axioms, of physics. Finally, the axioms
of physics could be made dependent on the theorems and the axiom
of metaphysics. Thus, one of the several branches at the top of the
tree – pathology – can be shown to be dependent on one of the
three principal branches – medicine – below; this latter branch is
supported by the trunk – physics – which, in turn, is nourished by
the roots – metaphysics. Consequently, pathology is dependent on
metaphysics. In a similar vein, everything can be seen to be dependent
on metaphysics. The notion of dependence allows for knowledge to be
increased by experiment; it is not the case that once the metaphysical
axiom is discovered the rest of the knowledge can be secured by
executing merely logical deduction. Thus, Descartes was in earnest
when, at the end of the Discourse on the Method, “he had called for the
public’s help in carrying out all the experiments necessary to ‘justify
my arguments.’ ‘It could take several centuries before we have thus
deduced from these principles all the truths that can be deduced
from them.’” (R, 177–178; AT IX, 2:20)
Now, if metaphysics was the most fundamental discipline, were
physics and mathematics dependent on it? Likewise, was morals, for
example, dependent on physics and hence in turn on metaphysics?
Or were they – physics, mathematics, and morals – just vaguely sup-
ported by metaphysics, not dependent on it? In the Principles of

17 At forty, yet innocent of mathematics, Thomas Hobbes accepted a nobleman’s in-

vitation to travel on the continent and tutor his son. One day, he found a copy of
Euclid lying open in the library of his distinguished employer at the page boasting of
Pythagoras’ theorem. His friend, John Aubrey: “He read the proposition. ‘By God,’
sayd he, ‘this is impossible.’ So he reads the demonstration of it, which referred him
back to such a proposition; which proposition he read. That referred him back to
another, which he also read.Et sic deniceps [and so one after another] that at last he
was demonstratively convinced of that trueth. This made him in love with geometry.”
(Quoted in William Kneale and Martha Kneale, The Development of Logic, 311.)
 III The Tree of Philosophy 19

Philosophy, Descartes wrote, “medicine and mechanics, and all the

other arts . . . can be fully developed with the help of physics” (CSM I,
289; AT VIIIA, 327). What is the nature of this help? This leaves un-
clear, too, whether other principles, besides those of physics, are also
required for developing all the other arts. Is physics just necessary for
these arts, or is it both necessary and sufficient?
There are, of course, various sciences at the top of the tree, such
as (to mention only the ones that Descartes himself acknowledged)
transmutation of metals, chemistry, anatomy, virtues of plants, astro-
nomy, botany, meteorology, geography, zoology, psychology, music,
and optics. The status of the axioms must be properly understood.
The axioms of the other sciences – say, the axiomsj of meteorology –
are fundamental relative to their own science, but not fundamental
in the whole scheme of knowledge. They are in some ways dependent
on at least one of the three sciences below them: For example, the
axiomsi of meteorology are dependent on the axioms of mechanics.
Again, the axioms of the three sciences of medicine, mechanics, and
morals are fundamental relative to their own respective fields, but
not in the entire scheme of knowledge. For example, the axioms of
mechanics are dependent on the axioms of physics. The latter ax-
ioms, in turn, are dependent on the axiom of metaphysics. The axiom
of metaphysics is absolute not only within metaphysics, but absolute
in the total scheme of knowledge, since ultimately that axiom is not
supported by, dependent, derived, or based on anything other than
itself.
The relation between the sciences at the top of the tree and those at
the bottom is not precisely clear. For example, are the sciences at the
top dependent on one of the three principal sciences below (which
in turn are dependent on physics and metaphysics)? Or would two
or more principal sciences be required? It is understandable if mete-
orology is dependent on mechanics; which in turn is dependent on
physics, and so on. But it is not intuitively plausible to think that pathol-
ogy would be solely dependent on mechanics; at the very least it may be
dependent on mechanics and medicine, which in turn are dependent
on physics, and so on. Thus, in Description of the Human Body, Descartes
says that “it is so important to know the true cause of the heart’s move-
ment that without such knowledge it is impossible to know anything
which relates to the theory of medicine” (CSM I, 319; AT XI, 245). But
20 The Prolegomena to Any Future Epistemology

the problem of the motion of the heart belongs to physiology, and so

clearly physiology, not one of the three branches below, may not be de-
pendent on just medicine, but may be dependent on both mechanics
and medicine. Descartes also left open the question of just how inde-
pendent the three principal sciences were. Indeed, was it essential that
they be independent, if conjointly they could serve as the axioms for
everything above them?
The various axioms of various disciplines, such as medicine, me-
chanics, and morals, are in some deeper sense not quite fundamental
yet. For as Descartes will show us, they too can be doubted. If so, they
must take their nourishment from the roots, namely, metaphysics. But
metaphysics itself consists of complex truths about complex disciplines
of complex objects, such as the physical world, time, space, numbers,
matter, mind, and God. What, if anything, will provide us with the
starting point that Descartes is looking for? Will the nature of this
philosophical axiom be akin to the other axioms higher up on the
tree of knowledge? Or will it be different?
We have arrived at the primary task of philosophy, namely: Is
there such a single metaphysical axiom – an axiom that, like the
axioms of geometry, cannot be questioned, cannot be argued for,
cannot be proved by anything else? An axiom that would support
all the axioms and theorems of all the other disciplines, includ-
ing those of metaphysics? Or are there, after all, several metaphysi-
cal axioms? If there is just one axiom, then it too would only have
to be understood in order to be perceived as clear, distinct, and
indubitable, and in such a manner that its truth, says Descartes,
would strike home more strongly than the truth of a mathematical
axiom when the latter is understood.18 Descartes’ tree of philoso-
phy makes it amply clear that such an axiom would be the foun-
dation of all human knowledge: everything resting on it, it resting
on nothing else. Were such an axiom to be found, it would be as
though a fulcrum had been discovered from which to move the
universe.
If we combine the rules of analysis, the rules of synthesis, and
the rules of hypothesis making and hypothesis evaluation, we get

18 Such a claim is repeatedly made: See, for example, CSMK, 23, AT I, 144; CSMK, 29,
AT I, 182; CSMK, 53, AT I, 350; and CSM II, 416, AT X, 522.
 III The Tree of Philosophy 21

the method Descartes would have advocated to a scientist with the

following admonition: “[A]ll the things which can fall under human
knowledge are interconnected in the same way. And I thought that,
provided we refrain from accepting anything as true which is not, and
always keep to the order required for deducing one thing from an-
other, there can be nothing too remote to be reached in the end or
too well hidden to be discovered.” (CSM I, 120; AT VI, 19)
There are a few central things to be observed. First, there is Leibniz’s
old saw that Descartes’ rules of method were “like the precepts of
some chemist; take what you need and do what you should, and you
will get what you want.”19 This is not entirely deserved. At no place
I am aware of did Descartes give one set of rules for one discipline,
and another set of rules for another discipline: say, one method for
doing philosophy, and a different method for doing optics. What is
truly remarkable is that one method is offered – rightly or wrongly – as
sufficient for all of the disciplines. The picture one gets is that of a unity
of method across disciplines. No matter how different the subject matter
under investigation, Descartes’ method is supposed to yield the right
results.20 In any event, in the light of Descartes’ rules for the making
of science, Leibniz’s statement is surely suspect.
Given their universal applicability, it is hardly surprising that his
rules were of such a general nature; they could not have been other-
wise. One might as well complain that Newton’s and Descartes’ physics
were overly general and did not draw nice distinctions between lunar,
sublunar, and heavenly phenomena. But if the proof of the physics was
in its success, the proof of the method lay no less in its success. Descartes
claimed to have used this method in medicine, physics, morals, and a
variety of sciences such as optics, geometry, and meteorology. Practice –
practice – with the method will enable the scientist to learn what the
rules mean, and how best to employ them. The rules cannot be un-
derstood just formally.
Second, by doubting and digging deeper, Descartes is arriving at
axioms that are simple. Simplicity is hard to define, but one of the things

19 Leibniz, Die Philosophischen Schriften von G. W. Leibniz, Volume 4, 329.

20 For further details, see L. J. Beck, The Method of Descartes: A Study of the Regulae,
287–307. Gerd Buchdahl’s Metaphysics and the Philosophy of Science. The Classical Origins:
Descartes To Kant, Chapter 2, is still a standard reference for anyone doing serious work
on Descartes’ method.
22 The Prolegomena to Any Future Epistemology

that Descartes maintains is that it is easier to intellectually apprehend

the truth of the respective axioms of a given field of knowledge than
it is to understand its more complex theorems. Thus, it is easier to
see the truth of a Euclidean axiom, such as “ Things that are equal to
the same thing are equal to one another,” than it is to see the truth of
a more complex theorem, such as Pappus’ theorem. So, presumably,
the axiom of metaphysics will have to be simple, too.
Third, the axioms are of a general nature. An axiom of mathematics
speaks of the relation between part and whole, not of this particular
part and that particular whole, such as a particular cell in a colony of
cells. An axiom of passions speaks of wonder generally, not of wonder
about this or that, such as wonder about the starry heavens above
and the moral law within. An axiom of physics speaks of the motions
of objects in general and not of the fixed stars, or the sun, or some
particular planet, say, Jupiter. This is an enormously significant point,
because eventually we must inquire into the nature of the axiom of
metaphysics. Must it too be of a general nature? Can it be of a general
nature?
Fourth, the direction of the movement of thought between axioms
and theorems plays a significant role. As he attempts to move toward
the axioms, Descartes is using the method of analysis: His method leads
him to discover new truths. But once the discoveries have been made,
and we attempt to explain an old truth or a new one with the help of
the axioms, we are using the method of synthesis. Here we might well
use the logic of the scholastics, which we would not use in the method
of analysis.
The rationalist thread in Descartes’ method is vital to the central
claim of his metaphysics. It is what concerns me in this book.

IV. Method, Morals, and Bootstraps

“I had to uproot from my mind,” says Descartes, “all the wrong opin-
ions I had previously accepted, amass a variety of experiences to serve
as the subject-matter of my reasonings, and practice constantly my self-
prescribed method in order to strengthen myself more and more in
its use.” (CSM I, 122; AT VI, 22). Should he now apply his method
in every field of endeavor, every science? That would be quite an im-
proper use of the method. For “observing that the principles of these
 IV Method, Morals, and Bootstraps 23

sciences must all be derived from philosophy, in which I had not yet
discovered any certain ones, I thought that first of all I had to try to
establish some certain principles in philosophy.” And, Descartes as-
serts, “this is the most important task of all” (CSM I, 121–122; AT VI,
21–22).
Here a paradox is in the making. It is crucial that Descartes learn
to apply his method in order to strengthen his mind; he cannot just
jump into doing philosophy. The practice with the method is impor-
tant. ‘[A]s I practised the method I felt my mind gradually become
accustomed to conceiving its objects more clearly and distinctly; and
since I did not restrict the method to any particular subject-matter, I
hoped to apply it as usefully to the problems of the other sciences as
I had to those of algebra” (CSM I, 121; AT VI, 21). So I shall call this
the mathematical method. Indeed, in the Rules for the Direction of the Mind,
Descartes had declared:

So frequently was I successful in this that eventually I came to realize that I

was no longer making my way to the truth of things as others do by way of
aimless and blind inquiries, with the aid of luck rather than skill; rather, after
many trials I had hit upon some reliable rules of great assistance in finding
the truth, and I then used these to devise many more. In this way I carefully
elaborated my whole method. (CSM I, 35; AT X, 403–404)

But if he applies the mathematical method elsewhere first, with-

out having discovered a philosophical axiom, a certain principle of
philosophy, he is guilty of using this method improperly. If Descartes
has not applied the mathematical method elsewhere before using it
in his philosophical inquiry, he will neither learn of the efficacy of the
method nor strengthen his mind for the epistemic task. Therefore, he
will either use the method improperly or use it unjustifiably, without
knowing its efficacy. How, then, is he to proceed? He must, I think,
raise himself by his own bootstraps.
Let me present this argument differently. With the moral code (by
which he will guide his life) provisionally established,21 Descartes has
simply completed the first step. He does not feel obliged to commence
his philosophical inquiry straightaway.

21 For the influence of Pierre Charron’s Traite de la Sagesse, which had been on the Index
since 1605, on Descartes’ view of ethics, see R, 44–48.
24 The Prolegomena to Any Future Epistemology

As a matter of fact, he tells us, he spent nine years in preparation.

This is the second step. He read books, met with men of letters, culti-
vated his reason, and above all solved problems in mathematics, which
sharpened his eye for truth and for clear and distinct ideas. He stud-
ied mathematics at length because “of the certainty and self-evidence
of its reasonings” (CSM I, 114; AT VI, 7) and because “mathemati-
cians alone have been able to find any demonstrations – that is to
say, certain and evident reasonings” (CSM I, 120; AT VI, 19). Indeed,
in the Rules for the Direction of the Mind, Descartes had urged that, at
the start, we confine our studies to arithmetic and geometry, for these
alone are “free from any taint of falsity or uncertainty” (CSM I, 12;
AT VI, 364). What did he hope to gain from this study? “From this,
however, the only advantage I hoped to gain was to accustom my mind
to nourish itself on truths and not to be satisfied with bad reasoning”
(CSM I, 120; AT VI, 19). In short, he believed that he would sharpen his
reason.
Now, Descartes says that he was very successful in using the mathe-
matical method in algebra, and since the method was not devised for
any particular discipline or subject matter, he felt he would be able
to apply it in other fields equally successfully. “Not that I would have
dared to try at the outset to examine every problem that might arise,
for that would itself have been contrary to the order which the method
prescribes” (CSM I, 121; AT VI, 21). The problems to be tackled, in the
beginning, would have to be simpler and more manageable problems.
There is a necessary order that must be followed when an ideal seeker
aims to instruct himself (CSM I, 185; AT IXB, 13). As part of that order,
says Descartes – and here comes a crucial bit – the ideal seeker “should
study logic” (CSM I, 186; AT IXB, 13).
Not scholastic logic, to be sure, but, continues Descartes,

the kind of logic which teaches us to direct our reason with a view to discovering
the truths of which we are ignorant. Since this depends to a great extent on
practice, it is good for the student to work for a long time at practicing the
rules on very easy and simple questions like those of mathematics. Then, when
he has acquired some skill in finding the truth on these questions, he should
begin to tackle true philosophy in earnest. (CSM I, 186; AT IXB, 13–14)

Subsequently, he abandoned the study of letters and turned to the

practical world of courts and armies, savoring various experiences,
 IV Method, Morals, and Bootstraps 25

meeting people with diverse temperaments and ranks, and reflecting

on his experiences that he might derive profit from them. “For it
seemed to me that much more truth could be found in the reasonings
which a man makes concerning matters that concern him than in
those which some scholar makes in his study about speculative matters”
(CSM I, 115; AT VI, 9–10). While a scholar can afford to hold any
opinion with few if any real consequences, a man of the world has to
pay dearly for his false beliefs. What he learned through example and
custom, he treated lightly, for these were often in conflict even among
great communities and nations.22
Finally, with his mind sufficiently sharp and clear, Descartes takes
the third step; he embarks on the strenuous activity of systematic philo-
sophical inquiry.
What, precisely, did Descartes learn prior to his establishing the
cogito and, with it, the general rule of distinguishing the true from the
false? Not yet having established the general rule, how could Descartes
have been confident that he was training his mind, stocking it with
truths, performing good reasoning, successfully sifting good worldly
experiences from bad, profiting from his good experiences, and so on?
How could he rely on the truth of mathematics, however simple? How
could he trust the rules of logic, old or new? How would he, even as the
ideal seeker, know that he had acquired the right and correct skills?
Apparently, a clear mind, a good will, and good sense are not
enough; one also requires a good and guaranteed method. For it is the
method that will outline the procedure and certify the results. How-
ever, not only were his morals based on probability, so was his method.
Descartes was aware of the tentativeness of his method: “[P]erhaps

22 Descartes’ method, as one distinguished Kantian authority would have it, is to be

sharply distinguished from Kant’s method: “In Descartes’ work the grounding of
reason is closely linked to its political impotence. Only the repudiation of politics, and
more generally of criticism of action, allows him a meditative perspective from which
to discern and deploy the methods without self-stultification. Kant’s philosophical
enterprise evidently follows quite a different route.” (Onora O’Neill, Constructions of
Reason: Explorations of Kant’s Practical Philosophy, 5–6.) The fundamental difference is
this: The use and vindication of reason is a public matter, not a private one. “Reason
may be (in whole or, in part) ‘in’ each participant, but it cannot be discovered by
introspection: Kant insists that we are opaque, not transparent, to ourselves”(7). In
particular, see O’Neill’s discussion on the Kantian notion of “the sensus communis”
(24–27). See also Chapter 2, pp. 40–41, this volume.
26 The Prolegomena to Any Future Epistemology

what I take for gold and diamonds is nothing but a bit of copper and
glass” (CSM I, 112; AT VI, 3). Again:

It was never my intention to prescribe to anyone the method which he should

follow in his search for truth, but simply to describe the method which I used
myself: if it should be thought to be defective, it would be rejected; if good
and useful, others would use it too. . . . If someone should now say that it has
not got me very far, this is a matter for experience to determine. (CSM II, 419;
AT X, 525–526)

In Discourse on the Method, he could scarcely have been more tentative

about his method. He said that he was presenting that work “only as a
history” (CSM I, 112; AT VI, 4).
But then someone might argue – as, indeed, Pierre Gassendi might
well have argued – that instead of, say, the mathematical method, the
method to follow is the historical method.23 It might be argued that the
things to read are just the things that Descartes refused to consider
seriously: languages, fables, letters, histories, poetry, morals, theology,
and jurisprudence (CSM I, 112–115; AT VI 4–9). These will train the
inquirer, so a Gassendist might say, in a way in which indulging in mere
mathematics would not: indeed, the latter might stunt the natural
development and propensities of the human mind. Only after years of
intensive study in this method should an ideal seeker embark on his
philosophical quest, not before. Having embarked, the seeker might
discover, say, a stunning moral truth having as much power, clarity, and
certainty as the cogito, or perhaps having even greater power, clarity,
and certainty. And out of this primary moral truth could be recovered
an indispensable epistemic rule that would serve to establish other
important truths. His physics, medicine, and mechanics would then
come to rest on his metaphysics, and his metaphysics on the primary
moral truth (and not the other way around).
Is this not topsy-turvy? Surely, someone will point out, Descartes
did not approve of the historical method and defended that position.
Thus, he claimed that the moral writings of the ancient pagans, al-
though appearing to be magnificent, were nothing but “built only on

23 See, for example, Lynn Sumida Joy’s Gassendi the Atomist: Advocate of History in an Age
of Science. Strictly as an aside, Descartes dreamed on November 10, 1619, that poets,
because divinely inspired, often proclaim truths that are not only more profound
but better expressed than those of the philosophers.
 IV Method, Morals, and Bootstraps 27

sand and mud” (CSM I, 114; AT VI, 8). He also showed why, for his
purposes, he disvalued poetry and oratory, and so on. In short, we have
no reason to rely on the historical method.
But the certain criteria, the general rule, for distinguishing truth
from falsity is derived only after the cogito has been discovered, not
before. If Descartes could certainly distinguish the true from the false
at the stage of preparation, then the cogito would hardly be important.
In fact, the metaphysics of God and soul and the sciences could be
constructed with whatever principle or criteria Descartes was work-
ing with before he discovered the cogito. If discovering the general
rule, the criteria for distinguishing the true from the false, is indispens-
able to claiming certain knowledge, then Descartes’ advice to study
mathematics – as opposed to, say, history, fables, or poetry – was no
more than a mere conjecture about how best to prepare himself for
the philosophical task ahead.
Suppose that one is preparing for an elementary calculus exam. One
reads George Polya’s How to Solve It, Richard Courrant’s and Herbert
Robbins’s What is Mathematics? An Elementary Approach to Ideas and
Methods, and a standard textbook or two on calculus, such as George
B. Thomas, Jr.’s Calculus and Analytic Geometry. One studies functions,
limits, integration, transcendental and hyperbolic functions, vector
functions, partial differentiation, multiple integrals, infinite series, dif-
ferential equations, and so on. One is preparing one’s mind for a rig-
orous examination ahead. One is on the right track in studying these
books: They are acknowledged classics in the field; some of them, like
Polya’s book, deal with heuristics; they deal with important mathe-
matical problems whose solutions are generally acknowledged; these
works will teach the accepted general techniques of solving mathe-
matical problems, which will enable the reader to solve new and fresh
problems; and in these books, one will encounter problems that are
unsolved or unsolvable. The problem here is not the epistemological
one of discovering a general rule that will enable us to distinguish,
confronted by any proposition, mathematical or otherwise, whether it
is true or false. The ability to recognize mathematical truths is taken
for granted in such books. It would be irrelevant for an examination
in calculus if one studied instead the Book of Job, Petrarca’s On His
Own Ignorance and That of Many Others, or James Baldwin’s Go Tell It on
the Mountain (or Seneca’s On the Happy Life and Machiavelli’s The Prince,
28 The Prolegomena to Any Future Epistemology

which Descartes read with Princess Elizabeth) (CSMK, 256; AT IV, 263
and V, 202–203).
But it is begging the question against the historical method if one
studied mathematics as a way of preparing for the epistemological
exam, so to speak. Why should our epistemic abilities increase when
studying mathematics and atrophy when studying history or fable, and
not vice versa? It is, I think, that Descartes is here making a choice of
method that is solely based on probability. He is implicitly, but force-
fully, claiming that there is a higher probability of strengthening his
mind, and being successful in the epistemic enterprise, if he uses the
mathematical method than there would be if he were to use the histor-
ical method. Then he simply argues that the proof is in the pudding:
“If someone should now say that it has not got me very far, this is a
matter for experience to determine” (CSM II, 419; AT X, 526). He
may take a wrong turn and find that it leads nowhere. Alternatively,
he may find that he repeatedly makes the right moves – is successful,
for example, in algebra – and that this gives him confidence that he
is on the right track toward the epistemic goal. Well, was there not a
Chandoux to point out to Descartes that resting as it did on probability,
his mathematical method could be shown to be false?
One might say in Descartes’defense that one has to begin some-
where. And as we proceed in our task, we shall get a better idea of
how effective the method is. There is no other alternative. If one has
to fashion tools, one fashions good tools from crude ones, better tools
from good ones, and so on. This is, indeed, what Descartes at one point
recommends.

If, for example, someone wanted to practice one of these crafts – to become a
blacksmith, say – but did not possess any of the tools, he would be forced at first
to use a hard stone (or a rough lump of iron) as an anvil, to make a rock do as
a hammer, to make a pair of tongs out of wood, and to put together other such
tools as the need arose. Thus prepared, he would not immediately attempt to
forge swords, helmets, or other iron implements for others to use; rather he
would first of all make hammers, an anvil, tongs and other tools for his own use.
What this example shows is that, since in these preliminary inquiries we have
managed to discover only some rough precepts which appear to be innate in
our minds rather than the product of any skill, we should not immediately try
to use these precepts to settle philosophical disputes or to solve mathematical
problems. Rather, we should use these precepts in the first instance to seek out
 IV Method, Morals, and Bootstraps 29

with extreme care everything else which is more essential in the investigation
of truth.24 (CSM I, 31; AT X, 397)

What Descartes is saying is that we must inevitably begin with some

crude concepts of the mathematical method – hence, a nine-year
period of apprenticeship. With the help of these crude concepts, we
should be able to forge better epistemic tools. However, he might
caution, we should not rush into solving mathematical or metaphysi-
cal problems with these concepts or tools. With the help of these, we
should simply select, with extreme care, whatever is essential to the
discovery of fundamental truths. The point is well taken. But could
we not have suggested something similar had we begun with the crude
concepts of the historical method? That is the cardinal question. And
had we begun with that method, and its crude concepts, we would have
tried to fashion sophisticated concepts as we went along, and we could
have been just as cautious in applying those concepts to various other
issues and problems. After all, the truth of the cogito is independent
of the truth of mathematics or morals. Descartes could have arrived at
the cogito even if he had adopted the historical method. But Descartes,
without argument, settles for the mathematical method. This seems
plausible, if you accept the pragmatic approach. But Descartes was no
pragmatist.25
Furthermore, what if the mathematical method is false? What pow-
erful effect would a false idea have on the mind? Once the mind has
acquired it, can it go on to acquire any truth whatsoever without hind-
rance from this idea? If not, then we cannot afford to be unmindful
of a false idea in our moral code that we adopt for practical purposes.
For who can tell what havoc it would wreak later? The mind might

24 For a virtually parallel claim, see Benedict de Spinoza, The Chief Works of Benedict de
Spinoza, Volume 2, “On the Improvement of the Understanding,” 12.
25 Might it be plausible to argue that in studying mathematics the learner’s confidence
level increases, in studying a humanistic discipline it decreases, leaving the learner
baffled and ignorant? Well, this need not always be so. Consider, for example, the
Middle Ages: not much of studying mathematics therein, but a great deal of studying
the humanities. Was the confidence level generally low? In any society, there will be
some discipline that dominates: It may be a practical craft in one society, a theoretical
science in another, arts and poetry in a third. The confidence level of the members
of each society will to a large extent reflect its history and traditions. In any event,
Descartes offers no reason why it is with mathematics that he should commence rather
than with a humanistic discipline.
30 The Prolegomena to Any Future Epistemology

be imagined as a flexible storehouse whose capacity to acquire truths

increases as it receives and stores more truths, and whose capacity to
acquire truths diminishes as it receives and stores more falsehoods.
This is not Descartes’ analogy, but it is quite close in spirit to what
he says: “[E]rrors . . . may obscure our natural light and make us less
capable of heeding reason” (CSM I, 116; AT VI 10). Or this: “Someone
who . . . is stuffed full of opinions and taken up with any number of
preconceptions finds it difficult to submit himself exclusively to the
natural light” (CSM II, 416; AT X, 522–523).
The question, then, is this: If the historical method proposes one set
of crude notions with which to commence an epistemic inquiry, and
Descartes suggests an alternative set of equally crude notions, how can
we settle on either in the light of Descartes’ doubt? Descartes’ claim
that his notions are innate settles nothing. Those who propose the
historical method may question or doubt Descartes’ claim and argue
that their own crude notions are the ones that are really innate.
It is true that “those who proceed but very slowly can make much
greater progress, if they always follow the right path, than those
who hurry and stray from it” (CSM I, 111; AT VI, 2). But how is
his mathematical method justified as leading to the right path before
he has used it in philosophy? Descartes might say something like
this: “Begin with a method that is conjectured to be probably ef-
fective, given its initial success in mathematics. Use this method in
philosophy, and one will in all probability get at certain truth. This
will enable one to uncover a reliable epistemological principle, the
general rule. Henceforth, the procedure can be fully justified with-
out doubt in the following way: Each step after the general rule is
justified by appealing to the general rule. Subsequently, the discov-
ered epistemological principle, the general rule, may establish with
certainty the correctness of the steps previously taken. That is, the
general rule will determine that the moral code as well as the mathe-
matical method were in fact the correct ones to adopt at the start.”
This is exactly what I meant by the bootstrap approach in Descartes’
method.
Let us look at this matter from the point of view of the moral code.
Descartes provides himself with four such moral maxims. First, one
ought to obey the laws and customs of the country, believe in the
religion in which one was raised, and hold moderate opinions. Second,
 IV Method, Morals, and Bootstraps 31

one ought to be firm and decisive in one’s actions even when these are
based on the most doubtful opinions. Third, one should try to restrain
one’s wants rather than trying to change the order of the world in order
to satisfy one’s wants. Fourth, one should review the various tasks and
occupations men have in this life in order to choose the best. (CSM I,
122–125; AT VI, 23–28)
There is a problem worth noticing. What is the connection be-
tween Descartes’ moral code, followed with utter decisiveness, and his
epistemology? Suppose the much-maligned Turk,26 say of the seventh
century, were to adopt a different moral code. The Turk, following the
faith of his fathers, would be a polytheist, practice polygamy, and grant
allegiance to different social and market institutions; and Descartes
would follow the Christian faith, profess monogamy, and give alle-
giance to the social, economic, and market institutions prevailing in
the France in which he lived a thousand years later.27 But in other
respects, one can imagine the Turk to be like Descartes: The Turk is
politically conservative and unwilling to reform institutions built by
long custom and tradition; he is at home in what he finds. Question:
Would the Turk end up of necessity with exactly the same set of principal
beliefs as those of Descartes?
There are problems for Descartes on either answer. Suppose the
answer is yes, he would end up with exactly the same set of principal
beliefs as Descartes. (There is strong evidence to believe that Descartes
would have endorsed such an opinion; it would have testified to the
belief in the universality of human reason, a commonplace amid the
seventeenth century’s optimism.) But this answer is undermined by a
host of unheeded considerations. The Turk will begin with a different
concept of God from that 0f Descartes: Descartes finds the idea of one
good God self-evident; the Turk finds the idea of several gods clear and

26 The Turks “were at the time considered by all Europe to be the major danger”
(R, 210). “The sin that Turks and other infidels commit by refusing to embrace
the Christian religion does not arise from their unwillingness to assent to obscure
matters (for obscure they indeed are), but from their resistance to the impulses of
divine grace within them, or from the fact that they make themselves unworthy of
grace by their other sins” (CSM II, 105–106; AT VII, 148).
27 “In traveling through all these countries, Descartes extended his doubt to many
customs. Faithfulness to the religion of his childhood set him apart.” (R, 62) “I am of
the religion of my nurse,” said Descartes to the Protestant Revius (R, 208). So might
the Turk have traveled and doubted other customs and religions than his own.
32 The Prolegomena to Any Future Epistemology

indubitable. The Turk will have distinct notions of proof and reason,
and Descartes different ones of his own. The Turk may not learn to
doubt the reality of the external world simply because his senses have
deceived him in the past. The Turk, like some former-day Wittgenstein,
or lacking Descartes’ skeptical heritage, may not even be able to make
sense of that doubt. Indeed, the second moral maxim is so powerful
that it admonishes the Turk not to doubt his religious belief, namely,
polytheism. How could the second maxim fail to impinge on the sub-
sequent epistemic inquiry? How could a philosophical investigation
that led to theism be regarded as satisfactory by the Turk? Thus, so this
argument concludes, Descartes’ inquiry is merely geared to support
his initial beliefs, no more. Clearly, an affirmative answer to the origi-
nal question – would the Turk end up of necessity with exactly the same
set of principal beliefs as those of Descartes? – is in need of defense.
If the answer to that question, however, is no, the Turk will not end
up with the same set of principal beliefs as Descartes – a more likely
response – then Descartes needs an even more substantial argument
in favor of his moral code. He needs that argument as no less than a
fortress for his epistemology. For an inadequate moral code will con-
demn him to stray from “the right path,” as presumably it did the Turk.
But this Descartes cannot, in principle, do: For until Descartes has sub-
stantiated his method, which at this point he has not, he will have no
adequate means by which to determine the correctness of the moral
code. On the other hand, he cannot determine the correctness of his
method unless he has a moral code by which he can live. I conclude
that in his moral code, as in his method, Descartes is attempting to raise
himself by his own bootstraps. That is, he will provisionally accept a
moral code as the right one; he will subsequently discover the right
method; and then the right method will legitimize the moral code he
had initially adopted.28

28 For a similar pattern of argument, see, Jeffrey Tlumak, “Certainty and Cartesian
Method,” and James van Cleve, “Foundationalism, Epistemic Principles, and the
Cartesian Circle.” Whether these approaches are susceptible to the kind of objection
I level against Descartes, I leave to the reader to decide.
 2

The Problem of Epistemology

In the structure of a philosophical system, the form is at least as im-

portant as the content. Part of what I wish to demonstrate, but only
by implication, is that Descartes’ hold on later philosophy has a great
deal to do with its form, and perhaps less with its content. One can
quibble, for instance, about whether Descartes has given us an ade-
quate proof of the existence of God, or even of the existence of the
external world. One can doubt that his reasoning is sound about the
mind being a distinct and unique substance in its own right as matter,
or doubt that he is right about their essential properties being thought
and extension, respectively. But it would not follow that therefore his
way of doing philosophy is inadequate. Descartes has provided a form,
a way, a mode of doing philosophy that has not yet disappeared even
if no one takes seriously, for example, the idea that the pineal gland
in the brain is the meeting place of mind and matter. This chapter is
as much about the form of that kind of philosophizing as it is about
the specific issues in Descartes’ philosophy.
Let me, at least once, draw an explicit parallel between the form
of philosophy in Descartes and the form of philosophy employed by
some twentieth century philosophers of science: Descartes was after
not only truth, but also certainty; contemporary philosophers of sci-
ence are after truth or probable truth. Descartes used the method of
doubt to get rid of uncertain propositions; contemporary philosophers
of science recommend the use of experimental falsification to reject
scientific hypotheses. Descartes invoked a fundamental experience as a

33
34 The Problem of Epistemology

starting point; contemporary philosophers of science suggest invoking

paradigms and standard cases in the sciences as the point of departure.
Descartes laid down, in the light of that fundamental experience, a
strong and sweeping epistemic principle; contemporary philosophers
of science invoke a theory of corroboration in the light of paradigms
and standard examples. And finally, Descartes was after a deductive sys-
tem of knowledge; contemporary philosophers of science want some-
thing structurally quite similar, say, an inductive system of knowledge.
Descartes was among the first to provide us with a taxonomy of
problems – not just scientific problems, but philosophical problems
as well. Furthermore, he provided the conditions under which, for in-
stance, a solution is adequate and well understood. He argued that the
failure to solve a philosophical problem was oftentimes, in large mea-
sure, due to the failure to characterize the problem properly. Those
issues define the task of section I. In the next section, I distinguish be-
tween mental states and mental processes, as well as between first- and
second-order states and processes, the better to understand Descartes’
directive to dismantle what hitherto had stood as the structure of
knowledge. In section III, two models of doubt – the ordinary-doubt
model and the extraordinary-doubt model – are provided, and their
rationales underscored. In section IV, I make explicit the various steps
in the process of doubt, and argue against recent empiricist and causal
theories that purport to explain what Descartes was about in the First
Meditation.

I. Types of Problems
Let us begin somewhere in the back alley. I refer to rule 13 of Rules
for the Direction of the Mind. Here Descartes is focusing on the nature of
problems, the constraints on their solution, and the things that one
should do in order to have the best chance of solving them. Descartes
claims that, first, in every problem there must be something unknown.
Second, this unknown must be characterized in some way; such a char-
acterization will point us in one direction rather than in another. Thus,
the characterization is crucial, since a false characterization can put
us on a dead-end path. Third, the unknown can be characterized only
in terms of what is known, otherwise the problem could not even be
formulated or understood.
 I Types of Problems 35

These conditions hold for both perfect and imperfect problems.

For a problem to be a perfect problem, it must be determinate in
every respect. Its terms must be perfectly understood. Descartes of-
fers as an illustration the following: Suppose that someone raises the
problem, “[W]hat conclusions are to be drawn about the nature of
the magnet simply from the experiments which Gilbert claims to have
performed, be they true or false?” (CSM I, 52; AT X, 431) The point
is not to question the veracity or the reliability of the experiments, but
simply to deduce whatever we can from the knowledge of these ex-
periments. The three characteristics of every problem appear: First,
some of the properties of the magnet are unknown; second, the un-
known is characterized in some way (depending on how we understand
“nature”); and third, the unknown is constrained by the experimental
results.
In the previous rule, dealing with the same problem, Descartes
says,

But take someone who thinks that nothing in the magnet can be known which
does not consist of certain self-evident, simple natures: he is in no doubt about
how he should proceed. First he carefully gathers together all the available
observations [i.e., results of experiments] concerning the stone in question;
then he tries to deduce from this what sort of mixture of simple natures is
necessary for producing all the effects which the magnet is found to have.
Once he has discovered this mixture, he is in a position to make the bold
claim that he has grasped the true nature of the magnet, so far as it is humanly
possible to discover it on the basis of given observations. (CSM I, 49–50;
AT X, 427)

It is clear from the foregoing that Descartes believes that a per-

fect problem must have a unique solution. He seems to suggest that
in this work, but his examples leave the matter open. The magnet
case does not have a unique solution. One might have a variety of
characterizations of the unknown properties of the magnet, given the
observations, the data. Not that Descartes himself was unaware, as we
have already seen, of the problem of underdetermination by the data.
What Descartes is more likely to have had in mind here is a problem
in mathematics that does have a unique solution, such as, “What is the
sum of the interior angles of a triangle in a Euclidean plane?” Given
the nature of triangles, there is only one correct answer, namely, “It is
equal to two right angles.”
36 The Problem of Epistemology

On the other hand, the problem is imperfect if we are simply asked,

“What is the nature of a magnet?” Here the solution to the problem
is not conditioned in any respect. The terms are not perfectly under-
stood. Once again the three conditions of every problem are satisfied.
Note Descartes’ remark, “[W]e already understand what is meant by
the words ‘magnet’ and ‘nature’, and it is this knowledge which deter-
mines us to adopt one line of inquiry rather than another” (CSM I, 52;
AT X, 431). Presumably, a change in the meanings of the words, due to
the growth of scientific knowledge, would change the problem as well
as the line of inquiry. (Descartes thought that all imperfect problems
could be reduced to perfect problems, but he never completed that
part of his work. A conjecture on his behalf: If we let any imperfect
problem have as constraints any subset of the existing knowledge re-
lating to the thing we are investigating, then the imperfect problem
turns itself into a perfect problem.)
There is another demarcation of the class of problems. This is the
division between perfectly understood problems and those that are im-
perfectly understood. One should not suppose that a perfect problem
is simply a perfectly understood problem. There is more in the char-
acterization of the latter class of problems than in the characterization
of the former class. Thus,

We must note that a problem is to be counted as perfectly understood only if we

have a distinct perception of the following three points: first, what the criteria
are which enable us to recognize what we are looking for when we come upon
it; second, what exactly is the basis from which we ought to deduce it; third,
how it is to be proved that the two are so mutually dependent that the one
cannot alter in any respect without there being a corresponding alteration in
the other. (CSM I, 51; AT X, 429)

How do problems arise? They may arise due to the obscurity and
obfuscation of our language. Riddles belong to this class, such as the
riddle of the Sphinx, as well as those problems that turn on ambiguity
and vagueness of terms, such as the one about the anglers on the
shore who maintain, while holding the rod and line, that they no
longer have the fish they have caught but have the ones they have not
caught. The dispute of the learned is not infrequently about words,
such as the one about the definition of ‘space.’ Indeed, Descartes
declares that “[t]hese questions about words arise so frequently that,
 I Types of Problems 37

if philosophers always agreed about the meanings of words, their

controversies would almost all be at an end” (CSM I, 54; AT X, 434).
Almost all, but not all. There are genuine problems that are not the
result of careless use of language, or due to some inherent ambiguity
in the language. These are the problems in which one is asking if
something exists or, if it does exist, asking what is its nature, as in the
case of the problem of perpetual motion, or the problem of stellar
motion. Take the simpler problem posed by the statue of Tantalus
that Descartes once saw. The statue has a column at the top of which
is a figure of Tantalus. In this statue, water keeps pouring out of a
spout, but as soon as the fountain reaches Tantalus’ mouth, it runs
out. Descartes states that the important thing is to get to the essence
of the problem rather than to get bogged down in inessentials, such
as, how is the figure of Tantalus constructed? This, says Descartes,
is just a “coincidental feature” of the problem. Rather, “The whole
difficulty is this: how must the bowl be constructed if it lets out all the
water as soon as, but not before, it reaches a fixed height?” (CSM I,
55; AT X, 436) Thus, “no matter what the problem is, we must above
all strive to understand distinctly what is being sought” (CSM I, 54;
AT X, 434).
“[T]hree different sorts of questions,” says Descartes, should be
distinguished.

First, some things are believed through faith alone – such as the mystery of the
Incarnation, the Trinity, and the like. Secondly, other questions, while having
to do with faith, can also be investigated by natural reason: among the latter,
orthodox theologians usually count the questions of the existence of God,
and the distinction between the human soul and the human body. Thirdly,
there are questions which have nothing whatever to do with faith, and which
are the concern solely of human reasoning, such as the problem of squaring
the circle, or of making gold by the techniques of alchemy, and the like. Just
as it is an abuse of Scripture to presume to solve problems of the third sort
on the basis of some mistaken interpretation of the Bible, so it diminishes
the authority of Scripture to undertake to demonstrate questions of the first
kind by means of arguments derived solely from philosophy. (CSM I, 300;
AT VIIIB, 353)

Thus, on Descartes’ view there are problems that have genuine ra-
tional solutions, and there are some that have none, problems that
are solely the province of faith. As an example of the latter:“As for the
38 The Problem of Epistemology

question whether it is in accord with the goodness of God to damn

men for eternity, that is a theological question: so if you please you
will allow me to say nothing about it. . . . when truths depend on faith
and cannot be proved by natural argument, it degrades them if one
tries to support them by human reasoning and mere probabilities.”
(CSMK, 26; AT I, 153) But some problems of faith do have rational
solutions, such as the problem pertaining to the existence of God
and, against the Calvinists, the problem of understanding transub-
stantiation (CSMK, 88; AT I, 564). Finally, there are problems that
fall solely in the province of reason, where faith or scripture is in-
voked only mistakenly. “It is true,” – of course – “that we are obliged
to take care that our reasonings do not lead us to any conclusions
which contradict what God has commanded us to believe” (CSMK, 120;
AT II, 348).
My purpose in detailing this brief on the type of problems is to
ask, would Descartes have argued that a comparable situation prevails
in philosophy? That is, are there perfect and imperfect problems in
philosophy as well? Consider the philosophical problem, does God
exist? Is this a perfect problem? The term existence may be perfectly
understood. But is the term God clearly understood? Indeed, some
may argue that this problem may not be a genuine problem, for it
may conceal inherent contradictions.1 This gives rise to the possibility,
as in the case of mathematics, that some problems hitherto regarded
as genuine problems may later be shown to be unsolvable. Are there
unsolvable problems also in philosophy? Or are philosophical prob-
lems unsolvable only when they are couched in vague language?
(CSMK, 10–13; AT I, 76–82) How should philosophical problems
be characterized? What constraints should operate on them? By
what known factors should the philosophical unknown be character-
ized? Which philosophical problems, if any, are perfectly understood?
Which ones are imperfectly understood, and why? Finally, how will the
meanings of terms, or the conditions on the solution, direct the line
of philosophical inquiry? Interestingly enough, it was the particular
way in which the problem, can I know anything that I cannot doubt?,
was formulated that determined Descartes’ line of inquiry. Thus, he
doubted.

1 This was Leibniz’s well-known concern; see K, 128–131.

 II Directive to Dismantle 39

II. Directive to Dismantle

The motto of this section could well be, “The seeker after truth must,
once in the course of his life, doubt everything, as far as is possible” (CSM I,
193; AT VIIIA, 5). To begin to explain the nature and purpose of this
doubt, I resort to the frequent analogy in Descartes between rejecting
old, uncertain opinions in order to establish new ones or to secure old
ones on a firmer basis, and the dismantling of an old house in order
to build a new one:

And, just as in pulling down an old house we usually keep the remnants for use
in building a new one, so in destroying all those opinions of mine that I judged
ill-founded I made various observations and acquired many experiences
which I have since used in establishing more certain opinions. (CSM I, 125;
AT VI, 29)

And again:

For, since this knowledge is not enough to satisfy him, it must be faulty: I
would compare it to a badly constructed house, whose foundations are not
firm. I know of no better way to repair it than to knock it all down, and
build a new one in its place. For I do not wish to be one of those jobbing
builders who devote themselves solely to refurbishing old buildings because
they consider themselves incapable of undertaking the construction of new
ones. But, Polyander, while engaged upon this work of demolition we can use
the same method to dig the foundations which ought to serve our purpose,
and to prepare the best and most solid materials which will be needed for
building up these foundations. (CSM II, 407; AT X, 509)

The image that is conjured up is that of an architect who demol-

ishes a house by taking every part of the house – bricks and walls,
doors and window frames, tiles and roof, beams and stones – apart,
whether or not the parts are good and functional. The architect lays the
new foundation and then scrounges in the pile of dismantled things,
which once belonged to the house, to determine whether anything
in it is usable in the construction of the new house. Consequently,
one of three things can happen: The architect may find, first, that
no object is usable – they are all cracked, corroded, or damaged;
second, that some objects are usable and some not; or finally, that
everything is usable. In the last case, they would all be used to recon-
struct the new house, lie in equilibrium with one another, and rest
40 The Problem of Epistemology

on a fresh and solid foundation, a foundation that will be known

not to give way. As it turns out, it is the second possibility that is
eventually realized.
Here is a perceptive contrast with Kant’s invoking of the building
analogy.

Like Descartes, Kant uses metaphors of construction to explain his view

of philosophical method; but he starts with a more down-to-earth view of
building projects. The result is a quite different vision of philosophical pro-
cedures. We are to look back on what has been done in “Transcendental
Doctrine of Elements” as an estimate or inventory of our building materials,
which has instructed us about some constraints on what we can build. The
result is in some ways disappointing, especially when matched against the
rationalist ambition to build “a tower that would reach the heavens.” How-
ever, rationalism failed because it took no account either of the paucity of
materials or of the disagreements about the plan among the fellow workers.
It relied on the fiction of a unitary and authoritative architect, whose innate
ideas correspond to their real archetypes, to construct the edifice of human
knowledge.2

A full Cartesian response is still awaited. But, I venture, Descartes

would have asked of Kant the following questions: What makes for the
paucity of the inventory, namely, the human material? What leads to
the disagreement about the plan? Is the low estimate of the material
due to an inherent defect in the material or to an essentially eradicable
defect? If the former, then there is no way to resolve conflicts over a
plan, no matter how long, elaborate, or frequent the public use of
reason. If the latter, what makes the eradication of the defect possible,
other than the inherent nature of the individual reasoner and his
ability to make a correct decision? Then, in what way is that inherent
nature or ability to be described in non-Cartesian terms? Or, how can
this possibility rest on a non-Cartesian method? Must not the reasoning
parties ultimately share something in common if they are to arrive at
an agreement?
If the Kantian argument is that Descartes was guilty of not allow-
ing public use of reason in building his project because it would have
hastened that project, Descartes might have conceded the point. But he
might still have insisted that something like a community of Cartesian

2 Onora O’Neill, Constructions of Reason: Explorations of Kant’s Practical Philosophy, 11–12.

 II Directive to Dismantle 41

reasoners would be required if a final agreement is to be reached.

Perhaps we might profitably combine the two views. Here is an oft-
quoted remark from the Critique of Pure Reason: “The very existence of
reason depends upon this freedom, which has no dictatorial author-
ity, but whose claim is never anything more than the agreement of
free citizens, each of whom must be able to express his reservations,
indeed even his veto, without holding back”(A738–739/B7667–767).
The free citizen would profit from the Kantian idea of the public use of
reason – freely and openly conducted – no less than from the Cartesian
idea of exercising reason in private. Both are necessary, perhaps, in the
final analysis.
How he arrives at the foundation stone, we shall see later. What we
now need to do is to see how the architect examines the various objects
and determines their fitness for the construction of a new house. The
house corresponds to the system of beliefs; the various objects in the
house correspond to the various beliefs; and the design and structure
of the house correspond to the structure of, and the relationships
among, the beliefs. It is these particular beliefs, and the relationships
among the beliefs, that a system builder needs to examine.
Examine, then, the following propositions:

[1] Placebos can prevent pregnancy.

[2] Mount Everest is not the highest mountain in the world.
[3] Phenotypic variance is composed of three main sources of
variation: genotypic, environmental, and interaction.
[4] There is a piece of paper in front of me.
[5] 2 + 9 = 11

With respect to the foregoing propositions, if the reader is anything

like me, he will harbor extreme doubt about [1]. He will have some
doubt about [2], a doubt that can be settled by consulting a book
of maps. Of [3] he will be unsure rather than doubtful, and would
have to reread a book on genetics in order to determine its truth-value
(incidentally, it is true). Of [4] he will be quite sure. Of [5] he will be ab-
solutely certain. These propositions are arranged in a sequence begin-
ning with utter doubt about a certain proposition and ending with utter
certainty about another. In other words, the sequence of propositions
moves from one about which we have grave and genuine doubts, to
ones about which we have weak doubts, and finally to one about which
42 The Problem of Epistemology

we have no doubt at all. As Descartes will show us, there is room for one
more kind of doubt: exaggerated, pretend, hyperbolic, or metaphys-
ical doubt. This kind of doubt, its purpose served, he will laugh about
later. This style of doubt is meant to cast [4] and [5] in a different
light.
The following diagram corresponding to our beliefs represents de-
grees of doubt by the number of lines in each circle:

[1] [2] [3] [4] [5]

Degrees Of Doubt

Next, let us distinguish a mental state from a mental process. I shall

assume the term “mental state” to be a simple, undefined term.
A mental process is a collection of mental states. There are, of
course, myriad mental states, states that we are in when we com-
mand, desire, compel, ask a question, are happy, or are in pain or
deep sorrow, as King Lear was when he said, “Never, never, never,
never, never.” However, I shall confine myself to those mental states
that a thinker is in when he is entertaining a declarative proposi-
tion, that is, a proposition that is either true or false. Such propo-
sitions may range from “There is salt on the table” to the recent
pronouncements in high-energy physics. Some mental processes are
more coherent than other mental processes. There is a more rational
sequence of mental states in some mental processes, a less rational
sequence in others, as when one is free-associating, daydreaming,
or in a reverie. I shall deal with more or less coherent processes.
Thus:

[6] All white dwarfs are neutron stars.

[7] All collapsed stars are white dwarfs.
Therefore,
 II Directive to Dismantle 43

[8] All collapsed stars are neutron stars.

[6] [7] [8]

[9]

[9] The argument is valid.

My believing proposition [6], All white dwarfs are neutron stars, is

a mental state; on the other hand, my mental process is one consisting
of three mental states, namely, beliefs in [6], [7], and [8], respectively.
There is an argument, and corresponding to it is a mental process. One
might be doubtful when in a mental state corresponding to proposi-
tions [6], [7], and [8], respectively, and yet have no doubt when in a
mental state corresponding to proposition [9]. In fact, the premises
are all false, and so presumably one can be in a state of doubt with
respect to each of these three propositions. But the argument is a
simple one, a categorical syllogism, which the logicians call Barbara,
and it is valid: If the premises are true, the conclusion cannot be
false.
Finally, we must distinguish between first-order mental states and pro-
cesses and second-order mental states and processes. My second-order
mental state is one in which I believe the proposition

[9] The argument is valid.

which is about my first order mental process. It is entirely possible

that first-order states may be unclear and confused while the second-
order state, which is about the first-order state, is perfectly clear and
beyond doubt. Thus, [3] is dubious to me; but my second-order men-
tal state, namely, the belief that I do not know the truth-value of [3],
seems quite clear. So we have several possibilities if we commute a
44 The Problem of Epistemology

first-order mental state (or process) with a second-order mental state

(or process), defining each state (or process) as either clear or doubt-
ful. For instance, a first-order mental process may be doubtful, but
a second-order mental process about that first-order mental process
may be clear; a first-order mental state may be clear, but a second-order
mental state about that first-order mental state may be doubtful, and
so on.
Now, Descartes’ enterprise can be described in terms of the forego-
ing as follows. When an ideal investigator searches for truth, is there
any mental state or mental process in which he has no doubt3 – that is,
speaking pictorially, in which there are no lines in the diagram of that
mental state (or process) at all? Beliefs about [1] and [3] clearly could
not serve as candidates, since they are infected by doubts. Could be-
liefs in [4] and [5] serve as candidates? One would be hard put to find
clearer or more cogent beliefs. Descartes suggests that we should not
just rely on ordinary doubts to reach indubitability, but that we should
strive, as far as the human mind can be bent and stretched, to induce
doubts in ourselves by hook or by crook until we can doubt no more.
The task is to ensure that our doubt is sufficiently sweeping, that the
mind actually feels the pressure of doubt; the doubt should extend to
anything and everything until it arrives at a point that it cannot touch
or infect with doubt.

3 One might object to this formulation on the grounds that Descartes is not so much
after indubitable mental states (of whatever order) as he is after indubitable propo-
sitions. However, this objection overlooks the Cartesian view that language is merely
an instrument for dressing up one’s thought. What is primary is thought. Jill Vance
Buroker makes the point pithily: In Descartes, “thought is prior to language, that
words are merely external, conventional signs of independent, private mental states.
On this view, strictly speaking, linguistic utterances signify the thoughts occurring in
the speaker’s mind. Although the association between words and ideas is conventional
and thus arbitrary, language can signify thought insofar as both are articulated sys-
tems: there is a correlation between the structure of a complex linguistic expression
and the natural structure of the ideas it expresses.” Introduction, xxiii.
Given the correlation between the structure of a complex linguistic expression and
the natural structure of the ideas it expresses, one might say that given any mental state
(of whatever order) one can formulate a corresponding proposition. Having arrived
at one, one can easily arrive at the other. But the objection raises some nice issues.
For example, to what extent is the truth of the cogito language-dependent? Could a
speaker of the most rudimentary language – a language sans science, logic, mathe-
matics, or metaphysics – come to recognize the truth of the cogito? What minimum
linguistic resources would be needed for such a recognition?
 III Two Models of Doubt 45

This is not overstated. Descartes admonishes himself thus:

My habitual opinions keep coming back, and, despite my wishes, they capture
my belief, which is as it were bound over to them as a result of long occupation
and the law of custom. I shall never get out of the habit of confidently assenting
to these opinions, so long as I suppose them to be what in fact they are, namely
highly probable opinions. . . . I think it will be a good plan to turn my will in
completely the opposite direction and deceive myself, by pretending for a time
that these former opinions are utterly false and imaginary. I shall do this until
the weight of preconceived opinion is counter-balanced and the distorting
influence of habit no longer prevents my judgement from perceiving things
correctly. (CSM II, 15; AT VII, 22)

Three remarks are in order. First, either his doubts are real or
his doubts are hyperbolic, a mishmash of these two kinds of doubt
not being possible. Now, he has certainly not demonstrated the for-
mer. But only if his doubts are real can he lose confidence in as-
senting to his customary opinions and genuinely tilt his will in the
direction he wants. If his doubts are hyperbolic, as he avers, then
the weight of preconceived opinion cannot, or will not, be counter-
balanced and the distorting influence of habit cannot, or will not,
be set aside. Second, the preconceived opinion cannot both be dis-
torting, preventing correct judgment, and be highly probable. Third,
it is not necessary that he turn his will in completely the opposite di-
rection and deceive himself. It is enough for his epistemic enterprise
if he can keep clearly distinguished in his mind real doubts and hy-
perbolic doubts, but keep enlarging the domain of hyperbolic doubts
until their domain can be increased no further. With each piece of
certain knowledge gained, that domain will shrink proportionally; or,
he will discover that the domain of hyperbolic doubts is simply here
to stay.

III. Two Models of Doubt

Let me explain the structure of what I shall call the ordinary-doubt model.
Consider again

[1] Placebos can prevent pregnancy.

What does it mean to say that I doubt it? It means that my knowledge of
the causal lines that constitute the world’s make-up, in particular, the
46 The Problem of Epistemology

causal lines that are responsible for conception, contravenes claim [1].
Essential to conception is the process of ovulation in women and the
entry of the sperm into the ovum. Ovulation is controlled by hormones
that are secreted by the pituitary gland. Sperm cells are produced in the
testes, stored in the epididymis, and passed through ejaculatory ducts.
Conception occurs in the oviduct, and if fertilization occurs, then the
egg becomes embedded in the uterine wall. Thus, if the fallopian tubes
or the oviducts are blocked, conception will not occur. If the passage to
the ovum is not blocked in any manner, then a sperm cell can find its
way there; often it will. Then it may fertilize the ovum, and conception
will occur. But a placebo has no known causal property, as do some
contraceptives, by which it can block that passage. Therefore, I doubt
that [1] is true.4
This model of ordinary doubt presupposes the correctness or
the reliability of several things about human sexuality mentioned in
the foregoing paragraph. For example, it presupposes the truth of
the following: Ovulation occurs; sperm are produced in the testes;
fallopian tubes are not blocked; and ejaculatory ducts are not blocked.
If these propositions too were questionable, and no reliance could be
placed on them, then I could not properly doubt in the ordinary way.
The denial of [1] would be just as much open to doubt as its affir-
mation. Should we conclude, then, that a model of ordinary doubt
consists of a doubted proposition that necessarily presupposes a frame-
work of propositions believed to be certain? Can a proposition be doubted
with no certain propositions to rely upon at all?5

4 I am not drawing any sharp distinction between doubt and disbelief. If disbelief is
a stronger notion than doubt – so that what is disbelieved is doubted, while what is
doubted may not necessarily be disbelieved – then I am dealing here only with doubt.
I proceed to show how genuine doubt leads by stages to hyperbolic doubt, and it is
by no means obvious, nor is it my task to show, that they lie on the same continuum.
This I take to be precisely Descartes’ enterprise.
5 “However, the menace of universal shamming is an empty menace” – Gilbert Ryle,
The Concept of Mind, 166. H. A. Prichard claimed, “We can only be uncertain of one
thing because we are certain of something else” (“Descartes’s Meditations”). Ludwig
Wittgenstein proclaimed, “The game of doubt presupposes certainty. If you tried
to doubt everything, you would not get as far as doubting anything” (On Certainty,
section 115). The last two are quoted in James van Cleve’s “Foundationalism, Epistemic
Principles, and the Cartesian Circle,” 108. Van Cleve’s own reaction to these assertions
is to argue that one must distinguish between the ground for certainty and the source of
certainty. “We may sum things up thus: doubt presupposes that something is certain; so
 III Two Models of Doubt 47

The answer to the last question – in the ordinary, practical context –

is no, it cannot. Each doubt must be supported, in that context; indeed,
the author of the “Seventh Set of Objections” insisted that any doubt
in any context must be so supported. But that is not what Descartes
was after. Descartes simply contended that if a mere possibility can be
established that a proposition is false, then it is, for the epistemological
purpose at hand, to be regarded as false. Thus, a doubt may be a meta-
physical doubt, yet it serves this purpose: to “reject as if absolutely false
everything in which I could imagine the least doubt, in order to see
if I was left believing anything that was entirely indubitable” (CSM I,
127; AT VI 31–32). I am unsure what precisely Descartes means by ab-
solutely false – or wholly false, as he says elsewhere – that is not captured
by false. One might say that Descartes was fashioning an extraordinary-
doubt model. This is a doubt that presupposes no true propositions.
“There may be reasons,” says Descartes, “which are strong enough
to compel us to doubt, even though these reasons are themselves
doubtful, and hence are not to be retained later on” (CSM II, 319;
AT VII, 473–474).
Let me illustrate the nature of this extraordinary doubt by con-
sidering three distinct cases: (a) the probabilistic case,(b) the deter-
ministic case, and (c) the mathematical case. (a) Suppose we wish to
determine what effect high blood pressure has on the human heart.
More specifically, we wish to determine if high blood pressure is the
cause of cardiac arrest. We take a sample of 5,000 (in routine cases,
sample sizes are much smaller) individuals varying in age, traditions
of eating, habits of exercise, profession, culture, social status, sex, and
religion. This guarantees sufficient diversity and ensures that there
is no bias in our sample; in short, our sample is a random sample.
But individuals in the sample have one thing in common, namely,
they all suffer from high blood pressure. These individuals, the con-
trol group, are not treated for that disease. We take another similar
group of 5,000 individuals, except these individuals are treated for
high blood pressure. We trace the life histories of the individuals in the

far, Prichard and Wittgenstein are right. But the ground on which one bases a doubt
need not be certain; on this point Descartes was right.” (110) This is, in fact, quite a
concession, as will become clear as this book unfolds; in any event, it does scant justice
to Descartes: For the details of my argument against van Cleve, see Chapter 4.
48 The Problem of Epistemology

two respective samples. In the first group, 650 individuals experience

cardiac arrest; in the second group, 150. Thus, the difference between
the two random samples is 650 − 150 = 500. If we want our test re-
sults to be significant at three standard deviations (most investigations
make do with only two standard deviations), then we require the dif-
ference between the two samples to be 500 (plus or minus 150). As
a matter of fact, the difference between our two samples is 500. Con-
sequently, our hypothesis is corroborated at a 99.7 percent level of
confidence.
Would this satisfy Descartes? No, it would not, because there is
a clear, but extremely small, probability that this was a chance oc-
currence, and that high blood pressure has no bearing on a per-
son’s suffering cardiac arrest. Even very, very high probability is not
the same thing as absolute certainty. And that is what Descartes is
seeking.
(b) Consider the deterministic case: Suppose that Newton’s three
laws of motion (or Descartes’ own version of them) plus the gravi-
tational law entirely determined the observable, physical world. Sup-
pose further that no observation yet known to man falsified the theory.
Would this satisfy Descartes? No, because an alternative to Newton’s
first law can be conceived. As it stands, the first law (in the modern
version) states: If an object is at rest, it will continue to stay at rest,
or if it is in motion, it will continue to move in a straight line, if un-
hampered by any external force. But to cast a mere minimal doubt on
this proposition, we need only to conceive the possibility of an object
moving in a circular motion when unhampered by any physical object.
In the days of Ptolemy, astronomers believed that the natural motion
of celestial objects was circular. So even the complete empirical ade-
quacy and success of a theory of a deterministic system is not the same
thing as certain knowledge.
Consider, finally, (c) the mathematical case. Here is the proof of a
simple mathematical theorem, namely, that in an equiangular triangle,
all the sides are equal:

[1a] ∠B = ∠C, by hypothesis

[2a] AD is the common side
[3a] BD = DC, by construction
[4a] ∴ ABD = ADC
 IV Two Models of Doubt 49

[5a] ∴ AB = AC

B D C

[6a] Similarly, AB = BC
[7a] ∴ AB = AC = BC

Would this satisfy Descartes (before the discovery of the cogito)? No,
because there might be a slip on our part unbeknownst to us. Or there
might be a deceiver who makes us think that we have a proof, when in
fact we do not. For example, we have invoked in this proof the principle
that when two things are equal to a third, they are equal to one an-
other. It was this principle that enabled us to infer [7A] from [5A] and
[6A]. But perhaps that principle, no matter how perspicuous, may be
false. Descartes says explicitly in the Discourse on the Method: “And since
there are men who make mistakes in reasoning, committing logical
fallacies concerning the simplest questions in geometry, and because I
judged that I was as prone to error as anyone else, I rejected as unsound
all the arguments I had previously taken as demonstrative proofs”6
(CSM I,127; AT VI, 32). In Principles of Philosophy, in paragraph five, en-
titled The reasons for doubting even mathematical demonstrations, Descartes
says, “Our doubt will also apply to other matters which we previously re-
garded as most certain – even the demonstrations of mathematics and
even the principles which we hitherto considered to be self-evident”
(CSM I, 194; AT IXB 6).

6 In parodying the Second Meditation, Hyperaspistes (an assumed name of one of

Descartes’ correspondents) had said that a sceptic could say, “Let the evil demon
deceive me as much as he can, he will never deceive me about this geometri-
cal proposition,” to which Descartes had responded that “anyone who says this
is by that token not a sceptic, since he does not doubt everything” (CSMK, 196,
note 5; AT III, 433–434). Presumably, then, the doubting of mathematical proofs
is very much in order if one is to be a Cartesian skeptic in search of the indu-
bitable. To be fair, however, the context in which Descartes writes this to Hyper-
aspistes is different; but that does not affect my purposes, or Descartes’ intentions,
here.
50 The Problem of Epistemology

IV. Doubt and Principles

In light of the foregoing, we can see the structure of Descartes’ own
doubt. It is this. It is impossible – or, as he puts it, “an endless task”
(CSM II, 12; AT VII, 18) – to examine each of his beliefs and determine
if he can find reasons to induce even the least doubt about those beliefs.
There are just far too many beliefs. But the beliefs rest on certain basic
principles, and if these principles can be undermined, then all of the
beliefs that rest on the principles can be made to collapse; they can
be “demolished.” Note that Descartes assumes without argument that
these so-called principles are finite. It appears that these principles are
associated with sense experiences and reason, and since the latter are
finite, it is tacitly assumed that the principles are finite.
What, in this context, did Descartes mean by principles? He must have
meant the five senses and the intellect. There are, so to speak, the sight
principle, the smell principle, the sound principle, the touch principle,
the taste principle, and the intellect principle. (I distinguish between
the intellect principle and the principle of intuition.) Otherwise, it is
difficult to see how Descartes could have avoided an endless task. For
let us suppose that numbers are the principles of arithmetic; lines, the
basic principles of geometry (prior to the Cartesian analytic geometry);
humors, the basic principles of medicine; earth, air, water, and fire,
the basic principles of geology; and we can easily extend the list. We
doubt the numbers, so arithmetic collapses; we doubt the humors, so
medicine collapses. If such a list of basic principles is not finite, then
the task is just as endless. But the knowledge of these basic principles
is derived from the sense organs or from the intellect. Consequently,
if the occasional testimony of these organs or of the intellect could be
impugned, the task would be finite and manageable. This is exactly
what Descartes attempts to do: to impugn the testimony of the five
senses and the intellect.
He now commences his doubt. The least, or minimal, doubt of
which Descartes speaks can be established in a variety of ways. First
step. Many of his beliefs – say, about far and near objects, or objects
that are big or small – are “acquired either from the senses or through
the senses” (CSM II, 12; AT VII, 18). But the senses have sometimes
patently deceived: For example, square towers, at a distance, have been
judged to be round. Thus, the sight-principle is impugned. Consider
 IV Doubt and Principles 51

once again the proposition

[4] There is a piece of paper in front of me.
How can I doubt it ? By challenging the testimony of the witnesses.
In this case, the witnesses are my five senses. These witnesses report
that the object before me looks like paper, smells like paper, feels like
paper, and so on. How can I challenge these witnesses? By pointing out
that they have frequently been unreliable; and since they have perjured
themselves in the past, so to speak, their testimony should be scratched
from the books. Indeed, “it is prudent never to trust completely those
who have deceived us even once” (CSM II, 12; AT VII 18). Thus it is that
the sound-principle, the taste-principle, the smell-principle, and the
touch-principle are impugned, too. Descartes did not aim to establish
that propositions like [4] are in fact false, but rather that they are to
be presumed false, if not proved true. If they are not proved true, he
would withhold assent to them.
Second step. Some beliefs seem too stable and solid to deny, even
though they are “derived from” the senses. For example, the belief
that I am awake, wearing sandals, sitting on the lawn, and looking at
an orange tree. Or that I have ears, eyes, hands, and a body. To deny
these claims is to border on insanity. At this point comes the argument
from dreams. My dream experience cannot be distinguished from my
waking experience with certainty: “I see plainly that there are never any
sure signs by means of which being awake can be distinguished from
being asleep” (CSM II, 13; AT VII, 19). (What if waking experiences
were always colored and dream experiences were always noncolored,
and each possessed the usual properties of clarity, coherence, and
continuity? Would we then be able to distinguish waking experiences
from dream experiences?) But if we demolish the distinction between
dream experiences and waking experiences, then it is possible that
my experiences of my body, physical objects, and physical states are
nonveridical. It is possible that I am all – and nothing but – mind or
mental states, and have no body or bodily states.7
Third step. Can anything be salvaged? My dream experiences are
like paintings. Now, an artist can paint or create “sirens and satyrs with

7 Of the very many expositions of the dream argument in the literature, none supersedes
Edwin M. Curley’s in Descartes against the Skeptics, Chapter 3, “Dreaming.”
52 The Problem of Epistemology

the most extraordinary bodies” (CSM II, 13; AT VII, 20), but there
are some things in the paintings that are not, and cannot be, the
creation of the artist, namely, colors, corporeal nature in general, its
extension, quantity, size, number, place, and time. Thus, an artist can
create an image of an animal that has the trunk of an elephant, the feet
of an ostrich, the head of a crocodile, the tail of a polar bear, and the
ears of a poodle. He can create composite objects, such as the idea of
the head, ears, and feet, but he cannot create simple things, such as the
idea of size, time, quantity, or place. These are called simple, general,
and universal things. Here Descartes relies on the (causal?) principle
that what is imagined ultimately “must have been fashioned in the
likeness of things that are real” (CSM II, 13; AT VII, 19). He never ques-
tioned the causal principle, which would have given him a natural entry
to doubting these things as well. In those pre-Humean days, that was
perhaps too much to question. “So a reasonable conclusion from this,”
says Descartes, “might be that physics, astronomy, medicine, and all
other disciplines which depend on the study of composite things, are
doubtful; while arithmetic, geometry and other subjects of this kind,
which deal only with the simplest and most general things, regardless
of whether they really exist in nature or not, contain something
certain and indubitable” (CSM II, 14; AT VII, 20). Perhaps these are
salvaged?
Fourth, and final, step. Whether in my dream experiences or in wak-
ing experiences, the truths of arithmetic, such as [5], always hold,
as do the propositions of algebra and geometry. These propositions
about the simplest and most general things do not make any existen-
tial claims, and are not dubious in a way in which propositions are
that make existential claims. Has Descartes, at last, found something
certain? Not quite. It is the model of extraordinary doubt that drives
Descartes to invent the hypothesis of an evil demon. Here, then, is that
(in)famous hypothesis:

I will suppose therefore that . . . some malicious demon of the utmost power
and cunning has employed all his energies in order to deceive me. I shall
think that the sky, the air, the earth, colours, shapes, sounds and all external
things are merely the delusions of dreams which he has devised to ensnare
my judgment. I shall consider myself as not having hands or eyes, or flesh, or
blood or senses, but as falsely believing that I have all these things. (CSM II,
15; AT VII, 23)
 IV Doubt and Principles 53

The demon’s powers of deception are fantastic. This hypothesis in

hand, Descartes can now contest [5] and its ilk.
Descartes first postulated the existence of an omnipotent God, his
Maker, who has “brought it about that there is no earth, no sky, no
extended thing, no shape, no size, no place, while at the same time
ensuring that all these things appear to me to exist just as they do now”
(CSM II, 14; AT VII, 21). Yet how could a good God even occasionally
deceive, let alone often or continually deceive? But we are deceived,
and this requires an account. Consequently, second, Descartes sup-
posed that he was created “by fate or chance or continuous chain of
events, or by some other means” (CSM II, 14; AT VII, 21). However,
the less powerful and random the cause of his existence, Descartes
concluded, the more, not less, liable he would be to error. Descartes,
third, supposed that he was being manipulated by some “malicious
demon of the utmost power and cunning,” and that this demon can
actually lead Descartes to believe that 2 + 9 = 11 when in fact it does
not add up to 11.
One can see that this is an extraordinary doubt. There is no presup-
position of a framework of true propositions. There is no argument
to establish the evil demon; there is no argument to establish that he
is bent on deceiving us; thus, there is no argument to establish that
one could believe propositions that are negations of the laws of math-
ematics as we know them. There is simply a fiat: There is a malicious
demon, and he can deceive a person about the laws of mathematics.
Such a stipulation would cast doubt on proposition [5]. But truths
such as [5] or the propositions of arithmetic, algebra, geometry, and
the like are known by the intellect. Consequently, the intellect prin-
ciple is impugned. The demon hypothesis has taken Descartes to the
outermost limit of human doubt.
I am unsympathetic to Sir Karl Raimund Popper’s criticism that if we
eliminate all of our existing knowledge in order to start afresh, there is
no reason why we would advance any further than did Adam and Eve.
Thus, inConjectures and Refutations: The Growth of Scientific Knowledge,
Popper says,

Some people say that . . . it is their greatest wish to clean the canvas
thoroughly – to create a social tabula rasa and to begin afresh by painting
on it a brand new social system. But they should not be surprised if they find
that once they destroy tradition, civilization disappears with it. They will find
54 The Problem of Epistemology

that mankind have returned to the position in which Adam and Eve began –
or, using less biblical language, that they have returned to the beasts. (344)

Nor is this just a social hypothesis; it is also an epistemological one:

In other words, you should study the problem situation of the day. This means
that you pick up, and try to continue, a line of inquiry which has the whole
background of the earlier development of science behind it; you fall in with
the tradition of science. It is a very simple and a decisive point, but nevertheless
one that is often not sufficiently realized by rationalists – that we cannot start
afresh; that we must make use of what people before us have done in science.
If we start afresh, then, when we die, we shall be about as far as Adam and Eve
were when they died (or, if you prefer, as far as Neanderthal man). (219)

Descartes wants to wipe the epistemic slate clean in order for us to write
on it again properly. This time we are to write on it, not just a medley of
conjectures and refutations, but rather systematic, certain truths. But
any attempt to get rid of existing knowledge – by the method of doubt,
for instance – would prove injurious to science and knowledge. Thus
far, Popper. Descartes’ response to Popper’s objection would be that
if a given scientific society were to start in the way in which Descartes
recommends that an ideal seeker should proceed, then, of course, in
the beginning such a society might do only as well as, or slightly better
than, our ancestors. But in the long run, with each such successive
society running on the Cartesian principles and bequeathing certain
truths to the next – truths determined by the criteria of certainty –
there would be no necessity for each society to start afresh. Descartes
would argue that each scientist in each society should, in principle,
be able to start afresh in order to see clearly and distinctly that the
principles of his science are well founded. Such certainty the scientist
cannot have unless he proceeds in a Cartesian fashion. This, of course,
does not diminish Popper’s central point against rationalists such as
Descartes: that seeking certain knowledge in the empirical sciences is
a pied piper’s dream.
Harry Frankfurt has argued, in Demons, Dreamers, and Madmen:
The Defense of Reason in Descartes’ Meditations, that the “principles”
Descartes was attacking were simply epistemic policies “to be followed
in determining whether or not to accept a belief ” (33–34). Descartes’
skepticism in the First Meditation, said Frankfurt, lead him to for-
mulate more and more precise epistemic policies; however, since
these policies ultimately relied on sensory evidence, they could be
 IV Doubt and Principles 55

regarded as falling essentially under a single principle. Thus, what

I call the sight-principle, the smell-principle, the sound-principle,
the touch-principle, and the taste-principle fall essentially under a
single principle, the sensory principle. (Presumably the intellect-
principle falls under it as well? Or not?) Given this single sensory
principle, Frankfurt must inevitably adopt the position that mathemat-
ical statements, discussed at least in the opening meditation, are to be
construed as empirical statements. Descartes’ final goal in the First
Meditation, as Frankfurt sees it, was to demonstrate the untenability
of any such epistemic sensory principle, however precisely formulated.
Margaret Wilson found Frankfurt’s view inadequate, especially past
the dreaming argument; but she was not quite happy either with the
way Frankfurt handled the arguments preceding the dreaming argu-
ment. She offered an alternative account of what Descartes meant by
“principles” (WM, 5–6, 38–42). Wilson’s causal account is specifically
offered as a rival to Frankfurt’s account of principles as rules of
sensory evidence. Her one significant reservation, for example, is
that “Descartes’ view is not merely that sensations provide us with
an inadequate basis for distinguishing true perceptions from false.
Rather, Descartes thinks the senses actually mislead us concerning
what the world is really like.” (WM, 39) Furthermore, she says, there is
no adequate textual support for the suppositions of Frankfurt: namely,
(a) that simple natures are regarded as derived from the senses, and
(b) that mathematical knowledge is dependent on existence in nature
and on our sensory apparatus. Descartes, she maintained, was just
simply noncommittal on these issues, at least in the First Meditation
(WM, 40).
Consequently, she offered her own causal account of principles and
said of it:

If this perspective is valid, the arguments of Meditation I are indeed directed

against different versions of one principle, but the ‘foundation’ attacked is
not, as Frankfurt holds, faith in the senses specifically; it is rather faith in the
truth-conferring nature of the immediate causes of our beliefs. . . . Descartes
does not need specifically to suppose that the cause of the belief in simple
natures or mathematical propositions is sensory in order to question that they
are founded on a truth-conferring causal process. (WM, 42–43)

There are three fundamental difficulties in Wilson’s approach. First,

her causal theory is an ontological theory, not an epistemic one. It is not
56 The Problem of Epistemology

inappropriate to ask, what is the basis for questioning our faith in the
truth-conferring nature of the immediate causes of our beliefs? To an-
swer that question, it is difficult to imagine Wilson, while staying within
the Cartesian strictures, falling upon an approach, that is distinct from
the Frankfurt, or Frankfurt-like, approach. One must perforce fall back
upon sensory evidence, at least up to the dreaming argument.
Second, if one is to doubt mathematical knowledge, then one must –
despite Wilson’s remarks to the contrary quoted earlier – assume that
Descartes was not noncommittal, but rather distinctly empiricist in
his views about mathematical knowledge. If not, it is difficult to see –
even on minimalist assumptions – to what mathematical knowledge
is causally hooked. Paying close attention to analogous passages in
similar places in the Discourse on Method also reinforces one’s belief
that Descartes did not hold an empiricist view of mathematics in the
First Meditation. This is a view for which I have also heard Gordon
Baker argue. Perusing Descartes’ Dualism, the basis of his and Katherine
Morris’s claim becomes clearer. The authors write:

Descartes repeatedly tackled the common error of confusing modes and sub-
stances, i.e., of failing to distinguish what is “modally distinct”. He [Descartes]
suggested that mathematicians had misconceived what they investigated
through taking numbers or shapes to be substances. They will achieve a clear
and distinct understanding of these things provided they scrupulously avoid
“tack[ing] on to them any concept of substance” and “do not regard order or
number as anything separate from the things which are ordered or numbered”
[CSM I, 211; AT VIIIA, 26]. (68–69)

Arguably, when Descartes speaks of 2 + 3 = 5, in the First Meditation,

he must mean two birds (or stones, or trees), added to three birds
(or stones or trees), yields five birds (or stones or trees) (CSM I, 212;
AT VIIIA, 27–28).
Let us ignore Frege’s devastating arguments against reading arith-
metical truths in this way; let us simply concentrate on what Descartes
himself says, for example, in the Fifth Meditation:

But I think the most important consideration at this point is that I find within
me countless ideas of things which even though they may not exist anywhere
outside me still cannot be called nothing; for although in a sense they can
be thought of at will, they are not my invention but have their own true and
immutable natures. When, for example, I imagine a triangle, even if perhaps
 IV Doubt and Principles 57

no such figure exists, or has ever existed, anywhere outside my thought, there
is still a determinate nature, or essence, or form of the triangle which is im-
mutable and eternal, and not invented by me or dependent on my mind.
(CSM II, 44–45; AT VII, 64)

These are the truths – not merely empirical ones – that Descartes set
out to demolish with the help of the evil genius, the better to show
how the latter would be rendered helpless in the face of the cogito,
thereby demonstrating the power of the truth of the cogito. Otherwise,
Descartes’ skepticism in the First Meditation would have been a rather
limited thing. (There remains the task, of course, of showing how the
Meditations on First Philosophy dovetails with Principles of Philosophy.)
Third, it is not clear how either Frankfurt or Wilson would be able
to account for Descartes’ challenging the principles of logic, explicitly
referred to in the Discourse on Method (for example, at CSM I, 119–
120; AT VI, 17–18) and strongly implied in the First Meditation by
the phrase “and other subjects of this kind” (CSM II, 14; AT VII, 20) –
“this kind” presumably referring to arithmetic and geometry.8 Perhaps
what I call the intellect principle might be considered an important
principle, too, that Descartes was just as much concerned to challenge;
and perhaps this might raise afresh the issue concerning the validation
of reason.9 It is true that the view I offer does not have the nice unity
that is to be found on either Frankfurt’s or Wilson’s approach. I call
for at least two distinct kinds of principles – a sensory principle and
an intellect principle; they can make do with just one. But that is
Descartes’ problem, not mine (although, I confess, I do not know
what this lack of unity destroys in substance).
Doubting seems to lead nowhere. The building has been disman-
tled, and nothing salvageable is in sight. This led the Jesuit Pierre
Bourdin (1595–1653) to ask despairingly, “Is there no sole surviving
timber from the great shipwreck that is to be hung up as an offering
in the temple of truth?” (CSM II, 317) There is.

8 I owe this citation to my student Truitt Richter.

9 See Chapter 4 of this volume.
 3

The Solution: Cogito

No one has portrayed better the deep and powerful skepticism that
prevailed during the sixteenth and seventeenth centuries, especially in
France, in the fields of science, mathematics, and religion than Richard
Popkin. Religious persecution was rampant, and epistemology was “a
new machine of war.”1 Given the extreme doubt regarding so many
matters, confidence in the possibility of knowledge was low, at best.
This was the skepticism that Descartes tried desperately to combat. He
wanted to combat it by showing that there was, at least, one certainty
on which an entire edifice of knowledge could be constructed. That
certainty was such that it could not be doubted, no matter how deep
and sweeping the doubt that the human mind could invent. On that
certainty, the confidence in human reason could be regenerated. So
Descartes thought.
In this matter, Descartes nicely contrasts with Saint Augustine. Here
is Saint Augustine in De Utilitate Credendi: “I could produce many ar-
guments to show that absolutely nothing in human society will be
safe if we decide to believe only what we can regard as having been
clearly perceived”2 (CSM II, 152; AT VII, 217). By contrast, Descartes’

1 Richard H. Popkin, The History of Scepticism: From Erasmus to Descartes, 69; see especially
Chapters 4–8.
2 John Compton suggests that the cited passage from Saint Augustine contrasts sharply
with what he says elsewhere, for example, in On the Free Choice of the Will. Among
the endlessly fascinating parallels between Saint Augustine and Descartes, he thinks,
may be this one: In the last-mentioned book, Augustine anticipates Descartes’ own

58
 The Solution: Cogito 59

revolution of reason was designed to show that nothing would be safe

if we decided otherwise – neither society, nor knowledge, nor for that
matter faith. “I am vain enough to think,” he said, “that the faith has
never been so strongly supported by human arguments as it may be if
my principles are adopted” (CSMK, 88; AT I, 564). What was that cer-
tainty on which Descartes staked so much? That is the subject matter
of this chapter.
Now, if he is searching for at least one solid certainty, the seeker after
that certainty must know the characteristics of a certainty in order
for him to recognize that he has stumbled onto the real thing, and
not its fake surrogate. Thus, in section I, I list six criteria that that
which is certain must satisfy. In section II, a famous experiment in
thought is described in order to discover the sought-after certainty. In
section III, in light of Descartes’ taxonomy of problems and solutions,
that thought experiment is evaluated. Section IV raises a philosophical
objection on the basis of a claim made by Descartes in a theological
context. Section V illustrates the scope and mode of doubt in the case
of the cogito by highlighting the use of doubt in other cases. In the
sixth and final section, from the experience of certainty, an epistemic
rule of cardinal importance is elicited.

reasonings about the certainty of the existence of thought, and with it the existence
of truth, and with this the existence of “The Truth.” Granting this, for the sake of
argument, I am left wondering how an Augustinian reconciliation might be effected
between the two works.
Stephen Menn, for example, has argued (Descartes and Augustine, especially 185–
194) that Augustine distinguished between understanding, belief, and opinion; hav-
ing faith in the correctness of the scriptures and struggling with them were necessary
conditions, but neither singly nor jointly constituted sufficient condition(s) – the
grace of God was also necessary (but not sufficient?) – for achieving understand-
ing. While ignorant, one ought to seek understanding from authority (as Augustine
did from Saint Ambrose, especially seeking the latter’s allegorical reading of the
scriptures), and thence one would be led to understanding. As Menn notes, this
leaves open the question concerning which authority one should follow – or, for that
matter, I should think, which revealed scripture. Moreover, if the authority relied
upon received his understanding as a result of God’s grace, might not the learner
receive it similarly? Or isn’t God’s grace necessary for understanding what the au-
thority instructs, given that one has picked the right authority? If necessary, then
why couldn’t that grace work on that individual without any authority? If not nec-
essary, what need of an authority? Descartes seems to escape these problems. Or he
at least seems to make the tension between faith and reason vivid (and so the ten-
sion between himself and Saint Augustine), with reason having a slight edge over
faith.
60 The Solution: Cogito

I. The Nature of the First Principle

But first, and briefly, we need to ask, what properties should that thing
that Descartes is looking for possess? In short, does Descartes have any
criterion or criteria for recognizing what he is after? I should warn
that Descartes himself does not proceed in this way. What I am imme-
diately interested in, in this section, is not so much a historical report
of Descartes’ intention as a philosophical reconstruction of his ideas
and what he ought to have said. Something, I think, of philosophical
interest emerges.
In fact, Descartes does list a few criteria. For example, he says that
the thing must be “transparently clear” (CSM II, 104; AT VII, 145);
it must be a “first principle” (CSM, II, 415; AT X, 521) or the “first
item of knowledge” (CSM II, 24; AT VII, 35); it must be “easiest to
become acquainted with” (CSM II, 407; AT X, 509–510); and it must
be the most easily known and “the simplest” (CSM II, 400–401; AT X,
496–497). These are some of the characterizations of what is the “first
thing we come to know if we philosophize in an orderly way” (CSM I,
194; AT VIII A, 6). There is a question about how far Descartes listed
these criteria prior to his discovery of the central truth of his philo-
sophical inquiry and the extent to which he listed them after his dis-
covery. The answer is that some of the criteria he listed prior to, and
some after, the discovery; and not all of the criteria were listed with
that first item of knowledge in mind. But it is arguable that all of
them could, and should, have been listed before he made his discov-
ery. If Descartes had not listed the criteria, how would he have rec-
ognized the thing he was looking for, if and when he stumbled onto
it? As we have already seen, Descartes himself said in characterizing
a perfectly understood problem that “[a] problem is to be counted
as perfectly understood only if we have a distinct perception of the
following three points: first, what the criteria are which enable us to
recognize what we are looking for when we come upon it” (CSM I, 51;
AT X, 429). Consequently, if searching for an absolute certainty is to
be characterized as a perfectly understood problem, we must at least
have criteria for distinguishing or recognizing “what we are looking
for.”
Let us then, at least tersely, state these criteria, cast in their
proper contexts, and explain their virtues. In The Search for Truth,
 I The Nature of the First Principle 61

Descartes said:

Go on, then, Polyander, and show him how far we can get with good sense,
and also what conclusions can be derived from our first principle. (CSM II, 415;
AT X, 521; my italics)

Whatever is the truth to be discovered, the first criterion is that

it must be a first principle, a first item of knowledge. Not necessar-
ily first in the way things are, but necessarily first in the way things
come to be known. The first principle will help in the derivation
of other truths, and it will not be one that is derived from an-
other truth, or from some other truths.3 The primacy of this truth –
hereinafter called the first principle – should be unquestionable, or
the epistemic enterprise will not succeed. Descartes is insistent, as
for example in Principles of Philosophy, that we “philosophize in an or-
derly way” and that when we do so, “the first thing we come to know”
(CSM I, 194; AT VIII A, 6; my italics) is the first principle. There
is something that we know first. There is a hierarchy, at least in
the order in which we come to know with certainty, the order in
which one item of knowledge takes precedence over others, and
necessarily so.
The second criterion of the first principle is that it should be
a clear and distinct notion. Here is Descartes in Meditations on First
Philosophy:

Do I not therefore also know what is required for my being certain about
anything? In this first item of knowledge there is simply a clear and distinct
perception of what I am asserting. (CSM II, 24; AT VII, 35; my italics)

Again, in The Search for Truth:

All the mistakes made in the sciences happen, in my view, simply because at the
beginning we make judgements too hastily, and accept as our first principles
matters which are obscure and of which we do not have a clear and distinct
notion. That this is true is shown by the slight progress we have made in the
sciences whose first principles are certain and known by everyone. (CSM II,
419; AT X, 526; my italics)

3 Using ‘derivation’ broadly in the sense of being dependent on; see Chapter 1,
section III.
62 The Solution: Cogito

Finally, in the “Second Set of Replies”:

Now some of these perceptions are so transparently clear and at the same time
so simple that we cannot ever think of them without believing them to be true.
(CSM II, 104; AT VII, 145)

The virtues of clarity and distinctness of the first principle are stated
directly in the former passage, by implication in the latter.4 The second
passage also indirectly supports the claim that the first principle should
be the easiest to be acquainted with, since if it were not, it could not
surely be known by everyone. By everyone, Descartes means persons
who are intelligent and mature, with or without any paraphernalia of
special, acquired knowledge – in short, ideal seekers.5
We arrive at the next two criteria:

But, Polyander, while engaged upon this work of demolition we can use the
same method to dig the foundations which ought to serve our purpose, and to
prepare the best and most solid materials which will be needed for building up
these foundations. So please join me in considering which, of all the truths men
can know, are the most certain and the easiest to become acquainted with. (CSM II,
407; AT X, 509–510; my italics)

The third criterion the first principle must satisfy is that it must be the
most certain. This criterion virtually defines this epistemic enterprise.
In fact, so strong is this requirement that Descartes thinks that when
we finally come to know the first principle, we shall find that it is even
more certain than an instance of the law of noncontradiction. He says,

[I]f you simply know how to make proper use of your own doubt, you can use it
to deduce facts which are known with complete certainty – facts which are even
more certain and more useful than those which we commonly build upon the
great principle, as the basis to which they are all reduced, the fixed point on
which they all terminate, namely, “It is impossible that one and the same thing
should exist and at the same time not exist.”6 (CSM II, 415–416; AT X, 522)

4 Gaukroger offers an interesting conjecture of the connections – “striking parallels,”

he calls them – between Descartes’ theory of clear and distinct grasp of an idea and the
psychological theory of cognitive grasp as found especially in the work of Quintilian, a
psychological theory whose origin lay in Aristotle and which had its early development
in the Stoics (G, 118–123).
5 See Chapter 1, section I, this volume.
6 To Mersenne he wrote on April 15, 1630: “At least I think that I have found how
to prove metaphysical truths in a manner which is more evident than the proofs
of geometry” (CSMK 22; AT I, 144). He made elsewhere similar claims: see CSMK,
 I The Nature of the First Principle 63

The first principle should also satisfy the fourth criterion, that of
being the easiest to become acquainted with. If the first principle is difficult
to get to know, we cannot rely on its certitude. The entire edifice of
knowledge would be shaky and could collapse. Descartes claimed, in
fact, that an ordinary inquirer after truth, who has reached the age
of discretion (CSM II, 406; AT X, 508) and has a modicum of insight
(CSM II, 409; AT X, 514), would be able to know the first principle.
Such an inquirer, an ideal seeker, would not require the resources of
learning preached in the Schools. So: The first principle must be the
easiest to be acquainted with.
Here is Descartes’ fifth criterion:

For the items of knowledge that lie within reach of the human mind are all
linked together by a bond so marvelous, and can be derived from each other by
means of inferences so necessary, that their discovery does not require much
skill or intelligence – provided we begin with the simplest and know how to
move stage by stage to the most sublime. (CSM II, 400–401; AT X, 496–497;
my italics)

This passage lists another criterion, namely, that the first principle
should be the simplest, a virtue hard to define. For the moment, I
wish to express a puzzle. If Descartes is to be taken literally, then the
first principle is an item of knowledge on a par with other items of
knowledge. Descartes is claiming that the first principle is hooked by
a proper inference to those other items of knowledge just as surely
as those other items are hooked to the first principle. They can be
“derived from each other.” If this is true, then calling anything the first
principle is moot, since it can be derived from some other item of
knowledge. Any such item of knowledge could serve as the starting
point from which all other items of knowledge could be derived.
Of rule 6 in Rules for the Direction of the Mind, Descartes said that it
contained “the main secret of my method.” Here is the rule’s caption:

In order to distinguish the simplest things from those that are complicated
and to set them out in an orderly manner, we should attend to what is

29; AT I, 181–182 and CSMK, 53; AT I, 350, although in the latter he was talking about
his proof of the existence of God. But since that metaphysical truth is dependent on
the cogito, it could not be more certain than the cogito. Consequently, it is arguable
that the cogito is known with greater certainty than the truths of mathematics, which
is precisely what Descartes maintains.
64 The Solution: Cogito

most simple in each series of things in which we have directly deduced

some truths from others, and should observe how all the rest are more, or
less, or equally removed from the simplest. (CSM I, 21; AT X, 381; italics
omitted)

This rule “instructs us that all things can be arranged serially in var-
ious groups, not in so far as they can be referred to some ontological
genus (such as the categories [e.g., the Aristotelian categories of
substance, quality, quantity, relation, etc.] into which philosophers di-
vide things), but in so far as some things can be known on the basis
of others” (CSM I, 21; AT X, 381). By parity of reasoning, Descartes,
I conjecture, would have asserted the following: All items of knowl-
edge can be arranged serially, and that arrangement will reflect the
order in which things can be known on the basis of others. The first
item in the series will be the simplest, and it will not be known, ex
hypothesi, on the basis of anything else. However, it will serve as a
basis on which succeeding items of knowledge in the series can be
known.
Now, all things can be divided into the absolute and the relative. The
absolute is pure, simple, and independent. “I call this the simplest
and the easiest thing when we can make use of it in solving problems”
(CSM I, 21; AT X, 381). The relative shares its nature, and so it can be
related, and deduced, from the absolute. But the distinction between
absolute and relative depends on a point of view. “For some things are
more absolute than others from one point of view, yet more relative
from a different point of view” (CSM I, 22; AT X, 382). For example,
from the point of view of simplicity, a universal is more absolute than
a particular; but from the point of view of dependency, a particular is
more absolute than a universal, since the latter depends for its exis-
tence on the particular. Descartes offers a different example: Species
is absolute relative to the individual, but relative with respect to the
genus.
Descartes could have effectively argued, then, that from an epis-
temic point of view, the first principle should be not only the simplest
but also more easily known than the rest of the items in the series.
Consequently, if a particular is more easily known than a universal,
then the first principle will have the character of being a particular. If
knowledge of a universal is dependent on knowledge of a particular,
 II The Thought Experiment 65

then, too, the first principle cannot be a universal. In short, the first
principle cannot be a universal. This is the sixth criterion for something
serving as the first principle. This will later prove to be very important.
There is a potential conflict between the requirements of simplicity
and particularity. The first principle must be simple, so it must be uni-
versal; the first principle must be particular, so it cannot be universal.
I resolve this conflict by suggesting that simplicity is a relative term.
What Descartes calls the simplest is that which we can make use of in
solving problems. Thus, I infer that for the problem at hand, if the first
principle is a particular, it will aid in solving problems in a manner in
which a universal first principle cannot.
To say the least, none of these criteria or properties is precise. For
example, what constitutes simplicity? Is the law of noncontradiction
simple? Is utmost simplicity a property possessed by only one object,
or by a host of objects? What is “easiest to become acquainted with”
(acquaintance not through the senses, but presumably by reason or
intuition)? Is anything known by discursive reasoning either simplest
or easiest? Will only an intuition do for that? What is the argument,
if there is one, for the possibility that an object can simultaneously
possess two or more properties, such as that of being the simplest and
that of being known most easily? What is a principle: Is it a proposition
or a perception? Most significantly, can any of this be known prior
to stumbling onto the first item of knowledge? Are these properties
not discerned after the discovery of the object, and upon subsequently
examining and reflecting on it?
It is useful to keep these cues and questions in mind as Descartes
now approaches the climax.

II. The Thought Experiment

Descartes did not plan to draw any practical consequences from his
doubt. He was not going to let his daily life be interrupted or interfered
with on the basis of his doubts. But he did want these metaphysical
doubts to be treated seriously, in the hope that they might point to
something that is not susceptible to such doubts. If there is no such
thing, then at least he would know that there is nothing that can be
known for certain. “I will proceed in this way until I recognize some-
thing certain, or, if nothing else, until I at least recognize for certain
66 The Solution: Cogito

that there is no certainty”7 (CSM II, 16; AT VII, 24). Now, Descartes
took these doubts seriously, hyperbolic though they were.

So serious are the doubts into which I have been thrown as a result of yester-
day’s meditations that I can neither put them out of my mind nor see any way
of resolving them. It feels as if I have fallen unexpectedly into a deep whirlpool
which tumbles me around so that I can neither stand on the bottom nor swim
up to the top. (CSM II, 16; AT VII, 23–24)

Descartes believed that such mental states of doubt were essential to

the proper reception of that sought-after proposition – or argument –
that cannot be doubted. It is essential that one also feel the psycholog-
ical pressure involved in actually coming to believe that one’s beliefs, or
one’s system of beliefs, are without foundation or justification. It is not
enough to have intellectually established these doubts, at a distance,
disinterestedly, as it were. As he says, “it is not enough merely to have
noticed this” (CSM II, 15; AT VII, 22). Descartes claims that he must re-
member these doubts, goad his mind in the direction of these doubts
when the mind slips back into its habitual ways of thinking and feeling.
The pressure of commonplace thoughts, due to repetition, law, and
custom, is not to be underestimated, because the mind will continue to
believe such thoughts to be “highly probable opinions,” which is “what
in fact they are” (CSM II, 15; AT VII, 22). There is an inherent “laziness”
in us. Thus, he must undertake an arduous task: The will must be
turned in the opposite direction, and Descartes must attempt to de-
ceive himself “until the weight of preconceived opinions is counter-
balanced and the distorting influence of habit no longer prevents my
judgement from perceiving things correctly”(CSM II, 15; AT VII, 22).
Descartes thinks that unless there is this strong initial intellectual
discomfort brought about by a pervasive doubt, a would-be knower will
never come to know certain truth. Or at least, his grasp of that truth
will not be what it could be. How he grasps that truth is as important
as what he comes to grasp.

7 There is nothing paradoxical in that remark. Descartes is merely saying that his lack
of certainty about an nth-order proposition is compatible with his having certainty
about an n +1th-order proposition. Thus, his recognizing for certain that there is
no certainty in any nth-order proposition is simply a certainty about an n +1th-order
proposition. Either Descartes hits upon a proposition like the cogito, or he comes up
empty-handed. In either case, he has at least one certainty.
 II The Thought Experiment 67

Here, then, in the Meditations on First Philosophy of 1641, comes the

heart of the cogito claim:

But I have convinced myself that there is absolutely nothing in the world, no
sky, no earth, no minds, no bodies. Does it now follow that I too do not exist?
No: if I convinced myself of something, then I certainly existed. But there is
a deceiver of supreme power and cunning who is deliberately and constantly
deceiving me. In that case I too undoubtedly exist, if he is deceiving me;
and let him deceive me as much as he can, he will never bring it about that
I am nothing so long as I think that I am something. So after considering
everything thoroughly, I must finally conclude that this proposition, I am, I
exist, is necessarily true whenever it is put forward by me or conceived in my
mind. (CSM II, 16–17; AT VII, 25)

Four years earlier, in Discourse on the Method, a similar passage, at a

similar juncture in the argument, had read:

I resolved to pretend that all the things that had ever entered my mind were
no more true than the illusions of my dreams. But immediately I noticed that
while I was trying thus to think everything false, it was necessary that I, who was
thinking this, was something. And observing this truth, “I am thinking, therefore
I exist” was so firm and sure that all the most extravagant suppositions of the
sceptics were incapable of shaking it, I decided that I could accept it without
scruple as the first principle of the philosophy I was seeking. (CSM I, 127; AT
VI, 32)

While three years later, in Principles of Philosophy, published in 1644,

the corresponding passage was formulated thus:

In rejecting – and even imagining to be false – everything which we can in any

way doubt, it is easy for us to suppose that there is no God and no heaven, and
that there are no bodies, and even that we ourselves have no hands or feet,
or indeed any body at all. But we cannot for all that suppose that we, who are
having such thoughts, are nothing. For it is a contradiction to suppose that
what thinks does not, at the very time when it is thinking, exist. Accordingly,
this piece of knowledge – I am thinking, therefore I exist – is the first and most
certain of all to occur to anyone who philosophizes in an orderly way. (CSM I,
194–195; AT VIIIA, 7)

And in each case, without further ado, Descartes sets about to dis-
cover the properties of this ‘I’.
How are these passages to be understood? Let us draw on our dis-
tinctions. Are the passages susceptible to the ordinary-doubt model
68 The Solution: Cogito

analysis, or will only the extraordinary-doubt model do? Is the cogito –

I am thinking, therefore I exist – a mental state or a mental process? Is it
a first-order or a second-order type of thought? Finally, does the cogito
satisfy the six criteria in virtue of which it is to be accorded the status
of the first principle of the philosophy that Descartes was seeking? Is
it a perfect solution to a perfect problem? The answers to all of these
questions will not be finally settled here; but we must begin to answer
them now.
The cited passages cannot be susceptible to the ordinary-doubt
model, since to doubt a proposition, according to that model, is to
presuppose the truth of some other propositions. Descartes explic-
itly vows to doubt everything that can be doubted: the existence of
such things as the sky, earth, minds, and bodies. Thus, extraordinary-
doubt is involved, but in this case it has a curious feature. At this
point, I need you, the reader, to conduct the thought experiment that
Descartes is urging. Attempt to carry out the extraordinary doubt,
no matter how silly it seems to you. Well, then: Commence doubting
that the sky exists, that numbers exist, that there are mathematical
demonstrations, and so on. Provide yourself with reasons for raising
minimal doubts, reasons for rejecting the proposition that the sky ex-
ists or that there are mathematical proofs and propositions. If you
are unable to find reasons for rejecting a proposition, then do what
Descartes did: Raise the specter of an evil genius who is bent on deceiv-
ing you. Carefully notice and register the properties of your mental
states, whether they are doubtful or certain, clear or confused, distinct
or otherwise, and note as well the intensity, quality, and character of
these properties.
Now, either at the various stages in your attempting to give reasons,
or at the stage where you have arrived at that about which you have
reasoned, attempt to question your own existence. Attempt to doubt
your own existence, giving yourself reasons parallel to those you gave
yourself for doubting other things. “I do not exist,” say to yourself,
“because ——–.” Inquire why no reason is sufficient. You will find that
any attempt, at any stage, at providing even a solitary reason for the
least conceivable amount of doubt regarding any proposition, about
either yourself or anything else, will confront you with your own exis-
tence, thus notarizing – with you as the notary – the proposition that
 II The Thought Experiment 69

you exist. You will “notice” the ineluctable truth of the proposition,
namely, that you, the doubter, exist. “I am thinking, therefore I exist”
will come home to you as no other truth does. It is the coming home
of this truth that I need for you to latch onto, with whatever properties
are attendant upon it. The extraordinary doubt ruins each belief, pre-
supposing no other truth, and yet constantly and inevitably leads to the
cogito.
The focus should be on the doubting of the proofs or proposi-
tions, rather than on the proofs or propositions themselves, because
clearly which proofs or propositions are doubted is quite irrelevant to
establishing the truth of the proposition that you exist. The method of
doubt is meant as a purge, purification, and preparation: The doubt’s
“greatest benefit lies in freeing us from all our preconceived opin-
ions, and providing the easiest route by which the mind may be led
away from the senses. The eventual result of this doubt is to make
it impossible for us to have any further doubts about what we sub-
sequently discover to be true.” (CSM II, 9; AT VII, 12) The doubt is
certainly meant as a purge, to rid the mind of any beliefs that can
be doubted. The doubting process will purify the mind and rid it of
the intrusion of the senses. To speak metaphorically, the doubt en-
ables Descartes to eliminate the background noise produced by the
senses, by withdrawing the senses, the better to hear the voice of
reason. Thus completely unencumbered, reason will perceive clearly
what it could not otherwise perceive. Finally, the doubt will prepare
the mind to witness, or notice, that in the act of doubting, it can-
not – cannot in principle – purge the belief in the doubter’s own
existence.
This should also answer the question, is the cogito a mental state
or a mental process? It is, in the final analysis, a mental state. It is a
mental state in which the truth of the cogito is intuitively perceived.
But it is a final state that is linked to a mental process. The final men-
tal state can come about as the culmination of a variety of mental
processes, none of which is essential. If the cogito were a mental pro-
cess, would it be some definite process? That is, is there one and only
one route to recognizing the truth of the cogito? There is not. From
Descartes’ perspective, no particular mental process is essential, but
it does not follow that one could do without any of them. There are
70 The Solution: Cogito

three streets leading to my house: It is not essential to take any one

particular street, but it does not follow that one could do without
taking any one of them. Similarly, there are several mental processes
leading up to the cogito: It is not essential to take any one particular
mental process, but it does not follow that one could do without tak-
ing any one of those mental processes. Any such mental process is
simply a matter of preparing the mind to receive the first truth, as it
were.
The point can be illustrated by imagining possible worlds based
on the following propositions expressed in a mental state. M1 There
is an earth is The Thought1 : There is an earth. M stands for a mental
state, the subscript numbers the thought,8 and the remainder is the
proposition.

M1 There is an earth.
M2 There is a sky.
M3 There are numbers.
M4 There are human bodies.
M5 There is a heaven.
M6 There are minds.
..
.
Mn There is a God.
Mo I am thinking, therefore I exist.

We can now imagine the following possible worlds (using the ‘−’
as a sign for it is not the case that; thus, −M2 is The Thought2 : It is not the
case that there is a sky.).

8 “[M]oreover I have various thoughts which I can count” (CSM II, 30; AT VII,
44). This is from the Third Meditation, where Descartes is trying to give an ac-
count of how all of his ideas (ideas of “corporeal and inanimate things, angels,
animals and finally other men like myself” [CSM II, 29; AT VII, 43]) – save the
idea of God – can be explained in terms of ideas that give him a representation
of himself. This line is rarely cited, if ever, but it raises an interesting question: If
Descartes can assume that he can count his thoughts, is he then not able also to
count how many total thoughts he has? If that is the case, is he not assured of at
least some simple truths of arithmetic? But the existence of God has not yet been
demonstrated, and the truths of arithmetic have already been questioned in the
First Meditation. This would leave mysterious his ability to successfully count his
thoughts.
 II The Thought Experiment 71

w 1: {M1 , M2 , M3 , . . . . . . . . . , Mn } —➤ M0
w 2: {−M1 , M2 , M3 , . . . . . . . . . , Mn } —➤ M0
w 3: {M1 , −M2 , M3 , . . . . . . . . . , Mn } —➤ M0
w 4: {M1 , M2 , −M3 , . . . . . . . . . , Mn } —➤ M0
..
.
wi: {M1 , −M2 , −M3 , . . . . . . . . . , Mn } —➤ M0
..
.
wk: {−M1 , −M2 , M3 , . . . . . . . . . , Mn } —➤ M0
..
.
w n: {−M1 , −M2 , −M3 , . . . . . . . . . , −Mn } —➤ M0

If one worked with just n propositions, one would have 2n possibili-

ties. One can increase the possibilities by introducing new propositions
without limit, or one can decrease the possibilities by eliminating the
propositions without limit, so long as there is at least one proposition.
Each possibility would lead to Mo . The symbol ‘ ’ is simply a sign that
vaguely means leading up to, and could, depending on the discussion,
take on a more definite meaning later. Notice, too, that I think in Mo ,
namely, The Thoughto : I am thinking, therefore I exist, is quite redundant,
since it is some thought, a mental state, that expresses it; thus, The
Thoughto : I exist could do just as well.
Now, imagine a doubter, D1 , living in world, w 1 . The earth, sky, bod-
ies, minds, and numbers exist; in short, it is our world as we ordinarily
know it. D1 ’s ordinary life is geared in the belief that the propositions
M1 through Mn are true. His thought and talk, act and will, and beliefs
and performances make that clear. But D1 succumbs to Descartes’ ad-
vice and invents reasons to doubt the propositions that he normally
believes in. He discovers that in the very act of doubt, he discovers
the truth of the cogito. By contrast, in another world, w k , say, there is
a doubter Dk : In this possible world, nothing is true save the follow-
ing three propositions (and whatever is implied by them) – numbers
exist, the doubter exists, and God exists.9 The doubter’s ordinary life

9 In a letter of August 1641, written to Hyperaspistes, Descartes had speculated,“I have

no doubt that if [the soul] were released from the prison of the body, it would find [the
ideas of God, itself, and self-evident truths] within itself” (CSMK, 190; AT III, 424).
72 The Solution: Cogito

(whatever is ordinary in such a possible world!) is lived out in accor-

dance with his belief in those propositions. Once again, his thought
and talk, act and will, and beliefs and performances make that clear, al-
though it may be extremely difficult for us to imagine what these would
be like, since Dk lacks a physical body, and hence our form of life. Now,
Dk too succumbs to Descartes’ advice about engaging in doubt at least
once in one’s life and invents reasons to doubt the propositions that he
normally believes in. Yet, as a doubter, he finds reasons that give him
minimal doubt about whether numbers and God exist. (Of course, he
could not doubt the proposition that he, the doubter, exists, without
guaranteeing his existence.)
And so on for other doubters in other possible worlds.
If we trace the mental histories of the doubts of the individuals
in various possible worlds, and the reasons that they give for these
doubts, we shall surely find them to be quite different. For one thing,
unlike D1 , an individual of our world, Dk , will have no concept of the
external world, and no concept of place, color, or shape. Yet, in each case,
the doubter in each possible world will arrive at the same final cogito-
state, namely, Mo . Qualitatively, and in every relevant way, if Descartes
is right, there should be no difference between the cogito-state of any
two possible worlds. Mo is true in every possible world, it is a necessary
truth.
One might wonder: How can the cogito be foundational for knowl-
edge of our actual, determinate world, if it is certain in such a wide
variety of worlds (not that Descartes has any positive theory of possi-
ble worlds)? It does not seem to be slanted, in other words, toward
our world and our knowledge problems, so how can it be helpful?

There is also this from Pierre Gassendi (1592–1655), the author of the “Fifth Set of
Objections”: “For I ask you, what progress do you think you would have made if, since
being implanted in the body, you had remained within it with your eyes closed and
your ears stopped and, in short, with no external senses to enable you to perceive this
universe of objects or anything outside you” (CSM II, 217; AT VII, 310). Descartes
replied thus: “I do not doubt that the mind – provided we suppose that in thinking it
received not just no assistance from the body but also that it received no interference
from it – would have had exactly the same ideas of God and itself it now has, with
the sole difference that they would have been much purer and clearer” (CSM II, 258;
AT VII, 375). Consequently, the example imagined here is quite in keeping with the
Cartesian spirit.
 II The Thought Experiment 73

Therein, precisely, lies the beauty and power of the cogito. One might
say that it is an empirical necessary truth, a first truth in all possible
worlds (in which there are truth seekers). But the common starting
point of the inquirers in all of these possible worlds notwithstanding,
what they know about their respective worlds will eventually diverge.
One will claim that his physical world is informed by a gravitational
law, another that his world is informed by a gravitational law but of a
different order of magnitude, a third that there is no gravitational law
at all in his world. What is so incredibly fascinating is that the knowl-
edge of such remarkably different worlds can begin on so slender a
basis.
This way of viewing the matter may be an answer to just the kind
of query that Montaigne might be thought of as constructing. Thus:
“Now if there are several worlds, as Democritus, Epicurus and almost
the whole of philosophy have opined, how do we know whether the
principles and laws which apply to this world apply equally to the
others? Other worlds may present different features and be differ-
ently governed” (M, 96). Much later, this is followed by the following
challenge: “For whatever Nature truly ordained, we would, without any
doubt, all perform, by common consent: not only all nations but all
human beings individually would be deeply aware of force or compul-
sion when anyone tried to make them violate it. Let them show me
just one law with such characteristics: I would like to see it.” (M, 161)
Descartes’ cogito would be such a “law”: applicable, I have argued, not
only among all men, in all nations, at all times, but in every world –
even in an inconceivable world that defied our laws of logic? “Or, had
I not been such a commonsensical chap,” says David Lewis, “I might
be defending not only a plurality of possible worlds, but also a plu-
rality of impossible worlds, whereof you speak truly by contradicting
yourself.”10 While I quite prefer Lewis’ common sense, it would be
enormously interesting to see how the denial of the cogito would be
true in an impossible world, not just simply claimed to be true by fiat
or stipulation. I am not unmindful of the fact that Montaigne is speak-
ing here of a moral rule as a Natural Law. The generalization is very
much in keeping with the spirit of Montaigne’s point.

10 David Lewis, On The Plurality of Worlds, 1.

74 The Solution: Cogito

Now Descartes takes these doubts seriously. That is, he thinks that
he should compel his mind to concede the soundness of the reasoning
that leads to utter skepticism, a concession that leads him to intellectual
despair and to an understandable desire to find a way out. But it is
difficult to see why this is necessary. May I not casually entertain the
possibility that each proposition put before me is false? As nearly as I
can judge, the process of serious doubt is not essential to establishing
the truth, even the necessary truth, of the cogito. But Descartes is after
bigger things. He wants to establish not only that the cogito is necessarily
true every time he conceives it, but also that it is the only truth of
which Descartes could be certain at the start. Whereas, if I am right,
an individual can correctly perceive the truth of the cogito and yet leave
open the possibility that there might be other propositions of which
he could be equally, or even more, certain. Or again, he might believe
that there is no other such proposition.
Is the cogito a first-order or a second-order mental state? Is it a
first-order or second-order mental process? If we map the proposi-
tional content of each distinct mental state (represented by a circle,
with or without shading) onto a proposition, then we might formulate
some of the questions as follows: Is the cogito a proposition? Is it an ar-
gument? Is it valid? Is it sound? For now, I continue to speak in terms
of mental states and mental processes. Consider, then, the following
few possibilities:

Case 1. First-order mental state

s1 s2 s3 .... .... sn .... sc

Case 1 represents the possibility where the thinker goes through

a series of mental states, s 1 , s 2 , s 3 , . . . , s n , and is lead to s c , where s c
is the state in which he, the thinker, acknowledges the truth of the
cogito. As before, a circle shaded with lines expresses a mental state
in doubt about the proposition it is entertaining. In this case, all
of the mental states prior to the cogito-state are represented by cir-
cles shaded with lines. The propositions entertained by these mental
 II The Thought Experiment 75

states are doubted, as did Descartes in his hyperbolic doubt: the

model of extraordinary doubt. The earlier mental states simply lead
up to the cogito-state; they are a preliminary to, a framework for, the
cogito-state. The earlier mental states have no essential hook-up with
the final cogito-state. The mental state s c is free of doubt; the other
mental states are infected by at least a minimal doubt. In the state
s c , the thinker is compelled11 to recognize and assent to the truth
of the cogito; but in the other mental states, he is free to withhold
assent.

Case 2. First-order mental process

s1 s2 s3 .... . . . . sn sc

Case 2 represents the possibility where a series of mental states leads

up to the cogito-state, but the earlier states are essential to that state.
That is, the earlier mental states ineluctably lead to the final cogito-state.
What precisely is the nature of this ineluctability, this compulsion, is as
yet left undefined or undetermined, and is represented by the symbol
‘ ’.

Case 3. First-order mental process

s1 s2 s3 .... . . . . sn sc

Case 3 is similar to Case 2 in all respects, except that the earlier

mental-states are all free from doubt, too.

11 To read this compulsion as no more than psychological, although it is at least that, is

a serious mistake; see also Chapter 4, especially 127 and note 24. For further support
of the idea that “certainty is always at least partially an evidentiary notion, and never
a purely psychological or purely a semantic one,” see Jeffrey Tlumak, “Certainty and
Cartesian Method,” 40, 43–44.
76 The Solution: Cogito

Case 4. Second-order mental state

s1 s2 s3 .... . . . . si sn

⊥
sc

Case 4 represents the possibility in which the earlier mental

states, s 1 , s 2 , s 3 , . . . , s i are related to s n as premises are related to
the conclusion of an argument. Here, s c , the final cogito-state, is
a second-order mental state of a first-order mental process, s 1 , s 2 ,
s 3 , . . . , s i  s n . The symbol ‘’ is a sign for a mental state acknowl-
edging valid inference. In this case, it signifies that s n is acknowledged
to be validly inferred from s 1 , s 2 , s 3 , . . . , s i . In the second-order mental
state, s c , the thinker recognizes in a compulsive way that the inference
from s 1 , s 2 , s 3 , . . . , s i to s n is valid. In short, it is in s c that the thinker
recognizes the truth, the absolute certainty, of the cogito.

Case 5. Second-order mental state

s1 s2 s3 .... . . . . si sn
⊥

sc
 II The Thought Experiment 77

Case 5 is identical in all respects to Case 4, save in one. In Case 4,

there was some element of doubt in all the earlier states; only the
final cogito-state was indubitable, without shading. In the present case,
Case 5, all the earlier states leading up to the cogito-state are indubitable
too; they are all represented by circles without shading. Alternatively,
only those earlier mental states, which are without shading, lead up to
the cogito-state in a way in which the shaded earlier states could not.
There are, of course, other possibilities,12 and the question is, how
can we best understand the cogito that represents Descartes’ view and
is also the least objectionable?
Descartes’ view is as follows. The cogito-state is a first-order state.
It is not about any other state. The individual doubter recognizes, in
the first-order state, the truth of the cogito. This is extremely difficult
to express, and once again I need to beg the reader to conduct the
thought experiment. Descartes says that the notions of truth, doubt,
thought, and existence are primitive notions, that is, notions that are not
definable in terms of other notions. (Another primitive notion that
Descartes needs, but that he never separately lists in this context, is the
notion of the indexical ‘I’.) They are also clearly understood by us.13
Thus in The Search for Truth he writes: “But someone who wants to
examine things for himself, and to base his judgements about them
on his own conceptions, must surely have enough mental capacity
to have adequate knowledge of what doubt, thought and existence
are, whenever he attends to the question, without having to be taught
the difference between them” (CSM II, 417; AT X, 523). He makes a
similar claim with respect to the notion of truth. On October 16, 1639,
he wrote to Father Mersenne that truth is

a notion so transcendentally clear that nobody can be ignorant of it. There

are many ways of examining a balance before using it, but there is no way to
learn what truth is, if one does not know it by nature. What reason would we
have for accepting anything which could teach us the nature of truth if we did
not know that it was true, that is to say, if we did not know truth? (CSMK, 139;
AT II, 597)

12 For example, substitute ‘’ for ‘ ’ in Case 1 and Case 2.

13 CSMK, 218; AT III, 665–666 provides a very useful backdrop to this paragraph and
for what immediately follows. See also CSMK, 191–192; AT III, 426–427 together with
the corresponding note 5.
78 The Solution: Cogito

Armed with the meaning of these four terms, notions that are in-
nate in us, the doubter in the cogito-state learns to join the particular
doubting thought with his particular existence. He must try to rear-
range the four innate concepts, linking them to his thoughts, in order
to formulate a proposition that he cannot doubt. Now, the doubter
cannot say that he knows even the following truth of identity, namely,
the concept of thought is the concept of thought, nor can he assert
the following logical truth, namely, the concept of doubt is included
in the concept of thought. For the thinker can invoke the demon who
will make Descartes doubt the truth of these statements, as the demon
made him doubt the truths of mathematics. At any rate, second, such
truths do not give him what he is looking for, if he is looking for the
first existential truth.14 The first truth then suddenly dawns on him.
He sees – notices, perceives, intuits, witnesses – that it is true that in
this particular case, his doubting now ensures his existence now. “You
will surely admit that you are less certain of the presence of the objects
you see than of the truth of the proposition ‘I am thinking, therefore
I exist.’ Now this knowledge is not the work of your reasoning or infor-
mation passed on to you by teachers; it is something that your mind
sees, feels, and handles.” (CSMK, 331; AT V, 137–138)
One might suggest that this experiment idea be explored in terms of
falsification.15 Maybe the cogito is not an argument or even an insight,
but a challenge to falsify. It says something like, “Falsify this!: I exist –
exclusively as a consciousness.” The experiment surrounding the cogito
shows that the “information” that would falsify all other propositions
cannot falsify this one. The cogito proper – the various I think, I exist
ensembles, with or without connecting therefores – might be just an el-
ement in this experiment that really cannot be understood separately.

14 “But because these are very simple notions [the notions of thought, existence, cer-
tainty, it is impossible that that which thinks should not exist, and so on], and ones
which on their own provide us with no knowledge of anything that exists, I did not
think they needed to be listed” (CSM I, 196; AT VIIIA, 8). Some scholars think that
therefore Descartes was searching for the first existential truth, and in that search
for the first truth, as well as in its substantiation, Descartes could have used any of
these “very simple notions.” This must be mistaken. Descartes proclaims to know no
truth at the end of the First Meditation, and not merely that he has no knowledge
of an existential truth. Descartes would have been pleased, at this stage of the in-
quiry, if he could have assured himself of a conceptual truth or two. His doubting of
nonexistential truth makes very little sense, if he was not concerned with these.
15 Following a cue from Catherine Wilson.
 II The Thought Experiment 79

This is, indeed, a happy suggestion and would at least in signifi-

cant part capture Descartes’ intentions. After all, the First Meditation
is a skeptical challenge and attempts to show that all other proposi-
tions can be “falsified.” The Second Meditation proposes the cogito
experiment; the cogito proposition is put severely to the test by the very
skeptical arguments of the First Meditation and is shown to be unfalsi-
fied – indeed, Descartes would claim it to be unfalsifiable. Here we run
into problems that would suggest caution: First, a proposition thus far
unfalsified, from a purely Popperian angle, is not a guarantor even of
probable truth, let alone of truth. Consequently, even the most severe
testing of the cogito statement, leading to nonfalsification, would not
start the Cartesian enterprise. Second, Descartes would need the cogito
to be at least an insight, an intuition, if not an argument. Third, there
is an interesting asymmetry: On the one hand, several propositions
are demonstrated to be false or purported to be false, without invoking
the cogito; on the other hand, the cogito experiment is meant to verify
one truth, without showing why particular propositions are, or may be,
false.
Finally, what about the claim that the cogito-state is a first-order state?
Here is a counterclaim: “It is not true that the cogito-state is a first
order-state; it is a second-order state. Recognizing that I am thinking
is a second-order-state: I have a propositional attitude to the proposi-
tion that I am thinking, which involves a first-order state. For example,
when I believe that 2 + 2 = 4, I am thinking that 2 + 2 = 4. Here my
belief is a first-order state. When I believe that I am thinking that 2 +
2 = 4, my belief is a second-order state. Correspondingly, when I am
thinking that ‘I am thinking, therefore I am’, clearly my thinking that is
a second-order state.” But this will not do, for a variety of reasons. First,
let us clearly identify the distinct states. The first-order cogito-state is, “I
am thinking, therefore I am.” The second-order state is, “I am thinking
that ‘I am thinking, therefore I am.’ ” But the latter is not Descartes’ cog-
ito; it is something else: it is about the cogito. Second, the objector might
contend that our formulation is mistaken. He might argue that “I am, I
exist,” is the first-order state; the second-order state is the cogito, namely,
“ ‘I am, I exist,’ is necessarily true.” However, the latter does not capture
the exact formulation of the cogito in the Second Meditation; the key
phrase –“whenever it is put forward by me or conceived in my mind” –
is missing. That phrase indicates Descartes’ knowing something in the
80 The Solution: Cogito

very act of conceiving it. Take the analogy with mathematics: When I
have clearly intuited 2 + 2 = 4 – recall the “self-evidence and certainty
of intuition” (CSM I, 14; AT X, 369) – am I in the second-order state?
If I am not in the state of certainty while in the first-order-state, what
element confers that certainty while in the second-order state? It is diffi-
cult to provide a satisfactory Cartesian answer to that question.16 Third,
if we allow the cogito to be a second-order state, do we open up the possi-
bility of the mind making a mistake – however slender the possibility –
as to what it is claiming its state to be about? If so, we shall rob the cog-
ito of the certainty that Descartes was after. Fourth, and importantly,
whether the cogito is first-order or second-order, it is still claimed to be a
state (not a process), a state to which there corresponds a proposition.

III. The Experiment Evaluated

At this juncture, I should like to bring in the resources of the first
section and raise some pertinent questions. Descartes, as we saw,
had defined a problem, a perfect problem, and a perfectly under-
stood problem. Let us call the problem – can I know anything for
certain? – the epistemic problem. I shall argue that the epistemic problem
of Descartes is a problem, a perfect problem, and a perfectly under-
stood problem. Descartes had maintained that every problem must be
about something hitherto unknown. The unknown should be charac-
terized in some way. The characterization must be cast only in terms of
what is known. Thus, one might now argue on Descartes’ behalf that
the problem, can I know anything for certain?, is a problem because it
possesses the three necessary features: First, something is unknown at
the start, namely, a certain truth. Second, the solution to the epistemic
problem is characterized in terms of several properties, such as simplest,
can be easily known, particular, and most certain. Third, characterization
of these properties is set in terms that are known. Thus, the epistemic
problem is a problem.
Next, the epistemic problem is a perfect problem. Descartes would
have argued it so. A perfect problem is determinate in every respect.
Its terms must be perfectly understood. The epistemic problem is
determinate in every respect, and its terms are perfectly understood.

16 The last section of Chapter 7 may prove relevant here.

 III The Experiment Evaluated 81

The central concepts are the concepts of truth, doubt, certainty, and
existence. Descartes claims to understand them. Furthermore, Descartes
would have claimed that given the boundary conditions on the epis-
temic problem, the solution – Descartes’ solution – is a unique one.
The boundary conditions on the epistemic problem are defined by the
six criteria that the first principle must possess. There cannot be any
other solution; if there were, then the problem would be an imperfect
problem. As Descartes put it in the Discourse on the Method, “[T]here
is only one truth concerning any matter, whoever discovers this truth
knows as much about it as can be known” (CSM, I, 121; AT VI, 21).
Whatever may be the truth of that statement as a general statement, it is
surely true that Descartes would have argued that the cogito is a unique
solution to a perfect problem.17 There is no other proposition that
could have served as a solution. This is shown by the exhaustive list of
propositions considered and doubted. Each proposition in the system
of belief has been doubted; each part of the building has been dis-
mantled. Only the foundation stone, the cogito, remained unscathed.
Finally, Descartes would have argued that the epistemic problem is
a perfectly understood problem, not an imperfectly understood one.
I quote again the relevant passage:

[A] problem is to be counted as perfectly understood only if we have a distinct

perception of the following three points: first, what the criteria are which
enable us to recognize what we are looking for when we come upon it; second,
what exactly is the basis from which we ought to deduce it; third, how it is to
be proved that the two are so mutually dependent that the one cannot alter
in any respect without there being a corresponding alteration in the other.
(CSM I, 51; AT, 429)

One can now appreciate why it was important to lay down the criteria
by which one could recognize the first principle. It would “enable us
to recognize what we are looking for when we come upon it.” The
universal doubt will be the basis from which we will deduce it. As

17 To be sure, Descartes has offered diverse formulations of the cogito; look no further
than Discourse on the Method and Meditations on First Philosophy. This, however, does
not detract from the claim about uniqueness. For Descartes offers no argument to
suggest that one formulation is inherently better, or captures his intentions better,
than another formulation. Where formulations are different, their differences are
insignificant; and where the differences are marginally significant, they should be
tidied up in order to bring them in line with one another.
82 The Solution: Cogito

Descartes said, “if you simply know how to make proper use of your own
doubt, you can use it to deduce facts which are known with complete
certainty” (CSM II, 415–416; AT X, 522). When Descartes speaks here
of ‘deducing’ a fact, he is speaking loosely, and not using ‘deducing’
in the sense used in formal logic. We need to show, of course, that if
we alter the criteria by which we recognize the first principle, our first
principle will no longer appear to be the first. Quite simply, let the
criterion of certainty be replaced by the criterion of high probability.
Then the first principle will be quite different, the epistemic problem
will not be a perfect problem, and the solution will not be a unique one.
The cogito is a perfect solution to a perfect problem. The six crite-
ria enable Descartes to recognize that the cogito is indeed the solution
he was looking for. First, his survey of the prospective claimants to
the first principle has been exhaustive and complete. It is the sought-
after first principle because “after considering everything thoroughly,”
Descartes could not find any proposition that he could not doubt save
the cogito. Second, the cogito is the most certain. Given that certainty
and minimal doubt are contradictory notions – the absence of one
entails the presence of the other – Descartes has shown that in the
cogito, any attempt to establish minimal doubt fails and hence leads
to certainty. Third, the cogito is also a clear and distinct notion. Both
of these properties are noticed when the doubter is in the cogito-state.
Fourth, the cogito is the easiest to become acquainted with. It presup-
poses no special knowledge; the terms of understanding it, or coming
to recognize it, are common currency. The doubts and dismantling
that precede the cogito may involve special fields – such as history and
mechanics, morals and mathematics – but the doubting itself does
not rest on any special knowledge of these fields. Thus an ordinary,
but ideal, seeker can easily become acquainted with it. Fifth, it is the
simplest. Essentially, only the grasp of simple notions – such as, truth,
doubt, certainty, and existence – are presupposed, together with the
ability to join them. But the joining takes place in a simple, single men-
tal state and does not involve a complex mental process. Finally, it is a
particular: That is, it does not rest on some deep and general proposi-
tions about time, thinkers, and existence; it is rather about this thinker
in this mode of doubt and of his existence in that moment. The cogito
satisfies all six criteria, and hence it is the first principle that Descartes
was seeking.
 IV The Eucharist Objection 83

In June or July 1646, Descartes wrote to Clerselier that

the word “principle” can be taken in several senses. . . . In the first sense, it can
be said that “It is impossible for the same thing both to be and not to be at the
same time” is a principle which can serve in general, not properly speaking to
make known the existence of anything, but simply to confirm its truth once
known. . . . This is of very little importance, and makes us no better informed.
In the second sense, the first principle is that our soul exists, because there is
nothing whose existence is better known to us. (CSMK, 290; AT IV, 444)

This reinforces the point of particularity in the last criterion, which

will prove significant later.
There is an inevitable paradox. At the start of his inquiry, Descartes
does not have a criterion – indeed, by the nature of the case, he cannot
have one, let alone have six – whereby he can judge whether he has
stumbled onto the right answer to the epistemic problem. For if he
has any criterion, he must be certain about it. If he is certain about
any criterion, such as the criterion of simplicity, then he need look no
further, for that criterion can serve as the first item of knowledge, the
first truth of his philosophy. On the other hand, if he does not have
a criterion, then by his own admission he will not know when he has
encountered the first principle, the first truth, if he should encounter
it. Perhaps Descartes is once again trying to raise himself by his own
bootstraps: He provisionally accepts the six criteria at the start, with
no certainty as to where he will be led by them. He is, in fact, led to
the cogito. In that state, and in that state alone, he is certain of its truth.
Thence, he is led to believe with certainty in the veracity of the criteria
with which he began his inquiry.

IV. The Eucharist Objection

There is an objection that is devastating for the cogito. The source of
the objection has to do not with the cogito but rather with Descartes’
view on the doctrine of the Eucharist. In the “Fourth Set of Ob-
jections,” Antoine Arnauld (1612–1694), the French logician and
theologian, wrote:

But what I see as likely to give the greatest offence to theologians is that accord-
ing to the author’s doctrines it seems that the Church’s teaching concerning
the sacred mysteries of the Eucharist cannot remain completely intact. We
84 The Solution: Cogito

believe on faith that the substance of the bread is taken away from the bread
of the Eucharist and only the accidents remain. These are extension, shape,
colour, smell, taste, and other qualities perceived by the senses.

By contrast, said Arnauld, Descartes “denies that these powers are in-
telligible apart from some substance for them to inhere in, and hence
he holds that they cannot exist without such a substance” (CSM II,
152–153; AT VII, 217–218).
What was Descartes’ reply? First, “[M]y saying that the modes are not
intelligible apart from some substance for them to inhere in should
not be taken to imply any denial that they can be separated from a
substance by the power of God; for I firmly insist and believe that
many things can be brought about by God which we are incapable of
understanding” (CSM II, 173; AT VII, 249). In short, God could have
created an attribute, a mode, or an accident (powers, to use Arnauld’s
word) that did not inhere in any substance. Second, God could have
created a new substance in place of the old. Specifically, in the case
of the Eucharist, the substances of the bread and wine could be trans-
formed into another substance in such a way that this latter substance
affects all of our senses in “exactly the same way as that in which the
bread and wine would be affecting them if no transubstantiation had
occurred”18 (CSM II, 175; AT VII, 251).
Now consider the cogito. A thought is an attribute that inheres in the
mind or soul of the person thinking that thought. Thus from the exis-
tence of the thought, the thinker infers his own existence. The thought
of the thinker is an attribute, an accident, a mode; the ‘I’ of that thinker
is the substance. But if the foregoing about the Eucharist is correct,
then there are two possibilities: First, the attribute of thought could
exist without the substance, ‘I’. Second, in place of the ‘I’ there
could be another substance. Indeed, such a possibility was touted by
Hyperaspistes, an unknown supporter of Gassendi, in August 1641:
“Moreover,” he said, “you do not know whether it is you yourself

18 Elsewhere Descartes says, “[T]here is nothing incomprehensible or difficult in the

supposition that God, the creator of all things, is able to change one substance into
another, or in the supposition that the latter substance remains within the same
surface that contained the former one. . . . It clearly follows from this that any given
surface must always act and react in the same way, even though the substance which
is beneath it is changed.” (CSM II, 177; AT VII, 255) For further interesting details,
see R, 136–137, 160, 213–214, 246 note 7.
 IV The Eucharist Objection 85

who think or whether the world-soul in you does the thinking, as the
Platonists believe” (CSMK 192, note 1; AT III, 403). This was no wild
conjecture, but rather had solid historical roots. “Augustine’s work on
the Trinity,” writes an eminent scholar of Saint Augustine,

had profound influence on subsequent western concepts of personality.

Porphyry had thought that all souls had a share in the “world-soul,” source of
all energy and vitality in the physical universe. The early Augustine used the
notion of a world-soul. The late Augustine never said there was no such entity,
but thought the young Augustine rash to assume that there was: “For us God is
not this world, whether or not there is a world-soul. If there is, God created it. If
not, the world cannot be anyone’s god,a fortiori not ours. But even if there is not
a world-soul there is a life-force obeying God working through the angels.”19

Parroting Descartes, one could say that this new substance (the world-
soul) affects all of our senses in the same way that the old substance –
the ‘I’ – would be affecting them if no change or transubstantiation had
occurred. In either case, the inference to sum is mistaken or invalid. In
the first case, that ‘I’ does not exist, and so the conclusion is false; in the
second case, there is blatant ambiguity of terms. The cogito collapses.20
Add other operating assumptions and we have the following possi-
bilities and consequences:

[1] There exists the thought, “I think,” but there is no ‘I,’ no thinker,
thinking that thought. An attribute exists independent of any

19 Henry Chadwick, Augustine, 94.

20 I have encountered the following intriguing objection: Descartes is arguing to the
existence of his self, and not necessarily to the necessary existence of a substance
underlying some attributes. There is no ontology of the self presumed yet at this
point. And thus to argue that Descartes cannot legitimately conclude to the existence
of a substance is irrelevant. This defense of Descartes can be squarely met on several
grounds. First, such a response to Arnauld would be odd, since Descartes himself in
his reply invokes the notion of substance. Second, Descartes would have questioned
the first sentence of the objection, given his 1648 reply to Henri de Roy (Henricus
Regius), Comments on a Certain Broadsheet: “I wrote that we cannot doubt that our
mind exists, because from the very fact that we are doubting, it follows that our mind
exists. . . . From this I concluded and demonstrated that we clearly perceive the mind
as an existing thing, or substance” (CSM, 301; AT VIIIB, 354; see also CSM, 300; AT,
VIIIB 352). Third, nowhere, either before or after the Meditations, did Descartes ever
regard a self as anything other than a substance. Finally, the Eucharist objection could
be read as if it made no overt assumption about the nature of the self. It could be
interpreted simply as making a claim about the ‘I’ parallel to the one that Descartes
makes about the substance, relying on Descartes’ assertion that “many things can be
brought about by God which we are incapable of understanding.”
86 The Solution: Cogito

substance. In this case, the ergo of the cogito cannot possibly lead
to the conclusion that the ‘I,’ or the thinker of that thought,
exists. The argument is invalid, and God is a deceiver.
[2] God is not a deceiver. The thought, “I think” must inhere in the
‘I,’ or the thinker thinking that thought. An attribute inheres of
necessity in a substance. God could not have done otherwise. In
this case, the argument is valid, but God is not omnipotent: He
could not alter at least one eternal truth. Moreover, Arnauld’s
objection, in view of the Eucharist, rises anew in full force.
[3] God is not a deceiver. He could have altered the eternal truth
that an attribute does not exist independent of a substance. The
thought “I think” contingently inheres in the ‘I,’ or in the thinker
of that thought. An attribute inheres contingently in a substance.
In this case, the cogito is a contingent truth and not a necessary
one, and the argument is invalid.21

I conclude: Either Descartes accepted the theological doctrine of

the Eucharist and in private cast aside the necessity of the truth of the
cogito, or Descartes retained the cogito as a necessary truth and espoused
the doctrine of the Eucharist only in public.

V. Doubt and the Cogito

That doubt plays a significant role in Descartes’ metaphysics is gener-
ally appreciated. But what specific role does it play in the case of the
cogito? Hitherto I have argued that the systematic doubt of the First
Meditation was meant to accomplish four things: first, to undermine
current beliefs; second, to underscore the correlation between the de-
gree of doubt required and the degree of simplicity of the proposition
to be doubted; third, to provide the ideal seeker with an arsenal for
challenging substantial claims when they are subsequently made; and
fourth and finally, to provide an indirect way of establishing the first
truth. I now want to bring these claims into the foreground and to show
how they highlight my interpretation of the cogito, and how they en-
able us to correctly understand a solitary, but outstandingly important,
passage.

21 I am grateful to Anthony Kenny for his encouragement on the Eucharist counter-

argument against Descartes provided in this section.
 V Doubt and the Cogito 87

The first claim, about undermining current beliefs, is simple

enough, so I let it pass without comment. The last claim is simple
enough, too, so I let it pass, with the comment that when all of our
current beliefs are shown to be false, then ex hypothesi any truth that is
established after that is the first truth, or the first principle for which
Descartes has been looking. The second point I save for later. I begin
now with the crucial third point and show how, in fact, doubt is used
in cases other than the cogito.
Having just established the truth of the cogito, in the Second Medita-
tion, Descartes says that “I do not yet have a sufficient understanding
of what this ‘I’ is” (CSM II, 17; AT VII, 25). As we know, Descartes
establishes that the essence of this ‘I’ is to think. But the pattern of
reasoning that leads him to that conclusion is important. He lists a
variety of hypotheses:

(h1 ) I am a man.
(h2 ) I am a rational animal.
(h3 ) I am a body (consisting of hands, face, arms, and the whole me-
chanical structure of limbs).
(h4 ) I am a thing that is nourished, moves about, and is engaged in
sense perception and thinking (“and these actions I attributed
to the soul”).
(h5 ) I am a soul (that is, a soul that is imagined to be tenuous, like
wind, fire, or ether).
(h 6 ) I am a thing that thinks.

Having listed these hypotheses, how does Descartes proceed?

Descartes’ announced strategy is this: “I will then subtract anything
capable of being weakened, even minimally, by the arguments now introduced,
so that what is left at the end may be exactly and only what is cer-
tain and unshakable” (CSM II, 17; AT VII, 25). The emphasis is mine.
The “arguments now introduced” are none other than the skeptical
arguments of the First Meditation, and these skeptical arguments are
precisely the ones invoked in rejecting various hypotheses concern-
ing the nature of this ‘I’. There is an exception, though. Hypotheses
(h 1 ) and (h 2 ) are rejected based on a bold assertion not even men-
tioned, to my knowledge, elsewhere in the Meditations on First Philos-
ophy, save in a single sentence in the Second Meditation. The asser-
tion is developed and substantiated, however, in The Search for Truth
88 The Solution: Cogito

(CSM II, 410–412; AT X, 515–518). It expresses the argument against

the scholastic way of solving problems by defining terms, à la Porphyry;
and consequently Descartes dismisses the first two hypotheses as rely-
ing on scholastic subtleties on which he does not wish to waste time.
The point, however, is that this can be rightly construed as a skeptical
argument, part of the general strategy. That is, no real solution is pos-
sible based on the definition of terms that in turn have to be further
defined. The explicitly developed skeptical arguments of the First Med-
itation, then, enable him to undermine, even if these are undermined
only minimally, such hypotheses as (h 3 ), (h 4 ), and (h 5 ). For example,
he rejects the hypothesis that the soul has nutrition or movement on
the ground that these attributes cannot occur without the body, and
that there are already “powerful and well thought-out reasons” for be-
ing skeptical about the existence of the body or any corporeal thing.
When, in the Sixth Meditation, Descartes sets out to prove the exis-
tence of the external world, or “that corporeal things exist,” his strategy
is not one whit different. He says, “To begin with, I will go back over all
the things which I previously took to be perceived by the senses, and
reckoned to be true; and I will go over my reasons for thinking this.
Next, I will set out my reasons for subsequently calling these things
into doubt. And finally I will consider what I should now believe about
them” (CSM II, 51; AT VII, 74). Once again, there is the listing of
hypotheses, eliminating some of them in the light of the skeptical ar-
guments of the First Meditation, and espousing the remainder as true.
So I submit that the method of doubt proposed in the First
Meditation is not just a way to undermine current beliefs. It is much
more. The function of the skeptical arguments is to provide the most
powerful weapons that can be devised to undermine proposed knowl-
edge claims. A proposed knowledge claim is a proper knowledge claim
when, and only when, it withstands the onslaught of the skeptical
arguments. If Descartes does not use some of these arguments in
the Third and Fifth Meditations, it is because there he is proving the
existence of God, and there is no traditional hypothesis that regards
God as a bodily being. Nor is there any occasion to doubt the truths
of mathematics, or any like truths, since they are created by God.
In the Fourth Meditation, no existential claim is proved. So it is
that Descartes’ strategy is prominent mostly in the Second and Sixth
Meditations.
 V Doubt and the Cogito 89

We are now prepared to examine the passage in the Second Medi-

tation that contains the cogito. I shall divide this long passage into four
parts. I divide it differently than the recent tradition does, in order to
exhibit that Descartes’ strategy is at work here just as well and just as
much as it was in the two other cases (where he concluded that his
essence is to think and that corporeal things exist).

(h 1 ) Is there not a God, or whatever I may call him, who puts into me
the thoughts that I am now having? But why do I think this, since I myself
may perhaps be the author of these thoughts? In that case am I not, at least,
something?22
(h 2 ) But I have just said that I have no senses and no body. This is the
sticking point: What follows from this? Am I not so bound up with a body and
with senses that I cannot exist without them? But I have convinced myself that
there is absolutely nothing in the world, no sky, no earth, no minds, no bodies.
Does it now follow that I too do not exist? No: If I have convinced myself of
something, then I certainly exist.
(h 3 ) But there is a deceiver of supreme power and cunning who is deliber-
ately and constantly deceiving me. In that case too I undoubtedly exist, if he is
deceiving me; and let him deceive me as much as he can, he will never bring
it about that I am nothing so long as I think that I am something.
(h 4 ) So after considering everything thoroughly, I must finally conclude
that this proposition, I am, I exist, is necessarily true whenever it is put forward
by me or conceived in my mind. (CSM II, 16–17; AT VII, 24–25)

Descartes is offering us the truth of the cogito, and that truth is

reached as a climax in (h 4 ). It is only when each hypothesis – (h 1 ),
(h 2 ), and (h 3 ), respectively – is offered and refuted that we arrive at
the unrefuted – indeed, irrefutable – claim of the cogito.
Descartes’ procedure is worthy of close examination. Descartes is not
offering a single claim supported by different arguments. He is doing
here tacitly what he does explicitly elsewhere. As in the other two cases,
he is offering different claims supported by different arguments; each
claim is falsified until he reaches (h 4 ). Thus, he offers (h 1 ). Some think
that Descartes is only asking questions, not making any claims (C, 82).
On the contrary, he is making the strong and clear, but implicit, claim
that I (Descartes) am something, namely, the author of my thoughts.
His reason for making this claim is that he is now having some thoughts.

22 There is a clear implication: I am something.

90 The Solution: Cogito

He makes the assumption that nothing comes from nothing, and re-
jects the conjecture that God is the author of his thoughts in favor of
the conjecture that he is himself their author. If he is the author, then
he is something. This is the first step. Perhaps it is best not to regard
this part of the passage as proposing a distinct hypothesis, but rather
to read it as a preliminary to what is immediately to follow.
This step is subjected to a doubt using the kind of skeptical doubt
presented in the First Meditation. If I am to exist as an author, I must
have senses and a body. (This is clearly implied by the following con-
junction of statements: “ . . . am not I, at least, something? But I have
just said that I have no senses and no body.”) So a skeptical argument of
the second degree is introduced, an argument that challenges the ex-
istence of senses and body. (The skeptical argument of the first degree
is the argument against solving problems by defining terms.) Notice
here the second point of the systematic doubt, namely, the degree of
doubt needed to cast an epistemic shadow on a proposition depends
on the simplicity of the proposition in question. The proposition chal-
lenged is a complex proposition: “I am an embodied author.” If (h 2 ) is
regarded as one of the arguments, among four, establishing the cogito,
as some have claimed (K, 56–58), then it is a poor argument. The
premises can be, and are, questioned or doubted. Thus even if the
argument is valid, we do not know it to be a sound one.
This leads to the next step: I do not have to have senses or a body.
I would still exist, so long as I have convinced myself of something.
The proposition to be questioned here – “I have convinced myself
of something” – is even simpler than the foregoing. So, now a third
degree of doubt is introduced, and it is induced in me by “a deceiver
of supreme power and cunning who is deliberately and constantly
deceiving me.” This deceiver can falsify that which I have “convinced
myself of.” For example, I may have convinced myself that there is
nothing in the world: no mind, no earth, no sky, and so on. Once again,
if (h 3 ) is regarded as one of the arguments, among four, establishing
the cogito, then it is an inadequate argument.
And then there is the final step. But first note the difference between
the two following claims:

(a) If I have convinced myself of something, then I certainly exist.

(b) I am [not] nothing so long as I think that I am something.
I paraphrase (b) as
 VI The General Rule and Truth 91

(b ) If I think that I am something, then I am not nothing.23

The antecedents of the two conditionals are quite different. So we

have reason to think that (h 2 ) and (h 3 ) are different. The proposition
to be doubted – “I think that I am something” – is the simplest, and
hence calls for extreme doubt, the fourth degree of doubt. When this
supreme doubt is allayed, then the cogito is established, and not before.
It is for this reason that Descartes says, “So after considering everything
thoroughly . . . ” He intends to say that when the cogito has been sub-
jected to each and every degree of doubt, and emerges unscathed at
the end of the most powerful doubt devised by Descartes, then the
first truth has been discovered, and not before. This first truth, as we
have seen, is elicited from performing a thought experiment: an ex-
periment in which the ‘I,’ through a thought, comes to realize that in
the very act of performing an experiment, devised to show that it, the
‘I,’ does not exist, it, the ‘I,’ is inevitably assured of its existence.

VI. The General Rule and Truth

Why is the cogito important, even if we grant that it serves as the
starting point, the foundation, of Descartes’ philosophy? What other
truths, if any, will it yield? Or what other types of truth will it yield,
which, in turn, will yield truths of the cogito type? The cogito is the
first existential truth, that is, the first truth about what exists in the
universe. The cogito is thus an ontological claim. Descartes’ next step
is not to establish other ontological claims on the basis of this one,
but something else. His next step is to establish the first epistemo-
logical truth. This epistemological truth will henceforth serve as an
insignia of truth. When ideas bear this insignia, then Descartes can
confidently assert them to be true. For example, using this insignia
Descartes will arrive at the second existential truth, namely, God exists.
Using this insignia, henceforth, he will prove the distinction between

23 Let me defend my reading. I have rendered ‘so long as’ as ‘if’ – everything else
remaining the same. It is obvious that ‘if’ does not capture the nuances of ‘so long
as’: for one thing, ‘so long as’ has the idea of something continually present, while
something else is occurring, and to that extent (b) and (b ) are different. Granting
the difference, then, two remarks are in order: First, how crucial is the difference for
the point that is being made? Second, can (b) or (b ) be made the same as (a)? If
not, the conclusion of the argument is not seriously affected.
92 The Solution: Cogito

mind and body, discover their respective essences, demonstrate the

existence of the external world, and so on. Without the insignia, he
is not in a position to assert or know anything to be true, even if it is
true.
With the appropriate context, the epistemological truth reads as
follows:

I am certain that I am a thinking thing. Do I not therefore also know what is

required for my being certain about anything? In this first item of knowledge
there is simply a clear and distinct perception of what I am asserting; this would
not be enough to make me certain of the truth of the matter if it could ever
turn out that something which I perceived with such clarity and distinctness
was false. So I now seem to be able to lay it down as a general rule that whatever I
perceive very clearly and distinctly is true. (CSM II, 24; AT VII, 35; my emphasis)

There is a stronger version of the principle, which I prefer, for it better

captures Descartes’ position. It reads:

I have drawn the conclusion that everything which I clearly and distinctly
perceive is of necessity true. (CSM II, 48; AT VII, 70)

But why of necessity? Because, says Descartes, of his nature. He is so

constituted that he cannot help believing what is clear and distinct
to be true. “Admittedly my nature is such that so long as I perceive
something very clearly and distinctly I cannot but believe it to be true”
(CSM II, 48; AT VII, 69). This must be of his essence. If he were
not so constituted, then he could not ever possess any knowledge,
any certainty. If it were just an accidental feature of Descartes, and
of man in general, that he just happened to believe to be true what
is clear and distinct, then there could be circumstances in which he
had a clear and distinct idea, but in which the accidental feature or
property of believing that idea to be true was absent, so that, in those
circumstances, he would either regard the idea as false or be indifferent
to the truth-value of the idea. There would be no insignia with which
to mark his perceptions or ideas. Thus, there would be no escape
from doubt. One need not fear that this would commit Descartes to
idealism: It is not saying that Descartes’ mind or thought makes the
clear and distinct propositions true. Thanks to God, he cannot help
believing clear and distinct propositions to be true, where truth is
defined independent of Descartes’, or anyone else’s, mind.
 VI The General Rule and Truth 93

Notice that in the general rule the epistemic and the ontological
converge: Ontologically, clear and distinct ideas are true, and epistemi-
cally, clear and distinct ideas are known to be true. (This is the point
in Descartes’ thought that has determined much of the subsequent
development of the history of philosophy. Some have reinforced this
convergence; others have denied it. Those who have denied it are
legion; they have claimed that no amount of piling up of evidence
[epistemic talk] can ineluctably lead to the truth [ontological talk],
the gulf between the two – evidence and truth – being what it is.) If it
were not of his essence, Descartes would have had no insignia of truth,
and yet unbeknownst to him his ideas may have been true. Alterna-
tively, his clear and distinct ideas may have been false, yet he could
have been so constituted, his essence could have been such, that he
could have mistakenly taken his clear and distinct ideas to be true; in
short, the general rule would have been false. The cogito serves as a
powerful anchor that holds these two claims, the ontological and the
epistemic, together in a fascinating way: First, whatever the insignia of
truth, the cogito cannot be false; and second, so clear and distinct is the
idea of the cogito that nothing else could serve, for a human being, as
the epistemic criteria.
When you go to solve a problem, says Descartes, do not forget your
yardstick.

Frequently people are in such a hurry in their investigation of problems that

they set about solving them with their minds blank – without first taking ac-
count of the criteria which will enable them to recognize distinctly the thing
they are seeking, should they come across it. They are thus behaving like a
foolish servant who, sent on some errand by his master, is so eager to obey
that he dashes off without instructions and without knowing where he is to go.
(CSM I, 54; AT X, 434)

Someone looking for an epistemic criterion, such as the general rule,

might easily be puzzled by this advice: If he already has the criterion,
the yardstick, he does not have to look for it; if he does not have the cri-
terion, how will he know that he has the right one if he should stumble
upon it? Perhaps Plato’s Socrates in the Meno could not escape from
this problem; but Descartes thinks he can. The cogito is not established
by any antecedent epistemic criterion; the criterion, the general rule,
emerges only later from the cogito.
94 The Solution: Cogito

What it is vital to recognize, the therefore and conclusion in the pas-

sages quoted earlier notwithstanding, is that Descartes is not explain-
ing or offering an argument – and rightly so – for the epistemological
truth. He is not grounding this epistemological truth on some other
epistemological truth, or on anything else, save the cogito. Indeed,
Father Mersenne had demanded that Descartes do just that:

Why should it not be in your nature to be subject to constant – or at least

very frequent – deception? How can you establish with certainty that you are
not deceived, or capable of being deceived, in matters which you think you
know clearly and distinctly? Have we not often seen people turn out to have
been deceived in matters where they thought their knowledge was as clear as
the sunlight? Your principle of clear and distinct knowledge thus requires a
clear and distinct explanation, in such a way as to rule out the possibility that
anyone of sound mind may be deceived on matters which he thinks he knows
clearly and distinctly. Failing this, we do not see that any degree of certainty
can possibly be within your reach or that of mankind in general. (CSM II, 90;
AT VII, 126; see also CSMK, 201–203; AT III, 474–479)

Mersenne had failed to see the fundamental nature of the general

rule, the epistemic truth. He had failed to recognize that the ideal
seeker after truth must see, intuit, the general rule as emerging from
the cogito, or else nothing could induce him to believe in the veracity of
the general rule. An attempt to explain the general rule would result
in either an infinite regress or a vicious circle. Let us suppose that the
general rule was explained by “a clear and distinct explanation,” E .
Now, the legitimacy of E can be questioned in just the same way as
the general rule was. Either E itself proposes a different insignia of
truth, S, or it proposes the insignia of the general rule of clear and
distinct ideas. In the latter case, we would be arguing in circles. In the
former case, we will require “a clear and distinct explanation” for S.
Then one questions the legitimacy of S, and so we offer a “clear and
distinct explanation,” T, for S’s legitimacy. Then T is questioned, and
so on. The infinite regress is unavoidable. Mersenne’s demand cannot
be met in principle.
Descartes probably foresaw that. When a person suffers from dropsy,
says Descartes, he feels thirsty, but a drink of water would be quite
harmful; here following an instinct implanted in us would lead to dis-
astrous consequences. “[I]t does appear that we are really deceived by
the natural instinct which God gave us” (CSM II, 102; AT VII, 143).
 VI The General Rule and Truth 95

But in the Sixth Meditation, Descartes did explain why this would not
reflect poorly on God, who is all-good. Now, could not God have simi-
larly implanted in us a faculty of judgment that occasionally leads us
astray, in the same way in which the feeling of thirst occasionally leads
us to do a harmful thing? Or is there a parallel story that preserves the
idea of an all-good God? Descartes answers: “In the case of our clear-
est and most careful judgements, however, this kind of explanation
would not be possible, for if such judgements were false they could
not be corrected by any clearer judgements or by means of any other
natural faculty” (CSM II, 102–103; AT VII, 143–144). We can prevent
harming ourselves in the state of dropsy by not drinking water, by
a clearer judgment about our sick condition and about the conse-
quences of drinking water for our body. But what explanation could
be clearer than the clearest and most distinct ideas of which we are
capable?
The beauty of the cogito is that it does not fall prey to the vicious circle
or infinite regress argument. The cogito is not justified by the general
rule. This is why the actual thought experiment and the undergoing of
the doubt by the ideal investigator are so crucial to Descartes’ method.
Descartes is not arguing that the cogito-argument is sound or the cogito-
statement true because it instantiates the general rule. This would put
the cart before the horse. Rather, the ideal investigator finds himself
first in the cogito-state (orcogito-process) and only subsequently recog-
nizes, while still in that state, that the general rule is true. If he does
not come to see it in that state, nothing will induce him to espouse the
general rule. If he does not espouse the general rule, he will not be
certain that he can, henceforth, know any truth.24
What, then, is the notion of truth that Descartes was after? There
is a tension in Descartes’ views. Let us call absolute truth, truth1 , the
God’s-eye view. Then there are relative truths, let us call them truth2 ,
the angel’s-eye view; and truth3 , the man’s-eye view. By parity of rea-
soning, there is truth4 , say, the horse’s-eye view; truth5 ,the bird’s-eye
view; truth6 ,the serpent’s-eye view, and so on. Descartes set out to
discover truth1 , but sometimes spoke as if he was interested only in
truth3 . A proposition may be true3 but not true1 . We may think that
the proposition “The blue bells are blue,” is true1 . But we would be

24 See also Chapter 4, section IV, and Chapter 8, section III.

96 The Solution: Cogito

mistaken; it is only true3 , not true1 , for material objects do not pos-
sess the property of being colored. It requires strenuous thought to
recognize, as did Descartes, that while the material world exists, a
truth of type 1, it is not true1 that they are colored; they only possess
properties, such as length, breadth, width, and duration, that can be
mathematized.
There is every reason to think that Descartes was after the God’s-eye
view, truth1 , insofar as he, a finite being, could attain it. That is, he
wanted to discover as many truth3 s as he could that were also truth1 s.
Yet, this is also what he said:

[A]s soon as we think that we correctly perceive something, we are sponta-

neously convinced that it is true. Now if this conviction is so firm that it is
impossible for us ever to have any reason for doubting what we are convinced
of, then there are no further questions for us to ask: we have everything that
we could reasonably want. What is it to us that someone may make out that the
perception whose truth we are so firmly convinced of may appear false to God
or an angel, so that it is, absolutely speaking, false? Why should this alleged
“absolute falsity” bother us, since we neither believe in it nor have even the
smallest suspicion of it? (CSM II, 103; AT VII, 144–145)

First, even granting the distinction between absolute and relative

truth, there must be some truths that are absolute even on Descartes’
view and not just relative ones, such as the proposition, “God exists.”
And what about the cogito? Is it not an absolute truth, too?
Second, inasmuch as Descartes talks about degrees of clarity and
distinctness, we can imagine the following possibilities. Let us range
these in the following order: C1 , C2 , C3 , . . . ,Cn . Let us suppose that
God has created sentient creatures of n orders.25 The first-order crea-
tures (say, the serpents) have the lowest degree of clear and distinct
ideas, namely, ideas of type C1 . The second-order creatures (say, birds)
have a slightly higher degree of clear and distinct ideas, namely,
ideas of type C2 ; the third-order of creatures (say, horses) have a
slightly higher degree of clear and distinct ideas, namely, ideas of
type C3 , and so on. (If the reader should properly object that since,
for Descartes, these creatures are mechanical objects and have no
mind, they could not have any ideas at all, then he should imagine

25 “For an infinite number of other creatures far superior to us may exist elsewhere”
(CSMK, 349; AT V, 168).
 VI The General Rule and Truth 97

that God created n varieties of Homo sapiens, and that He gifted each
variety with a different unique degree of clear and distinct ideas.)
These creatures are endowed with a nature such that when consid-
ering an idea of the highest degree of clarity and distinctness for
their order, they cannot help but assent spontaneously to that idea. If
Descartes were interested solely in relative truth, then he could claim
only that, given his nature, he cannot help believing that certain ideas
that are clear and distinct are true3 , but not true1 . This is innocuous
enough, and all creatures in each order can make a similar claim, re-
lying on a similar general rule of their own, namely, that whenever
ideas of clarity and distinctness of order i appear to them (creatures
of order i) they shall regard them as truei . This would seem to be
Descartes’ position, duly generalized, if one relied on the last quoted
passage.
If Descartes is interested in absolute truth, then of course he has to
show why the general rule that he relies upon should tell him anything
more than truth3 , rather than truth1 . Descartes, I think, is interested
in absolute truth. The notions of truth and falsity are implanted in
us, says Descartes; he does not say that the notions of relative truth
and falsity are implanted in us. Thus, it is natural to conclude that the
human mind cannot be deceived in taking certain ideas to be true
when these ideas are perceived to be clear and distinct. Could it be
that Descartes thinks that we need to know the existence of God – an
example of absolute truth – so that we are not deceived even about
relative truth?
But third, there is an interesting query to be raised even if Descartes
were interested only in relative truth. Suppose that each order of crea-
tures in God’s creation possessed the ability to have clear and distinct
ideas of all the degrees possessed by all the lower orders. For example, crea-
tures of order j would possess the clarity and distinctness of ideas of
orders 1 through j , and those of order i would possess the clarity and
distinctness of ideas of orders 1 through i, but not that of j . (I limit
myself to a single modality.) Each order of creatures, then, would have
its own outermost limit of clarity and distinctness. The serpent per-
ceives less clearly and distinctly than the bird, the bird less than the
horse, the horse less than man. I wonder: Will each order have its
own unique paradigm that displays the outermost limit to it, as the
cogito displays our outermost limit of clarity and distinctness to us? For
98 The Solution: Cogito

example, while a person pursues truth3 , the angels pursue truth2 . Will
the cogito also serve as a paradigm for the angels – will it also be the
first existential truth that they know?
Now, how could Descartes justify the claim that he has successfully
reached the outermost limit for the human order? We can imagine
the community of creatures of j order making do with only i degree
of clarity and distinctness, until a Descartes in their community comes
along and shows them what they can in principle achieve, namely,
j degree of clarity and distinctness. Indeed, for a long time we took
the deliverance of our senses as giving us true reports, until Descartes
showed us better. Will there arise another Descartes among us who
will take us one step forward toward a higher degree of distinctness
and clarity, as Descartes took us a step forward from what we were
used to before? Failure to demonstrate that the general rule takes
us to the outermost limit of which we as human beings are capable
would mean that we have no reason to be certain even of the relative
truth, let alone of the absolute truth. For we can be certain of the
relative truth only if we are at the outermost limit of what we can
clearly and distinctly perceive. Descartes, therefore, must claim that in
the cogito-state we have arrived at the outermost limit. My question is,
has Descartes shown that?
Fourth and finally, what about the criteria of clear and distinct ideas
themselves? There are plenty of people who go about their entire
lives, says Descartes, without perceiving anything sufficiently clearly
and distinctly to warrant a sound and certain judgment about it. A
judgment based on anything else would surely not do. Hence, it is
essential to understand what Descartes means by clear and distinct.

A perception which can serve as the basis for a certain and indubitable judge-
ment needs to be not merely clear but also distinct. I call a perception ‘clear’
when it is present and accessible to the attentive mind – just as we say that we
see something clearly when it is present to the eye’s gaze and stimulates it with
a sufficient degree of strength and accessibility. I call a perception ‘distinct’ if,
as well as being clear, it is so sharply separated from all other perceptions that
it contains within itself only what is clear. (CSM I, 207–208; AT VIIIA, 22)

What is at work here is more an analogy than a close-knit argument.

Imagine a simple object, such as a piece of wax, present to the eye’s
gaze under normal circumstances, such as ordinary daylight, normal
 VI The General Rule and Truth 99

eyesight of the viewer, and so on. The wax, under those conditions,
stimulates the eye with a sufficient degree of strength and accessibil-
ity. The eye “distinctly” perceives the wax, perceives it as distinct from
surrounding objects. In a similar way, when the senses are withdrawn –
or, so to speak, cast off – and the mind is to itself and fully attentive,
then a ‘distinct’ perception will stimulate it with a sufficient degree of
strength and accessibility. This is the case, for example, when the wax
is “not strictly perceived by the senses or the faculty of imagination
but by the intellect alone” (CSM II, 22; AT VII, 34). Descartes has
an even more distinct perception of himself upon having a more dis-
tinct perception of the wax. As he states it in the Second Meditation:
“Moreover, if my perception of the wax seemed more distinct after it
was established not just by sight or touch but by many other considera-
tions, it must be admitted that I now know myself even more distinctly”
(CSM II, 22; AT VII, 33).
In this passage, Descartes speaks only of distinctness, and not of
clarity. “Clarity” and “distinctness” are not the same thing. Descartes
offers an example of a perception – pain – that can be clear without
being distinct. When someone suffers a pain in the foot, he has a
very clear perception of the pain, but he may sometimes confuse it
with “an obscure judgement [he makes] concerning the nature of
something which [he thinks] exists in the painful spot” (CSM I, 208;
AT VIIIA, 22). Thus, running distinct perceptions together, or mixing
up a perception with a judgment about another perception, is not
conducive to having a distinct perception. Now, whereas what is clear
may not be distinct, what is distinct must be clear (CSM I, 208; AT
VIIIA, 22). Hence, the perception of the wax, and of himself, while
referred to only as being distinct, implies that such a perception is
clear, too.
Evidently, then, the cogito must be both clear and distinct. It must
be, first, a perception that will serve as a basis for a certain and in-
dubitable judgment. This perception is unique in this respect: It will
be present and accessible to the attentive mind and will stimulate it
with the highest degree of strength and accessibility of which a human
mind is capable; in short, it will take the mind to the outermost limit of
its epistemic abilities. This supports, too, the psychological casting of
the cogito in the present chapter. Second, it must be “sharply separated
from all other perceptions.” This lends at least preliminary support to
100 The Solution: Cogito

the hypothesis that the cogito is a mental state rather than a mental pro-
cess, and a first-order mental state rather than a second-order mental
state, since the latter cannot be sharply separated from the first-order
perception that it is about. Third and finally, the cogito must contain
“within itself only what is clear.”
 4

A Skeptic against Reason

The defense of reason has had a long history. Descartes played a pivotal
role in it, if for no other reason than that he sharply focused on the
problem (look no further than the start of the Meditations on First
Philosophy), firmly etched the separate domains of reason and religion,
and believed that his work would quietly, but firmly, restore reason to
its rightful place, not at the periphery of human thought but at its very
epicenter. This chapter attempts to construct a sequence of arguments
that a skeptic might pose against Descartes’ defense of reason.
I begin, in section I, with a brief look at another player in this his-
tory, not given much attention by philosophers: Michel de Montaigne.
Montaigne was Descartes’ precursor, with views on reason that have
much bearing on the work of his successor in France.1 In his book
Descartes against the Skeptics, Edwin M. Curley argues that the proof
of the existence of God is interlocked with the argument of the cog-
ito in this way: “Descartes would hold that even the proposition ‘I
exist’ is fully certain only if the rest of the argument of the Medita-
tions goes through. We must buy all or nothing.” (C, 95) Section II
is designed to demonstrate that Curley’s view would land Descartes in
a paradox from which it would be impossible for him to escape, and
would thereby wreck his entire system. But arguing against Curley is
not the primary purpose of this section. In it, I show that there is a new

1 Montaigne died on September 13, 1592; Descartes was born three and a half years
later, on March 31, 1596.

101
102 A Skeptic against Reason

Cartesian Circle, more devastating than the old one made famous by
Descartes’ contemporary Antoine Arnauld. One needs to confront
this new Cartesian Circle; it is, I think, of intrinsic interest. But more:
The new Cartesian Circle will be needed as a constant reminder later,
when various ways of understanding the cogito as an argument are dis-
mantled. It should also become amply clear why the Cartesian Circle
argument, usually discussed in relation to the Third Meditation, is
discussed here when dealing with issues prominent in the Second.
Sections III and IV are devoted to examining the common allega-
tion that Descartes committed an egregious error at the heart of his sys-
tem, infamously known as the Cartesian Circle. (He did not, in my view.
But I shall not argue that here.) Here I am concerned with refuting
an interesting proposal by Curley intended to free Descartes from that
objection. The virtue of my argument is its generality. If my argument
is right, a skeptic can be defended against Curley – or at least against
Curley’s Descartes. Section V presents an argument against James van
Cleve’s well-known way of escaping from the Cartesian Circle. The final
section shows how a skeptic might argue against a recent conjecture of
Stephen Gaukroger in answer to the question, “ What was the method
Descartes adopted in his deductive demonstrations in the Meditations?”
Whatever the historical worth of this conjecture, a skeptic could show
that it would have been something of an embarrassment for Descartes.

I. Why Natural Reason?

Michel de Montaigne:

God in his mercy may perhaps have deigned to protect those tender principles
of rough-and-ready knowledge of Himself which Natural Reason affords us,
amid the false imaginings of our dreams. But there are religions Man has
forged entirely on his own: they are not only false but impious and harmful.
(M, 82)

Natural Reason is taught by philosophers to be “the Comptroller-

General of everything within and without the vault of heaven; they
themselves say that it can embrace everything, do everything, and
is means by which anything is known or understood” (M, 116). But
Natural Reason is not to be trusted: It is an imbecile; it is sickly; it
is so inadequate, so blind; it is full of falsehood, error, defects, and
feebleness; it hobbles, limps, and walks askew (M, 86, 12, 13, 117, and
 I Why Natural Reason? 103

144). Nor was this merely notes for a program. Montaigne offered at
least two arguments: one remarkably Cartesian in its design, the other
more modern.2
The first argument: “Our discursive reasoning,” says Montaigne, “is
driven and shaken at the mercy of [the] influence” of the sun, moon,
and stars (M, 14). They “govern us by silent laws.” As a result of their
influence, not just one man or one king, but monarchies and empires
reel and collapse, and our discourse on virtues and vices, knowledge
and competencies is (mis?)guided. “If we are dependent upon the
disposition of the heavens for such little rationality as we have, how
can our reason make us equal to the Heavens? How can their essence,
or the principles on which they are founded, be subjects of human
knowledge?” (M, 15) So I infer: If the influence of the heavenly bodies
is baneful on us, it will make us think that reason is thus-and-so in
essence when reason is no such thing; if their influence is pernicious,
they may thwart our reason as easily as they engineer the collapse of
a king. How can we tell? Perhaps our reason is ill-structured, let alone
equal to the reason of the heavens. Descartes cannot counterargue
that the astrological hypothesis – namely, that the heavenly bodies
determinately influence the behavior of a person – is quaint, at best.
Like the improbable hypothesis of the evil genius that is designed
to cast doubt on the truths of mathematics, the former hypothesis is
offered to cast doubt on reason’s trustworthiness.
The second argument: “ The senses themselves being full of uncer-
tainty cannot decide the issue of our dispute. It will have to be Reason,
then. But no Reason can be established except by another Reason. We
retreat into infinity.” (M, 185) A reason cannot establish itself; it has
to be established by another. No unestablished reason is reliable. If no

2 Compare this to the critique of reason set ninety-eight years afterMeditations on First
Philosophy: “Reason first appears in possession of the throne, prescribing laws, and
imposing maxims, with an absolute sway and authority. Her enemy, therefore is obliged
to take shelter under her protection, and by making use of rational arguments to prove
the fallaciousness and imbecility of reason, produces, in a manner, a patent under
her hand and seal. This patent has at first an authority, proportion’d to the present
and immediate authority of reason, from which it is deriv’d. But as it is suppos’d to be
contradictory to reason, it gradually diminishes the force of that governing power, and
its own at the same time; till at last they both vanish away into nothing, by a regular and
just diminution.” David Hume, A Treatise of Human Nature, Book I, Part IV, section 1,
“Of Scepticism with Regard to Reason,”125.
104 A Skeptic against Reason

unestablished reason is reliable, then Reason is not reliable. There-

fore, Reason is not reliable. This preamble from Montaigne is drawn
to suggest that Descartes, too, should have worried, with the skeptics,
over the reliability of reason. Did he?
“I have always thought,” says Descartes, in the “Dedicatory Letter to
the Sorbonne,” which serves as the opening of the Meditations on First
Philosophy,

that two topics – namely God and the soul – are prime examples of subjects
where demonstrative proofs ought to be given with the aid of philosophy rather
than theology. For us who are believers, it is enough to accept on faith that the
human soul does not die with the body, and that God exists; but in the case of
unbelievers, it seems that there is no religion, and practically no moral virtue,
that they can be persuaded to adopt until these two truths are proved to them
by natural reason.3 (CSM II, 3; AT VII, 1–2)

Descartes goes on to say that, of course, we must believe in God be-

cause it is the cardinal doctrine of the Holy Scriptures, and that we must
believe in the Holy Scriptures because they are a gift of God. The rea-
soning is circular. For the believers it does not matter; but it is of some
consequence if one wishes to persuade the unbelievers, the skeptics.4
What, then, is this thing called Natural Reason, and why should an
unbeliever trust it any more than faith or scripture?5 It is often noted

3 Raymond Sebond put it this way: “It alleges no authority – not even the Bible – for
its end is to confirm what is written in Holy Scripture – and to lay the foundations
on which we can build what is obscurely deduced from them. And so, in our case, it
precedes the Old and New Testament.” (M, Appendix 2, xlii–xliii)
4 Lest I be as guilty as his biographers (R, ix), let me hastily cite the last words of
Maurice of Nassau, who, when admonished by a pastor to “make a kind of profession
of faith,” said: “I believe that 2 and 2 are 4 and that 4 and 4 are 8” (R, 62). Descartes
was interested in defending the faith, especially against skeptical mathematicians who
held statements of mathematics to be veridical while doubting the existence of God
(R, 79, 108–109). Among others, Descartes quarreled with Voetius, who found hy-
perbolic doubt unpalatable and by implication accused Descartes of being an atheist
(R, 147–148, 163–164, 169–170, also 171–173, 206). Descartes “wanted official ac-
knowledgment that he had never been suspected of atheism” (R, 170).
5 So when Rodis-Lewis claims, “Nevertheless, to avoid all polemic, and because ‘natu-
ral reason’ is in agreement with ‘the Christian religion,’ which “wants us to believe
in that manner,” (R, 102), a skeptic might respond, “That still needs to be demon-
strated; Descartes must assure us that he is beginning with reason that is neither
Christian, nor of any other religion, in its underpinning. It must be natural tout court.
Otherwise, the argument that Descartes was involved in circular reasoning would
stand undiminished.”
 I Why Natural Reason? 105

that Descartes’ doubt is not sufficiently sweeping: He doubts his senses,

his belief in an external world, mathematics, and other such things,
but he never doubts his reason. Why should not reason be just as much
subject to serious doubts as anything else? Indeed, the circularity of
reasoning, an unbeliever might argue, is no less evident here. To justify
reason, one has to produce good, strong reasons. How else can reason
be justified? But no matter how strong and good the reasons are in
support of Natural Reason, they are first and foremost reasons. So
one would have to say that one must believe in reason because there
are good and strong reasons in support of it, and conversely that one
must believe in these good and strong reasons because they are the
deliverances of reason.
Now, of course, the good and strong reasons can be redescribed in
such a way that they are no longer described as reasons. But then they
are either stronger or weaker than reasons. If weaker, the justification
is incomplete; if stronger, then they in turn have to be justified. At
this point, the matter is given up. There is a parallel problem in justi-
fying deductive and inductive logic, respectively. Deductive logic can
be justified either deductively or inductively; to justify deductive logic
deductively is to argue in circles, and to justify it inductively is not to
argue strongly enough. Similarly, inductive logic can be justified either
inductively or deductively; to justify inductive logic inductively is to ar-
gue in circles, and to justify it deductively is to argue too strongly.6 Nor
do I have a way of solving the problem of justifying reason – as against
faith, say. But Descartes cannot take that stand, if his initial skepticism
is to be comprehensive.
Indeed, Descartes’ failure to question his capacity to reason is puz-
zling, for his manner of doubting his senses would have provided him
with a near-exact model for doubting his reason. There are, I shall ar-
gue, three degrees of doubt. Descartes could have constructed doubts
about his reason that are analogous to the doubts about his senses.
His first round of doubt about the senses explains, and justifies, the
obvious claim that our senses sometimes deceive us; the second round
of doubt justifies a stronger claim, namely, that at any given time we
are unable to say whether they deceive us or not; and finally, the third
and strongest claim is made, namely, that we (may) have no senses

6 See Susan Haack, “The Justification of Deduction.”

106 A Skeptic against Reason

at all. Similarly, Descartes could have ventured the following: a first

round of doubt about reason, showing how using reason sometimes
misleads us; a second round of doubt to show that at any given time
we would be unable to tell whether we have reasoned correctly; and
finally, a third round of doubt to show that we (may) have no reliable
reason at all.
As we have seen, Descartes’ view of the ideal seeker evolved quite
gradually. It must have been an important problem for him, otherwise
his frequent remarks on the issue would be difficult to explain. Con-
sider, then, Descartes’ remark in Discourse on the Method: “The simple
resolution to abandon all the opinions one has hitherto accepted is not
an example that everyone ought to follow. The world is largely com-
posed of two types of minds for whom it is quite unsuitable.” (CSM I,
118; AT VI, 15) The first type consists of those who think themselves
too clever, are precipitate in their judgements, are unable to make an
orderly inquiry, and so on; the second type consists of those who have
just enough reason to recognize their lack of capacity to distinguish the
true from the false. And let us say that there is a third type, composed
of those who have just the right disposition, aptitude, and capacity –
the type that constitutes the ideal thinker.
Descartes needs some justification for discriminating the first two
types from the third. The first two types of people are known to
make mistakes in reasoning, plain and plenty. So that even if these
individuals occasionally reason correctly, it is best not to trust their
reasoning. “It is prudent,” someone seeking to justify the use of
Natural Reason might say, “never to trust completely those who have
deceived us even once.” But this issue does not challenge Descartes to
produce a justification for using Natural Reason, or for relying on the
third type of individuals, who, let us grant, occasionally make errors.
Could the demon be deceiving Descartes into thinking that the first
two types of individuals would not be good ideal seekers? Could the
demon be deceiving Descartes into thinking that the third type would
make for a good ideal seeker? How should the demon be answered?
Consider the following famous passage:

[H]ow could it be denied that these hands or this whole body are mine? Unless
perhaps I were to liken myself to madmen, whose brains are so damaged by
the persistent vapours of melancholia that they firmly maintain they are kings
when they are paupers, or say they are dressed in purple when they are naked,
 I Why Natural Reason? 107

or that their heads are made of earthenware, or that they are pumpkins, or
made of glass. But such people are insane, and I would be thought equally
mad if I took anything from them as a model for myself.7 (CSM II, 13; AT VII,
18–19)

Let us say that these constitute set1 of mad beliefs.

But Descartes’ doubt had led him to the claim that there is nothing
in the world: no sky, no earth, no minds, no bodies, no colors, shapes,
or sounds, no hands or eyes, and no flesh and blood. Yet, in the Synop-
sis he had described these as things “no sane person has ever seriously
doubted” (CSM II, 11; AT VII, 16). These, say, constitute set2 of mad be-
liefs. How should one distinguish one set of mad beliefs from another
set? How are we to distinguish set1 from set2 ? Not, surely, in terms of
the content of their claims: for any one claim in one set is as false as any
claim in the other set. If Descartes is to argue successfully, he has to
show us that there is a difference in the manner and structure of their
reasoning.
If the physical world is doubted, then the scientific conjecture that
the madmen have had their brains deranged by the vapors of melan-
cholia is untenable, or ought to be untenable, to the inquirer. Worse
yet, perhaps minds of the third type are infected by the noxious vapors.
How can one tell? Perhaps, then, a neutral terminology is required.
The first and second types of individuals have mental processes of
X-kind. Descartes and others belonging to the third type of individu-
als have mental processes of Y -kind. In terms of what are the Y -kind
mental processes to be preferred over the X-kind? Once again, why
should not the evil demon be persuading Descartes that the Y -kind of
mental process is to be preferred? Let us suppose that he who thinks
that he is a pumpkin, or a king, is at least consistent. That, surely, can-
not be ruled out in an a priori way. Of course, consistency is so minimal
a demand that it can easily be met by the most absurd collection of
statements. This is just the point, though. One should not eliminate the

7 Having just seen, and heard, the three witches, Banquo exclaims:
Were such things here, as we do speak about,
Or have we eaten on the insane root
That takes the reason prisoner?
Macbeth, act 1, scene 3, lines 83–85. See also act 1, scene 3, lines 122–124; and act 1,
scene 7, lines 66–68. It is useful to be reminded that Shakespeare wrote Macbeth in
1623.
108 A Skeptic against Reason

objection by claiming that inconsistency is written into the statements

of a lunatic, or of an individual belonging to one of the first two types.
Here is a different problem: How can Descartes tell that he is not,
at certain moments, seized by the X-kind of state or process? If that
were the case, at any given time Descartes could not tell whether he
was, at that moment, in an X-kind or a Y -kind state or process. This
problem is analogous to the second degree of doubt with respect to
the senses, namely, that for all we know, at any given time, our senses
may be deceiving us.
Perhaps this alone might have convinced Descartes that there is,
after all, no certainty. He had made room for just that possibility in
the opening paragraph of the Second Meditation: “I will proceed in
this way until I recognize something certain, or, if nothing else, until
I at least recognize for certain that there is no certainty” (CSM II, 16;
AT VII, 24). He might have concluded that his reasoning is simply a
teaching of nature: “Nature has apparently taught me to think this”
(CSM II, 26; AT VII, 38). If Descartes reasoned at all, it was inevitable
that he did so, but his reasoning was not therefore reliable. His reason-
ing would be a spontaneous, blind impulse that leads him to believe
certain things (CSM II, 26–27; AT VII, 38–40).
There is a different problem that needs to be addressed, and one
that falls within the purview of Descartes. But for all that, he did not
raise the question, let alone answer it. This is not a problem of how
to justify reason over faith, but rather a problem of how to arbitrate
between two doctrines concerning the origin and reliability of reason:
reason as a natural property of a person as against reason as a God-
given property of a person. At the start, Descartes relies on reason as a
natural property of a person: He thinks that he cannot do otherwise,
or else he will fail to convince the atheist, the unbeliever, and the
skeptic. With the aid of this natural property alone he establishes the
existence of God, and then claims that the natural property of a person
is really a God-given property of man. Hence, he can rely on it and
not err.
By contrast, consider, for example, Saint Augustine’s theory of rea-
son as illumination. Augustine argued that the light of God illuminates
our mind, and that it is this illumination that enables us to see clearly
and distinctly; were it not for this illumination, we would have no rea-
son to believe in what we perceive through the intellect.
 II “Buy All or Nothing” 109

There is no argument in support of reason as a natural property of

man. And, surely, it needs justification. Descartes imagines that it is
possible that he was created by “fate or chance or a continuous chain
of events, or by some other means.” But, he says, the less powerful
the cause that brought him to his present state, the more likely it is
that he is an imperfect being. This imperfection may be in his sense
experiences; perhaps they are defective. The imperfection may also be
in his reason; perhaps he errs more often than not when he uses his
reason. Perhaps truth is to be acquired through feeling, in a manner
after Blake or Wordsworth.
To put the point crudely, but in a more contemporary idiom: Hu-
man beings have evolved through natural selection; traits survive that
best fit the demands of the environment. Other traits perish. Our rea-
son is one of our traits: Our ability to reason may well have evolved
over time, although some think not.8 If so, our reason is minimally
imperfect, even as practiced by the scientists.9 We may be making lots
of mistakes. What reason have we to think that our Natural Reason is
powerful enough to yield metaphysical truths of the highest order? Evi-
dently, one must develop some argument in support of the claim that
reason as a natural property of a person can be relied upon. Failure
to do so would be a fatal omission in the system. Such a defense would
be required to sustain all the propositions from the cogito on up.

II. “Buy All or Nothing”

In this section, I offer a new Cartesian Circle (paralleling the old one
offered by Arnauld). This new Circle has, I think, profound impli-
cations for how the cogito should be read, whether as an argument
or an intuition. The immediately succeeding chapters will use the re-
sult obtained in the present section as a significant backdrop against
which their own claims are set. That result is my primary concern
here, but this new Circle is offered in the context of an argument
against Curley, my secondary concern. This bit of demolition, to-
gether with the new Circle, will bring out the importance of decid-
ing what precisely lies within the scope of Cartesian doubt, and thus

8 Brian Ellis, Rational Belief Systems.

9 See David Faust, The Limits of Scientific Reasoning.
110 A Skeptic against Reason

how the famous passage from the Third Meditation should be read.
The reading of that passage has an enormous implication for the
cogito.
Curley concludes his chapter on the cogito thus:

We must reverse a very common judgment about Descartes’ system. It is often

said that Descartes, in his quest for absolute certainty, is able to establish his own
existence, but unable to establish the existence of anything else. If what has
been argued here is correct, Descartes would hold that even the proposition
“I exist” is fully certain only if the rest of the argument of the Meditations goes
through. We must buy all or nothing. (C, 95)

Call this Curley’s Central Contention.10

I want to argue that Curley’s Central Contention commits Descartes
to an egregious error. This is how my argument goes. Suppose
that there is the following argument for “I exist.” Call it the cogito-
argument:

Premises: S1
S2
S3 Rules of Inference:R1 , R2 , . . . , Rn
·
·
·
·
Sn

∴ I exist.

10 I wonder why Curley needs to make so strong a claim. Wouldn’t going up to the
Third Meditation, which offers two proofs of the existence of God – or perhaps
just one proof, given what Descartes said to Johannes Caterus – suffice? How about
up to the Fifth Meditation, which offers the ontological proof ? Why is it that
we need to rely on the rest of the argument of the Meditations? What, in particu-
lar, does the Sixth Meditation have to add that is of vital concern to the present
enterprise?
While I argue against Curley, my argument applies to any other view tending in
the same direction. For example, it applies to Stephen Gaukroger, who maintains a
Curley-like thesis, thus: “The cogito presents us with a paradigmatic case of clarity and
distinctness of the kind that Descartes is seeking, but it does not legitimate the use
of clarity and distinctness as a criterion. That task rests ultimately with God, whose
existence we can deduce [sic] from the idea that we have of Him. God then acts as
the guarantor of knowledge, although how He acts in this way is not specified here.”
(G, 320)
 II “Buy All or Nothing” 111

Now a skeptic might argue against Curley’s Descartes as follows: If

Descartes were offering an argument for the cogito, as Curley alleges –
not a sound argument, not a proof, not a demonstration, not any such
thing – that argument minimally would have to be valid.11 But the
validity of the argument is directly dependent upon the soundness of
the rules of inference.12 Since the rules of inference, like the truths
of mathematics, could be questioned by postulating a most malignant
and powerful demon, the rules of inference cannot be demonstrated,
at this stage of the argument anyway (which is all of the First Meditation
and into the second paragraph of the Second Meditation). Therefore,
even Curley’s quite modest proposal seems to be based on the truth of
propositions – eternal truths, so-called – that Descartes has left quite
scathed by the end of the First Meditation. Curley’s cogito argument is
not even known, nor can it be known, to be valid. Not only Curley’s
particular version of the nondemonstrative argument, but no argu-
ment can be offered as a valid one, at this juncture in the Meditations,
to capture the cogito. But I intend to make a much stronger claim.
Curley thinks that the remedy lies in having a proof of the existence
of God. Once the argument for the existence of God is in place, then

11 I discuss in detail Curley’s way of understanding the cogito in the next chapter – see
the Fourth View – and so I omit it here.
12 In his reply to Pierre Bourdin, Descartes says, “Finally, when he is about to deploy
syllogisms in his formal presentation, and he extols them as “a method of conducting
our reasoning” which is to be contrasted with my own, his apparent intention is to
persuade people that I do not approve of syllogistic patterns of argument, and hence
that my method is not a rational one. But this is false, as is clear enough from my
writings where I have always been prepared to use syllogisms where the occasion
required it.” (CSM II, 355; AT VII, 522)
Well, what might such an occasion be? “Amongst others, the very syllogism whose
matter and form my critic pretends that I repudiate, is one which he copied down
from my own writings; for I use it at the end of my Replies in the Second Set of
Objections, proposition 1, where I demonstrate the existence of God” (CSM II, 371;
AT VII, 544). In his letter to Silhon (March or April 1648), Descartes wrote that
intuitive knowledge is quite different from discursive knowledge, and that it is the
latter we must engage in perforce when it comes to acquiring knowledge about God.
Thus, “to speak more accurately, using the natural (and consequently comparatively
rather obscure) knowledge of one attribute of God, to construct an argument leading
to another attribute of God. So you must admit that in this life you do not see, in
God and by his light, that he is unique; but you deduce it from a proposition you
have made about him, and you draw the conclusion by the power of argument, which
is a machine which often breaks down.” (CSMK, 332; AT V, 138–139) For a more
complete background, I beg the reader to consult Chapter 6, pp. 191–2, 203-7.
112 A Skeptic against Reason

the cogito can be justified, which in turn would be required to get

the proof of the existence of God going (or, at least, one of the three
proofs of the existence of God that Descartes offers in the Meditations).
Suppose, then, that there is the following argument for “God exists.”
Call it the God-argument:

Premises: K1
K2
K3 Rules of Inference: L 1 , L 2 , . . . , L n
·
·
·
·
Kn

∴ God exists.

Curley’s Central Contention argument cannot go through. Either

the rules of inference, R1 , R2 , . . . , Rn , of the cogito-argument are the
same as the rules of inference, L 1 , L 2 , . . . , L n , of the God-argument, or
they are different. (For the sake of simplicity, I ignore the alternative
in which they partially overlap.) We have this paradox: If they are the
same, then the rules are as much in need of justification in the proof of
the existence of God – that is, in the God-argument – as they were when
offered as part and parcel of the first argument, the cogito-argument,
which implied the truth of “I exist.” This is a simple case of question
begging.
Alternatively, if the rules of inference are not the same, then the
existence of God may verify the rules of inference, R1 , R2 , . . . , Rn , and
yet not verify the rules of inference, L 1 , L 2 , . . . , L n . One cannot say
that if, indeed, God does exist and can verify, by His goodness and
undeceitful nature, the eternal truths (namely, the rules of inference,
R1 , R 2 , . . . , R n ), then surely God justifies the rules of inference, L 1 ,
L 2 , . . . , L n , as well. Of course not. The careful thing to say is that if
Curley is right, then Descartes has simply not given us a proof of the
existence of God. For the rules of inference, L 1 , L 2 , . . . , L n , which are
being used in the God-argument, are just as much in doubt as were the
earlier rules of inference. The soundness of the rules of inference, L 1 ,
L 2 , . . . , L n , has to be established before they can be used in constructing
 II “Buy All or Nothing” 113

the proof, the God-argument. This he surely cannot do. This is a simple
case of justification failure.13
There is an extra puzzle. Curley has said, when arguing against
Harry Frankfurt’s book Demons, Dreamers, and Madmen, that even if the
premises were true, they would still not ensure the indubitability of the
conclusion (C, 82–84). But if there is no assurance that the premises of
an argument are true, whence the assurance that the conclusion is true,
let alone indubitable? (Was this the insight that compelled Frankfurt to
divorce truth and indubitability in his book?) Unless, of course, there
is a core supposition whose truth is guaranteed and from which, and
from which alone, flows the truth of “I exist”? But Curley is unwilling
to be saddled with the epistemic labor of justifying the premises. To go
a step further, if the cogito provides a crucial starting point for a proof
of the existence of God, and the cogito-argument is dubious, we should
regard the God-argument as merely an alleged proof.
There is a line of argument that might be suggested in defense. It
might be argued that Curley explicitly draws attention to two types of
indubitability, namely, normative as opposed to descriptive indubitabi-
lity (C, 82–83, 100–101). Until one has a proof of the existence of God,
one might have descriptive indubitability for the cogito-argument, but
after, and only after, that proof can one have normative indubitability
for that same cogito-argument. Curley says, referring to a passage cited
by a critic, “ The passage would thus be saying that because I cannot, in
fact, doubt the teachings of the natural light (when I attend to them),
I must, at least provisionally, treat them as though they ought not to be
doubted in examining whether there is a God.”14 Presumably, then,
once we have the proof for the existence of God, we would at the same
time achieve normative indubitability.

13 There is a parallel here with David Hilbert’s claim about proving the consistency of
arithmetic. Other branches of mathematics can be shown to be consistent if arith-
metic is assumed to be consistent. Hilbert himself has shown, for example, that
Euclidean geometry is consistent if arithmetic is. “In geometry and physical theory
the proof of consistency is accomplished by reducing it to the consistency of arith-
metic. This method obviously fails in the proof of arithmetic itself.” Quoted in Morris
Kline, Mathematics: The Loss of Certainty, 249. Hilbert’s claim was that if in proving the
consistency of arithmetic one used in the metalanguage (the language in which the
proof was carried out) the very rules of inference whose consistency was in question,
then, of course, arithmetic could be “demonstrated” to be consistent. But, one would
have done so only by begging the question.
14 This is suggested both in the text and in note 25 (C, 94–95).
114 A Skeptic against Reason

I do not see how. We run into the same problem. Once we have the
proof – if we have it – of the existence of God in hand, the rules of
inference, R1 , R2 , . . . , Rn , are now no longer descriptively indubitable,
but rather normatively indubitable. Even supposing that were so, this
would leave the rules of inference, L 1 , L 2 , . . . , L n , without any norma-
tive justification. Since these were used in the proof of the existence of
God, their indubitability was at best descriptive. Of course, the careful
thing to say is that if Curley is right, then Descartes has simply not given
us a proof of anything. For the rules of inference, L 1 , L 2 , . . . , L n , that
are being used in the God-argument are just as much in doubt as were
the previous rules of inference. We do not have a normative proof of
the existence of God, only a descriptive proof – if I may be permitted
a parallel distinction. Since God is the guarantor of the truth of every-
thing else, according to one reading of a passage I shall presently cite,
this spells epistemic doom.
Curley, like several other scholars, reads the following passage from
the Third Meditation of Descartes’ Meditations on First Philosophy as
including the cogito in its skeptical scope:
I spontaneously declare: let whoever can do so deceive me, he will never bring
it about that I am nothing, so long as I continue to think I am something; or
make it true at some future time that I have never existed, since it is now true
that I exist; or bring it about that two and three added together are more or
less than five, or anything of this kind in which I see a manifest contradiction.
And since I have no cause to think that there is a deceiving God, and I do
not yet even know for sure whether there is a God at all, any reason for doubt
which depends simply on this supposition is a very slight and, so to speak,
metaphysical one. But in order to remove even this slight reason for doubt, as
soon as the opportunity arises I must examine whether there is a God, and, if
there is, whether he can be a deceiver. For if I do not know this, it seems that
I can never be quite certain about anything else. (CSM II, 25; AT VII, 36)
If the correct reading of this passage is that I cannot be certain
about the cogito, about the truths of mathematics, and so on, unless I
have examined and successfully concluded – unless, that is, I know –
that there is a God, it would result in the eventual collapse of the
normative indubitability of R1 , R2 , . . . , Rn , and with it the collapse of
the cogito-argument and everything else.15

15 On Curley’s interpretation, Descartes was remarkably coy about telling us in a

forthright manner that his famous conclusion of the Second Meditation was pro-
visional, subject to the success of the enterprise of proving the existence of God, to
 II “Buy All or Nothing” 115

Many historians and philosophers are quite bothered by the just-

cited passage, and seem to argue, or at least to arrive at the same con-
clusion, in a manner not unlike Curley’s. The astonishing thing is that
very rarely, if ever, does one even mention, let alone discuss, a passage
that comes in the same meditation just three and a half pages later:

But what is my conclusion to be? If the objective reality of any of my ideas

turns out to be so great that I am sure the same reality does not reside in me,
either formally or eminently, and hence that I myself cannot be its cause, it will
necessarily follow that I am not alone in the world, but that some other thing
which is the cause of this idea also exists. But if no such idea is to be found
in me, I shall have no argument to convince me of the existence of anything
apart from myself. (CSM II, 29; AT VII, 42)

Descartes is here stating quite explicitly, and in the context of the

proof – the first proof – of the existence of God, that even if he had
no proof of the existence of God, he would still be convinced that he
himself existed, only he would have no argument to demonstrate that
something besides himself existed as well. He emphatically does not
say that because he has no proof of the existence of God, he cannot
be reasonably sure that he exists in the world, too.16

be undertaken in the Third and Fifth Meditations. Not only was he not coy, with
Bourdin he was particularly blunt. Here are two passages from his reply to Bourdin,
the author of the “Seventh Set of Objections.” “Note that my critic here admits that
I have made my first step in philosophizing, and for the first [sic] time established a
proposition as firm, by recognizing my own existence” (CSM II, 323; AT VII, 480).
One page later: “Thus I was right to begin by rejecting all my beliefs; and later on,
noticing that there was nothing which I could know more certainly or more evidently
than that I existed so long as I was thinking, I was right to make this my first [sic] asser-
tion.” Furthermore, if Curley is right, the order of philosophizing, which Descartes
so emphasized, is inexplicable. If the cogito and the general epistemic rule are unre-
liable without the proof of God’s existence, and the proof of God’s existence can, in
principle, go through without the other two, why did the proof of God’s existence
not directly succeed the metaphysical doubt? Descartes could then have laid down
his general rule, then the cogito, and then the proof about his essence. Why did not
Descartes proceed in that way? I submit that it was not odd for Descartes to do what
he did precisely because Descartes fully recognized that without the cogito and the
general rule, no subsequent part of his philosophy could be carried out. I aver that
even if the proof of the existence of a good God fails, the cogito would still stand
unscathed.
16 In a letter (June or July 1646), Descartes wrote to Clerselier: “[I]t is very useful
indeed to convince oneself first of the existence of God, and then of the existence
of all creatures, through the consideration of one’s own existence.” In the immediately
preceding paragraph, Descartes had identified the first principle [sic] as that our soul
exists (CSMK, 290; AT IV, 444–445).
116 A Skeptic against Reason

Curley has seriously underestimated the problem that Descartes

would face on his reading. It is not, as Curley thinks, merely a matter
of Descartes’ not coming up with an adequate proof of the existence
of God. If my argument is correct, Curley has made sure that Descartes
could not come up with such a proof. For any proof must presuppose
the soundness of the rules of inference that it uses; ex hypothesi, in the
absence of a proof of the existence of God, no rules of inference can
be demonstrated to be sound. Therefore, there can in principle be no
proof of the existence of God. On Curley’s reading, Descartes’ project
is inescapably circular.17 Descartes’ defense of reason is a failure. So
there really is nothing to buy. Caveat emptor! Since there cannot be a
proof of the existence of God, in the absence of rules of inference
known to be sound, there is no assurance for the cogito either. Curley’s
view of the cogito as an inference or an argument leaves Descartes with
nothing. So a skeptic might argue.

III. Attempting to Step Out of the Circle

Antoine Arnauld found Descartes guilty of circular reasoning.18 This is
what Arnauld famously wrote to Father Marin Mersenne (1588–1648):

I have one further worry, namely how the author avoids reasoning in a circle
when he says that we are sure that what we clearly and distinctly perceive is
true only because God exists.

17 Here is another oddity. In the Fourth Meditation, in the context of explaining the
source of his mistakes and the scope of his will, and giving examples to illustrate where
and how he makes mistakes and where and how he does not make them, Descartes
says: “For example, during these past few days I have been asking whether anything
in the world exists, and I have realized that from the very fact of my raising this
question it follows quite evidently that I exist. I could not but judge that something
which I understood so clearly was true.” (CSM II, 41; AT VII, 58) First, there is talk of
what exists in the world; second, the cogito is offered as a prime example of evident
knowledge; third, in this case the will is freest when it assents to the cogito; fourth,
there is no qualification; and fifth and finally, there is no mention of God.
18 This is the same Arnauld of whom Descartes wrote to Mersenne on March 4, 1641,
after he had received objections from several readers of his Meditations on First Philos-
ophy: “. . . M. Arnauld, who has put me greatly in his debt by producing his objections.
I think they are the best of all the set of objections, not because they are more telling,
but because he, more than anyone else, has entered into the sense of what I wrote.”
(CSMK, 175; AT III, 331)
It should be noted that while the “Second Set of Objections” was attributed to a
nameless set of “theologians and philosophers,” the objections were mainly authored
by Mersenne. Furthermore, Mersenne gave as crisp a statement of circular reasoning
as Arnauld did. (CSM II, 89; AT VII, 124–125)
 III Attempting to Step Out of the Circle 117

But we can be sure that God exists only because we clearly and distinctly
perceive this. Hence, before we can be sure that God exists, we ought to be
able to be sure that whatever we perceive clearly and evidently is true. (CSM II,
150; AT VII, 214)

In the Third Meditation, Descartes had formulated the crucial gen-

eral rule. It stated that “whatever I perceive very clearly and distinctly is
true” (CSM II, 24; AT VII, 35). But the general rule, Arnauld asserted,
could not be relied upon until the proof of the existence of God was
forthcoming. The proof of the existence of God, on the other hand,
could be a proof only if it were very clear and distinct – Descartes had
spoken of his proofs for the existence of God and for the distinction
between the mind and the body as “quite certain and evident” (CSM II,
5; AT VII, 4). Thus we have a short, inescapable circle. Making it as
closely parallel to the circular argument Descartes himself had iden-
tified,19 we can state the Cartesian Circle as follows: “It is of course
quite true that we must believe in the existence of a good God because
it is a doctrine delivered and endorsed by reliable clear and distinct
ideas, and conversely, that we must believe in the reliable clear and
distinct ideas because they are endorsed by the existence of a good
God; for since faith is the gift of God, He who gives us grace to believe
other things can also give us grace to believe that He exists and that
we can rely on clear and distinct ideas. But this argument cannot be
put to skeptics, because they would judge it to be circular.” Is there no
escape?
Enter Curley:

Descartes’s defense of reason is much stronger than it is generally given

credit for being. In particular, I shall argue . . . that Descartes is attempting
a rational defense of reason (not something else), and that this defense is
neither obviously nor subtly circular, that on a proper understanding of the
skeptical opposition, there is no reason in principle why Descartes’ defense
should not succeed. I shall then argue that, though the defense of reason
does fail in the end, it fails because Descartes’ arguments for God’s existence

19 “It is of course quite true that we must believe in the existence of God because it
is a doctrine of Holy Scripture, and conversely, that we must believe Holy Scripture
because it comes from God; for since faith is the gift of God, he who gives us grace to
believe other things can also give us grace to believe that he exists. But this argument
cannot be put to unbelievers because they would judge it to be circular.” (CSM II, 3;
AT VII, 2)
118 A Skeptic against Reason

are not good enough, not because the project itself is inescapably circular.
(C, 100)

How does Curley plan and prop his defense? To meet the skeptical
challenges laid down in the First Meditation, Descartes must adopt,
thinks Curley, “a subjective conception of proof.” “Descartes will accept
an argument as a proof if, as he is going through it, it compels his
assent, and if, at the end of the argument, he finds that he has no valid
ground for doubting the conclusion” (C, 115–116). This leads to a
peculiarly anti-Cartesian conclusion: A sequence of propositions may
well constitute a proof of p, and yet p may not be true. This version
renders Descartes no less a proponent of a coherence theory of truth
than does the version of Frankfurt in Demons, Dreamers, and Madmen.
The central function of the subjective conception of proof is less to
establish truth than it is to avoid arbitrary rejection, or establishment,
of claims. Thus, Curley reads anew not the notion of truth in Descartes,
but the notion of proof.20
So the burden of this argument lies in offering an adequate ac-
count of proof. The central notion in this account is the notion of the
assent-compelling proposition. Now, where a proposition fails to compel
our assent, we have no proof. But compelling our assent is not a suf-
ficient condition. Over and above our assent being compelled, we
must have “no valid or reasonable ground for doubting it.” The task is
to characterize the latter notion. Curley offers this as a preliminary
attempt:

(D) Someone has a valid ground for doubting a proposition (say p) if

and only if he can think of (that is, able to state when requested)
some other proposition (say q ) such that
(i) q is incompatible with p;
(ii) he can think of no assent-compelling proposition incom-
patible with q ;
(iii) q explains how he might have erroneously thought
p. (C, 119)

20 A reader in haste to know at least in outline the central thesis of this book, and
how it is arrived at, can skip the rest of this section and the next one for now, and
return to them later, without sacrificing too much by way of understanding what is to
follow.
 III Attempting to Step Out of the Circle 119

The three requirements are justified as follows. Condition (i) is

fairly plausible, for if we are to doubt the truth of one proposition
by another, they must at least be contraries, if not contradictories;
in short, they must be incompatible. Condition (ii) states a minimal
evidential requirement, and the following from Descartes is adduced
in support: “A reason may be valid enough to force us to doubt, and
nevertheless be doubtful itself, and so not to be retained. . . . [Such
reasons] are indeed valid so long as we have no others which induce
certainty by removing doubt.” (C, 120) Condition (iii) ensures that if
a skeptic “does not offer some conjectural explanation of my error,
then his attempt to cast doubt on my belief is not likely to be very
persuasive” (C, 88).
The criterion for what makes a proposition, p, dubious on the
strength of another proposition, q , is that the negation of the latter
proposition should not compel assent. Moreover, it is a virtue of this
view that what is doubtful at one stage of the inquiry need not be
doubtful at a later stage. For example, while the existence of the
physical world is doubted in the First Meditation as well as in all
the others through the Fifth Meditation, it is no longer doubted after
the Sixth. Likewise, the existence of God might be doubted in the
first two meditations, but not after the Third (or at any rate after the
Fifth).
But (D) is only a first approximation. Let us, then, consider the next
approximation.

(D’) Someone has a valid ground for doubting a proposition, p, if

and only if he can think of some other proposition, q , such that

(i) q is incompatible with p;

(ii) (a) if p is not assent-compelling, then he can think of no
assent-compelling proposition incompatible with q ;
(b) if p is assent-compelling, then q is also assent-
compelling,
(iii) q explains how he might have erroneously thought
p. (C, 120)

Curley deems (D’) a bit more satisfactory, particularly because of

the addition ensconced in clause (ii) (b). But first, consider clause
(ii) (a). It is straightforward enough. If I wish to retain p, and p is not
120 A Skeptic against Reason

assent-compelling, and q is incompatible with p, then I must find a

proposition, j , such that j is assent-compelling and j is incompatible
with q .
Now for clause (ii) (b). Curley claims, rightly, that if we have two
propositions, both of which are assent-compelling, it would seem arbi-
trary to opt for one over the other. We should rightly withhold assent, at
best. In the phrasing of Descartes in the First Meditation, “I must with-
hold my assent from these former beliefs just as carefully as I would
from obvious falsehoods” (CSM II, 15). But Curley makes a further
claim. He says, “The effect of this change is to make the evidential re-
quirement imposed on a valid ground of doubt vary with the ‘evidence’
of the proposition to be doubted” (C 120–121).
What change is that? Where a proposition to be doubted is assent-
compelling, any proposition incompatible with it, so long as it is also
assent-compelling, can serve as a ground for doubt. Where a proposi-
tion to be doubted is not assent-compelling, any proposition, no matter
how improbable, incompatible with it can serve as a ground for doubt
(C, 121). That the latter is a sufficient condition is understandable,
but it is not a necessary condition. For consider why Descartes rejects
the truths of mathematics in the First Meditation. He postulates the
existence of an evil genius who tricks him into believing that two plus
three is other than five. Now p, two plus three is five, is a proposition
of mathematics, so it is safe to say that it is assent-compelling. Proposi-
tion q is the hypothesis of an evil genius; being, at most, a contingent
proposition, it could scarcely be assent-compelling. So requirement
(ii) (b) clearly goes a-begging. (Perhaps this is just the difficulty that
Curley thought sprang from his amendment.)
Curley then proposes to modify (D’) by amending not (ii), but
rather the incompatibility requirement, (i). He proposes the following:

(D”) Someone has a valid ground for doubting a proposition, p, if

and only if he can think of some other proposition, q , such that
(i) q is incompatible with p or with some principle, r , which
provides the basis for his assent to p;
(ii) (a) if either p or r is not assent-compelling, then he can
think of no assent-compelling proposition incompatible
with q ;
(b) if both p and r are assent-compelling, then so is q .
 IV No Escaping from the Circle 121

(iii) q explains how he might have erroneously thought

p. (C 122)

The significant clause added in requirement (i) is “or with some

principle, r , which provides the basis for his assent to p.” Not only
would this include the premises, if there are any, but more significantly,
it would include second-order principles used in the subsequent, re-
flective assessment of arguments. (Might not second-order principles
be used in the argument itself – as, for instance, when principles of
inference are used?) Thus, if p is a “first principle,” then r could be
used to derive p from skeptical hypotheses; if p is the conclusion of an
argument, then r could be a principle governing the reliability of that
type of argument. This new definition, (D”), would enable Descartes
to draw a distinction he desired, namely, that an atheistic mathemati-
cian cannot have genuine knowledge of mathematics, which a theis-
tic mathematician can have. Thus, an atheistic mathematician would
eventually have no grounds for accepting the truth of any mathemat-
ical statement, since he can easily learn to doubt some principle, r –
such as, “If a proposition compels assent whenever considered, or fol-
lows by compelling steps from compelling propositions, it is true.” If
his doubt is weakly justified, as demanded in (D”), then he could not
claim genuine mathematical knowledge, in view of (D”) (ii) (a). A the-
istic mathematician, who has grounds for believing both in the truth of
a mathematical statement as well as in the principle, r , can invalidate
this knowledge claim only if he finds q to be at least as compelling
as p and r . But given the proof of the existence of God, he will not
be able to find any such q . This, in some detail, has been Curley’s
argument.

IV. No Escaping from the Circle

Our task now is to examine its adequacy. First, Descartes might have
wondered how Curley could have gotten behind his general rule –
“Whatever I perceive very clearly and distinctly is true” – to something
more basic. Thus, he would have asked, “Is (D”) clear and distinct?”
Or, “Is each part of (D”) clear and distinct?” If it is, then the final
reason for accepting (D”) is his general rule. If it is not, then clearly
(D”) is not good enough. What makes the reason good enough must
eventually be clear and distinct, and not something else. Or, at the very
122 A Skeptic against Reason

least, Curley must offer concepts with which he would replace ‘clear’
and ‘distinct’. He does not do so.
Second, can valid or reasonable grounds for doubting p be defined in
terms of assent-compelling ? If so, and if no alternative seems clear or
forthcoming, then a proposition’s capacity to compel our assent is
not only a necessary condition, but also a sufficient condition for the
concept of proof. Descartes’ objection then might well be that Curley’s
view is too subjective.
This view is at variance with one of Descartes’ important distinc-
tions. In the Third Meditation, Descartes had explicitly drawn a dis-
tinction between impulse and light : “blind impulse,” “natural impulse,”
or “spontaneous impulse,” on the one hand, and the “natural light”
on the other (CSM II, 26–27; AT VII, 38–39). The impulse results
from the prompting of Nature, nothing more. Thus, it is uncertain
whether the proposition to which we feel compelled to assent as a re-
sult of a blind impulse is true. Such can never be the case where we are
compelled to assent as a result of what is revealed to us by the natural
light. Descartes says, “When I say ‘Nature taught me to think this’, all I
mean is that a spontaneous impulse leads me to believe it, not that its
truth has been revealed to me by some natural light” (CSM II, 26–27;
AT VII, 38). Even if (D”) were otherwise correct, it seems to be unfaith-
ful to the text – and to another as well. In the Principles of Philosophy,
Descartes had drawn a distinction between moral certainty and meta-
physical or absolute certainty (CSM I, 289–291; AT VIIIA, 327–329).
Perhaps (D ”) might capture moral certainty, and not just psychological
certainty; but it does not capture, nor can it, the metaphysical certainty
that Descartes so earnestly sought.
Third, I now want to argue that (D”) could not save for Descartes
the principles of inference that he will need for constructing first and
further arguments. The first is the argument of the cogito (if, indeed, it
is an argument); the subsequent arguments concern his essence, the
existence of God, the existence of the physical world, and so on. My
argument proceeds as follows: Suppose L to be a theory of logical infer-
ence that Descartes (or Curley) prefers. But we know that Descartes
did have to contend with other theories of logical inference.21 Let
one of these be labeled L ∗ . At the very least, neither theory of logical

21 See Curley’s brilliant chapter on methods (C, Chapter 2).

 IV No Escaping from the Circle 123

inference is compelling. Or if one is compelling, then so is the other.

One might easily invoke the hypothesis of the evil genius to convince
one of the implausibility of a theory of logical inference.
Now, suppose (D”) were adequate. Let us then apply (D ”) to L ∗ .
The first clause is satisfied: L ∗ is incompatible with L.22 The second
clause is satisfied: Either both are assent-compelling or neither is, and
if the latter, then there is no assent-compelling proposition incompat-
ible with L ∗ . The third clause is satisfied: Curley himself has done a
remarkable job of explaining just that in an earlier chapter. Given the
minimal condition that a proposition must satisfy in order to make it
cast doubt on another proposition, I take it that L ∗ satisfies (D”) and
thus casts doubt on L. At the very least, we must withhold assent from
both L and L ∗ . My argument is sufficiently general. So, what holds
for L, holds for any other theory of logical inference. If no logical
theory can pass muster, there cannot be a proof of the existence of
God. With that, the cogito - argument collapses, too, since it is Curley’s
thesis that the latter argument depends on the former. My argument,
if correct, is important for this reason: We do not have to wait until a
proof of the existence of God is offered, so that we can find a lacuna
in the substance of that proof. Curley’s (D”) assures us, with one of its
consequences, that such a proof cannot be forthcoming.
Fourth, I want to demonstrate that (D”) collapses when it is applied
to itself. Thus it returns to the skeptic the weapons Descartes wants
(and Curley wants) to wrench from him. Imagine a rival theory of
“valid or reasonable grounds for doubting” a proposition – call it D ∗ –
to be:

(D ∗ ) Someone has a valid ground for doubting a proposition (say p)

if and only if he can think of (that is, can state when requested)
some other proposition (say, q ) such that:
q is self-consistent but is inconsistent with p.

One might argue that this is unsatisfactory because the third clause,
whereby it explains why the proposition in question errs, does not
exist. But I think that it need not exist, and that Curley could have

22 I have skipped the clause referring to r , which serves as a basis for the proposi-
tion under consideration; in this discussion, it adds nothing but complexity to the
argument.
124 A Skeptic against Reason

made his demand less stringent, in accordance with the aim of the
enterprise. Let me say a word about that aim.
One can offer a general explanation of the sort that Descartes did
offer in the Principles. There he considered three astronomical theo-
ries, namely, the theories of Ptolemy, Tycho Brahe, and Copernicus
(CSM I, 250–251; AT VIIIA, 85). All of these theories, save one, could
account for the phenomena. The remaining two were mutually incom-
patible. One might draw from this consideration a familiar claim: that
theories can be underdetermined by data. So if a theory has not been
proved – in some strong Cartesian sense, not in the sense of a merely
subjective conception of proof – we can rely on the possibility that
another incompatible theory could just as well account for the data at
hand. This shows, as antirealists say, that while a theory may well be
empirically adequate, that does not mean it is true.
Now, consider (D”). Substitute (D ∗ ) in place of q , and (D”) itself
in place of p. Such a substitution roundly defeats (D”). For consider:
The first clause is satisfied: (D ∗ ) is incompatible with (D ”). The second
clause is satisfied: Neither of them is assent-compelling, and there is
no assent-compelling proposition incompatible with (D ∗ ). I take the
third clause to be satisfied quite easily, and a plausible story can be told,
if a story needs to be told. (I have argued that the third condition is
too stringent and should not be required anyway.) Wouldn’t this have
given Descartes reason to reject (D”)? This, a skeptic might claim, gives
Descartes enough to be skeptical about reason.
Fifth, according to Curley, p is proved if there is no q that satisfies
(D ”). Acceptance of a notion of proof – the subjective conception of
proof – would show that Descartes can, contra Curley’s own recom-
mendation, wholly dispense with the proof of the existence of God,
yet accept the proof of the cogito, so that Curley’s reading of the Third
Meditation becomes untenable. Curley’s view is hung on a dilemma.
Let me explain this in detail.
Suppose (D”) is adequate. Any proposition, p, meets the require-
ment of certain knowledge provided it successfully resists being de-
throned by another proposition, q , that is incompatible with it and
that satisfies (D ”). Now, either the cogito is a proposition of which we
have certain knowledge or it is not. If the cogito is not successfully de-
throned by another proposition that satisfies (D ”), then it is certain
knowledge. But then we do not require the proof of the existence of
 IV No Escaping from the Circle 125

God. However, we saw that Curley insists that – a central theme of his
book – we must have the proof of the existence of God in order to be
assured of the proof of the cogito, so that in the absence of that proof,
the cogito is not such a piece of knowledge. We must buy all or noth-
ing, as Curley said. Presumably, the proposition that satisfies (D”), and
thus dethrones the cogito, is the hypothesis of the evil genius.23 This,
as we saw, was Curley’s reading of the relevant passage (CSM II, 25;
AT VII, 36) in the Third Meditation.
Curley finds Descartes’ proofs for the existence of God defective.
What is of enormous interest is how he finds them defective. He finds
them defective on substantive grounds. Thus, he finds that the principle
of causation, which Descartes presupposes, wanting – or he finds that,
at least since Hume, philosophers have found it wanting. He finds that
the argument unjustifiably presupposes a first cause, if not of the first
member of a series, then of the series of contingent beings. (Curley
does not quite ask this question, though we might: If there is nothing
wrong in there being an infinite regress, what is wrong in there being
an infinite regress in that which created a contingent series?) He finds
defective the talk about formal and objective reality, and with it the
associated talk of degrees of reality. To be sure, much of this is standard
fare among Descartes’ critics.
It is worth quoting Curley in full:

[Descartes] admits that indubitability is compatible with falsity. But he is un-

concerned about this, not because he is not interested in absolute truth, but
because he thinks that rejecting a proposition which is, in his sense, indubitable
is a patently arbitrary act. The skeptic does not want to be, and does not think
he is, patently arbitrary. Descartes’ argument is designed to show that by the
skeptic’s own standards, the skeptic would be acting arbitrarily if he rejected
the argument.
I think Descartes is right. There is no difficulty in principle about Descartes’
procedure in the Meditations. The really serious objection to his argument is
the substantive one that his arguments for the existence of a nondeceiving
God are just not compelling. (C, 118)

This, however, misses the most interesting problem or difficulty.

Even if one could find a proof of the existence of God – say, a version

23 See Appendix A for some connected arguments as they emerge from Jeffrey Tlumak’s
“Certainty and Cartesian Method.”
126 A Skeptic against Reason

of the ontological argument, which Curley respects more than other

proofs for the existence of God (C, 125) – the truth of the substance of
the premises in that argument would not be nearly enough. It would
not be enough because besides true premises, we would also require
sound rules of inference in order to move from premises to the con-
clusion. But given (D”), the proof would be impossible. Why? Because
(D”) could not recognize the soundness of any rules of inference – for
reasons given earlier. Thus (D ”) makes the proof of the existence of
God impossible.
Someone might say in Descartes’ defense: “Not necessarily, surely.
One might be able to come up with a set of rules of inference that
do satisfy (D”) and a proof for the existence of God that is sound
in the light of those rules of inference. Thus far we have just not
found any.” Suppose, then, this were possible (as I think it is not).
If so, those very rules of inference could be used to prove the cogito-
argument to be sound, without the demonstration of the existence of
God. Consequently, we do not have to buy all or nothing.
Finally, and most importantly, has Curley succeeded in showing
Descartes the way out of the Circle? Is there a parallel circular ar-
gument? Consider the following: “Distinguish between a theist logi-
cian and a nontheist logician. Like his counterpart in mathematics, a
nontheist logician can never have genuine knowledge, since, in the
absence of a proof of God’s existence, he can never be sure that
the principles on which his science of logic is founded are true.
Only the theist logician can be so certain. Thus: It is, of course, quite
true that we must believe in the existence of God and the reliability
of human reason because they satisfy the doctrine of (D ”), and con-
versely, that we must believe in the doctrine of (D”) because it comes
from God and is demonstrated by the reliability of human reason.”
This argument cannot be put to a skeptic, because he would judge it
to be circular – rightly, so far.

V. Another Failed Attempt

This section examines a different attempt to show how Descartes
might escape from the Cartesian Circle; it is an attempt by James
van Cleve, presented in his widely discussed paper “Foundationalism,
 V Another Failed Attempt 127

Epistemic Principles, and the Cartesian Circle.” This attempt, I shall

argue, also fails.
Van Cleve argues cogently against Alan Gewirth’s “The Cartesian
Circle” and Fred Feldman’s “Epistemic Appraisal and the Cartesian
Circle,” both attempts to avoid the Cartesian Circle in their own dis-
tinctive ways; Van Cleve then proceeds to offer his own way of exit-
ing from the Circle. Van Cleve distinguishes psychological, practical,
and metaphysical certainty.24 Psychological certainty is merely an irre-
sistible compulsion to believe; practical certainty is the sort of certainty
involved in ordinary knowing of the justified-true-belief variety; and
metaphysical certainty is an objective affair, implying truth. To answer
the question, “How can certainty about God possibly give rise to cer-
tainty about clear and distinct perceptions?,” Gewirth touted psycho-
logical certainty, while Fred Feldman touted practical certainty, such as
the kind of certainty we have about God; and from each of these claims,
Gewirth and Feldman, respectively, attempted to derive metaphysical
certainty about clear and distinct ideas. Van Cleve persuasively demon-
strates that from these watered-down concepts of certainty one could
not extract metaphysical certainty. Thus far, this is an exercise in Hume.
Van Cleve, however, goes on to offer his own alternative way of es-
caping from the Circle, one that I find objectionable as well. Following
Kenny, van Cleve, in reading the general rule – “Whatever I perceive
very clearly and distinctly is true”– draws the following distinction:

(A) For all P , if I clearly and distinctly perceive that P , then I am

certain that P .
(B) I am certain that (for all P , if I clearly and distinctly perceive that
P , then P ). (111)

“The difference,” says Van Cleve, “is that (A) says that whenever I
clearly and distinctly perceive any proposition, I will be certain of it
(the proposition in question), whereas (B) says that I am certain of
a general principle connecting clear and distinct perception with truth.
Clearly, (A) could be true even though (B) were false. (B) requires
that I have the concept of clear and distinct perception, but (A) does
not.” (111) “The point I have been insisting upon could be summed

24 For a very useful discussion of some of these distinctions see Edwin M. Curley,
“Certainty: Psychological, Moral, and Metaphysical.”
128 A Skeptic against Reason

up as follows: (A) is not a principle I have to apply in order to gain

knowledge; I need only fall under it.” (114)
In the cogito - state, I agree that the former doubter cannot know a
truth, even the first truth, by seeing that he has applied any principle,
because at this stage he knows of no principle to apply; he has doubted
them all. Even his doubt itself, as we saw, generates that truth. But
unless van Cleve’s doubter is remarkably un-Cartesian and philosophi-
cally lax, it is unfathomable how he could refrain from doing what
Kenny urged that he should do – and what van Cleve denied that he
need do (113, note 34):

If every other certainty is to be built upon the certainty afforded by clear

and distinct perception, then it is essential, if there is to be any certainty
at all of the type Descartes sought, that one should be able to be certain
that one is clearly and distinctly perceiving something. Moreover, it must be
possible to be certain of this independently of being certain of the truth of what
one perceives. For Descartes offers it as the sovereign methodological principle
for the avoidance of error never to make a judgment about anything that one
does not clearly and distinctly understand. (K, 197)

Kenny distinguishes what we may be uncertain about. (a) We may be

uncertain about whether or not our current mental state is clear and
distinct, and even if we are certain about that, (b) we may be uncertain
about whether or not our current clear and distinct mental state is true.
Consider, then, van Cleve’s three claims: First, there is a claim about
what kind of certainty Descartes sought: “The certainty [Descartes]
sought was certainty in a sense entailing both maximal evidence and
truth” (106). Second, there is a claim about what constitutes a reason
for doubt: “Reasons to doubt need not be certain; they need only be
epistemically possible. . . . I define epistemic possibility as follows: if P
is a proposition that S is considering at t, then P is epistemically possible
for S at t if and only if S is not certain at t of non-P .” (108) Let me
call this claim R ∗ . Third, there is a claim about what I need to know in
order to become certain of a proposition: “I maintain that, in order to
become certain of a proposition, I do not need to know that I am clearly
and distinctly perceiving it, nor that whatever I so perceive is either
certain or true. It is enough that I do clearly and distinctly perceive the
proposition. (A) says that this is enough. . . . It follows that nothing else
is necessary.” (113)
 V Another Failed Attempt 129

So much for the claims. Substitute not-(B) – the denial of (B) –

for reasons in R ∗ . Now, obviously, not-(B) is not clearly and distinctly
perceived at this stage (past the cogito, but before the completion of
the proof of the existence of God); in fact, van Cleve allows (B) to be
false even if (A) is true (by implication, even if P is true). But for the
doubter of the First Meditation, this constitutes a good reason to cast
P in doubt: For if (B) is false, he can invest no epistemic confidence
in any particular proposition, say P , that he may be entertaining at
the moment, because he can and should ask, “How can I justify my
belief in this proposition, P , which may well be misleading me now,
given the epistemic possibility of not-(B)?” Raising and answering such
a question is the only way to acquire maximal evidence for P . The
only thing that prevents him from doubting P is his refraining from
asking that pertinent question. Consequently, he may well fall under
the concept, yet not know – unless he knows the general principle –
that he has done so. (As a parallel: The doubter of the First Meditation
may fall under the concept of walking, but he surely does not know
that he falls under it unless he confronts and satisfies the skeptic’s
scruples.)
How could that doubter falling under a concept convince himself
that the mental state he is presently in is not merely the state of psy-
chological certainty rather than the state of metaphysical certainty?
Van Cleve says, “Moreover, to say (as I do) that we must be certain at
the outset of some clear and distinct perceptions is not to say (as I
don’t) that we must be certain at the outset of the proposition some
clear and distinct perceptions are true. Nor is it to say that we must be
certain of a more specific proposition of the form those clear and dis-
tinct perceptions that are F (e.g. bathed in the light of nature) are true.”
(114) Well, let us grant him some sort of certainty. Evidently, it is not
metaphysical certainty of the sort that Descartes sought. Let us call it
certainty∗ . With this in hand, van Cleve wishes to proceed to the proof
of the existence of God, and thence to the claim that (A) is true (115).
Even if van Cleve is right in thinking that this shows that as a start-
ing point (A) was not arbitrarily espoused (115–117), it does not help
him out of the Circle. If each premise in the proof of the existence
of God is known only with certainty∗ , then one knows that God exists
only with certainty∗ . Using van Cleve’s own form of argument (against
Gewirth and Feldman), one might say that whatever follows thereafter
130 A Skeptic against Reason

cannot have a certainty that goes beyond certainty∗ . We can never

free ourselves from the mire of certainty∗ to arrive at metaphysical
certainty.25

VI. How Not to Read the Meditations: A Skeptic’s Reply

There is one challenge that the foregoing must face, a challenge that
can be constructed from the works of Stephen Gaukroger, principally
from his book Descartes: An Intellectual Biography and his paper “The
Sources of Descartes’s Procedure of Deductive Demonstration in Meta-
physics and Natural Philosophy.”26 Let us begin by raising the question,
“What was the method Descartes adopted in his deductive demonstra-
tions in his Meditations?” Gaukroger answers that there are three signif-
icant alternatives to consider, two of which – the geometrical method
and the humanist-Aristotelian method – in his view present mistaken
ways of reading the Meditations; the third, the method of disputationes –
his own alternative way of reading – he deems satisfactory. But if this way
of reading is correct, it must have a significant impact on the reading
of the cogito, since the cogito is thought to be a demonstrative argument.
How might a skeptic respond to Gaukroger’s historical conjecture?
Consider the first method: the geometrical method. This method,
says Gaukroger, cannot offer the right way to read the Meditations.

25 The discussion on the Cartesian Circle has had a curious history. There are two so-
lutions available, and it is never the case that both views are held in roughly equal
regard. When the fortunes of one view goes up, those of the other decline. And which
solution holds the center of attention among Cartesian scholars depends on other
things, not always of intrinsic value. If the central thesis of this book is correct, then
perhaps the second view will be accorded the pride of place that it richly deserves.
As is well known, Descartes offered the second view in order to escape from the
Cartesian Circle in the Meditations itself. In the Fifth Meditation, he wrote: “Admit-
tedly my nature is such that so long as I perceive something very clearly and distinctly
I cannot but believe it to be true. But my nature is also such that I cannot fix my
mental vision continually on the same thing, so as to keep perceiving it clearly; and
often the memory of a previously made judgement may come back, when I am no
longer attending to the arguments which led me to make it. And so other arguments
can now occur to me which might easily undermine my opinion, if I did not possess
knowledge of God; and I should thus never have true and certain knowledge about
anything, but only shifting and changing opinions.” (CSM II, 48; AT VII, 69)
And it is surely interesting that when Descartes came to reply to his two critics,
Mersenne and Arnauld, the only solution he offered as a way out of the Circle (al-
leged, by his lights) was the second solution (CSM II, 89, 104–105, 150, 171; AT VII,
124–125, 146, 214, 245–246, respectively).
26 Hereinafter GS.
 VI How Not to Read the Meditations 131

Juxtapose Euclid’s Elements, Descartes’ Geometry, and his Meditations

and we find that each differs from the others in significant respects.
Thus, Euclid’s work begins with definitions and axioms and derives
theorems, trivial and not-so-trivial, from them. This is the stock de-
ductive model. But Descartes’ Geometry does no such thing. There the
reader is introduced, in a few pages, to how an arithmetical operation
can be represented geometrically; thereafter he is immediately thrown
into the midst of a serious and difficult problem, namely, Pappus’ locus
problem for four or more lines. Descartes’ concern in his Geometry “is
not deductive demonstration but rather problem solving and the elab-
oration of techniques for problem solving” (GS, 49). This, however,
says Gaukroger, is not an accidental feature of Descartes’ procedure –
he forswears the geometrical method as a matter of principle. Why?
Because Descartes had accused the ancients of using this method with
guile: The ancients were insinuating that they had been led to their
discoveries in the way in which they presented their results, and such
false advertising enabled them not only to gain undeserved recogni-
tion, but also to conceal their true method. As early as the Regulae,27
Descartes had admonished thus:

I have come to think that these writers themselves, with a kind of pernicious
cunning, later suppressed this mathematics as, notoriously, many inventors
are known to have done where their own discoveries were concerned. They
may have feared that their method, just because it was so easy and simple,
would be depreciated if it were divulged; so to gain our admiration, they may
have shown us, as the fruits of their method, some barren truths proved by
clever arguments, instead of teaching us the method itself, which might have
dispelled our admiration. (CSM I, 19; ATX, 376–377)

Descartes’ reasoning is not entirely perspicuous, if he is about what

Gaukroger claims he is about. First, what is “this mathematics” that
the ancients later suppressed? Is it just their method? Second, we can
distinguish between a method of discovery and the results to which it
leads. If all the results were barren, few would have been moved by the
ancient mathematicians; at the very least, this was not true of Descartes.
Consequently, some results must have been fecund and were no less
the fruits of the method of yore. Third, the ancient mathematicians

27 Various parts of this work were composed at different times.

132 A Skeptic against Reason

wanted to pass themselves off as clever for the results they were able
to discover; so they concealed their easy and simple method for fear
of being thought no more ingenious than they in fact were. Conse-
quently, there was a useful method – albeit an easy and simple one.
Fourth, there was a powerful desire on Descartes’ part in the Regulae to
make evident a method of proceeding in mathematics and science that
would speed up discovery. Descartes is implying that had the ancient
mathematicians revealed their methods, we, their successors, might
have been led to make similar discoveries after them (otherwise, the
analogy with the many inventors makes little sense). They thus failed
us, and for this they deserve to be chastised. Fifth and finally, the
Meditations seems to have far less in common with Descartes’ Geome-
try than with the Elements. Here the reader is not introduced, in a few
pages, to some philosophical preliminaries and thereafter immediately
thrown into the midst of a serious and difficult philosophical problem,
namely, how can we be sure the external world exists? It would clearly
be false to say of the Meditations that in it Descartes was not interested in
deductive demonstration but rather in philosophical problem solving
and the elaboration of techniques for philosophical problem solving.
Descartes approvingly cites in the Meditations the works of Archimedes,
Apollonius, and Pappus, and explicitly makes several comparisons be-
tween their geometries and his metaphysics; above all, he speaks re-
peatedly of “proofs,” “demonstrative proofs,” and “exact demonstra-
tions” and leaves no doubt that he is doing in his Meditations what
good mathematicians had done in their own works28 (CSM II, 4–6;
AT VII, 4–6). The lesson to be learned, then, is that Descartes was
not so much opposed to the geometrical method as he was, rightly, to
moving by guile. His method, he thought, would lead to interesting
results in metaphysics no less than in mathematics; but he would not
deny to the public, or at least to the able, access to that method.29
Consider the next method of discovery: the humanist-Aristotelian
method. Peter of Spain and Lambert of Auxerre had defined dialectic,

28 In a letter of July 31, 1640, to Huygens, Descartes wrote: “For I draw a comparison
between my work in this area and the demonstrations of Appollonius. Everything in
the latter is really very clear and certain.” (CSMK, 150; AT III 751) Since it would
take too much space to cite long passages from the “Second Set of Replies,” I ask the
reader to consult CSM II, 110–111; AT VII, 155–157.
29 These claims are made not only in Regulae, but also in the Discourse on the Method.
 VI How Not to Read the Meditations 133

following Aristotle, as “the art of arts, the science of sciences, possess-

ing the path to the principles of all methods” (GS, 50). This need not
be a method of discovery, but in the scholastic tradition it was widely
recognized as such. “The details,” writes Gaukroger, “of how the pro-
cedure was to be effected were generally left vague, however, and if
one takes this ‘dialectic’ to be something conducted in a systematic
way, starting from first principles, then it is hard to see what it could
be other than syllogistic” (GS, 50). But Descartes, in clear anticipation
of John Stuart Mill, had roundly rejected the syllogism as a means of
discovery. In rule 10 of Regulae, he forewarns against logicians’ using
the syllogism. Such logicians “are unable to devise by their rules any
syllogism with a true conclusion unless they already have the whole
syllogism, i.e., unless they have already ascertained in advance the
very truth deduced in that syllogism.” Plainly, it follows that neither
the discovery nor the justification of the cogito could be a syllogistic
argument – and Descartes knew no other.
Ramond Lull (1233–1315) proposed an alternative method, “the
art of finding the truth” (ars inveniendi veritatem). Lull’s aim was to de-
vise a universal language from the axioms of which new truths could
be generated, especially ones pertaining to the doctrines of the Trin-
ity and the Incarnation. Henricus Cornelius Agrippa’s De incertitudine
et vanitate de scientarium et artum (1527) shed Lull’s religious goals in
the devising and use of his method, and exhibited two key, almost
Cartesian, features. First, that method is a general and universal sci-
ence: It advocated starting from absolutely certain principles, armed
with an equally certain criterion of knowledge; and second, it offered a
way of ordering all knowledge. Gaukroger then argues that Descartes
had utterly rejected the Lullian approach by the late 1620s, “and there
is no question of it playing a role in, for example, the Meditations”
(GS, 51). Gaukroger further adduces Descartes’ letter to Father Marin
Mersenne (November 20, 1629) as evidence. In that letter, Descartes
had explicitly rejected the possibility of discovering anything new vis-
à-vis a constructed universal language. Such a language could serve
only as a vehicle for expressing truths already discovered, not one in
or through which new truths could be discovered. However, Agrippa’s
1527 treatise with its two features can be easily understood without
any reference to a universal language; indeed, it could be understood
even if one admitted the impossibility of a universal language. That
134 A Skeptic against Reason

being the case,to what extent should Agrippa’s method be regarded as

a way of discovering new truths, and what Cartesian arguments might
be offered against such a watered-down enterprise? But let me pro-
ceed to the third method, the method of disputationes: Gaukroger’s
own preferred alternative.
In 1513, the Lateran Council, held under Pope Leo X, condemned
any reading of Aristotle that led to the conclusion that there was no
personal immortality, and admonished philosophers and theologians
alike to rise in defense of the contrary claim. As a result, a new genre
of commentaries arose, commentaries by Suarez, Fonesca, Toletus,
and the Coimbra commentators. Unlike the medieval commentaries
on Aristotle that attempted to use Aristotle to provide Christianity
with a philosophical foundation, as in the works of Saint Thomas
Aquinas, these commentators simply reworked, revised, and rewrote
Aristotle.30 There was an extant body of Aristotle’s theory that was to
be reconstituted and there was an explicit religious agenda: Any re-
constitution that did not yield the required theological conclusion was
to be rejected; those that yielded the desired conclusion were to be re-
tained. Descartes was schooled in metaphysics by these commentaries.
Gaukroger adverts to three reservations about how this model was to be
applied to the Meditations (GS, 57–59). The reservations were that the
Meditations did not read like a textbook, nor did it belong to the genre
of devotional manuals; that there was no independently existing meta-
physical material that the Meditations set out to reconstitute; and finally,
that Descartes himself thought he was doing analysis in the Meditations.
But in the end, Gaukroger sets these reservations aside and concludes
that the disputationes “explicitly enable one to provide a defense of
doctrines whose truth one is already certain of, by showing how such
doctrines can be deduced from shared premises. And they do this in
a way that makes no reference at all to mathematical demonstration,
but which, on the contrary, is designed specifically for metaphysical,
theological, and natural-philosophical arguments.” (GS, 59)
I find this view deeply mistaken. My primary concern is to show that
if Descartes did what he is alleged to have done, then his philosophical

30 See Charles H. Lohr, “Jesuit Aristotelianism and Sixteenth Century Metaphysics,”

“Metaphysics,” and “The Sixteenth-Century Transformation of the Aristotelian Divi-
sion of the Speculative Sciences.” See also G, 59–61.
 VI How Not to Read the Meditations 135

enterprise is quite suspect – he gives the skeptics enough to cavil about.

If I am right, perhaps historians of ideas more astute than I will find
good reasons why they should confront Gaukroger’s historical con-
jecture head on. There are three reasons for my misgivings. First,
a skeptic or an atheist might claim that if the Meditations are noth-
ing more than an expression of what follows from shared premises –
premises shared by the believers, that is – then Descartes’ pleading
to the faculty of Sorbonne that with his arguments in the Meditations
he would convince the most hardened disbeliever was something of a
sham. Descartes had offered to convince the skeptics and the atheists
(in obedience to the dictates of the Lateran Council) by demonstrative
proofs based on natural reason; only then could he demonstrate, as
he himself foresaw, that his arguments were noncircular.31
There were several interpretations of Aristotle in circulation:
“Metaphysics had become an extremely contentious subject by the six-
teenth century, not so much because of the proliferation of Platonist
and other rivals to Aristotelian philosophy, but because of the prolifera-
tion of different and unorthodox interpretations of Aristotle.” (GS, 55)
Some interpretations were scarcely Christian in orientation, so that, at
a minimum, Descartes would have had on his hands the epistemologi-
cal problem of how to determine the truth of an Aristotelian interpre-
tation of a Christian variety from a non-Christian variety. Moreover, if
Descartes was justified in his approach, should the atheists have gone
about their business constructing an interpretation of Aristotle based
on their shared premises? How would the author of the First Meditation
have reacted to such an interpretation?
Second, in mathematics Descartes had drawn a sharp distinction be-
tween the analytic method and the synthetic method, or the method
of exposition. Once Descartes had solved a mathematical problem on
a satisfactory basis of analysis, his synthetic demonstration did not par-
ticularly worry him; he did this, claims Gaukroger, in his Geometry all
the time, where he mostly offered an analytic solution to a mathe-
matical problem, and only on occasion the synthetic proof (GS, 54).
Perhaps, then, in metaphysics Descartes had in mind a parallel distinc-
tion: Analytic problem-solving techniques were to be used in solving
a metaphysical problem – for example, does God exist? – and once

31 See the opening page of the Meditations, “Dedicatory Letter to the Sorbonne.”
136 A Skeptic against Reason

he had that solution, he could then offer a synthetic demonstration.

But he did not do that, for he thought ill of the synthetic or syllogistic
method. Why then did he even occasionally use it in mathematics?
Did he even occasionally use the method of disputationes in mathemat-
ics? Did Descartes have any such problem-solving techniques in meta-
physics? If he did, and did not share them with the public, he would
have been no less guilty of low cunning than the ancient mathe-
maticians he chided. If he did not have them, and had no genuine
analytical solutions to metaphysical problems, his guilt would have
been immeasurable.
But what constituted a genuine analytical solution? Imagine: In writ-
ing his Geometry, Descartes was not concerned about the order in which
the theorems were presented. He took whatever important mathemat-
ical problems struck his fancy and solved them analytically. Each the-
orem was presented in the order in which he solved it, there being
no particular order in which he solved them. Then, imagine that he
wrote another book called The Philosophical Foundations of Geometry. In
this second book, Descartes took the theorems of the first book and an-
alyzed each of them in order to uncover the basic axioms or principles
on which it rested. Next, he collected all of these axioms or principles
and attempted to simplify the list. The lot being simplified, he placed
them at the start of the book. Finally, he rearranged the entire order
of the theorems as they appeared in the first book in such a way that
no theorem in the second book came before any theorem it needed
for its own demonstration. Question: How was the justification of the
theorems in the first book in essence different from the theorems in
the second book? How could Descartes have claimed to have made a
discovery in the absence of a proof ?
Third, in light of Gaukroger’s thesis, it might be a very useful exer-
cise to conjecture how Descartes might have persuaded himself about
the genuineness of the solutions to the philosophical problems he
was engaged upon, and then to compare that conjecture with what
we actually find in his Meditations. It should not be at all surprising,
I submit, to find that the conjectures are very nearly identical. This
would be especially true of Descartes’ cogito. Descartes may well have
had an inkling of the truth of the cogito – indeed, he had more than
an inkling, given its role in the Discourse. But the more he thought
about the problem and the provisional solution, the better his sense
 VI How Not to Read the Meditations 137

of the cogito became, until it was finally embedded in the Second

Meditation.
I conclude with a cautionary note. The central claim of this chap-
ter is embedded in section II, wherein the argument for the new
Cartesian Circle is presented; some of the other sections treat of the
old Cartesian Circle and how to escape from it. The argument for the
new Cartesian Circle shows how Descartes’ attempt to prove the ex-
istence of God is foiled: In order to prove that God exists, Descartes
needs to justify the rules of logic; but in order to establish the rules of
logic, Descartes needs to establish that God exists. Descartes’ system is
riddled with this circular argument. The importance of the argument,
for the main thesis of this book, lies less in showing why Descartes’
proof for the existence of God needs to be amended than in show-
ing that before the Third Meditation, the rules of logic have no proper
Cartesian anchor; they can be open to serious doubt. Caution: The
conclusion of this argument is limited. It is not intended to throw ev-
erything reason touches into disrepute; it is only intended to show that
whatever else may be reliable as a deliverance of Natural Reason – say,
Descartes’ intuition – no reliance on the rules of logic has been, or can
be, justified, even by the end of the Second Meditation. Naturally, our
next task is to show what impact, if any, this has on Descartes’ cogito.
 5

The Five Ways

Often a system of philosophy rests on a central claim. The rest of the

system is so tightly interwoven with that central claim that changing it
changes the system, challenging it challenges the system. The system is
like a spider’s web: Touch the spider, and the whole web reverberates.
The central claim of the logical positivists of early twentieth-century
Vienna, and of their immediate successors, was the principle of verifi-
cation: A sentence of a language was meaningful if and only if it was
either analytically true or empirically verifiable. On the basis of such
a central claim, a new philosophy was defined; metaphysics was elimi-
nated; a new critique of ethics and theology was propounded; an alter-
native account of personal identity was suggested; and a fresh view of
the nature of logical and mathematical propositions was touted.1 With
the alteration of the principle of verification, the system was altered.
When a final successful assault was launched on said principle, the
system as a whole tottered. As a consequence, metaphysics began to
flourish again.
In Descartes’ philosophy, the cogito is the central claim. Conse-
quently, it is quite important to be as clear as possible about the precise
nature of this claim. Thus: Is the cogito an argument? Why is this prob-
lem so important? Why is it necessary to settle the issue of whether
or not the cogito is an argument? After all, one might argue, we have
captured a very interesting truth: What does it matter whether that

1 Still worth reading is A. J. Ayer’s Language, Truth and Logic, first published in 1936.

138
 I The Five Ways . . . 139

truth is encased in an argument (a sequence of appropriately related

propositions) or in a single proposition? The reason is that the answer
to the question, “Is the cogito an argument?”, affects our understanding
of other issues in Descartes, including Descartes’ views on syllogism,
discovery, and explanation; his views on the nature and purpose of
questioning the truths of logic and mathematics; and his views on the
role of memory and will.
In section I, I set up five distinct ways of reading the cogito as an
argument. To my knowledge, the rest of the ways of reading the cogito
as an argument are just variations upon these five themes. Then, in the
next section, I discuss another way to read the cogito, a famous view
that reads the cogito as a performance. I have two aims in presenting
these claims. First, each way illuminates a certain part of this dark and
complex problem, even if it eventually fails as a whole. Second, and
more importantly, despite their overt differences – different failures,
different insights – each way of reading the cogito as an argument bears
a simple and striking resemblance to the others. This core similarity
is defective. The defect is isolated in this chapter, and is shown to be
irreparable in the next.

I. The Five Ways . . .

Reconsider the famous passage in the Second Meditation:

But there is a deceiver of supreme power and cunning who is deliberately and
constantly deceiving me. In that case I too undoubtedly exist, if he is deceiving
me; and let him deceive me as much as he can, he will never bring it about
that I am nothing so long as I think that I am something. So after considering
everything very thoroughly, I must finally conclude that this proposition, I am,
I exist, is necessarily true whenever it is put forward by me or conceived in my
mind. (CSM II, 17; AT VII, 25)

Conspicuous by its absence is the word ‘therefore’. Hence, one might

assume that Descartes was not offering an argument. But there are
at least two considerations that militate against such an assumption.
First, there is an equally famous passage in Discourse on the Method that
serves as a counterpoint:

But immediately I noticed that while I was trying thus to think everything false,
it was necessary that I, who was thinking this, was something. And observing
140 The Five Ways

that this truth, “I am thinking, therefore I exist” was so firm and sure that all the
most extravagant suppositions of the sceptics were incapable of shaking it,
I decided that I could accept it without scruple as the first principle of the
philosophy I was seeking. (CSM I, 127; AT VI, 32)

Also, consider the following passage in The Search for Truth, which,
although posthumously published, was written (probably) in summer
of 1641, when Descartes was intensely occupied with the issues raised
and discussed in the Meditations on First Philosophy. Here it is, in the
words of Eudoxus, the mouthpiece of Descartes:

You cannot deny that you have such doubts; rather it is certain that you have
them, so certain in fact that you cannot doubt your doubting. Therefore it is
also true that you who are doubting exist; this is so true that you can no longer
have any doubts about it. (CSM II, 409–410; AT X, 515)

A little later, here is Eudoxus again:

I quite share your view, Epistemon, that we must know what doubt is, what
thought is, what existence is, before being convinced of the truth of this infer-
ence, “I am doubting, therefore I exist,” or what amounts to the same thing,
“I am thinking, therefore I exist.” (CSM II, 417; AT X, 523)

The many ‘therefore’s, and at least one use of ‘inference’, lend support
to the claim that Descartes’ cogito passages had better be construed as
an argument.
There is a second reason. Descartes had explicitly regarded the cogito
passages as an enthymematic argument, a syllogism with a missing
general premise, such as, “Whatever thinks, exists.”2 Without it, the
argument is invalid. With the general premise supplied, the argument
is valid and reads as follows:

[1] Whatever thinks, exists

[2] I think

[3] ∴ I exist

Once such an argument has been set up, philosophers have worried
over the following sorts of problems: What is the relationship between

2 Descartes, Conversation With Burman, 90–91. For details see the following discussion.
 I The Five Ways . . . 141

the general proposition and the particular? Which one comes first?
Does, as Gottfried Wilhelm Leibniz thought, the general come first,
because without it we could not know the particular? Or, as John
Locke thought, does the particular come first, from which the gen-
eral is adduced? But if Leibniz is right, then Descartes has not said
anything novel in the conclusion that he has not already said in the
premises (WM, 57–58). On the other hand, if Locke is right, the ar-
gument is merely inductive, and it is a probable syllogism of the sort
Descartes had expressly condemned (CSM I, 11; AT X, 363–364). An-
other problem that has been raised is: Is existence a predicate? An
elegant problem first introduced by Kenny is whether Descartes’ cogito
argument is compatible with his ontological argument (K, Chapters 4
and 7).
Thus we have been led to suppose that the cogito is an argument.
This approach fails, I think, because it fails to pay attention to what
are seemingly harmless elements in most of these passages; it fails to
take into account such things as the role, if any, of memories at this
stage in the argument in Meditations on First Philosophy; and above all,
it fails to take into account what was said in the First Meditation. It
can be demonstrated that whatever Descartes may have said about the
cogito, expressly or implicitly, he is philosophically preempted, unless
the First Meditation is a sham, from regarding it as an argument.
There is a passage in Willard Van Orman Quine’s Methods of Logic
that nicely summarizes the history of the attempts to capture the cogito
in the net of an argument.

When we want to analyze and appraise a logical inference which someone has
propounded, we have to take such suppressed premises into account. At this
point two problems demand solution simultaneously: the problem of filling in
details of a logical deduction leading from premises to desired conclusion, and
the problem of eking out the premises so that such a deduction can be con-
structed. Solution of either problem presupposes solution of the other; we can-
not set up the deduction without adequate premises, and we cannot know what
added premises will be needed until we know how the deduction is to run.3

There are, in all, five ways in which the cogito has been considered as an
argument, with varying premises and with different runs of deduction.

3 W. V. O. Quine, Methods of Logic, 186.

142 The Five Ways

First, it has been regarded as a syllogistic argument. The cogito is

put in the form of an argument with which Descartes would have been
familiar. Second, the cogito is considered to be an enthymematic syllo-
gistic argument. “Whatever thinks, exists,” or some variant of it, serves
implicitly as the major premise in the latter argument, explicitly in
the former. These two are, historically speaking, the most common
representation of Descartes’ view. Third, the cogito has been consid-
ered to be an enthymematic nonsyllogistic argument, invoking the
aid of first-order logic or quantification theory. So considered, the
cogito consists (usually) of two or more premises, and the missing
premises are filled out in various ways. Fourth, the cogito is consid-
ered to be a one-premise argument, with “Whatever thinks, exists,”
serving not as a premise but rather as a rule of inference. Fifth and
finally, the cogito is construed as a particular kind of argument that
invokes the notion of presupposition. I shall argue that while each
one of these attempts to read the cogito as an argument has pecu-
liar problems of its own, each is foiled by a singular, irreparable
difficulty.

A. The First and Second Ways

I shall consider the first two views together, since they both purport
to regard the cogito as a syllogism, the uninteresting difference being
that one is fully elaborated, the other is enthymematic.4 The miss-
ing implicit major premise of the latter syllogism, “Whatever thinks,
exists,” or some variant of it, occurs explicitly in the former. Now,
suppose that Frans Burman’s (1628–1679) report of a conversation

4 Gordon Baker was kind enough to suggest, in conversation, that the central thesis
of this book, namely, that the cogito is not an argument, is not implausible. But so
deep-rooted is the tradition of regarding it is an argument that when one consults
the book Descartes’ Dualism, one finds that the authors, Baker and Morris, repeatedly
speak of the cogito as if it were an argument. They do not recognize any particular
argument pattern that the cogito is alleged to be, but they speak of premises, argument,
entailment, inference, and validity in connection with the cogito. For example, they
speak of “admissible premises in the argument schema of the Cogito,” (34); “Thus, pace
Hobbes, ‘Cogito, ergo sum res cogitans’ is a valid inference,” (65); “these reports play a
parallel role as the premise for the Cogito” (72); and other examples can be found on
76 note 43, 115, and 117. In fairness, their book is on Descartes’ famous two-substance
doctrine, and it sets out to demolish a different legend about Descartes’ philosophy;
it is not principally about the cogito.
 I The Five Ways . . . 143

with Descartes is accurate:

Before this conclusion, I think, therefore, I am, one can know the major, whatever
thinks is, because it is really prior to my conclusion, and my conclusion rests
on it. And so in the Principles the author says that it is prior, because it is always
implicitly presupposed and prior. But I do not on that account always know
expressly and explicitly that it is prior, and I do know my conclusion first,
because I attend only to what I experience in myself.5

This seems unproblematic enough: The major premise is always in the

background, even if the conclusion is reached first. In the final analysis,
we have a syllogistic argument. There are at least three problems with
this. First, if we are searching for the first principle, the first truth,
then the cogito is arrived at first, the major premise is reckoned next.
Note that the cogito is regarded as a truth, a first principle, a first item
of knowledge, a proposition, an inference, a piece of knowledge, a
consequent, and a conclusion, but it is never (?) called an argument.
Indeed, Descartes is explicit that the major premise is dependent on
establishing the conclusion:
When someone says “I am thinking, therefore I am, or I exist,” he does not
deduce existence from thought by means of a syllogism, but recognizes it as
something self-evident by a simple intuition of the mind. This is clear from
the fact that if he were deducing it by means of a syllogism, he would have to
have had previous knowledge of the major premiss “Everything which thinks
is, or exists”; yet in fact he learns it from experiencing in his own case that it is
impossible that he should think without existing. 6 It is the nature of our mind

5 Descartes, Conversation With Burman, 90–91. Anyone – such as Kenny, for example –
who claims that cogito ergo sum is an argument, and that this is shown by the appearance
of “ergo,” might consider the following: First, few seem to have noticed that even in
this passage Descartes is referring to I think, therefore, I am as the conclusion! And a
conclusion is a single proposition. Second, the most famous cogito passage from the
Meditations on First Philosophy does not contain a conclusion indicator. Kenny acknowl-
edges this, but he claims that the “cogito” is also missing. On the contrary, I take it to
be not only present in that passage, but elaborately present in the following guise:
“whenever it is put forward by me or conceived in my mind.”
6 To anyone who, like Kenny, sees this as a clear instance of Descartes’ endorsing the
view that the cogito is a one-premise argument, I reply: First, Descartes does not say –
not even implicitly – that it is an argument. He simply remarks on what an individual
might learn from his own case, leaving it entirely open whether what he learns, what
convinces this individual, is based on a deductive argument or on an intuition of a
single proposition. Second, Descartes is insistent that the cogito cannot be a syllogism.
From this assertion, it will not do to infer that it is a nonsyllogistic argument. Third,
Descartes avers that we can know general truths only from particular ones. Fourth, if
the cogito were an argument, it would leave the general principles of inference entirely
144 The Five Ways

to construct general propositions on the basis of our knowledge of particular

ones. (CSM II, 100; AT VII, 140–141)

This, as Bernard Williams rightly points out, is no mere psychological

observation; it is surely intended as something more. But here we dis-
agree on what that more is. Williams goes on to say that “if we suppose
that Descartes believed for one instant that, as a matter of logic, the
proposition that everything that thinks exists can be established by in-
duction from one’s own case – and that he used this as an argument to
deny that the cogito is a syllogism – we should be obliged to say that he
was mistaken.” (WC, 91).
Why would we be so obliged? How else is the general proposition,
“Whatever thinks, exists,” to be known? Descartes himself often took
the inductive approach, not only in the empirical sciences but also in
mathematics: “[S]ay,” says Descartes,

I wish to show by enumeration that the area of a circle is greater than the
area of any other geometrical figure whose perimeter is the same length as
the circle’s. I need not review every geometrical figure. If I can demonstrate
that this fact holds for some particular figures, I shall be entitled to conclude
by induction that the same holds true in all the other cases as well. (CSM I,
27; AT X, 390)

I suggest, then, that it is not at all implausible, but rather quite likely,
that Descartes would infer the truth of “Whatever thinks, exists” by re-
lying on the truth of a particular statement, such as “I think, therefore,
I exist,” even if we regard “Whatever thinks, exists” to be on all fours
with a mathematical proposition or an eternal truth.
The truth of a proposition is known either by sense experience or
by reason. Sense experience, at this stage, cannot be relied upon; this

undefended. Fifth and finally, Descartes claims, in this very passage, that the cogito is
an intuition.
To Father Mersenne, Descartes wrote on October 5, 1637, complaining that Fermat
had misunderstood him, “He thought that when I said that something was easy to
believe, I meant that it was no more than probable; but in this he has altogether
mistaken my meaning. . . . [W]hen I say that something is easy to believe I do not
mean that it is only probable, but that it is so clear and so evident that there is no need
for me to stop to prove it.” (CSMK III, 74; AT I, 450–451; my italics) How might one
explain this letter on the view that the cogito is a one-premise argument? Might one
say that it only goes to show that the cogito is not “so clear and so evident” after all, so
that it is in need of proof ? Or might one simply deny that an immediate inference is
a proof ?
 I The Five Ways . . . 145

leaves us with reason as the only source of knowledge. If reason were

to validate that general proposition, prior to the cogito, then of course
the cogito would hardly be the first truth that Descartes was looking for.
At the very least, there could be other truths, like the general truth,
any one of which could serve as the first truth. One way in which the
defense has then proceeded is to argue that Descartes was not looking
for any truth, but rather for an existential truth. Descartes is cited:

And when I said that the proposition I am thinking, therefore I exist is the first and
most certain of all to occur to anyone who philosophizes in an orderly way, I
did not in saying that deny that one must first know what thought, existence
and certainty are, and that it is impossible that that which thinks should not
exist, and so forth. But because these are very simple notions, and ones which
on their own provide us with no knowledge of anything that exists, I did not
think they needed to be listed. (CSM I, 196; AT VIIIA, 8)

But Descartes is simply mistaken; either that, or the First Meditation

is remarkably deceptive. When Descartes doubts the truths of sense ex-
perience or the truths of mathematics (which he does both in Principles
of Philosophy and in Discourse on the Method), he should not invoke the
evil genius, or something like it, since mathematical truths, as well as
the principles of logic and the rules of inference, do not typically yield
any existential truth. If Descartes is searching for an existential truth,
why should he doubt the truths that purport to be harmless, nonex-
istential truths? What is the point in denying these? Contrariwise, if
he is searching for any certain and necessary truth, then why should
the cogito, rather than a host of logical and mathematical truths, be
regarded the first truth ?
“I think,” says Margaret Wilson,

it is quite possible Descartes simply could not make up his mind. On the one
hand he wanted to say that we can know the (necessary) truth of “If I think,
I exist,” without having heard or thought of any abstract principle such as
“Whatever thinks exists.” This does seem plausible, and it suggests that (as
Locke might say) we know the instance without knowing the principle. On
the other hand, he wanted to say that the (necessary) truth of “If I think,
I exist,” is not strictly independent of the truth of “Whatever thinks exists” or
the validity of “if . . . a, exists a.” This seems plausible also, and tends to suggest
(as Leibniz might hold) that we must in some sense, explicitly or implicitly,
know the first principle. (WM, 57)
146 The Five Ways

Wilson correctly observes that the universal principle is presupposed

in the Principles of Philosophy, occurs in the Discourse on the Method post
cogito, and is not mentioned at all in the Meditations on First Philosophy.
Could Descartes hold the Leibnizian view? Under the circumstances
at the end of the First Meditation, I think not. The universal principle
is either implicit or explicit. If it is implicit, Descartes cannot rely on
it: For there is no reason at all to believe that that universal principle
can be relied upon and is not subject to the dictates of the evil one.
If it is explicit, then Descartes requires that there be a reason, too, to
hold onto it. But this reason can hardly be forthcoming where even
the truths of logic and mathematics are not held sacrosanct. It is suffi-
cient to cite Descartes in the Second Meditation: “I do not now admit
anything which is not necessarily true.” This makes Wilson’s following
remark puzzling: “One can recognize that the inference is sound prior
to achieving express recognition of any universal principle” (WM, 57).
But how? Shouldn’t the ideal seeker say, “This inference seems sound.
But it relies on a universal principle which I do not, yet, know to be
true. How, then, can I hold the inference to be sound? For I do not
want a mere sound argument, or for that matter just a valid argument.
I want an argument that I know to be sound.” The question, therefore,
of Descartes’ admitting, relying, resting on, presupposing, taking for
granted, or the like, propositions that are implicitly or explicitly held,
but without justification, simply does not arise. There is, given the
epistemic enterprise, no alternative for Descartes but to opt for the
Lockean view.
Imagine: There is a thinker, D , who undergoes methodical doubt;
he not only challenges all his explicit and consciously held belief, but
also asserts that he should not rely on anything that invokes implicit
principles; he abandons everything. At this stage, he also abandons,
as a natural outcome of his doubt, the belief that there is any other
thinker save himself. Consequently, he prevents himself from mak-
ing any generalization concerning thought, doubt, or existence about
any other thinker besides himself. He commences his procedure of
doubt, introducing the evil genius to provide him with a rationale for
doubting things that he could not otherwise doubt. As he engages in
this methodical doubt, he discovers the cogito: Namely, no matter how
much the evil one deceives him at this moment, t0 , he cannot be de-
ceived into thinking that he, D , does not exist, since if he is deceived,
 I The Five Ways . . . 147

then assuredly he exists. End of t0 . This D regards as a necessary truth

about himself. In the next moment of time, he reflects that this is surely
true of himself as a general truth, namely, that whenever he, D , thinks,
he exists. End of t1 . In the next moment, he recognizes (but how?) that
even though he does not know that any other thinker besides himself
exists, it is true that if there were another thinker, he too would in-
evitably exist whenever he thought. End of t2 .
Suppose, now, that D had died at the end of t0 . Would Descartes have
said that there was something missing in the light of which D could
not be said to know the cogito as the first principle, the first truth? What
did D miss? One might say, “Well, he missed the universal principle.”
Suppose that D dies at the end of t1 . He now has a limited general
principle, a general principle true only of him. Would Descartes have
argued that that was insufficient, too? In what way is it inadequate?
One might say once again, “Well, he missed the unrestricted univer-
sal principle.” Finally, suppose that D dies at the end of t2 . Now, at
last, as some of the commentators would have it, Descartes would be
satisfied.
At the very least, what this shows is that the limited generalization
is based on a particular instance, and that the full generalization is
based on the limited generalization. It is difficult to see how, if at all,
Descartes could have proceeded correctly to the full generalization
without the particular instance or the limited generalization. Leibniz
would, of course, counter that at t0 Descartes had implicit knowledge of
the unrestricted general principle, “Whatever thinks, exists,” without
which he could not be said to have knowledge of the cogito. But implicit
knowledge, as we have just seen, is simply not sufficient for Descartes
to have the kind of knowledge that he is seeking at the end of the First
Meditation. If one contemplates what condition has to prevail before
Descartes can explicitly know, have justified belief in, the unrestricted
universal generalization, it turns out that he must first know the truth
of the particular. This imaginary case demonstrates that Descartes has
full and correct knowledge of the cogito without the general principle.
To demand anything more is to put weight on the full general claim
that it is quite unable to bear.
The aim of this example is to show that if in this singular case the
universal principle is not invoked, and yet a thinker has correct knowl-
edge of the cogito, then the universal principle is inessential. Second,
148 The Five Ways

if the thinker does not have knowledge of the first principle, what
in particular is his failure? Third, could he have arrived at the truth
of the universal principle without first recognizing the absolute truth
of its first instance? One might say that the particular instance cannot
demonstrate its necessity in a way in which the universal can. But what
makes for the necessity of the universal proposition that leaves the
necessity of the solitary instance in doubt? There is no such additional
element, and the claim that the universal alone would confer necessity
upon the singular instance needs much argument and illustration.
There is a second problem facing this view: Descartes wanted his
premises, in a deduction, to be not only true, but also certain and neces-
sary. His theory of deduction, which I shall consider in detail later, says
so explicitly, and it was strongly implied in his method of analysis. The
axiom of metaphysics cannot be an ordinary, contingent truth. What,
then, is the status of the minor premise, “I think,” in the syllogism?
Even if it is conceded that it is “transparently clear,” it would be false to
accord it the status of a necessary truth rather than that of a contingent
truth. This flies in the face of Descartes’ theory of deduction.
Third, if the cogito is a syllogistic argument, then Descartes unjustly
relies on the principles of inference, such as the rule of universal
instantiation and modus ponens, which have not been – and at this stage
cannot be – shown to be certain or even reliable. Thus, the inference
should be held suspect.

B. The Third Way

In “Cogito Ergo Sum: Inference or Performance?,”7 Jaakko Hintikka has
argued (without espousing the view) that the cogito can be formulated
as an argument in quantification theory or first-order logic, thus:

[1] Ba ⊃ (∃x) (x = a)
[2] Ba

[3] ∴ (∃x)(x = a)8

7 Hereinafter H.
8 H,111–114.
 I The Five Ways . . . 149

In the above argument, Bx = x is a thing that thinks; a refers to ‘I,’

the thinker – in this case, Descartes. Now,

[4] Ba ⊃ (∃x)((x = a) & Bx)

is a provable formula in first-order theory, and it unproblematically

entails [1]. Then, given modus ponens, Descartes can conclude [3],
namely, “I exist.” Hintikka thinks that Descartes’ reply to Gassendi,
who had maintained that Descartes could just as well have said,“I walk,
therefore I am,” is “not very clear,” partly on the ground that Bx can
be any predicate and [4], from which [1] can be derived, will be just as
provable. “This already makes the interpretation [1] suspect” (H,112).
But Descartes’ reasoning is fairly clear. If he is looking for an argument
at all, he is not looking for just a valid argument. He is also looking for
a sound argument. If so, the premise “I walk” is not certain: Descartes
may be dreaming that he is walking while he lies comfortably in bed.
But the premise “I think” serves his needs, since to doubt that he
thinks, or to be deceived into thinking that he is thinking, are no less
veridical instances of thoughts.
But Hintikka thinks that “the gravest objection” against the argu-
ment is that it is question-begging. He explains this in the following
way: Formula [1] is provable in first-order quantification theory, but
at a price. First-order quantification theory makes existential presup-
positions, namely, that all singular terms (names, pronouns, definite
descriptions, and so on), such as a, refer, denote, or designate an
existing individual. Consequently, since a cannot fail to refer in the
system, and it refers to ‘I’ (or whoever the thinker happens to be), it
follows trivially that “I exist.” But if we consider systems that do not
make such existential presuppositions, we find that we are unable to
draw “I exist” as a conclusion in such a system.
“My point,” says Hintikka, “may perhaps be illustrated by means
of an example constructed for us by Shakespeare. Hamlet did think a
great many things; does it follow that he existed?” (H, 114) The rhetor-
ical question has little force. Consider the following two arguments:

[A] Hamlet thinks. [B] Hamlet thinks.

∴ Hamlet exists. ∴ There is a thought.

150 The Five Ways

Hintikka thinks that [A] is clearly invalid, since, on his view, Hamlet
“did think a great many things,” but nonetheless Hamlet does not
exist. But consider [B]. [B] is invalid, too, on the ground that there
is no actual thought of Hamlet’s in the way in which there have been
actual thoughts since Adam. To say that Hamlet’s thoughts lie in the
same domain as actual human thoughts is, to say the least, obfuscating.
Perhaps they lie in the domain of fictional thoughts, where there are
also other thoughts such as those of Chichikkov, Hedda Gabler, and
Faust. This renders [B] valid; but it takes the sting away from Hintikka’s
claim, since [A] is no longer invalid. Hamlet thinks and exists in a
possible world: the fictional world.
There is a host of examples in the literature of the cogito as an
argument. I arbitrarily outline one, which is all but stated in Kenny:

[1] Whatever object has a property, [Metaphysical Premise]

then that object exists.
[2] Whatever substance in which the [Universal Instantiation, 1]
attribute of thought inheres, that
substance exists.
[3] If I am a substance in which the [Universal Instantiation, 2]
attribute of thought inheres, then
I, a substance, exist.
[4] I, a substance, inhere the attribute [Premise]
of thought.

[5] ∴ I, a substance, exist. [Modus Ponens, 3, 4]

In so many philosophers who have claimed that the cogito is an ar-

gument, a causal principle, or an ex nihilo nihil fit principle, is never
stated as one of the premises. Yet if one thinks that the cogito is an
argument, surely one must at least entertain the possibility that a
causal principle might well be one of the premises of the alleged ar-
gument. For Descartes explicitly says, “Is there not a God, or what-
ever I may call him, who puts into me the thoughts I am now hav-
ing? But why do I think this, since I myself may perhaps be the
author of these thoughts?” (CSM II, 16; AT VII, 24) It never occurs
to Descartes that thoughts could be produced out of nothing, out of
nowhere.
 I The Five Ways . . . 151

This is a separate claim from the one that both Bertrand Russell
and A. J. Ayer were to make later, that there might simply be author-
less thoughts. This was essentially the view of Georg Lichtenberg when
he criticized Descartes for claiming that the ‘I’ of “I exist” in the conclu-
sion (for the sake of argument) is unjustifiably assumed in the premise
of the cogito. Descartes, so goes the litany, was entitled only to the asser-
tion, “There are thoughts.” Consequently, that view would challenge
the implication of premise [2], and argue that one needs to demon-
strate why any attribute needs to inhere in any substance at all. Bernard
Williams offers a criticism of such views,9 but I should like to concen-
trate on a version offered by Baker and Morris.
Baker and Morris argue that Lichtenberg’s is a seriously flawed read-
ing of Descartes, since it ignores Descartes’ logic and metaphysics,
which prohibit him from making any such move in his language-game.
They write,

It invalidates the common criticism that Descartes begged the crucial question
in formulating the premise as ‘I think’ (‘cogito’) rather than as ‘Thinking
occurs’ (‘cogitatur’). In his view, there can be no judgement about thinking
(or the occurrence of a thought) whose logical subject is not a substance. To
predicate a mode of thinking of a thinking thing is the minimal move in the
language-game. A thought (or an act of thinking) isn’t a substance, and there
is no such thing as a mode which doesn’t inhere in a substance. . . . Descartes’
conception leaves no logical room for contrasting ‘cogito’ with ‘cogitatur’ (as
Lichtenberg did in accusing Descartes of committing a fallacy in drawing the
conclusion ‘sum’).10

As exegesis, this is not only entirely appropriate, but also makes a

very plausible case. What makes their view interesting, however, is that
Baker and Morris use arguments that are now familiar both in philos-
ophy of language and in philosophy of science and apply them to the
history of philosophy, and to Descartes in particular. Thus, their view
faces an objection similar to the ones often raised, say, in philosophy of
science. Their view blocks the possibility of progress in philosophy.11

9 See BW, 95–101.

10 Gordon Baker and Katherine J. Morris, Descartes’ Dualism, 65–66; see also 24, note 1,
and 209, as well as note 23.
11 See, in particular, Descartes’ Dualism, 201–219 for a discussion of a wealth of examples
of fundamental concepts.
152 The Five Ways

If one is constrained to speak only in terms of language-games, then

no view can be fundamentally criticized as being wrong-headed; one
is allowed to say only, “This move is not permitted in that language-
game.” The modern view of logic and metaphysics should not be al-
lowed to judge the philosophical systems of the past, and, of course,
conversely. Consequently, what happens to philosophical progress?
Is it simply that one philosophical system replaces another, and we
are no more entitled to speak of truth and progress in philosophy
than we are in science (say, on Thomas Kuhn’s view of science)?
When a revisionary metaphysics replaces a descriptive one, shall we
say that it is just a replacement, and that we are no closer to the truth
about the basic features of the universe than we were before? This
is a desperate view, I believe, and must be embraced only as a last
resort.
But, if the foregoing is anywhere near right, then Lichtenberg,
Russell, and Ayer can assert that they were finding fault at the heart of
the metaphysical system that Descartes was offering, not merely offer-
ing an interpretation of what Descartes was saying.
As a proper interpretation of the cogito, this full version of the ar-
gument has the following problem facing it. So far as the ideal seeker
knows, no metaphysical truths except the cogito have been established
at this stage in the investigation. If that is the case, Descartes has no rea-
son to rely on an argument one of whose premises – namely, [2], if not
also [1] – may be untrue. This would render the argument as yet not
known to be sound. Such arguments, formulated by Curley, Hintikka,
and Kenny, among others, share the defect of relying on principles
of logic that have not yet been shown or proven, at this juncture, to
be sound: That is, we have no guarantee that in using them we may
not move from true premises to a false conclusion. In the formulated
version not only modus ponens is involved, but also the rule of univer-
sal instantiation, in proving [5] from [1]; and neither rule has been
demonstrated to be sound. This is not to say that this version of the
argument may not be (indeed, it is) a good explanation12 – an expla-
nation that comes long after the cogito – but it is not a correct version
of the cogito, which comes in the opening paragraphs of the Second
Meditation.

12 See Chapter 8, 306.

 I The Five Ways . . . 153

Several critics, especially Frankfurt, have made much of a remark

of Descartes:

I know a man who once, when falling asleep, heard the clock strike four, and
counted the strokes as ‘one, one, one, one’. It then seemed to him that there
was something absurd about this, and he shouted: ‘That clock must be going
mad; it has struck one o’clock four times!’ Is there really anything so absurd
or irrational that it could not come into the mind of someone who is asleep
or raving? (CSM II, 306; AT VII, 457)

So a skeptic might parody someone who thinks that the cogito is an

argument, thus: “I know a man who once, when falling asleep, read
the cogito argument having four premises, and counted the premises as
‘one, one, one, one’. It then seemed to him that there was something
absurd about this, and he shouted: ‘That argument must be mad; it
has repeated the first premise four times.’ Is there really anything so
absurd or irrational that it could not come into the mind of someone
who is asleep or raving?” Indeed, a few lines later Descartes says, “There
are no limits to what a dreamer may not ‘prove’.”
There is a lovely problem facing Descartes. The problem is designed
by Kenny:

Descartes says that his system rests on two principles, his own existence and
God’s existence. If my account is right, he cannot have both these principles in
quite the way he wants them. If what does not exist can have properties, then
he can perhaps prove God’s existence, but not his own. If what does not exist
cannot have properties, he can perhaps prove his own existence, but he cannot
prove God’s existence from God’s essence without begging the question. The
cogito and the ontological argument cannot both be valid. (K, 170–171)

Kenny shows their incompatibility in the following brief: Quite like

Alexis Meinong, Descartes maintains not only that that which ex-
ists can have a nature, but also that that which is not has a nature,
too. To begin with, God is a datum and has a nature; it is part of
his nature, essence, or property to exist. Therefore, God exists. But
then the ‘I’ of a thinker can be a possible subject with an essential
nature of thought; such a thinker can subsist without existing. Since
it is not the essence of said thinker to exist, his existence cannot
be proved. My purpose is not to examine the details of Kenny’s ar-
gument, but simply to focus on its main property: namely, that both
the cogito and the ontological proof for the existence of God are
154 The Five Ways

arguments, and that both arguments cannot be valid; at most, only

one can be.
Now, if my reading of the cogito is correct, it shows Descartes a way
out of the Kenny dilemma. The occasion that gives rise to the dilemma
has been undercut. The cogito is not, and cannot be, an argument, while
the proof of the existence of God is, and must be, an argument. This,
of course, does nothing to show that Descartes’ ontological argument
is sound, either in Kenny’s version or in any other. It only aims to show
that the cogito and the ontological proof of the existence of God are
not obstacles to one another.

C. The Fourth Way

The fourth formulation of the cogito as an argument is owed to Edwin
M. Curley. The best entry into the discussion is Curley’s partial criti-
cism, and partial support, of Frankfurt’s view as presented in his book
Demons, Dreamers, and Madmen. Frankfurt had argued that Descartes’
aim was to conclude not that “I exist” is true, but only that it is indu-
bitable.13 But the notion of “indubitable” is normative, as opposed to
descriptive, in this sense: A proposition is normatively indubitable if it
could not be reasonably doubted.
We misunderstand Descartes’ enterprise, cautions Curley, if we see
Descartes as interested in offering a proof: “[T]he cogito, while it is
inference, is not intended to be a demonstration or proof of sum. It is
not essential to a valid inference that the premisses of the argument
be known with certainty to be true, though this is essential to a proof,
or demonstration, of the classical Aristotelian sort.” (C, 84) To say that
a proposition is proved is to say that the argument constituting a proof
eventually begins with some proposition in that proof. The premises
cannot go back to infinity; they must terminate in a first proposition
or propositions. But the skeptics had averred that, ultimately, to rest
on a first proposition is to rest on an arbitrary premise. For one could
just as well have chosen other premises as starting points. These leave
unanswered questions about their justifiability. Couldn’t these first

13 I find this to be an extraordinary view, one that cannot be sensibly foisted upon
Descartes. Confronted with the claim that a proposition was indubitable and yet
not true, Descartes would have been amused. In what is an otherwise extraordinary
book, Sources of the Self, Charles Taylor reads Descartes in this way, too. I share Bernard
Williams’ reaction that such a view is “totally implausible historically” (W, 35, note 2).
But, for the sake of argument, I am not contesting that view here.
 I The Five Ways . . . 155

propositions be regarded as self-evident? “Descartes, I suggest,” says

Curley, “does not want to rely, in his published works (that is, making
an exception for the Regulae), on the contention that his first principles
are self-evident.”14 (C, 85).
Since Descartes does not wish to countenance the skeptic’s criti-
cism of arbitrariness, a mere subjective certainty, he offers an argu-
ment. This he does by a negative procedure: the famous procedure
of doubting. It is important to understand the nature of the doubt. It
is not essential that what leads to the doubt be substantiated; we do
not have to believe in the veracity of the doubt. This frees us to keep
pushing at the limits of what can be doubted; hence, our doubt esca-
lates from being reasonable to being metaphysical, hyperbolic. Yet our
standards should not be so high that they cannot, in principle, be met.
In short, our task of discovering the first truth must not be an impossi-
ble one. What, then, is a reasonable doubt? “The one most relevant to
the cogito is that a reasonable ground for doubting a proposition must
offer some explanation of how it is that we might erroneously believe
the proposition even if the explanation is only conjectural” (C, 86).
So Descartes sets up the method of doubt. It is a method informed
by a variety of hypotheses: My senses are not trustworthy; I am dream-
ing; a demon of incredible power is bent on deceiving me. Each of
these hypotheses involves the supposition that I am thinking. In short,
if each of these hypotheses is to explain how I err, each must involve
my thinking. “But that supposition entails my existence. So ‘I exist’ is
inferred from ‘I think’” (C, 86). Here Curley lists the two virtues of his
view of the cogito. First, the proposition “I think” is not a premise of a
proof. If it were a proof, one would have demanded how we can know
the truth, or be certain, of the proposition “I think.” But Descartes
need take no epistemic responsibility for ensuring the truth of the
premises. Indeed, Curley shows that were Descartes to take respon-
sibility for demonstrating that he is certain that he is now thinking,
he would find himself bound to an infinite task.15 Second, the cogito
is not question-begging, because “I exist” is not the conclusion of
a proof. If it were, then, as the skeptics had argued, and as Curley

14 In the absence of the citation of a single line explicitly denouncing the value or
reliability of intuition, this is a fairly strong historical claim.
15 I have skipped this neat part of the argument, since it is not essential to my enterprise
(C, 86–87).
156 The Five Ways

so admirably illustrates, the conclusion would be contained in the

premises.16
There is, however, this stubborn exegetical fact, thinks Curley,
that “Descartes does often suggest that ‘I think’ is peculiarly certain”
(C, 87). This lends credence to the claim that “I exist” is, after all,
the conclusion of a proof or demonstration. But if someone were to
show that a proof or demonstration is possible, something that Curley
himself tries to show, that in itself would not refute Curley’s claim. For
Curley is not claiming that a proof or demonstration is impossible,
only that it is possible to do without it. Here is the crucial passage:

For it seems to me that the claim of “I think” to be a first principle can be

defended in the same way that the claim of “I exist” can, that is, a proposition
entailing “I think” is a common element in any reasonable skeptical hypothesis
which might be invoked as a ground for doubting that I think. Any supposition
that explains how my thinking might be erroneous must entail not only that I
exist, but also that I think. So “I think” would also survive systematic doubt, and
the two propositions are on a par. And if that is so, then we have the making
of a demonstration of “I exist.” (C, 88)

He then immediately continues:

This is why Descartes sometimes is tempted to treat “I think” as certain and “I

exist” as the conclusion of a demonstration. But to Descartes there would be no
advantage in using such an argument for the certainty of “I think” to establish
the certainty of “I exist.” For if the fact that a proposition is immune to rea-
sonable doubt is sufficient to make its acceptance as a first principle rational,
then we might as well take “I exist” as a first principle to begin with. (C, 88)

Curley thinks that “I think, therefore I am” is an argument. But,

he claims, instead of regarding “Whatever thinks, exists” as a missing
premise, we ought to regard it as a rule of inference that permits us to
move from “I think” to “I exist.” Thus it would appear that in a neat
and concise way, he avoids some of the difficulties attendant upon the
first two ways of viewing the cogito as an argument.
If the cogito were a syllogistic argument, then it would be a very spe-
cial type of syllogism. The major premise would be “Whatever thinks,
exists,” and this type of premise, says Curley, would require careful
attention. Such a premise would be what Descartes called an eternal

16 C, Chapter 2.
 I The Five Ways . . . 157

truth, a common notion, or a truth that we would normally say is

necessary. Add to it the fact that because it is a major premise in an
Aristotelian syllogism, it has existential import, and we would have
to face some pretty unpalatable consequences. It is a commonplace
theorem in modal logic that if p is necessary, and p entails q , then q
is necessary. Since “Whatever thinks, exists” is necessary, then via the
existential import of that major premise, it would follow that it is nec-
essary that there are thinkers. We would find ourselves saddled with
too many necessary existential propositions (C, 34). Furthermore, if
we are trying to establish the existence of a thinker via the cogito, this
will hardly do (WC, 93).
From this Curley concludes that “Whatever thinks, exists” had bet-
ter be construed as a rule of inference rather than as a statement.
Any necessary truth, such as the one under consideration, should be
regarded as a rule of inference, so as to avoid the troublesome impli-
cation of too many things existing necessarily. This also squares with
Descartes’ assertion, says Curley, that knowledge of eternal truths does
not give us knowledge of any existing thing (C, 34; see also CSM I, 196;
AT VIIIA, 8).
Now there is a modal principle that Curley invokes. It is this. If
a valid argument has a premise that is necessarily true, that premise
can be eliminated without affecting the validity of the argument: If p
and q entail r , and p is necessarily true, then q alone entails r . “So if
an argument which is not, as it stands, formally valid, were to be ren-
dered formally valid by the addition of a premise which is necessarily
true, Descartes would be right to think the exercise an empty one.
If the original argument requires only the addition of a necessarily
true premise to achieve formal validity, then the original argument
was already valid.” (C, 34; see also 32) Curley concludes that the ma-
jor premise is either a necessary truth and hence redundant, or not a
necessary truth and hence in need of support.
Here is Curley’s overall view in his own words:

The reading of Descartes which seems to me to do the least violence to the texts
would run as follows: Yes, there is a general maxim whose correctness is a nec-
essary condition of the validity of the inference from thought to existence; this
maxim is a necessary truth, not merely a contingent one; so it is best regarded,
not as a suppressed premise, but as a rule of inference; the argument, there-
fore, need not take the form of a syllogism, and it is artificial to put it in that
158 The Five Ways

form, since what comes first in the order of discovery is the realization that the
inference is valid in the particular case; but later, reflecting on the inference,
we recognize that justifying it requires us to justify the maxim. (C, 91)

So much for the preamble. Let us turn to examine Curley’s view.

First, among the very common notions or eternal truths is the follow-
ing, “that it is impossible that that which thinks should not exist.” This
eternal truth can be reformulated as “Necessarily, whatever thinks,
exists.” Of these simple notions, Descartes says, there are ones “which
on their own provide us with no knowledge of anything that exists”
(CSM I, 196; AT VIIIA, 8). Curley claims that we should use such an
eternal truth as a rule of inference. But one could use it as a premise,
too: For if eternal truths have no existential import, how do we over-
populate the ontological slum by using them as premises instead of
artificially construing them as rules of inference?
Descartes’ texts support the reading that one need not make exis-
tential commitments in a general statement of the type serving as a ma-
jor premise.17 Descartes had maintained in the Fifth Meditation that
from “the fact that I cannot think of a mountain without a valley, it does
not follow that a mountain and valley exist anywhere, but simply that a
mountain and a valley, whether they exist or not, are mutually insepa-
rable” (CSM II, 46; AT VII, 66–67). Likewise, whether triangles exist or
not, they have certain properties, such as that the sum of their interior
angles is equal to two right angles. Consequently, it would appear that
similar things might be said about minds or thinkers, or about exis-
tence. Whether thinkers exist or not, so long as there is a thought, there
is a thinker. In the Second Meditation, Descartes had convinced him-
self “that there is absolutely nothing in the world. . . . no minds . . .”
(CSM II, 16; AT VII, 25), no bodies, no earth, no sky, no world. But
from that it would not follow that the eternal truth “Whatever thinks,
exists,” or alternatively, “that it is impossible that that which thinks
should not exist,” is not true, even necessarily true. One can conclude
that one need not fear the consequences of existential import; one
could read the major premise in the Aristotelian syllogism as having
no existential import.18 Using eternal truths as premises in such an

17 I find Kenny’s reading of this point congenial; see K, 146–151.

18 Bernard Williams also argues that the cogito cannot be a syllogistic argument, in
part because of the existential import of the major premise, which would already
 I The Five Ways . . . 159

argument is more in keeping with Descartes’ intentions than using

them as rules of inference.
Second, Curley thinks that the cogito is an argument, even if that
argument is not to serve as a proof or demonstration. But minimally,
a skeptic might say, surely the argument must be valid? If so, then it
will be subject to the skeptical strictures. It is true that the stricture
here is not that Curley has stopped at an arbitrary stopping point, with
no more than a subjective conviction at hand. Rather, the stricture is
that if the argument is valid, then the truth of the conclusion must be
contained or presupposed in the conclusion. Why in this specific case
the stricture does not apply, Curley does not say.
Third, the modal principle that Curley invokes is controversial. To
repeat the modal principle in question: If p and q entail r , and p is
necessarily true, then q alone entails r . The restricted version of the
principle, championed by Peter Thomas Geach,19 reads: If p and q
entail r , and p is necessary, and q and r are contingent, then q alone
entails r . While the restricted version has been courted by Curley,
both versions of the principle have been rejected by Alan Anderson
and Nuel Belnap.20
How, then, should Curley’s argument be re-read? Perhaps as some-
thing like the following: The cogito is an argument where p (Whatever
thinks, exists) and q (I think) entail r (I exist), and since p (Whatever
thinks, exists) is necessary, and q (I think) and r (I exist) are contingent,
then q (I think) entails r (I exist). This does not treat the eternal truth
“Whatever thinks, exists” as a rule of inference. It is rather treated as
a premise in an argument, a premise that can be dispensed with in ac-
cordance with the restricted modal principle. Nor is there any gain in

presuppose the existence of thinkers. There is, surely, a slip in Williams’ remark, “The
eternal truth ‘In order to think, it is necessary to exist’ presupposes no existential
proposition; but, by the same token, it cannot play the role of major premise in
a syllogism” (WC, 93). Why? There are adequate interpretations of the Aristotelian
system of categorical syllogism that make no existential presupposition; such a system
has a restricted set of valid schemas. It is worth noting, with Williams, that one can
construe the eternal truth as “If one thinks, then one exists,” without knowing how
that conditional statement is to be taken. For one thing, it presupposes that existence
is a predicate. Yet the difficulties that Williams goes on to raise would have been, as
Williams himself recognizes, alien to Descartes’ thinking.
19 Geach, Logic Matters, 182. Only the letters used for variables have been altered for
the sake of uniformity.
20 Anderson and Belnap, “Enthymemes.”
160 The Five Ways

this. The restricted modal principle, like the unrestricted one, would
have been construed by Descartes as an eternal truth, and both would
be just as much in need of defense as any other eternal truth, such as
“Whatever thinks, exists.”
Be that as it may, there is a significant step in Geach’s proof of the re-
stricted modal principle that Curley fails to relate. It is this: “q entails r
if and only if there is an a priori way of getting to know that ‘If q then r ’,
which is not a way of getting to know either whether q , or whether r .”21
This is Geach’s truth condition of an entailment statement. It follows
that “I think” entails “I exist” provided that there is an a priori way of
getting to know “If ‘I think’, then ‘I exist’,” which is not a way of getting
to know either whether “I think,” or whether “I exist.” How could that
statement – “If ‘I think’, then ‘I exist’” – be known in a way that would
satisfy the Cartesian epistemic strictures? But suppose, per impossible,
that it can be known, without knowing whether “I think,” or whether
“I exist”; where then is the need for using it as a rule of inference?
Curley had argued that if “If p, then q ” is a necessary statement, then
it is not essential for deriving q from p. He invoked Geach’s restricted
modal principle. But Geach’s proof for the restricted version of the
modal proof relies on showing that “If p then q ” is a necessary truth. We
would be back to the original question: How could the statement – “If
‘I think’, then ‘I exist’” – be known to be a necessary truth that would
satisfy the Cartesian epistemic strictures? Whatever the accuracy of
Geach’s theory as a theory of entailment, it is of little aid to Curley.22

21 Peter Thomas Geach, Logic Matters, 179–180.

22 Von Wright’s theory of entailment, presented in Logical Studies, which Geach held
in high esteem, reads as follows: “A entails B, if and only if, by means of logic, it
is possible to come to know the truth of A ⊃ B without coming to know the false-
hood of A or the truth of B.” “By means of logic,” says von Wright; “by a priori
means,” says Geach, in Logic Matters. There are several inadequacies. As Anderson
and Belnap point out, the expression ‘come to know’ is loose; Strawson, in “Review
of von Wright,” found a parallel difficulty. Von Wright’s response was to say that
there were different meanings of ‘come to know,’ and that, at any rate, the sit-
uation was unsatisfactory. Anderson and Belnap also considered the proposal of
T. J. Smiley, in “Entailment and Deducibility,” but found that that proposal leads
to abandoning the transitivity of entailment. My purpose here is not to contest
any theory of entailment, or to plead for a particular one. It is rather to show
that one ought not to invoke so controversial a principle of entailment to sup-
port what is purported to be the first and simple truth – a truth Polyander could
acknowledge! – that is to serve as a foundation for other truths in the metaphysics of
Descartes.
 I The Five Ways . . . 161

Fourth, consider what I call the problem of justification: “[W]hat

comes first in the order of discovery is the realization that the inference
is valid in the particular case; but later, reflecting on the inference, we
recognize that justifying it requires us to justify the maxim” (C, 91).
I do not understand what Curley means by saying that we discover
the inference to be valid in the particular case. Surely, there is no
recognition of a valid inference in this case, or in any other, unless the
maxim, or the rule of inference, has been demonstrated to be sound:
that is, that it is a rule that never allows a move from true premises
to a false conclusion. In short, there is no discovery of the cogito, or
of any other argument, until the rules of inference are sound, and
antecedently known to be such.
Fifth and finally, Curley complains that to cast the cogito in the form
of a syllogism is artificial; but it is no less artificial to cast “Whatever
thinks, exists” in the role of a rule of inference. If anything, it is more
so. It is not just our contemporary understanding of rules of inference,
such as the modus ponens, modus tollens, rules of association, and so on,
that makes us think that “Whatever thinks, exists” is so different from
these other rules of inference. The rules of inference of the Stoics,
which the skeptics criticized, and of which Descartes was aware, are also
quite different from “Whatever thinks, exists.” It bears repeating, with
Curley, that Descartes had no clear notion of proof. So it is something
of a historical problem to decide to what extent Descartes would have
espoused talk about rules of inference.23

23 Ian Hacking has argued that while it was Leibniz who first gave us the notion of proof
as we understand it today, nearly all the elements that go to make up the Leibnizian
notion of proof were present in Descartes; see Ian Hacking, “Proof and Eternal
Truths: Descartes and Leibniz.” If Hacking and Y. Belaval, whose Leibniz: Critique de
Descartes is the basis of Hacking’s interpretation, are right, then the invocation of
rules of inference would have greatly puzzled Descartes. This does not undercut my
very sympathetic response to Gaukroger’s detailed criticism of Hacking’s view of the
connection between truth and proof in Descartes and the doctrine of eternal truths.
Gaukroger’s reading would lend further credence to my claim that the cogito could
not be an argument, at least at this stage of the argument; see Gaukroger, Cartesian
Logic: An Essay on Descartes’s Conception of Inference, 26–31.
Now, one might argue – as did Catherine Wilson – that there is something in
Hacking’s notion, namely, that Descartes did think of a mathematical proof not for-
malistically but phenomenologically. Mathematical arguments involve this powerful
compulsion to go from step A to step B, even where there is no syllogistic basis
for doing so, and no suppressed premise can be supplied. The mathematician just
sees where to go. The cogito seems to attempt to create just such a compulsion to
162 The Five Ways

D. The Fifth Way

The fifth way of reading the cogito as an argument is owed to Bernard
Williams. He has argued that the cogito is an inference – the proposition
inferred is “I exist,” and it is said to be inferred from the proposition
“I think” – but it is an inference of a special kind, not a syllogistic
inference. That there is undoubtedly an inference, Williams thinks, is
due to the fact that Descartes, in the relevant passages, uses terms like
‘it follows’, ‘infer’, and the like. Moreover, Williams maintains that this
way of reading the cogito as an argument, based on a special kind of
inference, faithfully represents Descartes’ own view.
Consider a typical subject–predicate sentence of the form Fa, where
F is a predicate and a is a proper name, a pronoun, or a definite de-
scription. Now, characteristically, or commonly, Fa presupposes that a
does not fail to refer, or designate: in short, that there exists an entity
referred to by a. The relation of presupposition is quite different from
the relation of implication: If p implies q , then if q is false, p is false. But
if p presupposes q , then if q is false, p is neither true nor false. In the
language of Sir Peter Strawson, in his Introduction to Logical Theory, in
such a case no statement has been made. As an example, if I say, “The
tenor in my house is fat and wealthy,” and there is no tenor in my house,
then the subject term, “the tenor in my house,” fails to refer or desig-
nate any such entity, and the sentence, “The tenor in my house is fat
and wealthy,” is neither true nor false. In order for the sentence to have
a truth-value, at a minimum the subject term must have a reference.
Williams claims that just such a principle of inference, the relation
of presupposition, gives us what we need in order to understand why

move, but does so even while holding “mathematics,” as a whole body of theory, in
suspense.
An adequate response to this proposal will emerge only at the end of the next
chapter. Meanwhile, let me raise three questions. First, how is Descartes to distinguish
a sound compulsion from an unsound one, even if from a purely phenomenological
point of view? Second, at this stage in the Meditations on First Philosophy, Descartes has
dismissed a simple mathematical truth, namely, 2 + 3 = 5; how, then, are good com-
pulsions to be distinguished one from another, such that the compulsion induced
by the cogito underwrites its truth but the other compulsion does not, yet, support
a simple mathematical truth? Third, what would be the relationship between the
sort of compulsion that guarantees the truth of the cogito and the general rule, also
guaranteeing truth, that comes only in the Third Meditation? Indeed, why is the
general rule even necessary in the presence of said compulsion? See also Gaukroger,
Cartesian Logic, 50–51.
 I The Five Ways . . . 163

the cogito is an argument and the nature of the inference involved in

it. He states that if “‘ p presupposes q ’ means that q must be true for p
to be either true or false, it follows a fortiori that q must be true for p
to be true; in other words, if p is true, q is also true. And this I take to
be the principle of the cogito.” (WC, 96) Williams must mean that the
principle of inference in the cogito-argument that legitimizes the move
from the premise, “I think,” to the conclusion, “I exist,” is the relation
of presupposition. In other words, if the relation of presupposition
is a sound principle of inference, then if “I think” is true, the subject
term, ‘I’, cannot fail to have a reference. Consequently, if the cogito
is asserted, then the relation of presupposition guarantees the truth
of the statement “I exist.” This is the first principle that Descartes was
seeking.
Williams further maintains that “there appears to be a close relation
between Descartes’ ‘eternal truth’ or ‘maxim’ and a rule of language.
And language – we should not forget – has at no time been eliminated
by doubt: ‘Descartes, and even more his reader, begins to meditate in a
universe of existing discourse.’” (WC, 97) Presumably, then, there is a
close connection between the eternal truth expressed by the relation of
presupposition and the rules of language that govern the use of certain
kinds of sentences to make statements in that language. Thus, the
eternal truth, expressed in the relation of presupposition, guarantees
the truth of the proposition “I exist,” when the sentence “I think” is
correctly used in ordinary language.
There are several difficulties in understanding the cogito along the
lines recommended by Williams. First, there is no way to determine
the truth of Fa without first determining whether or not a refers or
designates. To expand on the point: Let us suppose that there is a log-
ical relation between the sentence Fa and the sentence, “There exists
something referred to, or designated by, a.” It still would not follow
that we can extract from it a principle of inference, namely, an infer-
ence that enables us to come to know on the basis of our knowing p.
For it is written into the relation of presupposition that to know that
Fa is true is already to know the truth of the sentence, “There exists
something referred to, or designated by, a.”
Williams thinks that this objection trades on a simple confusion:
the confusion of two distinct principles, namely a principle of logic,
which is what he is offering, and a principle of epistemology, which is
164 The Five Ways

what his critic demands. “It is extremely useful in logic,” says Williams,

to be able to say, for instance, that there is a principle allowing us to infer “ p”

from “ p and q ” even on the assumption that it is in fact impossible to know
the truth of “ p and q ” without first knowing the truth of “ p”. I do not see why
we should not also be justified in considering the relation of presupposition
between “I think” and “I exist” as the basis of a principle of inference and
hence of the cogito itself in so far as it is an inference – not, of course, in the
sense of a syllogism with a suppressed major premise. (WC, 97)

This underestimates the objection. Consider the following from

Strawson:

Suppose someone says “All John’s children are asleep”. Obviously, he will not
normally, or properly, say this, unless he believes that John has children (who
are asleep). If John has no children, then the sentence, “All John’s children
are asleep,” does not have a truth-value, although it is perfectly meaningful.
But to say that the man who uses the sentence in our imagined case fails to say
anything either true or false, is not to say that the sentence he pronounces is
meaningless. Nor is it to deny that he makes a mistake. Of course, it is incorrect
(or deceitful) for him to use this sentence unless (a) he thinks he is referring
to some children whom he thinks to be asleep; (b) he thinks that John has
children; (c ) he thinks that the children he is referring to are John’s. We might
say that in using the sentence he commits himself to the existence of children
of John’s.24

So, too, we might say the following: “The man who says, ‘I think’,
and is using that sentence to make a statement, and not to test his
voice, for example, is incorrect or (deceitful) unless (a) he thinks that
‘I’ is referring to himself; (b) he thinks that he is acquainted with a
thought; and (c ) he thinks that the particular thought in question with
which he is acquainted when he says, ‘I think’, does belong to him. If
these conditions are not satisfied by anyone saying, ‘I think’, then he
has failed to make a statement; he has said something which has no
truth-value, even though the sentence he speaks is meaningful.” But
Williams’ critic can now point out that if condition (a) is satisfied, the
relation of presupposition can be safely dispensed with; it will not add
anything to what we can know.
One must of course insist on Williams’ distinction between the prin-
ciples of logic and the principles of epistemology, but one must equally

24 P. F. Strawson, Introduction to Logical Theory, 175.

 I The Five Ways . . . 165

insist that what is at stake, in Descartes, is the principle of epistemology.

After all, the primary epistemological problem is, what can I know for
certain? And Williams’ critic can make his point thus. Consider the
following argument, he might say:

I think and I exist.

Therefore, I exist.

This instantiates the principle of logic that Williams referred to,

namely, the “principle allowing us to infer ‘ p’ from ‘ p and q ’.” No one
need dispute the validity of this argument, nor the soundness of the
principle of inference. But it would be a poor thing if it were offered
as a reason for accepting the conclusion, since what is asserted in
the conclusion is already asserted in the premises. Descartes could not
claim to know with certainty that he exists on the basis of the premises,
since, in this case, the conclusion is blatantly contained in the premises.
For an exactly parallel reason, the critic is unsatisfied with the rela-
tion of presupposition, not because it is not a correct principle of infer-
ence, but rather because it adds nothing to our knowledge. Descartes
must offer us a reason for accepting the conclusion, if the cogito is
an argument, such that that reason does not already presuppose the
truth of the conclusion. But that is exactly what the principle of pre-
supposition does in this case. Step 1: Establish that “I think” is true.
Step 2: State, in this case, an instance of the relation of presupposition,
namely, “If ‘I think’ is true or false, then the subject term, ‘I’, must have
a reference; in other words, ‘I exist’ must be true.” Step 3: Conclude
that “I exist.” The critic’s point is not to question the correctness of
Step 2, but rather to point out that Step 1 simply cannot be carried
out without first establishing the truth of Step 3.
The second objection: There is no need to be certain of the premise,
“I think.” Descartes could just as well have said, I am walking, or I am
breathing, and that would have been enough. That sentence has a
truth-value: It could be true, if Descartes was indeed walking or breath-
ing, or false, if he was dreaming and lying in bed, in the one case, or if
he did not have physical body, in the other. In either case, he would be
using the sentence to make a statement, and that, via the relation of
presupposition, would presuppose the existence of the individual de-
noted by ‘I’. But Descartes explicitly wanted a true premise, and argued
166 The Five Ways

against claims that meant the premises could be false. Williams’ view
turns a crucial part of the text into a puzzle.
Thus, consider the following passage from a letter that Descartes
wrote to Reneri for Pollot in April or May 1638: “When someone says
‘I am breathing, therefore I exist,’ if he wants to prove he exists from
the fact that there cannot be breathing without existence, he proves
nothing, because he would have to prove first that it is true that he is
breathing, which is impossible unless he has also proved that he exists”
(CSMK, 98; AT II, 37–38). This makes it difficult to sustain the claim
that Williams’ view was also Descartes’ view. First, insofar as someone
is making the statement that “I am breathing, therefore I am,” and
not just uttering a sentence to test his voice, then the falsity of the
claim means nothing, since in making the statement the speaker has
guaranteed the referent of ‘I’. Second, Descartes himself would object,
in a manner similar to Williams’ critic, by saying that “unless he has
[first] also proved that he exists,” his claim means nothing.
The third objection is, of course, the most fundamental. It is that
Williams makes Descartes presuppose an eternal truth at a stage in
the discussion when no truth of mathematics, or of logic, is accepted.
They are all under the cloud of doubt. Whether rightly or wrongly,
Descartes would not have linked the truths of language as intimately
to the truths of logic as Williams has him do. Descartes never ques-
tioned the truths of language: For him, these truths amounted to no
more than our knowing or understanding the meaning of words. On
the other hand, he at least attempted to seriously question the truths
of mathematics and logic, or the so-called eternal truths.25 If there was

25 It will clearly not do to say that Descartes does not call all the truths of logic into doubt
when he calls the truths of mathematics into doubt. Even if God can make contradic-
tories true together, that doesn’t mean that we can doubt the truth of the principle
of noncontradiction. At most he can raise a second-order doubt about logical princi-
ples. There is much evidence, especially in the letters, for the second sentence in the
foregoing remark (CSMK, 23; AT I, 145–146; CSMK, 25; AT I, 152; CSMK, 235; AT IV,
118; CSMK, 358–359; AT V, 224). But from the point of view of the critic, there are
severe problems facing Descartes. First, one might easily argue that the same could
be said to be true of mathematical principles; yet Descartes did not prevent himself
from doubting those in the First Meditation (pace Frankfurt). Second, second-order
doubt about logical principles is of no help to Descartes; it is not as if having doubted
them at the second-order level, he could then safely use the logical principles in his
first-order reasoning. The sweep and power of the hyperbolic doubt in the First Med-
itation ensures that however minuscule the doubt, at whatever level, and of whatever
 I The Five Ways . . . 167

a link in Descartes’ mind between these truths, it needs explaining why

he never thought he had reason to doubt the truths of language, when
he had provided himself with reasons to doubt the eternal truths.
The fourth objection: There are a variety of ways of understand-
ing a certain class of subject–predicate sentences, not all of which
involve the relation of presupposition. There is, of course, Bertrand
Russell’s theory of definite descriptions; there is David Kaplan’s read-
ing of Russell; and finally there is van Fraassen’s and Lambert’s view
of presupposition. So long as this is controversial, Williams’ principle
of presupposition will not support the cogito as well as Descartes might
have thought. It would not be difficult for the evil genius to persuade
Descartes that, given the rival logical theories of subject–predicate sen-
tences, there is no conclusive reason to settle for Williams’ relation of
presupposition over other theories. It should best be left untouched,
at least initially, while Descartes is searching for a first principle of
which he is absolutely certain; that principle cannot be tainted by the
possibility that it relies on something less than absolutely certain, like
the relation of presupposition. For example, Russell’s theory would
make the statement, “I think,” false when ‘I’ fails to have a referent,
rather than neither true nor false, as on Strawson’s or Williams’ view.
There is a fifth and final objection (which affects not only Williams
but also Hintikka and Curley). It lies in Descartes’ explicit hypothesis
about the ideal seeker. Descartes repeatedly says that the first principle
can be understood by an ordinary person, guided by the natural light
of reason. Such an individual is represented by Polyander, who has
little, if any, knowledge of logic and mathematics. But on Williams’
view, as well as on Hintikka’s and Curley’s, Descartes’ ideal seeker has
to be a pretty sophisticated thinker in order to make the inferences
that he is supposed to make. And yet, it was Williams who wrote:

Descartes does not suppose that his right to claim a reader’s attention lies in
any sacramental, traditional or professional position. His authority to show us
how to think lies only in this, that he has himself, as he supposes, uncovered

nature, it undermines the proposition in question. So let me suggest a dilemma for

anyone who would claim that Descartes treated principles of logic separately from
principles of mathematics: If logic stays unquestioned, so should mathematics; and if
so, a significant part of the First Meditation remains quite inexplicable. On the other
hand, if mathematics is questioned, so should logic be questioned; if so, the validity
of the cogito argument would be impossible to determine.
168 The Five Ways

methods of simple, clear-headed and rational inquiry which all reasonable peo-
ple can conduct if they clear their minds of prejudice and address themselves
in a straightforward way to the questions. No special training, no religious dis-
cipline, no knowledge of texts or of history is needed in order to do this. He
was disposed to think, in fact, that such things could be an actual obstacle.26

Question: Could an ideal thinker be one ignorant of the theory of

presupposition and the laws of logic and inference? He could not,
says Williams. This seems to me to be quite contrary to Descartes’ own
avowed aim of making the cogito clear “even to women.”27 My aim
is not to parade Descartes’ male chauvinism, a seventeenth-century
commonplace. It is rather to insist that Descartes’ enterprise not be
lost sight of. He was offering a claim that he frequently felt could be
understood by the least sophisticated. Perhaps Descartes was wrong
in this: But then we need a separate argument to show this, and until
such an argument is forthcoming, we must not dismiss the notion of
the ideal seeker that was a central part of his epistemic view.

II. . . . Plus One

Written nearly thirty years ago, Jaakko Hintikka’s “Cogito, Ergo Sum:
Inference or Performance?” argues that the cogito is a performance, a
denial of which leads to existential inconsistency. Both terms, perfor-
mance and existential inconsistency, need to be defined. But first, I begin
with an intuitive idea of what Hintikka famously offers.
He asks us to consider sentences such as, “De Gaulle does not exist,”
“Descartes does not exist,” and “Homer does not exist.” When uttered
by De Gaulle, Descartes, and Homer, respectively, each of these sen-
tences is self-defeating. It is self-defeating in the sense that the ut-
terance of that sentence leads the listener (who may be the speaker
himself ) to a belief exactly the opposite of that which the speaker
intends to convey. By contrast, when uttered by a third party these

26 Bernard Williams, “Introduction,” ix.

27 Here is the context: “The certainty and evidence of my kind of argument for the
existence of God cannot really be known without distinctly recalling the arguments
which display the uncertainty of all our knowledge of material things; and these
thoughts did not seem to me suitable for inclusion in a book [Discourse on the Method]
which I wished to be intelligible in part even to women while providing matter for
thought for the finest minds” (CSMK 86; AT I, 560).
 II . . . Plus One 169

sentences do not have that failing. Thus, De Gaulle’s saying, “De Gaulle
does not exist,” is odd, absurd, inconsistent; my uttering the same sen-
tence, even in the presence of De Gaulle, is not. It is the purpose of
ordinary discourse that when one utters a declarative sentence one
intends to convey a piece of information. But that purpose is defeated
or destroyed in certain special cases, such as the case of De Gaulle.
Now, in order to convey information to others, I have to do something
that is heard, seen, or felt by others; but in my own case, I do not have
to do any such thing: A thought is sufficient.
Hintikka claims that the cogito-sentence, “I think, therefore I am,” is
just such a proposition, a piece of information. It is a declarative sen-
tence; when its negation is uttered by a speaker, say, by Descartes, it de-
feats its very purpose. In ordinary circumstances to an ordinary listener,
the denial of the cogito-sentence would convey that Descartes exists,
rather than what Descartes intends to convey, namely, that Descartes
does not exist. “This transition from ‘public’ speech-acts to ‘private’
thought-acts, however, does not affect the essential features of their
logic” (H, 119).
It is time to define terms. First, existentially inconsistent:

The notion of existential inconsistency may be defined as follows; let p be a

sentence and a a singular term (e.g., a name, a pronoun, or a definite descrip-
tion). We shall say that p is existentially inconsistent for the person referred to
by a to utter if and only if the longer sentence
p; and a exists

is inconsistent (in the ordinary sense of the word). (H, 116)

Second, to elucidate performance:

The inconsistency (absurdity) of an existentially inconsistent statement can in

a sense be said to be of performatory (performative) character. It depends on an
act or “performance,” namely on a certain person’s act of uttering a sentence
(or of otherwise making a statement); it does not depend solely on the means
used for the purpose, that is, on the sentence which is being uttered. (H, 118)

Third, consider statement. The notion of existential inconsistency ap-

plies to a statement, and not to a sentence. A statement is more than
a sentence: A sentence is a grammatical entity; in order to character-
ize a statement, on the other hand, we need to specify at least three
things, namely, the speaker, the sentence he utters, and the context or
170 The Five Ways

circumstance in which he utters it. Thus, “De Gaulle does not exist”
is a perfectly good sentence, but it is not existentially inconsistent or
even ordinarily inconsistent. However, if we specify that the speaker is
De Gaulle, that the sentence is “De Gaulle does not exist,” and that the
circumstances are ordinary ones in which De Gaulle wishes to convey a
piece of information, then we can say that we have an existential incon-
sistency on our hands. In the case of the cogito, Descartes is the speaker;
the sentence is “Descartes does not exist”; and Descartes is his own au-
dience, intending to convey to himself the truth of the sentence. In
this, of course, he fails.
“We may now appreciate,” says Hintikka,

the function of the word cogito in Descartes’s sentence as well as his motives
in employing it. It serves to express the performatory character of Descartes’s
insight; it refers to the “performance” (to the act of thinking) through which
the sentence “I exist” may be said to verify itself. For this reason, it has a
most important function in Descartes’s sentence. It cannot be replaced by any
arbitrary verb. (H, 123)

Let us examine Hintikka’s interpretation of the cogito. First, a minor

point: I simply see no existential inconsistency, or at least Descartes
should not, in De Gaulle’s claiming that “De Gaulle does not exist,”
as Hintikka maintains, as opposed to De Gaulle’s saying, “I do not
exist.” The reason is simple enough: If De Gaulle, like Descartes, were
searching for a certainty and was at the moment in the throes of doubt,
he could genuinely doubt that he was De Gaulle. It is implied that De
Gaulle knows that he is De Gaulle, and that when De Gaulle says, “De
Gaulle does not exist,” he does so in full knowledge of who he is. But
that, at least, is not certain at the end of the First Meditation.
But I want to go a step further. Not only is there no inconsistency
in De Gaulle’s claiming, “De Gaulle does not exist,” there is also none
in his saying, “I do not exist.” I think it is possible for someone to
genuinely assert, “I do not exist.” Imagine someone who has gone
through the First Meditation, its arguments deeply impressed upon
him. He comes to genuinely doubt everything he knows; at the very
end of that meditation (and not knowing what is to come shortly there-
after) he wonders if he exists, and he can find no reason to allay his
doubt. A thinker in the throes of doubt need not have discovered the
route to the cogito: He could have become permanently embedded
 II . . . Plus One 171

in doubt (or, Hume-like, he would have dismissed those doubts as

soon as he went out of his study). It is only after he reads the Second
Meditation and reflects upon it, invokes the demon argument, and
so on, that the thinker is relieved to learn that his existence can be
demonstrated.28
Imagine: Someone doubts the existence of the external world.
There is a satisfactory “proof” of the existence of the external world,
but no one has yet made that philosophical discovery. What does he
do in the meantime? Well, he either takes its existence for granted
and asks for a proper analysis of sentences referring to the external
world, much as G. E. Moore did; or he continues to search for the
proof; or he ignores the problem. Similarly, someone doubts his own
existence. There is a satisfactory “proof” of the existence of the self,
but no one has yet made that philosophical discovery. What does he do
in the meantime? Once again, he either takes its existence for granted
and asks for a proper analysis of sentences referring to the self; or he
continues to search for the proof; or he ignores the problem.
For someone who thinks that the problem of whether or not he
exists is a genuine philosophical problem, it is not implausible to con-
tend that, for all he knows, he does not exist. If you find this puzzling,
it is because you already have a philosophical theory mitigating your
doubt. Hintikka is worth quoting on this: “It is not strictly true to say
that an inconsistency arises from Descartes’s attempt to think that he
does not exist or to doubt that he does. Someone else may think so;
why not Descartes himself?” So far, so good. But immediately Hintikka
continues: “He can certainly think so in the sense of contemplating a
‘mere possibility.’ What he cannot do is to persuade anybody (including
himself) that he does not exist; wherefore he cannot try to profess
(to others or to himself) that he does not exist without defeating his
own attempt.” (H, 123, note 26) This is so unlike Descartes. It is, surely,
enough that Descartes can envisage a “mere possibility”; that is enough
to make him doubt his existence. Descartes doubts the existence of the
physical world on just this “mere possibility”; why not doubt his own
existence? In fact, if he can envisage the possibility of his nonexistence,
how can he take the knowledge of his existence for granted?

28 For an elegant telling of this point, see Jorge Louis Borges, “Borges and I” and “The
Circular Ruins.”
172 The Five Ways

Let me turn to the second objection. For all the talk about per-
formance, there is an “inference” after all. Were Descartes to enlist
Hintikka’s view, Descartes would have to maintain the following argu-
ment, without which he would have no reason to believe that he exists:

[1] If I, as both speaker and listener, in the ordinary circumstances,

intending to convey a piece of information, were to claim, S, “I
do not exist,” then I must make a statement that is existentially
inconsistent (that is, necessarily false).
[2] But if a statement, S, entails a necessary falsehood, then not-S is
necessarily true.
[3] I do claim that “I do not exist.”

[4] ∴ Not-S is necessarily true. That is, “I exist” is necessarily true.

The foregoing is an argument making use of several principles of
logic and inference. Descartes cannot reach the conclusion without
premise [3], a premise that I have just argued is in need of justifica-
tion. Moreover, will an attempt to understand and justify [3] return us
to Hintikka’s thesis, thus forcing us to argue in circles?
Third, if uttering or saying to myself, “I do not exist,” is existentially
inconsistent when the intention is to convey a piece of information,
then uttering its negation, “I exist,” is surely existentially tautologous
in character. But then, what is to be gained from that statement, which
could count as the first truth? Why is an existentially tautologous state-
ment, preferred over any other simple tautologous statement, such as,
“Either I exist, or I do not exist”?
Fourth, Hintikka’s definition of existential inconsistency does not
serve his purpose, much less Descartes’. The definition is too broad.
Consider the following statements uttered by me:

[1] I cannot speak a word of English.

[2] I am a liar.
[3] I wish for nothing.
[4] I cannot think a thought.
[5] I cannot utter a sentence.

When each of these statements, p, is uttered by me, a, it, strictly

speaking, satisfies Hintikka’s definition of existential inconsistency: “ p;
 II . . . Plus One 173

and a exists” is ordinarily inconsistent. In these cases, “a exists” is

not necessary, for there is something about each statement which,
when uttered to perform a speech act, renders it “absurd.” This shows
that Hintikka needs to put certain restrictions on what is to count
as p. Otherwise, on Hintikka’s view of Descartes, Descartes could have
uttered any sentence [1] through [5], and there would have been
an ordinary inconsistency, but not the existential inconsistency that
Hintikka needs to establish. As nearly as I can tell, p can only be the
statement, “a does not exist.” But then,

p; and a exists

reduces to

a does not exist; a exists

Surely, that is trivial.

Fifth, Hintikka had tried to show that the cogito is an argument:
Hamlet has many thoughts, but assuredly Hamlet does not exist. So
Descartes’ having many thoughts would not entail that Descartes exists,
unless his existence is presupposed. Or consider the following ar-
gument, said Hintikka: Homer was a Greek, or he was a barbarian
(non-Greek). If he was a Greek, then he existed; if he was a barbar-
ian (non-Greek), then he existed. Therefore, Homer existed. This,
Hintikka had argued, is a poor argument, since it, too, blatantly as-
sumes the existence of the very thing that the argument purports to
prove. “The celebrated Homeric question cannot be solved on paper”
(H, 115). Alternatively, one can see that the case of Hamlet shows
the inference to be fallacious. Thus, either Hamlet is the prince of
Denmark, or he is not. If he is the prince of Denmark, then he exists;
if he is not the prince of Denmark, then he exists. Therefore, Hamlet
exists.29 So far, Hintikka.
Imagine that Hamlet had the thought, “I do not exist,” and that
Shakespeare had made Hamlet, in a different soliloquy, go through
Hintikka’s philosophical argument that leads Hamlet to the thought
that he does exist after all. It would be a poor inference from this
to conclude that, therefore, Hamlet exists. Hintikka needs existential
presuppositions just as much as those he argues against. He is not

29 See Appendix B.
174 The Five Ways

defended by the following: “If Hamlet exists, and Hamlet says, ‘I do not
exist’, then, and only then, is an existential inconsistency committed
when Hamlet makes that statement.” But this is only a conditional
statement, and it is on all fours with the statement, “If I exist, and I
say, ‘I do not exist’, then, and only then, is an existential inconsistency
committed by me when I make that statement.” Those statements are
uncontroversially true. If the antecedent is false, then the conditional is
trivially true; if the antecedent is true, then the consequent is true. The
problem, of course, is to establish the truth of the antecedent, “I exist,
and I say, ‘I do not exist’.” This cannot be established in Hintikka’s way.
Sixth, Hintikka says:

It seems to me that the most interesting interpretation one can give to [the
cogito] is to say that Descartes realized, however dimly, the existential incon-
sistency of the sentence “I don’t exist” and therefore the existential self-
verifiability of “I exist.” Cogito, ergo sum is only one possible way of expressing
this insight. Another way actually employed by Descartes is to say that the
sentence ego sum is intuitively self-evident. (H, 121)

In the next paragraph he says that the cogito

cannot be a logical inference in the ordinary sense of the word. . . . In

Descartes’s argument the relation of cogito to sum is not that of a premise
to a conclusion. Their relation is rather comparable with that of a process to its
product. The indubitability of my own existence results from my thinking of it
almost as the sound of music results from playing it. (H, 121–122)

This will not do. The ‘therefore’ is puzzling: The opposite of ex-
istential inconsistency is existential tautology, hardly existential self-
verifiability. One is a logical notion, the other an epistemic one. Next,
as I have just tried to show in reconstructing Hintikka’s claims, the
individual thinker can scarcely do with a single intuition, but needs an
entire argument. Finally, this leaves no room for doubt to play a role
in Descartes’ argument. The doubting, or the possibility of deception,
is not essential. The deceiving demon could be set aside as so much
color, an insubstantial aspect of Descartes’ doctrine of the cogito. One
could simply have begun with, “I don’t exist,” and reached the conclu-
sion that one did exist. To this extent, Hintikka’s suggestions are quite
out of harmony with Descartes’ actual enterprise.
Seventh and finally, Hintikka has claimed that the cogito is so singular
that it will permit no generalization from it. But here is the proposition
 II . . . Plus One 175

from Meditations on First Philosophy: “This proposition, I am, I exist is nec-

essarily true whenever it is put forward by me or conceived in my mind.”
This lends itself to a simple generalization from which the quoted sen-
tence can be derived. Thus: “For any individual, this proposition, I am,
I exist is necessarily true whenever it is put forward by the individual
or conceived in the mind of the individual.” Perhaps this is the right
representation of the compact proposition, “Whatever thinks, exists.”
“In a sense, therefore,” says Hintikka, “Descartes’ insight is not gener-
alizable. This is of course due to its performatory character.” (H, 127)
This is to confuse the epistemic problem of how one can know the
truth of the cogito with the logical problem. Regardless of whether the
cogito is a performance, one could generalize, and Descartes was right
to think that he could do so.
Each of these five plus one ways of understanding the cogito as an
argument is riddled by several difficulties. But there is one difficulty
so deep and so irreparable that we had better conclude that the cogito
is not an argument. This brings me to the heart of my thesis.
 6

Cogito: Not an Argument

Perhaps nowhere in this book is the Sulmo principle more steadfastly

used than here. The principle states that “a philosopher’s system must
not be reconstructed and judged until after his death.”1 In order to re-
construct his system, we may bring together elements that the philoso-
pher himself did not bring together – for lack of time, or insight, or
whatever. Such a reconstruction is permissible so long as it is con-
sistent with his major views. Only such a picture is worthy of being
examined.
I have brought together theses that Descartes maintained at various
periods, and none of which – at the very least, no central thesis – he
recanted. For example, I have brought to bear on the cogito Descartes’
theory of deduction. The theory was cast in Rules for the Direction of the
Mind, written as early as 1628, perhaps even earlier, never discussed in
detail again, and never published during Descartes’ life. Even before
it was finally published in 1684 in a Dutch translation,2 it had already
profoundly influenced Logic or The Art of Thinking, famously known as
Port-Royal Logic, first published anonymously in 1662.3 The problem I

1 See the Preface.

2 A Latin version appeared in 1701, a better version than the Dutch version in scholarly
opinion, and it is that version that sits in AT. For the backdrop to Port-Royal Logic, see
Appendix C.
3 “In setting out their philosophical foundations, the authors borrow whole arguments
from Descartes’ Rules for the Direction of the Mind, Discourse on the Method, and Meditations,
occasionally acknowledging their source.” Buroker, “Introduction,” xx.

176
 I The Preliminaries 177

have set myself is to determine the truth-value of a counterfactual. It

goes something like this: If Descartes had given us in the first medi-
tation (or in the first three parts of Discourse on the Method) his theory
of deduction, his metaphysical doubts, the doctrine of the creation of
eternal truths, and the theory about memory, could Descartes have
regarded the cogito as an argument? Not just anywhere, but at the right
place: Could he have, for example, regarded the cogito as an argument
at the end of the third paragraph in the Second Meditation? Or at the
end of the fourth paragraph in Part IV of Discourse on the Method?
In section I, I delineate the preliminaries. In section II, I provide
the core argument, the general defense of my central claim about the
cogito. Then, in section III, I prove – I hope I am not overstating my
case by using that word – that Descartes could not claim, and assuredly
ought not to have claimed, that the cogito is an argument. Finally, in
section IV, I offer Descartes’ theory of deduction, and the skeptical
background, to further support this claim.

I. The Preliminaries
Let me begin, then, by noting several things as preliminaries to estab-
lishing the thesis, namely, that the cogito in not an argument. There
are, in all, seven prefatory remarks. First, in the very paragraph be-
fore the one that contains the cogito passage in the Meditations on First
Philosophy, Descartes says, “I will suppose then, that everything I see is
spurious. I will believe that my memory tells me lies, and that none
of the things that it reports ever happened.” (CSM II, 16; AT VII, 24)
So memory cannot be trusted, and the cogito claim, if an argument,
cannot essentially rely on memory.
Second, having just finished the First Meditation (recall that the cog-
ito passage in the Second Meditation is less than four paragraphs away),
“the malicious demon of the utmost power and cunning” is uppermost
in Descartes’ mind. The cogito passages are burdened with it. Thus, the
passage from the Meditations on First Philosophy explicitly mentions “a
deceiver of supreme power and cunning”; and the passage from Dis-
course on the Method refers to “the most extravagant suppositions of the
skeptics.”
Consequently, third, we need to be clear about the task to which
the malicious demon of utmost power, cunning, and deception was set
178 Cogito: Not an Argument

to perform in the First Meditation. For the sake of accuracy, it must

be noted that the malicious demon hypothesis is introduced only in
the final paragraph of that meditation; the earlier supposition was
that an omnipotent God was engaging in an act of deception. But
here the demon hypothesis is substituted for the hypothesis of God,
because God is all-good and cannot deceive; and so the demon is
made to perform the tasks once assigned to the deceiving God. Thus,
the malicious demon is to deceive Descartes into thinking that all of the
sciences that study composite things – physics, astronomy, medicine,
and all other such disciplines – are false; he will deceive Descartes into
thinking that there is a sky, earth, air, colors, shapes, extended things,
size, place, and sound; that he, Descartes, has hands, eyes, flesh, blood,
and senses, when, in fact, he has none of these things at all, nor do any
objects exist. Descartes may have moral certainty (CSM I, 130; AT VI,
37–38) about these things; what he does not have, yet, is metaphysical
certainty.
“What is more, since I sometimes believe that others go astray in
cases where they think they have the most perfect knowledge, may I
not similarly go wrong every time I add two and three or count the
sides of a square, or in some even simpler matter, if that is imaginable?”
(CSM II, 14; AT VII, 21) More significant for our purpose, one of the
principal tasks of the malicious demon is to make Descartes doubt
the truths, even the simple truths, of mathematics – so pervasive is his
hyperbolic doubt. But if the truths of mathematics are to be doubted,
the doubting of the truths of logic cannot be far behind. And it was
notorious that Descartes was very impatient with scholastic Aristotelian
logic.4 At the very least, Descartes’ doubt would have been, at best, an
incomplete doubt if the laws of logic and the principles of inference
were not included in the scope of his doubt.
Fourth, consider this:

Where knowledge of things is concerned, only two factors need to be consid-

ered: ourselves, the knowing subjects, and the things which are the objects of
knowledge. As for ourselves, there are only four faculties which we can use for
this purpose, viz. intellect, imagination, sense-perception, and memory. It is
of course only the intellect that is capable of perceiving the truth, but it has
to be assisted by imagination, sense-perception and memory if we are not to

4 See the last chapter, section II.

 I The Preliminaries 179

omit anything which lies within our power. (CSM I, 39; AT X, 411; see also,
CSM I, 32; AT X, 398–399)

The assistance of imagination and sense perception cannot be wholly

relied upon, since they “botch things” up. Thus, at the start, unless
there are arguments to the contrary, one should rely only on the in-
tellect and memory.
Fifth, in brief, consider Descartes’ account of intuition and
deduction.

By ‘intuition’ I do not mean the fluctuating testimony of the senses or the

deceptive judgement of the imagination as it botches things together, but the
conception of a clear and attentive mind, which is so easy and distinct that there
can be no room for doubt about what we are understanding. Alternatively, and
this comes to the same thing, intuition is the indubitable conception of a clear
and attentive mind which proceeds solely from the light of reason. Because it
is simpler, it is more certain than deduction, though deduction, as we noted
above, is not something a man can perform wrongly. (CSM I, 14; AT X, 368)

Now consider, in brief, Descartes’ account of deduction: “deduc-

tion, by which we mean the inference of something as following nec-
essarily from some other propositions which are known with certainty”
(CSM I, 15; AT X, 369). Descartes’ use of the notion of deduction is
different from ours. We can have a legitimate deductive inference from
false premises; not so for Descartes. For him, the premises have to be
known with certainty. However, there is this in Descartes: “Now none
of the conclusions deduced from a principle which is not evident can
themselves be evident, even though they may be deduced from the
principle in an evident manner” (CSM I, 183; AT IXB, 8). This seems
to hint at the possibility of an inference from false premises; but its
central point is to focus on the unreliability of the conclusion if the
premises are untrue or uncertain.
Descartes’ “method cannot go so far as to teach us how to per-
form the actual operations of intuition and deduction, since these are
the simplest of all and quite basic. If our intellect were not already
able to perform them, it would not comprehend any of the rules of
the method, however easy they might be.” (CSM I, 16; AT X, 372) This
cannot mean that deduction is not analyzable. For when Descartes
speaks of the chain of deduction or inference, he must presuppose an
understanding of intuition without which deduction could not get off
180 Cogito: Not an Argument

the ground. But no understanding of deduction is presupposed in the

understanding of intuition.
Sixth, in an utterly crucial passage in the Rules for the Direction of
the Mind, written in 1628, thirteen years before Meditations on First
Philosophy, Descartes says that it is necessary to distinguish between
intuition and deduction:

There may be some doubt here about our reason for suggesting another mode
of knowing in addition to intuition, viz. deduction, by which we mean the
inference of something as following necessarily from some other propositions
which are known with certainty. But this distinction had to be made, since
very many facts which are not self-evident are known with certainty, provided
they are inferred from true and known principles through a continuous and
uninterrupted movement of thought in which each individual proposition is
clearly intuited. This is similar to the way in which we know that the last link in
a long chain is connected to the first: even if we cannot take in at one glance
all the intermediate links on which the connection depends, we can have the
knowledge of the connection provided we survey the links one after the other,
and keep in mind that each link from first to last is attached to its neighbour.
Hence we are distinguishing mental intuition from certain deduction on the
grounds that we are aware of a movement or a sort of sequence in the latter but
not in the former, and also because immediate self-evidence is not required
for deduction, as it is for intuition; deduction in a sense gets its certainty from
memory.5 (CSM I, 15; AT X, 369–370)

And again:

As we have said, conclusions which embrace more than we can grasp in a single
intuition depend for their certainty on memory, and since memory is weak and
unstable, it must be refreshed and strengthened through this continuous and
repeated movement of thought. (CSM I, 38; AT X, 408)

In a deduction, more than one proposition is involved. Hence,

there is a passage or a movement from one proposition to the next; the
greater the number of propositions, the longer the movement. Where
movement is involved, a thinker must inevitably rely on his memory to
hold the links that bind the propositions together in his mind. But in
an intuition, there is no movement; there is no link; hence, there is no

5 Elsewhere, Descartes speaks of a number of “separate inferences” “between the con-

clusion and the primary and supremely simple proposition” and of the difficulty of
keeping in view “all the connections together” (CSM I, 22–23; AT X, 383–384).
 II The Core of the Claim 181

memory involved. If memory is unreliable, as just noted in the fourth

preliminary point, then deduction can scarcely be relied upon.
Seventh and finally, Descartes was explicitly searching for, and
claimed to have found, a single, solitary, and self-evident principle,
which would serve as the foundation of his philosophy. Here are exam-
ples of such a claim: From the Principles of Philosophy: “Accordingly, this
piece of knowledge [“. . . this inference” – French version] – I am think-
ing, therefore I exist – is the first and most certain of all to occur to anyone
who philosophizes in an orderly way” 6 (CSM I, 195; AT VIIIA, 7). From
the Discourse on the Method: “And observing that this truth ‘I am thinking,
therefore I exist,’ was so firm and sure that all the most extravagant sup-
positions of the sceptics were incapable of shaking it, I decided that I
could accept it without scruple as the first principle of the philosophy
I was seeking” (CSM I, 127; AT VI, 32). Unquestionably, Descartes is
seeking the first indubitable proposition from which the rest would
follow.7 Nowhere in Descartes is the geometrical model more evident.
In short, he claims to have discovered the philosophical axiom. With
this, I bring the seven prefatory remarks to a close.

II. The Core of the Claim

Together, these Cartesian claims make it indisputable that the cogito
cannot be construed as an argument. Why? If the cogito is an argument,
then it should have several elements. In the case of the cogito, there
must, in some versions, be at least three propositions – two premises
and a conclusion – and two links: a link between the premises, and
a link between them and the conclusion. There must be an uninter-
rupted movement of thought from the premises to the conclusion, a
movement where each proposition is clearly intuited. Self-evidence is
not required, but the conclusion must follow necessarily from some
other propositions that are known with certainty. In the case of the

6 Arnauld’s Port-Royal Logic offers this clarification: “Wishing to abbreviate their speech,
people created an infinity of words all signifying both an affirmation, that is, what is
signified by the substantive verb, and in addition a certain attribute to be affirmed.
All verbs besides the substantive are like this, such as ‘God exists,’ that is, ‘is existent,’
‘God loves humanity,’ that is, ‘God is a lover of humanity.’ When the substantive verb
stands alone, for example when I say, ‘I think, therefore I am,’ it ceases to be purely
substantive, because then it is united with the most general attribute, namely being.
For ‘I am’ means ‘I am a being,’ ‘I am a thing.’” (AN, 83)
7 Again, keeping in mind Chapter 1, section III.
182 Cogito: Not an Argument

cogito, this means that “Whatever thinks, exists” and “I think” are prin-
ciples that are true and known with certainty, and that “I exist,” the
inferred proposition, if known certainly, is known only derivatively.
Finally and most importantly, one has to remember the various proposi-
tions and their links: “[E]ven if we cannot take in at one glance all the
intermediate links on which the connection depends, we can have the
knowledge of the connection provided we survey the links one after
the other, and keep in mind that each link from first to last is attached
to its neighbour.” This we must do because “deduction in a sense gets
its certainty from memory.” But at this stage of the argument in the
Meditations on First Philosophy, memory is held to be explicitly unreli-
able, and therefore we cannot trust our memories of various links and
connections. Any argument would be essentially unreliable; at the very
least, it could not be known to be valid.
But isn’t the cogito argument simple enough? Cannot all of it be
held in the mind simultaneously? Isn’t each proposition sufficiently
self-evident to warrant confidence in the validity of the argument?
The answer is no. Descartes gives as an example a simpler argument
that he considers to be an inference. “The self-evidence and certainty
of intuition,” he says, “is required not only for apprehending single
propositions, but also for any train of reasoning whatever. Take for
example, the inference that 2 plus 2 equals 3 plus 1; not only must we
intuitively perceive that 2 plus 2 make 4, and that 3 plus 1 make 4, but
also that the original proposition follows necessarily from the other
two.” (CSM I, 14–15; AT X, 369) (Does not this rely on some general
principle – for example, that two things that are equal to a third are
equal to one another? Is not such a proposition a correlative of “What-
ever thinks, exists”?) But where there is inference, there is movement
of thought; if there is movement of thought, then there must be re-
liance on memory, and since memory is in doubt, the correctness of
the inference is in doubt.8
The foregoing passage also makes it clear that there is a deductive
chain of knowledge. In that chain or hierarchy, certain propositions

8 One can, of course, blandly deny that where there is an inference, there is memory.
One would make such an assertion, because it is the only way to have it both ways:
namely, to maintain that the cogito is an argument and that the inference to the conclu-
sion does not rely on memory. However, we have Descartes’ own words quoted earlier:
“deduction in a sense gets its certainty from memory.” There is no evidence that ‘deduc-
tion’ here refers only to mediate inference, and not to immediate inference as well.
 II The Core of the Claim 183

are self-evident; other propositions rely on these as theorems do on

axioms. If Descartes’ explicit aim is to arrive at a certainty from which,
Archimedes-like, he can move his entire philosophical system, then the
cogito understood as an argument can hardly serve that aim. Something
else – say, the general proposition, “Whatever thinks, exists,” or the
singular proposition, “I think” – must serve as the axiom, and the cogito
as the theorem. There is no warrant in the texts for attributing such a
view to Descartes. Indeed, as we saw from passages in the Meditations on
First Philosophy and the Discourse on the Method, Descartes’ explicit aim
is to find the first principle. Whereas on the view that the cogito is an
argument, the cogito is a philosophical theorem, not a philosophical
axiom.
In any event, Descartes himself settles this point in his “Appendix
to the Fifth Set of Objections and Replies”:
The author of the Counter-Objections claims that when I say “I am thinking,
therefore I exist” I presuppose the major premiss “Whatever thinks exists,”
and hence I have already adopted a preconceived opinion. Here he once
more misuses the term “preconceived opinion.” For although we can apply
the term to the proposition in question when it is put forward without attention
and believed to be true only because we remember that we judged it to be true
previously, we cannot say that it is always a preconceived opinion. For when
we examine it, it appears so evident to the understanding that we cannot but
believe it, even though this may be the first time in our life that we have
thought of it – in which case we would have no preconceived opinion about it.
But the most important mistake our critic makes here is the supposition that
knowledge of particular propositions must always be deduced from universal
ones, following the same order as that of a syllogism in Dialectic. Here he
shows how little he knows of the way in which we should search for truth.
(CSM II, 271; AT IX A, 205–206)

The passages representing the cogito are, in fact, far too complex to
be represented in the way in which they have been represented. Thus,
in the syllogism just cited,
[1] Whatever thinks exists.
[2] I am thinking.

[3] ∴ I exist.

the representation skips the important qualifier. The qualifier occurs

in various forms. For instance, in the Discourse on the Method it occurs
in this way: “I noticed that while I was trying thus to think everything
184 Cogito: Not an Argument

false . . .” In the Meditations on First Philosophy, it occurs as follows: “I am,

I exist, is necessarily true whenever it is put forward by me or conceived
in my mind.”
The syllogism, if one is to be constructed, must take into account
these propositional attitudes. The syllogism would then read:

[1a] Whenever I think, or put forward, or utter, or conceive in my

mind, “I am, I exist,” it is necessarily true.
[2] I am thinking.

[3] ∴ I exist.

This argument is plainly a non sequitur. Since the proposition, “I

am, I exist,” occurs in [1a] in the context of a propositional attitude,
[3] does not follow from [1a] via [2].
The word ‘thought’ in Descartes covers a multitude of mental states.
It is, in most of its uses – and in the cogito passages especially – elliptical.
It stands for “I think x.” Here is a passage from Principles of Philosophy,
Part I:

9. What is meant by ‘thought’. By the term ‘thought’, I understand everything

which we are aware of as happening within us, in so far as we have awareness
of it. Hence, thinking is to be identified here not merely with understanding,
willing, and imagining, but also with sensory awareness. For if I say “I am
seeing, or I am walking, therefore I exist,” and take this as applying to vision
or walking as bodily activities, then the conclusion is not absolutely certain.
This is because, as often happens during sleep, it is possible for me to think
I am seeing or walking, though my eyes are closed and I am not moving
about; such thoughts might even be possible if I had no body at all. But if I
take ‘seeing’ or ‘walking’ to apply to the actual sense or awareness of seeing or
walking, then the conclusion is quite certain, since it relates to the mind, which
alone has the sensation or thought that it is seeing or walking. (CSM I, 195;
AT VIIIA, 7–8)

Thus, whenever “I think” occurs, it is elliptical; it stands for “I think

I am walking,” “I think I am breathing,” “I think I am seeing a tree,”
and so on. The syllogism with premise [1] is thus incompletely stated.
Premise [1] is elliptical; it stands for statements such as “Whatever
thinks it is walking, exists” or “Whatever thinks it is seeing a tree, exists.”
As nearly as I can determine, there cannot be just a thought; it has to
be a thought of something or other.
 II The Core of the Claim 185

In rule 3 in Rules for the Direction of the Mind, Descartes said, “[L]et
us now review all the actions of the intellect by means of which we are
able to arrive at a knowledge of things with no fear of being mistaken.
We recognize only two: intuition and deduction.” (CSM I, 14; AT X,
368) If all our acquisition of knowledge can be explained in terms of
intuition and deduction, and the cogito is unquestionably knowledge,
then how was it acquired? Was it acquired by intuition or by deduction?
We may consider a different perspective by reconsidering the fol-
lowing from an earlier chapter:

Case 4. Second-order mental state

s1 s2 s3 .... . . . . si sn
⊥

sc

Case 5. Second-order mental state

s1 s2 s3 .... . . . . si sn
⊥

sc
186 Cogito: Not an Argument

Now evidently, as Descartes is engaged in the method of doubt, he

cannot be certain of any proposition that he entertains prior to his
establishing the cogito. But granting that the cogito is an argument, one
must also admit that such an argument not only must be valid, but also
must be sound – that is, its premises must be true. If so, Case 4 scarcely
represents the mental process of the thinker. The only alternative left is
Case 5. But here there is an obstacle. Since the cogito is the end result of
the argument, it simply cannot serve as the first item of knowledge that
Descartes had clearly and repeatedly said it was. That privilege must
now be accorded to s 1 or s 2 or s 3 or . . . , but emphatically not to s c .
This is in clear violation of the first criterion that the absolutely certain
truth must satisfy.9
Quite clearly, the cogito cannot be listed as a deduction. It must be an
intuition, something simple, alongside a host of mathematical things
that we clearly and self-evidently see.

III. The Proof

The simplest, strongest, most conclusive argument in favor of my view
that the cogito cannot, and ought not, be construed as an argument,
I have saved for this section. Here is the proof. Descartes’ alleged
argument, when made explicit, runs in at least one common version
as follows:
[1] Whatever thinks, exists. Premise
[2] I think. Premise
[3] If I think, I exist. 1, Universal Instantiation

[4] ∴ I exist. 2, 3, Modus Ponens

Note that [1] explicitly occurs in Descartes in the Principles of Philosophy

and is formulated in this way: “He who thinks cannot but exist while he
thinks.” Such a principle is labeled by Descartes as a common notion,
an axiom, an eternal truth (CSM I, 209; AT VIIIA, 24). I grant that the
argument is valid; I overlook the problem that existence is treated as a
predicate; and I dismiss, as did Descartes, questions concerning how we
can know the meaning of words. The question is, how can Descartes

9 See Chapter 2, section III.

 III The Proof 187

know that this argument is valid? At this point in his philosophical

investigations, he cannot know this.
At this juncture in the Second Meditation in Meditations on First
Philosophy, the demon has not relented. The existence of the exter-
nal world, the deliverance of his senses, whether he has a body, the
principles of mathematics, and by parity of reasoning the principles of
logic and inference are all in doubt. Ergo, Descartes has no principle
of inference to rely on to show that his alleged cogito-argument is valid.
It is of no avail to say: While men may make mistakes, “none of the
errors to which men – men, I say, not the brutes – are liable is ever
due to faulty inference; they are due only to the fact that men take for
granted certain poorly understood observations, or lay down rash and
groundless judgements” (CSM I, 12; AT X, 365). For all that Descartes
knows at this stage of the inquiry, the rule of universal instantiation and
the rule of modus ponens are false or “rash and groundless judgements.”
The demon may well be deceiving him into thinking that these princi-
ples of inference, like the cognate principles of mathematics, are true
when in fact they are not.
Quite simply, knowing that an argument is valid presupposes knowl-
edge of the principles of inference; so if one cannot be sure that one
is not being deceived about the veracity of certain principles of logic
and inference (which show the argument under consideration to be
valid), one cannot know that the argument is valid. If it is true that an
argument “is a machine which often breaks down,” as Descartes wrote,
probably to Silhon, in March or April 1648 (CSMK, 332; AT V,139) –
a passage rarely, if ever, cited in the literature – then in this situation
the thinker has every reason to suspect the validity of the argument,
let alone its soundness. It is indeed the case that Descartes could not,
at this stage of the argument, settle for even that much-despised prob-
able syllogism (CSM I, 11; AT X, 363–364) – watering down his aim
to mere probable belief rather than knowing – for even the correct-
ness of a probable syllogism must presuppose principles of inductive
inference in virtue of which some arguments are more probable than
others.
It is worth noting the context in which Descartes spoke of the ma-
chine of deduction breaking down. Silhon is interested in the nature
of knowledge about God. He distinguishes the nature of that knowl-
edge in beatific vision from our present knowledge of God. The latter,
188 Cogito: Not an Argument

Silhon claims, is intuitive knowledge. Descartes denies this. Descartes

writes:

Intuitive knowledge is an illumination of the mind. . . . Whatever we can know

of God in this life, short of a miracle, is the result of reasoning and discursive
inquiry. It is deduced from the principles of faith, which is obscure, or it
comes from the natural ideas and notions we have, which even at their clearest
are only gross and confused on so sublime a topic. Consequently, whatever
knowledge we have or acquire by way of reason is as dark as the principles
from which it is derived, and is moreover infected with the uncertainty we
find in all our reasoning. (CSMK, 331; AT V, 136–137)

As if this were not enough, Descartes continues in the next paragraph

thus: “You will surely admit that you are less certain of the presence of
the objects you see than of the truth of the proposition ‘I am think-
ing, therefore I exist.’ Now this knowledge is not the work of your
reasoning . . . ; it is something your mind sees, feels, and handles.”
Clearly, then, Descartes knows that when it comes to the cogito he
cannot, should not, rely on principles of logic or reasoning; intuitive
knowledge must suffice. When it comes to God, intuitive knowledge
cannot suffice.10
Consider, next, an extraordinarily strange passage from The Search
for Truth, where he says,

For is there anything in what you have said which is not exact, which is not
validly argued, which is not correctly deduced from what has gone before?
All these points have been stated and worked out not by means of logic, or a
rule or pattern of argument, but simply by the light of reason and good sense.
When this light operates on its own, it is less liable to go wrong than when
it anxiously strives to follow the numerous different rules, the inventions of
human ingenuity and idleness, which serves more to corrupt it than render it
more perfect. (CSM II, 415; AT X, 521)

He had expressed similar reservations earlier in the Discourse on the

Method, Part II. “When I was younger,” he says there,

my philosophical studies had included some logic. . . . But on further exam-

ination I observed with regard to logic that syllogisms and most of its other
techniques are of less use for learning things than for explaining to others the

10 To avoid undue repetition, I urge the reader to consult Chapter 4, note 12, for the
other aspect of the background under consideration here.
 III The Proof 189

things one already knows or even, as in the art of Lully, for speaking without
judgement about matters of which one is ignorant. And although logic does
contain many excellent and true precepts, these are mixed up with so many
others which are harmful or superfluous that it is almost as difficult to distin-
guish them as it is to carve a Diana or a Minerva from an unhewn block of
marble.11 (CSM I, 119; AT VI, 17)

In Part IV of Discourse on the Method, he is even more explicit. Here

is a statement that comes just one sentence before the statement of
the cogito. “And since there are men who make mistakes in reasoning,
committing logical fallacies concerning the simplest questions in ge-
ometry, and because I judged that I was as prone to error as anyone
else, I rejected as unsound all the arguments I had previously taken as
demonstrative proofs” (CSM I, 127; AT VI, 32). Given the systematic
doubt he was engaged in, it would clearly be essential to his enterprise
to discard all of these principles of logic, so that the bad ones did not
stay inadvertently in the good lot and contaminate his reasoning. And
so he decided that “in place of the large number of rules that make up
logic” (CSM I, 120; AT VI, 18), he would follow his famous four rules
of philosophical procedure. But these four rules, as we have seen, are
rules of the method of analysis and synthesis, not rules of inference.

11 The history of traditional logic as it was practiced between 1350 and 1600 is inter-
esting in its own right. It becomes more interesting when viewed from the vantage
point of the topic under discussion. For example, E. J. Ashworth argues that among
the questions raised by logicians during this period were those concerning “the re-
lation of the theory of consequences to syllogistic”; these questions were about “the
status of conversion, the status of consequences in syllogistic reduction and the status
of the syllogism itself” (“Traditional Logic,” 169). For example, (i) Johannes Dolz,
pursuing the standard procedure of the Parisian treatise on syllogism, had offered
an apparent counterexample in his treatment of Barbara (164); (ii) for second and
third figures, in trying to state the rules of conversion the threat of circularity posed
itself (169); (iii) the status of the fourth figure was “the most burning issue” (170);
(iv) the discussion of insolubles slowly dwindled, in part because the logicians of this
day thought them to be merely puzzles. “Although they wrote a great deal about
them,” Ashworth is quoting Spade, “there is no hint that they thought the paradoxes
were crucial test cases against which their whole logic and semantics might fall” (165,
note 145). Finally, (v) conflicting views about the paradoxes of material implica-
tion appeared (168–169). Ashworth’s final lines are worth quoting: “Moreover, many
textbooks simply adopted a position without enquiring whether it could be given
a logical justification. Syllogistic may have been regarded as central to logic, but it
was not always handled well” (172). Assuredly, Descartes could have detailed more
powerful reasons for doubting the system of logic he was familiar with than he did
for doubting the simple truths of arithmetic in the First Meditation.
190 Cogito: Not an Argument

This point gains further credence when connected to the famous

doctrine about the creation of eternal truths. This doctrine was de-
signed to answer the question, can any limitation be put on the powers
of God? The answer: No, God can do anything, including alter the laws
of logic and mathematics. To his longtime friend Father Mersenne,
Descartes wrote on April 15, 1630:

The mathematical truths which you call eternal have been laid down by God
and depend on him entirely no less than the rest of his creatures. Indeed to
say that these truths are independent of God is to talk of Him as if he were
Jupiter or Saturn and to subject Him to the Styx and the Fates. Please do not
hesitate to assert and proclaim everywhere that it is God who has laid down
these laws in nature just as a king lays down laws in his kingdom. (CSMK, 23;
AT X, 145)

Solon, the lawgiver, laid down one sort of laws, just and humane, that
bound and governed the Athenian community. But quite conceivably
he could have changed or altered those laws. He could have given
them a different set of laws, in which, as in the laws of Draco before,
enslavement was permitted for debt, and death was the penalty for
almost all criminal offenses. Now, God is a giver of laws, too. Descartes is
maintaining not merely the more plausible doctrine that God is a giver
of natural, empirical laws that he could change at will. For example,
God might have laid down that force, mass × acceleration, is instead
mass × velocity2 (as Leibniz had conjectured vis viva to be). He is rather
maintaining the doctrine that God is also a giver of mathematical laws,
eternal truths, which he could also alter at will. For example, God
√
might have laid down that the area of a circle, πr 2 , is π i, where i
√
is the imaginary number, −1. His will is absolutely free and is not
subject to anything, let alone to the Styx or the Fates.
In a letter of May 2, 1644, probably to Father Mesland, Descartes
wrote:

I turn to the difficulty of conceiving how God would have been acting freely
and indifferently if he had made it false that the three angles of a triangle
were equal to two right angles, or in general that contradictories could not be
true together. It is easy to dispel this difficulty by considering that the power
of God cannot have any limits, and that our mind is finite and so created as to
be able to conceive as possible the things which God has wished to be in fact
possible, but not be able to conceive as possible things which God could have
 III The Proof 191

made possible, but which he has nevertheless wished to make impossible. The
first consideration shows us that God cannot have been determined to make it
true that contradictories cannot be true together, and therefore that He could
have done the opposite. The second consideration assures us that even if this
be true, we should not try to comprehend it, since our nature is incapable
of doing so. (CSMK, 235; AT IV, 118; see also CSMK, 23; AT I, 145–146; and
CSMK, 25; AT I, 152)

More straightforwardly, in a letter of July 29, 1648, probably to Arnauld,

he wrote:

But I do not think that we should ever say of anything that it cannot be brought
about by God. For since every basis of truth and goodness depends on his
omnipotence, I would not dare to say that God cannot make a mountain
without a valley, or bring it about that 1 and 2 are not 3. I merely say that he
has given me such a mind that I cannot conceive a mountain without a valley,
or a sum of 1 and 2 which is not 3; that such things involve a contradiction in
my conception. (CSMK 358–359; AT V, 224)

If limits are to be placed, therefore, they are to be placed not on the

power of God but rather on the powers of the human mind, which
cannot conceive how 2 and 3 can add up to anything other than 5.12
Pierre Bourdin had objected that even if Descartes could make him
doubt many things, he could not bring Bourdin to doubt the validity
of Barbara. Barbara is a valid syllogism of the form “All M is P ; all S is
M; therefore, all S is P .” Says Bourdin:

Even if you give me every possible reassurance I shall still be very afraid of
overdoing my distrustful attitude if I renounce and forswear as false such
long-standing and virtually innate beliefs as “A syllogism in Barbara has a valid
conclusion.” (CSM II, 317; AT VII, 471–472)

Far from repudiating any such doubts, or invoking the idea that it
is possible for God to alter the laws of syllogistic logic, Descartes
claims that Bourdin is simply fearful of the maxim to doubt, and
fear does not constitute an argument against the maxim. If anything,

12 Rodis-Lewis puts the point thus: “According to Descartes, a mathematics different

from Euclidean geometry is conceivable for God but not for man” (R 104). A note
of caution: Non-Euclidean geometry is at least consistent, whether conceivable by
man or not, in a way in which 2 + 3 = 13 is not. Descartes, clearly, was after a much
stronger point: namely, that what is inconsistent, and not just inconceivable for man,
is conceivable for God.
192 Cogito: Not an Argument

it counts in favor of the maxim, since one has no reason, except

one’s inexplicable fear, to rely on the syllogism in question. Indeed,
Descartes cites Bourdin’s own example back at him: “But a man who
adds two and three together can be deceived by such a demon, as is
shown by the example my critic himself has produced concerning the
man who counted one o’clock four times” (CSM II, 320; AT VII, 476).
Note that there was nothing unusual about challenging the truths
of logic and mathematics. So strong was the skeptical atmosphere with
respect to mathematics in the early part of the seventeenth century
that a demand for the defense of the alleged “queen of the sciences”
seemed natural. Wilhelm Languis wrote De Veritatibus Geometricis in
1656, defending the truths of geometry against the sceptics Sextus
Empiricus and Marcum Meibornium. Marin Mersenne devoted most
of his Verite des Sciences to demonstrating the considerable number and
variety of mathematical truths, as the best means of “overthrowing
Pyrrhonism.”13
This is not to say that Descartes is wholly against the syllogistic form
of reasoning.14 As we shall see, the occasion for using the syllogism
is after the discovery of the cogito has been made and not before. Thus,
Descartes did not want to be seen as relying on the principles of logic
at the time when he was doubting and was in the process of discovering
the cogito.
Next, consider the following fairly self-explanatory passage:

And when we become aware that we are thinking things, this is a primary no-
tion which is not derived by means of any syllogism. When someone says “I
am thinking, therefore I am, or I exist,” he does not deduce existence from
thought by means of a syllogism, but recognizes it as something self-evident
by a simple intuition of the mind. This is clear from the fact that if he were
deducing it by means of a syllogism, he would have to have had previous
knowledge of the major premiss “Everything which thinks is, or exists”; yet
in fact he learns it from experiencing in his own case that it is impossible
that he should think without existing. It is in the nature of our mind to con-
struct general propositions on the basis of our knowledge of particular ones.
(CSM II, 100; AT VII, 140–141; see also CSMK, 330–332; AT V, 135–139)

13 Popkin, The History of Scepticism: From Erasmus to Descartes, 87.

14 CSM II, 355; AT VII, 522. See Chapter 4, note 12, where Descartes’ remark is quoted
in full.
 III The Proof 193

Descartes is saying that [1] cannot be justified unless the truth of

the cogito is already recognized. That is, “I think I exist” must first be
perceived clearly and distinctly in the cogito-state for it to serve as a
support for [1]. The critics then point out that Descartes is involved in
a clear case of petitio principii. This would be true, as Descartes himself
foresaw, if one were offering an argument. So this is another reason
why the cogito should not be considered an argument. Descartes’ view
is in fact much stronger. His view is that it is of our nature, our essence,
that we cannot know any generalization unless we begin with instances.
We cannot know any general proposition without knowledge of partic-
ular ones. Since [1] is a generalization, then essentially it must appear
to us later than, at least, one of its instances.
One final argument: Having established the truth of the cogito,
Descartes asked himself what, if anything, he could believe, given the
cogito-state. His answer was the powerful general rule, namely, that
whatever appears to him as clear and distinct is true. This rule was
to be his epistemic guide; it would provide him the indispensable in-
signia of truth. By implication, Descartes was saying that he would not
accept anything as true that was not clear and distinct. Descartes’ pro-
cedure would force him to measure each premise of the argument by
using the general rule as the yardstick, as well as using that rule to de-
termine if each inference was sound. If the measure provided by the
general rule is satisfactory, the cogito-argument is satisfactory; if not,
then not.
But here is the rub. If the cogito were an argument, then the general
rule could not be used to determine its validity, because the general
rule emerges from the cogito; the correctness of the general rule be-
comes evident when one sees the correctness of the cogito. Therefore,
without substantiating the general rule and becoming certain that it is
correct, the rules of inference cannot be determined to be sound. If
the rules of inference cannot be determined to be sound, the validity
and soundness of no argument can be determined; ergo, the validity
and soundness of the cogito-argument cannot be determined. Conse-
quently, either Descartes has no grounds for accepting the cogito as
even a valid argument, let alone a sound one; or Descartes did not
intend the cogito to be an argument at all.
It is easy to see the difficulty in which Descartes is placed if he is
construed in the cogito passages as offering an argument. The existence
194 Cogito: Not an Argument

of God, let alone the goodness of God, has not yet been demonstrated
in the Second Meditation. That will be demonstrated in the Third
Meditation. Then, given the doctrine of the creation of eternal truths,
God could have created the laws of logic such that in them the principle
of universal instantiation and the principle of modus ponens are false.
If they are, then no matter what Descartes’ finite mind can conceive
or not conceive, he cannot be assured of having found even a valid
argument.
If Descartes had been confronted with the foregoing objections, he
would have granted them. But he would have insisted that his cogito
principle was still indubitably true. He would have said something like
this: “Imagine as strange a logical and mathematical world as you wish,
a world in which the laws of logic and mathematics are vastly differ-
ent and inconceivable to the human mind. Even in this strange world,
where the malicious, powerful demon is deceiving me about the truths
of logic and mathematics, he cannot deceive me unless I exist. That
deception, or any similar deception, in any possible (or impossible?)
world, would guarantee my existence. This is what the cogito aims to
show; the argument that casts doubt on the principles of logic and
mathematics makes it even more evident.” It is only against a back-
ground such as this that we can make sense of Descartes’ otherwise
rather puzzling, and repeated, remark that the cogito is more certain
than the truths of logic and mathematics.
But there is a dilemma facing Descartes that I am unable to resolve.
Either memory plays a role in the cogito-state of a thinker, or it does
not. If memory plays a role, then of course the certainty of the cogito
is spurious, because memory, at the present stage in the investigation,
is not yet known to be reliable. On the other hand, if memory does not
play a role, then in the cogito-state the thinker cannot reliably remem-
ber the earlier skeptical arguments against relying on the truths of
mathematics or the deliverances of sense experience.15 But the failure

15 One might argue that the “state” that is produced by this long process requires of
the thinker that he remember – that he “include” somehow in the present state – the
past experiences he has been through. Must he not remember what the extraordinary
doubt has shown to be doubtful? In short, does not certainty in the present require,
for its possibility, that it include a faithful memory of what has failed to pass muster
in the past and is now seen to be quite different from the present “truth”? But if
memory is doubtful, we are left hanging. I should add to my answer that follows in
 IV Skepticism and the Theory of Deduction 195

to remember the skeptical arguments entails that in the cogito-state the

thinker has, at best, reason only to acknowledge the cogito as a truth –
but hardly as the truth, the first truth, or the truth on which everything
else should rest. On February 22, 1638, Descartes wrote to Vatier: “The
certainty and evidence of my kind of argument for the existence of
God cannot really be known without distinctly recalling the arguments
which display the uncertainty of all our knowledge of material things”
(CSMK, 86; AT I, 560). One might well ask, “Could the certainty and
evidence of the cogito as the first truth really be known without a distinct
memory of the arguments that display the uncertainty of mathematics
and of all our knowledge of material things?” Hardly.

IV. Skepticism and the Theory of Deduction

With an aim to show that it cannot ultimately be done, I wish to
construct a case for reading the cogito both as an intuition and as an
argument. The case rests on one way of reading Descartes’ theory of
deduction. The governing point is that Descartes allows whole deduc-
tions to be intuited. If the cogito is an argument, then nothing prevents
its being intuited in one fell intellectual swoop.16 First, we consider
the type of case in which principles of inference are involved, then the
type of case in which they are not. The case then can be supported
in the following way. A deduction consists of a sequence of steps plus
principles of inference. The final step is the conclusion; the others
are premises. Now, memory is involved as we move from one step to
the next. Since memory is unreliable, we keep repeating, slowly and
carefully, the steps. And if we keep repeating the steps in our minds,
we shall eventually be able to dispense with our memory and see the
whole argument in one intellectual intuition.

the next sentence that (a) at best this concern would show that our not remembering
what the extraordinary doubt has shown to be doubtful would not enable us to claim
that the cogito is the first truth that we have chanced upon; (b) perhaps that Descartes
should have drawn a distinction between a memory that he cannot rely upon as being
veridical and a state in which he has no memory at all; the latter state would be rather
difficult to handle, but a case in which he was not entirely sure of his memory –
in which his memory was merely apparent – would not rob him of knowing of this
certainty in the present; and (c) a lot more needs to be said in order to demonstrate
that certainty in the present requires, even for its possibility, a veridical memory of what
has failed to pass muster in the past.
16 Kenny assures me that this represents his position exactly.
196 Cogito: Not an Argument

Thus, reconsider the following argument:

[1] Whatever thinks, exists. Premise (Intuition 1)

[2] I think. Premise (Intuition 2)
[3] If I think, I exist. 1, Universal Instantiation
(Intuitions 3, 4, and 5)

[4] ∴ I exist. 2, 3, Modus Ponens (Intuitions 6,

7, and 8)

The argument for the case is that, in accordance with Descartes, we

intuit several things: Specifically, we intuit the first proposition (Intu-
ition 1); the second proposition (Intuition 2); the third proposition
(Intuition 3); the first rule of inference, namely, the rule of universal
instantiation (Intuition 4); the application of that rule of inference to
this argument (Intuition 5); the second rule of inference, namely, the
rule of modus ponens (Intuition 6); and the application of the second
rule of inference in this argument, resulting in the fourth proposition
(Intuition 7). Finally, we intuit the conclusion of the argument (Intu-
ition 8). We need to run over these steps again and again, until memory
plays virtually no part and we can finally see the entire argument in a
single intellectual intuition.17
This procedure is valid even if, as has been contended, Descartes’
theory of deduction does not involve (the second type of case) any
explicit or implicit principle of inference. One simply intuits that the
conclusion follows from the premises. Reconsider the following pas-
sage: “Take for example, the inference that 2 plus 2 equals 3 plus 1:
not only must we intuitively perceive that 2 plus 2 make 4, and that 3
plus 1 make 4, but also that the original proposition follows necessarily
from the other two” (CSM I, 15; AT X, 369). To set it up formally:

[1] 2 plus 2 make 4. (Intuition 1)

[2] 3 plus 1 make 4. (Intuition 2)

[3] ∴ 2 plus 2 equals 3 plus 1. (Intuition 3, and then Intuition

4, namely, that the conclusion
follows from the two premises)

17 See Stephen Gaukroger’s Cartesian Logic: An Essay on Descartes’s Conception of Infer-

ence, 50.
 IV Skepticism and the Theory of Deduction 197

Somewhere, Descartes recognizes the law that two things equal to a

third thing are equal to one another, and consequently that law ought
to figure in the foregoing deduction.
What lies in parentheses are justifications: Notice that they contain
nothing beyond intuitions. There are no principles of logic or infer-
ence that occur therein. One goes over these premises and conclusions
again and again until one can grasp the entire argument in one single
intuition. Thus, one might conclude from considering the two types
of cases that the kind of skepticism that I am suggesting does not apply
to Descartes after all. If my argument were to question the reliability
of intuition, then of course the claim of this case would collapse;18 but
then the cogito would collapse as well. The significant thing to note
is that the cogito can now be seen as an argument without falling prey
to the kind of objection that I have leveled against the traditional
interpretation. The latter could be set up formally as follows:

[1] Whatever thinks, exists. (Intuition 1)

[2] I think. (Intuition 2)
[3] If I think, I exist.19 (Intuition 3 that [3] follows
from [1])
[4] ∴ I exist. (Intuition 4, and Intuition 5,
namely, that the conclusion
follows from the two premises)

This case seems to undercut nicely my argument against the tra-

ditional interpretation. If there are no rules of inference, there
is nothing, save intuitions, to be skeptical about. Being skeptical
about intuition is not only utterly un-Cartesian, it is ruinous for any

18 Descartes must suppose, I am claiming, that intuition in particular cases is in fact reli-
able. But must he also suppose that all intuitions are reliable (that he knows a general
principle) without which intuition in particular cases cannot be reliable? Descartes,
I surmise, would answer as follows: Let us suppose that the meditator knows the gen-
eral rule and only on the basis of it makes the claim that this particular intuition is
reliable. “How, then,” Descartes would ask, “would the meditator demonstrate that
the general rule (at this stage in the Meditations) is reliable? Either intuitively or de-
ductively. To demonstrate deductively would be to push the justification a step back;
one would have to justify the premises in that argument. To do so intuitively would
be to beg the question.”
19 Descartes would have contended – as did Port-Royal Logic (AN, 99–101) – that condi-
tional statements in a deduction or demonstration would have to be necessary truths.
For their relevance and importance to the argument here, see Appendix C.
198 Cogito: Not an Argument

interpretation of the cogito. This interpretation seems to be lent further

support by Descartes:

But when we think of the process of deduction as we did in Rule Three, it does
not seem to take place all at once: inferring one thing from another involves
a kind of movement of our mind. In that passage, then, we were justified in
distinguishing intuition from deduction. But if we look on deduction as a com-
pleted process, as we did in Rule Seven, then it no longer signifies a movement
but rather the completion of a movement. That is why we are supposing that
the deduction is made through intuition when it is simple and transparent,
but not when it is complex and involved. (CSM I, 37; AT X, 407–408)

One might say that the cogito-argument is not complex or involved,

but rather simple and transparent, and that, given what Descartes has
just said, it can be known through intuition rather than deduction; it
is not necessarily to be known by a deduction,20 and it does not rely
on memory.21
This objection must be met if my claim is to survive.22 To that end,
first, let us distinguish, as Descartes could have but did not, between

20 “But the deduction seems in all important respects to be simply an intuition, al-
beit an intuition whose content is a relation between other intuitions” (G, 117). A
page later, Gaukroger adds that Descartes “seems concerned above all to restrict
intuition to an absolutely instantaneous act, so that if there is any temporal se-
quence of any kind, no matter how brief, we are dealing with deduction rather than
intuition.”
21 “Memory, in any genuine sense,” writes Gaukroger, “would seem to play no real role
in the deduction from 2 + 2 = 4 and 3 + 1 = 4 that 2 + 2 = 3 + 1. And why does
he specify that remote consequences are known only through deduction? Could it be
that the consequence in the example, which is far from being remote, is known not
by deduction but by intuition? No: it is the example that Descartes himself gives of
a deduction, and the only example at that.” (G, 117–118) If one reads ‘remote’ as
literally as does Gaukroger, it raises an insolvable difficulty. But if one simply reads
‘remote’ as being ‘away from’ the first principles – which is exactly how it is juxtaposed
in the passage – the difficulty evaporates. Only first principles are known by intuitions;
other propositions derived from these first principles, i.e., those propositions that are
away or ‘remote’ from first principles, are known by deduction. Otherwise, one has
to make plausible the claim that Descartes, while claiming intuition and deduction to
be of paramount importance, did not bother to ask the obvious question, “Where do
I place the boundary line that demarcates remote consequences from consequences
that are near?”
22 It may appear that the objection developed here against my view is indeed the one
that Descartes gave to Burman, or something quite similar to it. (CSMK, 334–335; AT
V, 148–149). This is only an appearance. A more perceptive reader will immediately
focus on the following: “And as for as long as he does pay attention to [the axioms in
the proof of God], he is certain that he is not being deceived, and he is compelled
 IV Skepticism and the Theory of Deduction 199

primary intuition and secondary intuition. Consider any of the pre-

ceding arguments. Each is analyzable into several parts, and each part
is known through an intuition. Thus, there are at least four intuitions
involved. These four intuitions must then be coalesced into a single
intuition: The argument as a whole, in accordance with Descartes’
method of analysis, is analyzable into distinct propositions, where each
of the steps of the argument (with or without the rules of inference)
is not further analyzable. The knowing by intuition of atomic proposi-
tions (that is, propositions not analyzable into further propositions)23
I shall call primary intuition. By contrast, any other intuition I shall refer
to as secondary intuition.24

to give his assent to them.” This claims effectively that each axiom is assured of its
indubitability – without the proof of the existence of God – so long as the axiom is
being carefully attended to. What assures the truth of these axioms is, of course, the
general rule, namely, whatever is clear and distinct is necessarily true. This rule is
dependent entirely on the cogito, and not on God. For if it were dependent on God,
the circularity objection would be unavoidable (vide Chapter 4, section II).
23 To be a bit more precise: First, I am not denying that atomic propositions can be
analyzed, only that they are not analyzable into further propositions as compound
propositions can be so analyzed. The last formal argument in the text, for example,
can be thought of as a compound proposition, and it is clearly analyzable into four
other propositions, each of which is to be known independently by an intuition. The
cogito claim, by contrast, is not such a compound proposition. It may have parts – a
subject and a predicate – or it may simply be a concatenation of names, à la Wittgen-
stein, but it does not have as parts other propositions. Second, Descartes has no
theory about what is an atomic or elementary proposition and why it is unanalyz-
able into other propositions. Bertrand Russell says somewhere that when he pressed
Wittgenstein to give him an example of an atomic proposition, the latter replied that
he was not interested in merely empirical matters! Max Black argued that “[w]e can
produce no elementary propositions and would not recognize them if we had them”
(A Companion to Wittgenstein’s Tractatus, 11). Descartes, on the other hand, claimed
that there was an elementary proposition in his system. It was unanalyzable, and it was
recognizable: That proposition was the cogito. So I shall rest with this. If there can be
only propositions that are analyzable into other propositions, then the cogito cannot
be rescued as an argument, because the cogito argument could not ever be stated in
full; each premise of that cogito argument would have to be further analyzable into
other propositions, which in turn. . . . Descartes’ Tree of Philosophy would have to be
uprooted. Third and finally, there might be a persistent worry that Descartes’ intu-
itions are not sufficiently clearly individuated. Notice: Such a difficulty would plague
not only the account of the cogito offered in this book, but any account (since all
accounts must at a minimum invoke intuitions). But insofar as Descartes maintains
that there are atomic propositions (whether he has a theory about them or not),
one might claim that primary intuitions can, indeed, be individuated: In Descartes,
primary intuitions are intuitions about atomic propositions.
24 “Notice, however, that as well as consisting in a grasp of a necessary connection be-
tween two limiting terms, which is what deduction reduces to, intuition can also
200 Cogito: Not an Argument

Second, Descartes draws a distinction between the process of de-

duction and the completed process of deduction. This is a significant
distinction for our purposes, and it seems to me to coincide with the
divide between the method of analysis and the method of synthesis.

consist in a grasp of a single proposition. Later in the same paragraph, Gaukroger

adds: “Although intuitus disappears from Descartes’ vocabulary in his later writings,
this general conception does not, and indeed its crowning achievement is the cogito.
The cogito is effectively an intuition of a basic premiss which, because of its indubitabil-
ity and self-evidence, can be grasped independently of anything else, including rules
of inference. It forms the starting point for knowledge and the paradigm for knowl-
edge in that . . . it is a grasp of a single proposition.” (Cartesian Logic: An Essay on Descartes’s
Conception of Inference, 50–51; my emphasis) The cogito would be a primitive intuition,
on my view.
Judging from the just-quoted passage, one might think that Gaukroger was also
peddling the view that the cogito is not, and cannot be, an argument. But one would
be wrong. Here is the central passage from his later book: “Many commentators have
been misled into thinking that Descartes is maintaining here that the cogito is not
an inference. Quite the contrary, . . . intuitus is the paradigm form of inference. If
the cogito is an intuitus, then it is necessarily an inferential judgment of some kind.
And what makes the cogito so important is that it is the paradigm form of intuitus,
which is in turn the paradigm form of inference.” (G, 341) A bit later, Gaukroger
removes any residual doubt: “There is a widespread view that Descartes thought that
awareness of one’s own mental states was in fact constitutive of the uniqueness of the
human cognition, and this view has been reinforced by an interpretation of the cogito
whereby my grasp of my own existence is an instantaneous act of self-consciousness,
rather than an inference or judgement. We have seen that such an interpretation is
mistaken: The cogito does involve inference and judgment.” (G, 349)
Returning to the earlier work, Gaukroger says, “In construing deduction in terms
of intuition rather than rules of inference, one thing that Descartes is doing is rul-
ing out any attempt at analyzing inferential steps: in the limiting case, there are no
such steps. Inference cannot be analyzed on Descartes’s view because it is simple
and primitive.” (Cartesian Logic: An Essay on Descartes’s Conception of Inference, 51.) One
might then construe inference as being reliable for the following reasons: First, it is
unanalyzable; second, it is known through intuition. One might then think that such
a deduction might be used to construct the cogito as an argument. This, in my view,
would be a mistake. First, let us grant that some particular deduction is primitive,
not derivable from any other. But to come to know that deduction intuitively, one
would antecedently have to know its correct application in at least one instance (for
Descartes, though not for us). However, in that instance, one would have to come
to know at least one true proposition that serves as a premise and from which, using
that deduction, a true conclusion is derived. Thus, knowing intuitively that deduction
would be an instance of a secondary intuition.
Second, suppose that we could know the soundness of the deduction prior to any
application. Why, then, should it, and not the cogito, be treated as the first truth? After
all, if the cogito is an argument, we must first be intuitively assured of the deduction
on which the cogito is based. But the deduction cannot be treated as the first truth
unless we assume that Descartes distinguished between eternal truths: Some (such
as mathematical propositions) he doubted; some (such as rules of inference) he
did not.
 IV Skepticism and the Theory of Deduction 201

One offers a synthesis once the analysis is complete; that synthesis re-
flects the completed process of deduction. Here there is no movement
of the mind, and intuition and deduction can collapse into one. But
insofar as one is doing the analysis, there is a movement of the mind;
memory plays an essential role,25 and we must draw a distinction be-
tween intuition and deduction.
The application of this distinction to the cogito is clear. Insofar as
Descartes has made the discovery of the first truth, analysis is complete;
he can now offer it as an argument, reflecting a completed process
of deduction, and such an argument might be intuited. By contrast,
so long as Descartes is searching for the first truth, analysis is not
complete; there still remains the distinction between intuition and
deduction. Thus, the discovery of the cogito can only be an intuition,
not a deduction.
The key question to be faced, then, is, must the cogito be known only
by a primary intuition? Or, can it, at least, be known by a secondary in-
tuition? The answer must be that it cannot be known by a secondary
intuition. There is a world of difference between knowing the full

25 I take the following to be strongly supportive: “Scholastic thought generally treated

being conscius of something as a time-relative cognitive capacity. Descartes clearly held
this view. He defended this principle: at the time of thinking something, I must be
conscius of so thinking, but I may later forget what I was thinking earlier and thereby
lose this conscientia. Given that conscientia is a form of knowledge and that knowledge
of any kind is a capacity to make judgements, this principle can be rephrased: at the time
of thinking something, I must have the capacity to say what I am thinking, but I may
subsequently, sooner or later, lose this particular cognitive capacity through lapse
of memory.” (Gordon Baker and Katherine J. Morris, Descartes’ Dualism, 109–110)
Explicating a phrase in this passage, Baker and Morris state in note 134, “Arguably
Descartes conceived the present moment to be temporally extended and to be dif-
ferentiated from the past by virtue of the fact that our knowledge of what is present
is independent of memory.”
Consider the cogito to be an argument. Let us say that it is conceived in the present
moment, t, which is temporally extended. Are there parts of that argument about
which similar things can be said? For example, could we say that a premise or a rule of
inference of that argument was conceived in the present moment, t , which is tempo-
rally extended too? (Never mind that the notion of “the present moment” becomes
rather vague, given that t is a subset of t, and yet both are present moments!) If so, is
it – the premise or the rule of inference – liable to be lost through lapse of memory
or to become something about which we may be uncertain, given the unreliability of
the memory? What if the cogito were a nonargument, a statement, conceived in the
present moment, t , which is also temporally extended? In which present moment,
t or t , does a subject run greater epistemic risk? If the answer is that he does so
in t – in that moment, he has more to remember – then Descartes is better off not
making the cogito something that can be conceived only in t.
202 Cogito: Not an Argument

argument intuitively after each of the propositions (including princi-

ples of inference, if any) have been known intuitively, and knowing it
before knowing any of its constituent propositions intuitively. But then
the secondary intuitions presuppose primary intuitions.26 When that
much is granted, we run into an intractable problem for the traditional
interpretation, namely, which of the several intuitions is to be regarded
as the first truth? By the nature of the case, they are all equally good con-
tenders for that position! So even if Descartes would have applauded
G. H. Hardy’s remark – proofs are “gas, rhetorical flourishes designed
to affect psychology . . . devices to stimulate the imagination of pupils” –
this would still not be innocuous for the cogito-argument.
My third argument against this approach is its reliance on an unar-
gued premise that the cogito-argument is simple and not what Descartes
calls “complex and involved.” Immediately following the last quoted
passage, Descartes says that when the argument is complex and in-
volved, “we call it ‘enumeration’ or ‘induction’, since the intellect
cannot simultaneously grasp it as a whole, and its certainty in a sense
depends on memory, which must retain the judgments we have made
on the individual parts of the enumeration if we are to derive a single

26 Consider Descartes’ own example of deduction, the deduction of 2 + 2 = 3 + 1 from

2 + 2 = 4 and 3 + 1 = 4. In order to intuit the first truth, he must clearly first intuit
the truth of 2 + 2 = 4 as well as the truth of 3 + 1 = 4. The latter two would be
examples of primary intuitions, the first would be an example of a secondary intuition.
The inference itself would be a secondary intuition inasmuch as it relies on intuiting
some other truths, not necessarily intuiting some other rules of inference.
For the point in question, among others, perhaps the letter that Descartes wrote
to Huygens on July 31, 1640, is worth close scrutiny: “For I draw a comparison between
my work in this area and the demonstrations of Apollonius. Everything in the latter
is really very clear and certain, when each point is considered separately; but because
the proofs are rather long, and one cannot see the necessity of the conclusion unless
one remembers exactly everything that has gone before, you will hardly find a single
person in an entire country who is capable of understanding them. Nevertheless,
because those few who do understand them vouch for their truth, everyone believes
them. Similarly, I think that I have fully demonstrated the existence of God and
the non-material nature of the human soul; but this depends on several successive
arguments, and forgetting the smallest element will make it impossible to understand
the conclusion.” (CSMK, 150; AT III, 751–752) First, the proofs for the existence of
God and for the immateriality of the human soul are long; second, therefore, the
reliance on memory – which is presently in doubt – is indispensable in Descartes’
two proofs; and third, Descartes explicitly avows that those proofs are dependent on
several successive arguments. This dependence clearly necessitates the drawing of
the distinction between primary and secondary intuitions.
 IV Skepticism and the Theory of Deduction 203

conclusion from them taken as a whole” (CSM I, 37; AT X, 408). Now,

admittedly, Descartes is referring here to induction, and to the kind
of complexity it begets, rather than to the complexity of an argument
with a fairly complex structure. But I think the substance of the claim
should remain unaffected. If so, there would be a loss of certainty,
since, together with other things, “the intellect cannot simultaneously
grasp it as a whole.”27
Fourth, if any argument can be intuited, there is the problem of why
the cogito should be regarded as the “first truth.” For example, with the
cogito providing an essential step in the argument, Descartes is led to
prove the existence of God. Why should not this entire argument be
intuited as well, and be regarded as the first truth? Indeed, some have
thought that we have no truth in the meditations until such time as
the existence of a good and nondeceiving God has been proved28
(C, 95).
What bearing, if any, does the history of logic have on the prob-
lem of interpreting the cogito? Is there, for example, a real historical
background against which we can make our conjecture – that the cogito
should not be regarded as an argument – more plausible? Why is it that
Descartes does not rely, or ought not to have relied, on proof or prin-
ciples of inference? There is such a historical background. There were
two skeptical texts that had been widely known since the sixteenth
century, namely, Cicero’s Academica and Sextus Empiricus’s Outlines
of Pyrrhonism and Against Mathematicians. We may, therefore, assume
that Descartes was familiar with them or had encountered them in
some form, particularly since he wished to be the scourge of skepti-
cism. Descartes had already argued, in conformity with the traditional
criticisms, that the syllogism was question-begging – see, for exam-
ple, rules 10 and 13 of Rules for the Direction of the Mind. What I wish
to do here, briefly, is to set up the skeptical challenge, translate it
into the context of the cogito, and then claim that Descartes could not

27 Can we save the day by distinguishing, with A. K. Stout, between retentive memory
and reminiscing memory? (See “The Basis of Knowledge in Descartes,” 185.) I do
not think so, because none of Stout’s arguments undercuts my fundamental con-
tention that there is an important and irrevocable distinction between intuition and
deduction in Descartes. The essential difference lies in the following: Memory is
indispensable in deduction; it is not so in intuition.
28 See my argument against Curley in Chapter 5, section II.
204 Cogito: Not an Argument

adequately have responded to it with the resources available to him

in the Second Meditation at the point where he had established the
cogito.
In Outlines of Pyrrhonism, in his attack on the Stoic conditional ar-
gument, Sextus asks us to consider the following two arguments,29
Argument [A] and Argument [B]:

Argument [A]: Argument [B]:

[1a] If it is day, it is light. [1b] It is day.
[2a] It is day.

[3a] It is light. [2b] It is light.

Argument [A] is the proof for the inference made in Argument [B].
Now, says Sextus, either “It is light” follows from “It is day,” as in
Argument [B], or it does not. If it does, then the first premise, [1a],
of Argument [A] is redundant. Since redundancy is involved, then by
the light of Stoic logic, the inference is invalid. If “It is light” does not
follow from “It is day,” then the first premise, [1a], of Argument [A]
is false. Thus, the alleged proof will be invalid or will contain a false
premise.
Sextus then goes on to apply similar arguments against the
Aristotelian categorical syllogism. He argues in this way:

[I]n the argument – “The just is fair, but the fair is good, therefore the just is
good,” either it is agreed and pre-evident that the fair is good, or it is disputed
and is non-evident. But if it is non-evident, it will not be granted in the process
of deduction, and consequently the syllogism will not be conclusive; while if
it is pre-evident that whatsoever is fair is also without exception good, at the
moment of stating that this particular thing is fair the fact that it is good is
likewise implied, so that it is enough to put the argument in the form “The
just is fair, therefore the just is good,” and the other premiss, in which it was
stated that the “fair is good” is redundant.30

The skeptical argument, against Stoic or Aristotelian logic, is aimed at

undercutting the possibility of proof.

29 Sextus Empiricus, Outlines of Pyrrhonism, II, para. 159. In what follows I am indebted
to Stephen Gaukroger’s Cartesian Logic: An Essay on Descartes’s Conception of Inference,
especially 11–18; and C, 30–31. I refer the interested reader to the very useful bibli-
ography on ancient skepticism regarding inference supplied by Gaukroger.
30 Sextus Empiricus, Outlines of Pyrrhonism, II, para. 163.
 IV Skepticism and the Theory of Deduction 205

It does not require much effort to apply Sextus’ argument to the

case of the cogito:

Argument [C]: Argument [D]:

[1c ] If I think, then I exist. [1d] I think.
[2c ] I think.

[3c ] I exist. [2d] I exist.

Argument [C] is offered as a proof of the inference made in Argument

[D]. So, Sextus might say, either “I exist” follows from “I think,” as in
Argument [D], or it does not. If it does, then the first premise, [1c ], of
Argument [C] is redundant. Since redundancy is involved, then by the
light of Stoic logic, the inference is invalid. If “I exist” does not follow
from “I think,” then the first premise, [1c ], of Argument [C] is false.
But there is no interpretation I know that maintains that Argument
[D] is valid.31 Thus, the alleged proof will be invalid or will contain a
false premise.32
Or, if the cogito were cast in the mold of Aristotelian logic, Sextus
might have said:

In the argument – “I think, but whatever thinks exists, therefore I exist,”

either it is agreed and pre-evident that whatever thinks exists, or it is dis-
puted and is nonevident. But if it is nonevident, it will not be granted in the
process of deduction, and consequently the syllogism will not be conclusive;
while if it is pre-evident that whatever thinks also without exception exists, at

31 A Kenny-like counterposition would go as follows: “Cogito ergo sum is an argument,

but it is an argument that does not require the use of memory; the following of
sum from cogito is an object of immediate intuition.” Anyone purchasing the just-
mentioned argument from Sextus Empiricus ought to find the foregoing defense
puzzling. First, the validity of the said argument has not been, and cannot at this
stage be, assured. Second, there is no justification for the claim that there is no use
of memory, in contrast to Descartes, who alleges that there is, and must be, the use of
memory in such an argument. Third, nowhere does Descartes distinguish between
“deduction with memory” and “deduction without memory” (without a caveat, such
as “completed process”), let alone provide a justification for the latter. If there were
such a justification, it would clearly be trumped by “self-evidence is not required for
deduction, as it is for intuition; deduction in a sense gets its certainty from memory”
(CSM I, 15; AT X, 369–370); or we would be saddled with the task of explaining
away the inconsistency. Finally, it is difficult to fathom the precise difference, if any,
between the immediate intuition of the Kenny-like argument and the immediate
intuition of a single proposition. See also Appendix C.
32 See also Lisa Jardine, ‘Humanistic Logic,’ 180, as well as the next note.
206 Cogito: Not an Argument

the moment of stating that this particular thing thinks the fact that it exists
is likewise implied, so that it is enough to put the argument in the form,
“I think, therefore I exist,” and the other premise, in which it was stated that
“whatever thinks exists,” is redundant.

It is to be borne in mind that Descartes had discarded the syllogism

as a tool of discovery; perhaps it was satisfactory as a vehicle for ex-
plaining things, but it was worse than useless in enabling us to make
new and fresh discoveries in science or in philosophy. Consequently,
Descartes could not have used the syllogistic approach in his discovery
of the cogito.33
It is not quite to the point that Descartes did not, if he did not,
subscribe to the conditional logic of the Stoics, or the Aristotelian

33 Some social historians might argue that mathematics was actually viewed as very prob-
lematic and doubtful by seventeenth-century skeptics. For instance, it was associated
with magic and cabalism and was “obscure”; to some people who casually picked
up a book, its formulas would resemble weird incantations and magic signs, not to
mention the fact that it was associated with demonic machines. Making mathematics
a paradigm of transparency was a social task. Descartes seems preeminently to con-
tribute to this task, though this involves at the same time discovering truths that are
“more certain” than those of mathematics and even logic. He makes mathematics
nonoccult, one might say, with the help of this theory of intuition and his rejection
of other forms of learning. Now I am no social historian of logic and mathematics,
but given that bit of social history, Descartes would have had even more reason for
doubting the principles of logic than I have indicated.
Let me turn to a historian of logic to support that claim, (that Descartes would
have had even more reason for doubting the principles of logic), this time from the
side of internal history, not social history. “A humanist treatment of logic,” writes Lisa
Jardine, “. . . takes the view, therefore, that any significant study of argument (the
subject-matter of logic/dialectic) must concern itself equally with argument (strictly,
argumentation) which is compelling but not amenable to analysis within traditional
formal logic. It is this fundamental difference of opinion over what is meant by
‘compelling’ argument which accounts for the dogmatic insistence (on ideological
grounds) of the scholastic (and of the historian of scholasticism) that the humanist
is a ‘grammarian’ or a ‘rhetorician’.” (“Humanistic Logic,” 175–176) Or consider
this: “As long as distinguished thinkers like Valla proposed a radically different ap-
proach to reasoning, namely a sceptical acknowledgement that an argument must
always fall short of certainty, and that greater or lesser degrees of likelihood were the
object of the logician’s attention, they could hold their own against their scholastic
adversaries” (“Humanistic Logic,” 192).
The point is that these humanist logicians, with their concern for oratio and
exegesis, need not be right, but that Descartes was required to take these as pos-
sible objections against his favored ideas about logic, consequence, and validity, at
least in the First Meditation, so that by the time he got to the Second Meditation to
display the truth of the cogito, he could not rely, or ought not have relied, on the tools
of logic (which, of course, on my view, he did not).
 IV Skepticism and the Theory of Deduction 207

logic. The point is that Descartes should have entertained the skeptical
arguments on logic very seriously. These arguments did not simply
invoke the evil demon, as Descartes was wont to do with respect to
mathematics. Rather, they were offered as a serious skeptical challenge
to anyone who would invoke the laws of logic or the principles of
inference. At the end of the First Meditation and at the point in the
Second Meditation where he had established the cogito, Descartes had
simply no resources with which to answer the skeptic. Thus, any use he
made of the principles of inference would have been dubious, at best.
I conclude that the traditional interpretation of the cogito as an
argument leaves quite baffling how Descartes could have overlooked
so simple a challenge to his first truth presented as an argument –
or, if he was aware of it, why he insisted on regarding the cogito as an
argument.
There is an interesting conjecture by Harry Frankfurt. The ‘I’ of
the First Meditation, he argues in Demons, Dreamers, and Madmen, is
a naive realist, not our ideal thinker. Consequently, when Descartes
challenges the truths of logic and mathematics in that meditation, he
is challenging those claims as expressed by a naive realist. One might
add they are the claims of a crude empiricist who confuses “2 + 2 = 4”
with an empirical claim, such as “2 apples plus 2 apples are equal to
4 apples.” Frankfurt thinks that Descartes never seriously challenged
the truths of logic and mathematics, because they are clear and distinct
ideas, and such ideas are never challenged by Descartes.
With that as a backdrop, Frankfurt then proceeds to offer one of
the several arguments that can be construed as the cogito-argument.
One can no more doubt the rules of inference, so one might argue
with Frankfurt, than one could doubt the truths of mathematics. This
is the cogito-argument that Frankfurt presents.
There are at least four things wrong with Frankfurt’s view. First, the
truths of mathematics have been challenged everywhere, as a sort of a
prolegomena to the cogito: for example, in the Discourse on the Method,
The Search for Truth, and in Descartes’ letters. Descartes repeatedly
claims that he wishes to establish a truth that is more certain than
the truths of mathematics. Second, the general rule of accepting clear
and distinct ideas as true is not established until the Third Meditation,
hence one cannot invoke that principle anytime earlier than the Third
Meditation. Third, it follows as a corollary that the ‘I’ of the First
208 Cogito: Not an Argument

Meditation is not a crude empiricist, any more than the ‘I’ of the Sixth
Meditation is. Indeed, the conceptual questioning of time, place, and
so on leaves little room to doubt that the kind of thinker that Descartes
has invoked is sufficiently sophisticated to challenge well-established
truths. Fourth and finally, and by far most significant, the principles
of inference are nowhere near established at the time when the cogito
is presented in the Second Meditation. That being the case, the naive
realist has every reason to be skeptical of the truths of logic and the
principles of inference here, for they are no more than rules of thumb,
as in the First Meditation.
I conclude with a small, but not insignificant, textual clue. The first
few pages of the Meditations on First Philosophy consist of the Dedicatory
Letter to the Sorbonne, the Preface to the Reader, and the Synopsis.
A careful examination of these pages reveal that when Descartes speaks
of “demonstrative proofs,” “exact demonstration,” “proof by natu-
ral reason,” “demonstrated by reasoning” – or of truths “proved” or
“demonstrated”34 – he speaks of proofs of only five truths: proof of the
existence of God, proof of the immortality of the soul, proof of the
distinction of the human mind from the body, proof that the exter-
nal world exists, and finally, proof that humans have a body. None of
these phrases referring to proofs is ever used for the cogito. Nor is this
notion of proof in metaphysics isolated from the notion of proof in
mathematics, for twice in the course of these pages Descartes explic-
itly draws a parallel between these metaphysical proofs and proofs in
geometry, and once he draws a parallel between his work and the geo-
metrical writings of Archimedes, Apollonius, Pappus, and others. The
cogito is mentioned precisely once. And then Descartes refers to it as
an “exercise” (CSM II, 9; AT VII, 12), not as a proof. No sooner is that
done than he reverts in the very next sentence to talking about proofs
and compares his method explicitly to that of the geometers: “Hence
the only order which I could follow was that normally employed by ge-
ometers, namely to set out all the premisses on which a desired propo-
sition depends, before drawing any conclusions about it” (CSM II, 9;
AT VII, 13). So the cogito cannot be an argument; it has to be a proposi-
tion. Or, in the language of the geometers – or of seventeenth-century
logicians – the cogito is not a theorem; it is an axiom.

34 For instance, see CSM II, 3, 4, 5, 12; AT VII, 2, 3–4, 12–13.

 7

The Content of the Cogito

The strength of an interpretation lies not only in the interpretation of

a core idea, but also in how well it coheres with Descartes’ other central
theses and ideas. The more that these – interpretation, conjectures,
and consequences – dovetail with one another, make other texts more
plausible, make surprising connections with other issues in other texts
and the claims of contemporary philosophers, make cogent the analy-
sis of the source of errors, make plausible the solution of philosophical
problems, the greater is the strength of that interpretation. This is what
I aim to do in this chapter and the next. I make conjectures, offer vari-
ant readings of Cartesian texts, and draw consequences from claims
thus far developed, in the hope of showing that my interpretation of
the cogito preserves not only much of what Descartes said about the
cogito, but also what he explicitly maintained on other matters.
In section I, I pad up the proof of the core claim, namely, that the
cogito is not an argument, by conjecturing what might have led some
to think that it was. Section II provides, as a follow-up, the outline of a
solution to the problem of the content of the cogito. Finally, section III
attempts to explain why the ‘I’ is elusive – it is because one is seeking
the content of the ‘I’ in the wrong place.

I. A Source of the Mistake

Let me venture a guess about what has perhaps gone wrong
when philosophers have insisted that the cogito is an argument.

209
210 The Content of the Cogito

Someone might have said: Consider the stream of consciousness

of a thinking being. Let us divide this stream into his thoughts:
T1 , T2 , T3 , . . . , Tg , . . . , Tt , . . . , Te , . . . , Tn . The Ts are the thoughts; in
particular,Tg is the thought, “Whatever thinks, exists”; Tt is the thought,
“I think”; and, Te is the thought, “I exist.” We delineate the thoughts
in this way so that we can give a one-to-one correspondence between
the three propositions, P1 (Whatever thinks, exists), P2 (I think), and
P3 (I exist) in the cogito argument and the three thoughts, Tg , Tt , and
Te , respectively. More generally, we map the various thoughts onto the
various propositions that express the content of the thoughts. Then
we get the situation in which we see a movement of thought in that
thinking being: namely, from Tt to Te , or from Tg to Te via Tt . Thus,
we see a movement in thought in the mind of the thinking being cor-
responding to the inference of the argument that we see on paper.
Thus:

Tg is the thought Whatever thinks, exists.

Tt is the thought I think.
Te is the thought I exist.

This entire mode of representing the cogito, by no means explicit, is

wrong. First, this picture is mistaken since it forces the thinking being
to rely on his memory: He must remember not only his thoughts, Tg
and Tt , but also the links between them, in order to ensure that he has
carried out the movement of thoughts correctly to the thought Te .
Second, there is no thought in isolation that corresponds to the
proposition Te , “I exist.” Take the famous wax example. There is a
piece of wax before Descartes. It has a certain taste, scent, color, shape,
and size; it is hard and cold to the touch, and when knocked it makes
a sound. Descartes puts it by the fire and the wax alters radically: the
taste disappears, as do the scent, color, shape, and size. It is no longer
hard and cold; it does not make any sound when knocked. It is never-
theless the same piece of wax. Clearly, Descartes does not know this on
the basis of his senses or by the faculty of his imagination, but “by the
intellect alone” (CSM II, 22; AT VII, 34).
If we were to represent this in terms of the foregoing, we would
begin by enumerating a sequence of properties, something like the
following: P1 (This piece of wax has a pleasant taste), P2 (This piece of wax
has the scent of the flowers), P3 (This piece of wax has an off-white color),
 I A Source of the Mistake 211

and so on up to Pn . Then there will be the familiar corresponding

thoughts. Thus, T1 would correspond to P1 ; T2 would correspond to
P2 ; and so on. Nowhere is there just a thought, C w , corresponding to
the proposition Pw (This is a piece of wax), whereno property of the wax
that the mind perceives with the help of the senses or the imagination
is present. There is always “This is a piece of wax that is white,” or “This
is a piece of wax that is cold,” or “This is a piece of wax that is hard,”
and so on.
Similarly, there is no thought, Te , in the mind of the thinker cor-
responding to the proposition P3 (I exist). “I regard,” wrote Descartes
(quite likely to Mesland) on May 2, 1644, “the difference between the
soul and its ideas as the same as that between a piece of wax and the
various shapes it can take. Just as it is not an activity but a passivity in
the wax to take various shapes, so, it seems to me, it is a passivity in the
soul to receive one or other idea, and only its volitions are activities”1
(CSMK, 232; AT IV, 113–114). Descartes is getting us to see that the
‘I’ already exists in the thought; it could not be otherwise. Properly
reformulated, the suggestion would go as follows:

(T1 + Te) (T2 + Te) (T3 + Te) . . . . (Tg + Te) (Tt + Te) . . . . (Tn + Te)

The Thinker and His Thoughts

Each circle represents the thought in question, the point or the dot
in the circle, •, the thinker of that thought. There is a major difference,
of course, between the wax case and the cogito case. Tw (This is a piece of
wax) can be associated only with other Ts that are about other aspects

1 A nice argument could be developed out of a passage in a letter to Mesland, written on

February 9, 1645; but it is too involved to develop here (CSMK, 241–244; AT IV, 166).
Descartes is speaking of the conditions under which the body of a person is “numerically
the same body, so long as it remains joined and substantially united with the same soul”
(CSMK, 243; AT IV, 166). This would raise the question of the conditions under which
the soul of a person is numerically the same soul. The answer to that question would
be something akin to what now follows.
212 The Content of the Cogito

of the wax pertaining to its color, shape, size, taste, fragrance, and so
on. It cannot be made a necessary part of every thought. By contrast,
Te is a part of every thought, but it is not given to the mind by the
senses or the faculty of the imagination, any more than is the wax. Te
cannot be isolated in pristine purity.
Here Descartes would have agreed with Hume: When he introspects
his mind, he cannot see himself stumbling over a thought that is his
pure self, however hard he might try.

For my part, when I enter most intimately into what I call myself, I always stumble
on some particular perceptions or other, of heat or cold, light or shade, love or
hatred, pain or pleasure. I never catch myself at any time without a perception,
and never can observe anything but the perception. When my perceptions are
remov’d for any time, as by sound sleep; so long am I sensible of myself, and
may truly be said not to exist.2

Hume’s mistake, Descartes would have said, was to assimilate the entity,
‘I’ to the entities observable by sense experience. In fact, the ‘I’ is
cognized “by the intellect alone.” Here is Descartes’ formulation:

For even if all the accidents of the mind change, so that it has different objects
of the understanding and different desires and sensations, it does not on that
account become a different mind; whereas a human body loses its identity
merely as a result of a change in the shape of some of its parts. And it follows
from this that while the body can very easily perish, the mind [“or the soul of
man, for I make no distinction between them” – added in the French version]
is immortal by its very nature. (CSM II, 10; AT VII, 14)

One need not nurture the claim about immortality, as Descartes him-
self did not in the book, in order to maintain the more modest thesis
of the singleness of the mind that persists so long as the body persists.
Thus, in a telling passage in a letter to Mersenne of July 1641,
Descartes wrote:

In the case of the soul the matter is even clearer. As I have shown, the soul is
nothing but a thing which thinks, and so it is impossible for us ever to think of
anything without at the same time having the idea of our soul as a thing capable
of thinking of whatever we think of. It is true that a thing of such a nature

2 David Hume, A Treatise of Human Nature, 165. It is not frequently enough noticed
that Hume was not entirely pleased with his own views on personal identity: See the
Appendix to the Treatise, paras. 20–21.
 I A Source of the Mistake 213

cannot be imagined, that is, cannot be represented by a corporeal image. But

that is not surprising, because our imagination is capable of representing only
objects of sense-perception; and since our soul has no colour or smell or taste,
or anything which belongs to the body, it is not possible to imagine it or form
an image of it. But that does not make it any less conceivable; on the contrary,
since it is by means of it that we conceive all other things, it is itself more
conceivable than all other things put together. (CSMK, 186; AT III, 394; see
also CSMK, 357; AT V, 221)

When Gassendi had objected that Descartes was carrying too much
unnecessary baggage, that he could have done what he wanted to do
by assuming any of his actions, Descartes had replied, “When you say
that I ‘could have made the same inference from any one of my other
actions’ you are far from the truth, since I am not wholly certain of any
of my actions, with the sole exception of thought. . . . I may not, for
example, make the inference, ‘I am walking, therefore I exist’ except
in so far as the awareness of walking is a thought.” (CSM II, 244; AT
VII, 352) In a letter to Reneri for Pollot (April or May 1638), he had
said something quite similar:

When someone says, “I am breathing, therefore I exist” if he wants to prove

he exists from the fact that there cannot be breathing without existence, he
proves nothing, because he would have to prove first that it is true that he is
breathing, which is impossible unless he has also proved that he exists. But
if he wants to prove his existence from the feeling or the belief that he has
that he is breathing, so that he judges that even if the opinion were untrue
he could not have it if he did not exist, then his proof is sound. For in such
a case the thought of breathing is present to our mind before the thought of
our existing, and we cannot doubt that we have it while we have it. To say, “I
am breathing, therefore I exist,” in this sense, is simply to say, “I am thinking,
therefore I exist.” (CSMK, 98; AT II, 37–38)

Thus, the ego cannot be asserted to exist in isolation; it can exist only
in the context of a particular thought.
But then: Assume any thought of a general proposition, Tx , and any
thought of a particular proposition, Ty , which, in conjunction with
the general proposition, tx , is used to derive the cogito argument. Deny
these thoughts. Assume thoughts about any laws of logic and principles
of inference, such as universal instantiation and modus ponens, Tz (both
required for the cogito -argument). Deny them, as assuredly one can,
on the still operating demon hypothesis and in conformity with the
214 The Content of the Cogito

doctrine of the creation of eternal truths. One could, I believe, still

arrive at the truth of the cogito. In short, “I think that ‘Not Tx and Not
Ty ’ and ‘Not Tz ’, therefore I exist,” is as good a cogito proposition as can
be formulated.
Even such a proposition as that would necessarily have been true;
it would have been the simplest; its discovery would not have required
much skill or intelligence; it would have been indubitable; and it would
have served Descartes as a starting point for the rest. But had he offered
an argument, then Descartes would have been guilty of having “built
upon sand instead of digging further down to find hard rock or clay”
(CSM II, 408; AT X, 513).

II. The Content of the Cogito

The point or dot, ‘•’, in the diagram, representing the ‘I’, is convenient
and picturesque, but it is a poor substitute for a philosophical argu-
ment. Gassendi, the author of the “Fifth Set of Objections,” wrote:
“Moreover, when you think you somehow perceive this underlying
‘something’, how, may I ask, do you do so? Do you not perceive it as
something spread out and extended? For you do not conceive of it
as a point.” (CSM II, 190; AT VII, 272) Descartes’ cogito has raised the
problem of the content of the cogito (WC, 103, and for details, 103–
107). What, exactly, does the ‘I’ refer to, and what does “I exist” assert?
The ‘I’ seems to be a descriptionless referent. The statement “I exist”
seems to be a typically empty proposition, devoid of content. If that is
the case, it is argued, since only analytic statements are thought to be
empty or devoid of content, it is unavoidable that “I exist” is an analytic
statement, and hence cannot be an informative statement serving as
the first truth from which other informative truths would follow.
Contemporary metaphysics may provide a clue or two toward a bet-
ter understanding of the content of the cogito. There are two things
that we need to consider: first, the notion of primitive thisness as an
indispensable metaphysical category, and second, the property of
self-ascription. When someone raises the question of the content of
the cogito, he is inevitably working within the Leibnizian framework.
Leibniz held that in the final analysis everything can be defined, with-
out residue, in terms of basic qualities, or primitive suchnesses. From
God’s point of view, every individual in the universe has a certain set
 II The Content of the Cogito 215

of properties such that it, and only it, possesses these qualities. By
contrast, it has been recently proposed that the notion of primitive
thisness might be indispensable, too.
The individual thinker, as he examines the sense content of his
mind, cannot identify himself with any of that sense content. He has
doubts about a myriad of things, but the one thing that he cannot
doubt, while in that mode of doubt, is that he exists. I suggest that
Descartes would have approved of the idea that the ‘I’ cannot be de-
fined in terms of a set of qualities, characteristics, properties, and
so forth, but can only be denoted, referred to, or designated by the
thinker, and that this relation is to be captured by a primitive this-
ness. The point is pithily put in Discourse on the Method, thus: “this “I” –
that is, the soul by which I am what I am –” (CSM I, 127; AT VI, 33).
The ‘I’ in the cogito-state in Meditations on First Philosophy frames it in
this way:

I know that I exist; the question is, what is this ‘I’ that I know? If the ‘I’ is under-
stood strictly as we have been taking it, then it is quite certain that knowledge
of it does not depend on things of whose existence I am as yet unaware; so it
cannot depend on any of the things which I invent in my imagination. (CSM
II, 18–19; AT VII, 27–28)

The ‘I’ is captured by a thisness, on which qualities can then be pegged.

The argument of this section, I believe, entirely squares with one
of the principal theses of Baker and Morris about Descartes’ two-
substance theory and the Aristotelian logical theory that comes at-
tached to it, and it dovetails as well with Descartes’ view of conscientia.3
Quite briefly, the authors argue that the basic logico-metaphysical
structure of Descartes’ philosophical system involves a substance (des-
ignated by the subject term of the Aristotelian categorical proposition)
and a mode (designated by the predicate term of that proposition).
All judgments are of subject–predicate form; singular judgments dif-
fer from both particular and universal judgments; and each judgment
must be uniquely analyzable into subject and predicate.
What I am attempting to discover, in particular, is this: What is this
Cartesian substance designated by “I” in the singular proposition “I am
a being”? Descartes, alas, says very little about it. Let me put my point

3 Baker and Morris, Descartes’ Dualism, especially 59–69, 100–112.

216 The Content of the Cogito

in terms of conscientia. Conscientia is possessing knowledge (scientia) of

the soul and its operations (operationes). My concern is not with how
one acquires conscientia about the operations of the soul – operations
such as, borrowing from Saint Thomas Aquinas, voluntary actions,
acknowledging what we have done, judging what we are to do, making
judgments about our moral character; or, borrowing from John Locke,
operations such as compounding, comparing, and abstracting ideas –
as it is with how one acquires conscientia about the soul. “Conscientia
is knowledge of the soul and its attributes (or its operations),” write
Baker and Morris. “It is self-knowledge, and it is confined to attributes
of the soul (as opposed to properties of the body). Consequently, it
must be manifested in making true singular judgments about the soul’s
operations.”4 This, of course, cannot be read literally. Descartes cannot
maintain that having full and complete knowledge even of all the
operations of the soul would be equivalent to having knowledge of
the soul. His metaphysics, as Baker and Morris might say, would not
permit him to make that move in the language-game. For one thing,
the notion of substance would become redundant. It would be safer to
say that there are two types of knowledge: first, knowledge of the soul,
and second, knowledge of the operations of the soul. I am interested
in the first type of knowledge.
This much is clear: ‘I’ is a mental substance, independent of any
other substance, but God-dependent. This should at least partially ac-
commodate Robert Merrihew Adams’ reservations, to be listed later.5
But these are general characteristics, possessed by all rational beings –
given the absolute use of conscientia6 – and, therefore, would not dis-
tinguish Descartes from any of us.
“In Descartes’ thinking, this agent-centred aspect of conscientia is
given a more rigid form. Given that modes must inhere in substances,
and given that any singular judgement must have a substance as its
logical subject, the logical form of any judgement that exhibits my
conscientia must be that I (the soul) have a mode of thinking.” Having
listed several instances of conscientia, the authors continue, “I can’t
make a true judgement about any mode of thinking ‘that I find within

4 Ibid., 110.
5 See p. 218, this volume.
6 Baker and Morris, Descartes’ Dualism,105–106.
 II The Content of the Cogito 217

me’ unless I think that this mode of thinking is a property of myself

(i.e. my soul). For this reason, I–thoughts must be the minimal moves
in the language-game of exhibiting conscientia.”7
Why is this a problem in discussing the cogito? Here is why. In the
cogito-state, the ‘I’ has no knowledge of any of its operations; the only
thing that the ‘I’ claims to know, in that instant, is that “I exist.” Ex-
isting does not seem very much like an activity of the soul – on a par,
say, with making a promise, dreaming, judging, wishing, and so on. If
the ‘I’, therefore, has no content, it would be very difficult to expli-
cate the content of the cogito. The ‘I’ that is the soul of one person
(using the term in its ordinary sense, not in the Strawsonian sense) is
surely different from the ‘I’ of another person. (Furthermore, since
both are existing, even if existence were a predicate, it would not
distinguish between these two persons). If one could specify the differ-
ence between two ‘I’s in the cogito state, then one could give content
to the ‘I,’ and thence to the proposition “I exist.” But, surely, not
otherwise.
Nevertheless, there is an ‘I’ – “this puzzling ‘I’ which cannot be
pictured in the imagination” (CSM II, 20; AT VII, 29) – a qualityless,
descriptionless ‘I,’ referred to by a thisness. Gassendi objected that
Descartes had told them only what this ‘I’ is not, and not what it is, and
demanded that Descartes give them “a kind of chemical investigation”
in terms of spirits, tartar, the distillate, and other ingredients, which
would manufacture this ‘I,’ what this ‘I’ really was that had the property
to think (CSM II, 192–193; AT VII, 276). Descartes rejoined that it was
the wrong kind of question to ask. It was wrong because it was asking
for a description: something like the proportion of salt, sulphur, and
mercury that went into the making of this ‘I.’ Descartes was, however,
careful enough to point out that “the more attributes of a given sub-
stance we know, the more perfectly we understand its nature” (CSM
II, 249; AT VII, 360). That knowledge of attributes is necessary, for
the knowledge of a substance does not mean that the substance is not
something different and distinct from its attributes. It is different, and
that is Descartes’ view. It was to this that A. J. Ayer objected: “But when
I tell myself that I exist, I do not identify myself by any description:
I do not identify myself at all. . . . There is nothing more to me than

7 Ibid., 110–111.
218 The Content of the Cogito

what can be discovered by listing the totality of the descriptions which

I satisfy.”8 Contemporary metaphysics is more on the side of Descartes
than, understandably, was Ayer.
It does not follow, nor do our intuitions allow, that once we have
divested primitive thisness from primitive suchness, then any suchness,
quality, or characteristic can be pinned to a primitive thisness.

The conclusion, that there cannot be any purely qualitative necessary condi-
tion for the possession of any given thisness, is absurd, however. It implies that
you and I, for example, could have been individuals of any sort whatever –
plutonium atoms, noises, football games, places, or times, if those are all in-
dividuals. If we cannot trust our intuition that we could not have been any of
those things, then it is probably a waste of time to study de re modalities at all.
If there are any transworld identities and non-identities, there are necessary
connections between thisness and some suchnesses.9

Whatever may be the case otherwise, we know that Descartes would

have acknowledged this. He would not have claimed that you and I
could be any of those things just listed: At a minimum, we have to be
thinking beings. In Descartes’ view, there is a necessary connection
between the ‘I’ denoted by a thisness and a set of suchnesses, in par-
ticular, thought. “But what then am I? A thing that thinks. What is
that? A thing that doubts, understands, affirms, denies, is willing, is
unwilling, and also imagines and has sensory perceptions.” (CSM II,
19; AT VII, 28) These are all suchnesses, and no amount of specifying
them in detail would succeed in getting at Descartes’ ‘I,’ any more
than it would succeed in getting at the ‘I’ of you or me. Descartes in
his doubt captures his ‘I’ by a primitive thisness (a thisness that cannot
be defined in terms of suchnesses). He can later peg the various such-
nesses onto this ‘I’: A necessary condition of his being this ‘I’ is that he
must possess the property of thinking. So when someone insists that
we give content to the cogito, we answer that the ‘I’ cannot be defined
away in terms of primitive suchnesses, which is essentially what that
demand amounts to. So it is contentless in the strict sense that it has
not been defined in terms of suchnesses, but this does not warrant the
conclusion that the cogito is therefore without content.

8 A. J. Ayer, “I Think, Therefore I Am,” 85.

9 Robert Merrihew Adams, “Primitive Thisness and Primitive Identity,” 24. See note 5
to this chapter.
 II The Content of the Cogito 219

Let me try a different path. Consider the following from John Perry:

An amnesiac, Rudolf Lingens, is lost in the Stanford Library. He reads a num-

ber of things in the library, including a biography of himself, and a detailed
account of the library in which he is lost. . . . He still won’t know who he is, and
where he is, no matter how much knowledge he piles up, until that moment
when he is ready to say, “This place is aisle five, floor six, of Main Library,
Stanford. I am Rudolf Lingens.”10

The point is that Rudolf Lingens can acquire much by way of knowl-
edge, but at some point self-ascription has to take place; Lingens has
to self-ascribe the property of being in a certain perceptual situation.
David Lewis proposes a more difficult case:

Consider the case of two gods. They inhabit a certain possible world, and they
know exactly which world it is. Therefore they know every proposition that is
true at their world. Insofar as knowledge is a propositional attitude, they are
omniscient. Still I can imagine them to suffer ignorance: neither one knows
which of the two he is. They are not exactly alike. One lives on the top of the
tallest mountain and throws down manna; the other lives on the top of the
coldest mountain and throws down thunderbolts. Neither one knows whether
he lives on the tallest mountain or on the coldest mountain; nor whether he
throws manna or thunderbolts.
Surely their predicament is possible. (The trouble might perhaps be that
they have an equally perfect view of every part of their world, and hence
cannot identify the perspectives from which they view it.) But if it is possible
to lack knowledge and not lack any propositional knowledge, then the lacked
knowledge must not be propositional. If the gods came to know which was
which, they would know more than they do. But they wouldn’t know more
propositions. There are no more to know. Rather, they would self-ascribe more
of the properties they possess.11

Lewis concludes that “[s]ome belief and some knowledge cannot be

understood as propositional, but can be understood as self-ascription
of properties.” To apply this insight to the case of the cogito: At first sight,
one is likely to dismiss it out of hand as being not only implausible, but
also simply false when applied to Descartes. Descartes was looking for
propositional knowledge, and if Lewis is right, then what Descartes has is

10 John Perry, “Frege on Demonstratives,” 474–497.

11 David Lewis, Philosophical Papers, Volume 1, 139.
220 The Content of the Cogito

knowledge of a self-ascription, which is nonpropositional knowledge –

not a first truth.
There are two steps that for a long time, in no small measure owing
to Descartes himself, have been fused as one. There is, first, the non-
propositional knowledge of self-ascription of the property of existing.
Second, and only then, there is the formulation of the proposition
that is to serve as the first truth – namely, “I am thinking, therefore
I exist” is articulated. With only a minor modification, the Lewisian
argument can be made to apply in the case of the cogito. Let us sup-
pose there are two persons, D and D ’, in a possible world, w . They
know every proposition that is true in w . Insofar as knowledge is a
propositional attitude, they are omniscient. Still, I can imagine them
to suffer ignorance: Neither one knows which of the two he is. They
are not exactly alike. Unlike the Lewisian case, in which the two gods
are vastly different and do vastly different things, the two persons in
w are very similar and do very similar things. Indeed, they differ in
exactly one thing. Aside from every other similarity, each of them has
engaged in Cartesian doubt (by the very same route, by the very same
reasoning), but only one of them – say, D – has arrived at the cogito,
“I am thinking, therefore I exist”; the other has not. After the cogito,
they once again do similar things (consequently, in w neither D nor
D ’ has proceeded beyond the stage described in the third paragraph
of the Second Meditation). Neither of them knows who it is that has
arrived at the cogito-proposition, although each of them knows that one
and only one of them has. Until there is self-ascription, there is not,
as Descartes himself had intended, the knowledge of that first truth.
This argument is a much weaker version of Lewis’ argument, and so it
must be correct if Lewis’ argument is sound. Once that first step has
been carried out, the first truth can then be trumpeted out, and the
game of deduction begins – but not until then.
The ‘I’ of Descartes cannot be defined by the essential property
of his mind, namely, thought, for that essential property defines the
mind of, say, Spinoza, as well. The ‘I’ of Descartes could not be the
‘I’ of Spinoza, even if Descartes and Spinoza had in common every
suchness. Imagine a possible world, w 0 , in which there are two men,
D and D ’. Whatever suchness D possesses, D ’ possesses, and vice versa.
Imagine a possible world, w , which is an extension of w 0 , in which only
D exists but not D ’ and another possible world, w ’ in which D ’ exists
 II The Content of the Cogito 221

but not D . Surely, so the argument goes, w is different from w ’: for in

one, D exists but not D ’, and in the other, D ’ exists but not D . Yet,
ex hypothesi, w is identical to w ’ in terms of suchnesses. Two things can
be different without that difference being a difference in suchnesses.
Such is the way in which the Leibnizian principle of the identity of
indiscernible is defeated. It is the thisness that would capture the ‘I’
of Descartes, as it would the ‘I’ of Spinoza; and so to demand that I be
defined in terms of suchnesses is to work implicitly with a disjunction:
Either the ‘I’ in the cogito can be defined in terms of suchnesses, or the
‘I’ in the cogito is without content. There is reason not to accept that
disjunction.12

12 Pertaining to the arguments of the last several pages, John Compton wrote: “My
concern would be better put in terms of ‘what’ it is that the cogito, as an intuition,
reveals. For, as a matter of experiential fact, I can certainly coincide with my activity
of thinking and have the experience of self-existence.” (Quoted remarks of John
Compton in this chapter are from a private communication of June 1, 1999.) I fear
that Descartes did not say much about what that ego is – and given his view, I surmise,
he could not have said more about it – whose existence the cogito is supposed to have
established. But I am reasonably confident that Descartes would have vehemently
denied the possibility that Compton speaks of when he says, “I can certainly coincide
with my activity of thinking,” for reasons very much of the sort Baker and Morris have
advanced. “We cannot,” Compton continues, “leave the matter in the hands of John
Perry and David Lewis – there must be more than a self-ascription or a ‘thisness’
to give content to the proposition ‘I am thinking, therefore I exist.’ We are after
‘propositional knowledge’ after all.” Otherwise, Compton concludes, we seem to be
left “with an unanalyzed ‘subject’ and an unanalyzed process of ‘thinking’.” I take
Perry’s and Lewis’s argument to demonstrate that there is a difference between the
two gods, and that that difference cannot be stated in terms of a description. There
is no propositional knowledge of that difference to be had. I surmise that Descartes
would have accepted that the propositional knowledge he begins with is not a propo-
sition about the nature of that ‘I’ it is rather about the existence of that ‘I’.
“‘Thisness’ just is not enough here,” says Compton. “No doubt the individuality
of the existent self cannot be defined by ‘suchnesses’. But defining is one thing and
characterizing or ‘pointing’ to what we experience as our individual existence is an-
other. There must be some way of characterizing this ‘I’ experientially . . . ‘intuitively,’
and without using an appeal to the senses, . . . Husserl tried to characterize the ego
experientially as the ‘pole’ of conscious ‘intentional’ activity and describes its ‘self-
constitution’ through time. What might have Descartes said about this?”
One might reply as follows: (a) Pointing and thisness are quite intimately linked;
with pointing one might capture the thisness of the object, although capturing the
thisness of the object need not always involve pointing. But characterizing an object
is quite different from pointing to it and seems prima facie to be not different from
describing or listing its suchnesses. Once Compton concedes that individuality may
not be defined by suchnesses, how might suchnesses characterize an object in a way
that enables us to get at what Perry and Lewis want to get at, namely, the individual?
222 The Content of the Cogito

There are two other, and separate, consequences that I wish to

explore. The first is the issue of the identity of the ‘I’ over an interval of
time;13 the second is the importance of the cogito to the general rule via
the case of the wax. Let us begin with the first issue. “But what then am
I?” (CSM II, 19; AT VII, 28) asks Descartes. He answers that the ‘I’ is that
which thinks, and to think is to affirm or deny, to doubt or understand,
to will or otherwise, to imagine or have sensory perceptions, to love or
hate. These are all quite different activities of the ‘I’ and could not be
performed simultaneously: For instance, the ‘I’ could not both affirm
and doubt at the same time. Thus, an interval of time is required for
the performance of these distinct activities. For example, the ‘I’ may
affirm the cogito at one instant, doubt that it has a body in the next, will
to solve a mathematical problem a moment after that, understand its
solution in the next few moments, and so on. That ‘I’ is the “one and
the same I”14 (CSM II, 19; AT VII, 28). By what argument or criterion
is Descartes able to judge that it is one and the same ‘I’?
In the “Sixth Set of Objections,” compiled by Father Mersenne,
we have this: “[W]hen you say you are thinking and that you exist,
someone might maintain that you are mistaken, and are not thinking
but are merely in motion, and that you are nothing else but corporeal
motion” (CSM II, 278; AT VII, 413). But the idea, replies Descartes,
of corporeal motion is essentially different from the idea of thought,
and therefore the two cannot be one and the same. If we are to regard

(b) What is the Husserlian “pole” of conscious intentional activity that is not cap-
tured by the Cartesian ego engaged in thinking, understanding, willing, affirming,
denying, doubting, and so on, which are paradigmatically intentional activities? The
parallel question to ask of Husserl would be, how might one characterize the indi-
viduality of this pole experientially, intuitively, without using an appeal to the senses?
Or, for that matter, the self-constitution of that pole through time? I hope the last
section will go a small way toward further answering some of Compton’s concerns
discussed in this chapter.
13 Descartes had not much to say on the issue of the identity of the ‘I’ over time. At
least since Locke, that question has been of considerable interest. While it is clearly
a deficiency in Descartes not to have raised and answered that question, I am not
sure I would go so far as some who maintain that what Descartes does establish is
consequently not quite important – not to mention the importance of the way in
which he establishes the claim of existence.
14 I agree with Bernard Williams and Margaret Wilson that one needs to distinguish
the autobiographical ‘I’ of the respondent to the objections to Meditations on First
Philosophy (he is Descartes) from the ‘I’ of the investigator in Meditations on First
Philosophy (he could be any investigator).
 II The Content of the Cogito 223

them as one and the same thing, we may do so “either in virtue of

the unity or identity of their nature, or else merely in respect of unity
of composition” (CSM II, 285; AT VII, 423). For example, the idea of
shape is not the same as the idea of motion, but we perceive clearly that
the same substance that is capable of taking shape is capable of motion.
The idea of volition is not the same as the idea of understanding, but
we perceive clearly that the same substance is capable of both volition
and understanding. This is possible in virtue of the unity of nature.
By contrast, the idea of flesh is different from the idea of bone, and
they are found in the same animal, the same substance, in virtue of the
unity of composition. Such, then, is the case with respect to corporeal
motion and thought, we have different and distinct ideas of them, and
they belong to the same substance, not in virtue of the unity of nature,
but rather in virtue of the unity of composition. Finally, says Descartes,
“it is self-contradictory to suppose that things we clearly understand
as different and independent could not be separated, at least by God”
(CSM II, 287; AT VII, 425).
Has Descartes then successfully shown that it is one and the same
‘I’ that underlies the activities of volition, affirmation, denial, under-
standing, perceiving, and so on? I begin with Descartes’ last claim first.
At this stage, the existence of God, let alone of a good God, has not
been proved, and so God cannot be effectively marshaled in the ar-
gument. Thus, for all we know, God could have, but did not, separate
what we perceive to be clearly different and distinct.15 But Descartes’
distinction between the unity of nature and the unity of composition
does not answer the kind of question I am raising, even if we accept
his answer to his critic.

15 Compton suggests an interesting route that the argument might take: “Perhaps it
is that certainty of this identity awaits the certainty of the existence of God. For,
Descartes does hold that God is necessary to sustain each being in its existence from
moment to moment.” Two qualifiers before following this suggestion: First, the issue
is not epistemological in nature; we are not asking how we can know the identity
for certain; it is rather an ontological question about the nature of identity. Second,
while it is true that God preserves or creates (Descartes does not distinguish these
concepts in this context [CSM II, 33; AT VII, 49], each creature must already have an
identity that God preserves or creates. Might Compton then suggest that the question
of the identity of the individual is only settled at that individual’s creation by God,
and in no other way? What then would it be for God to create this individual, or to
re-create this individual from one moment to the next, rather than some other?
224 The Content of the Cogito

Look at it from the point of view of an infant. “I do not doubt that

the mind begins to think as soon as it is implanted in the body of an
infant, and that it is immediately aware of its thoughts, even though it
does not remember this afterwards because the impressions of these
thoughts do not remain in the memory” (CSM II, 171–172; AT VII,
246). As yet, no argument has been established to justify reliance on
memory, so memory cannot tie in a single knot the various activities,
and then hook it onto the ‘I’. What, then? Perhaps, at each moment
in the womb, a distinct infant, housed in the same body, perceives this
‘I’, which is himself. Could there then be different ‘I’s for different
moments, a fresh mind supplied, at each fresh instant, to the same
body? Why is that an incoherent claim?16
To turn to the second issue: The cogito, as I have argued before,
provides the first and sole support for the much-needed general rule,
namely, that whatever is clearly and distinctly perceived is true. If any
subsequent claim must have the general rule’s insignia, in order to
be regarded as true, that insignia will bear marks of the cogito. This
is further supported by Descartes’ view of the content of the cogito
via the wax case. While the wax case is not essential, it is nevertheless
crucial. For with it, Descartes is enabled to make a comparative case.
Thus, the wax that was before cold, hard, white in color, and tasted of
honey, no longer has those properties when placed near the fire; it is,
instead, the wax without the same color, taste, smell, and feel. But the
properties of white color, sweet taste, honey smell, hardness, and so on
are known through sense perception or imagination. If he were to rely
on them, Descartes would have concluded that the wax has changed.
But, of course, it has not; it is “the same wax.” Descartes concludes:
“I must therefore admit that the nature of this piece of wax is in no
way revealed by my imagination, but is perceived by the mind alone”
(CSM II, 21; AT VII, 31).
But with this argument, Descartes wished to establish not just that
human perception requires the human mind, but also this: “Surely
my awareness of my own self is not merely much truer and more
certain than my awareness of the wax, but also much more distinct

16 See Chapter 3, section IV: the argument against Descartes based on his claims about
the Eucharist.
 II The Content of the Cogito 225

and evident”17 (CSM II, 22; AT VII, 33). The general rule claims that
whatever is clear and distinct is true; according to Descartes’ argu-
ment, the cogito-state is more clear and distinct than the perception
of the wax, which is clear and distinct enough; at the very least, he
implies that nothing else is as clear and distinct at this stage. What
degree of clarity and distinctness is required, then, before the general
rule is applied? It is, or ought to be, the highest degree of clarity and
distinctness, the degree noticed in the cogito-state. This, I shall argue
much later, leads to the undoing of Descartes.
Finally, there is a curious bit in the Second Meditation. When
Descartes begins to consider the wax case, he says: “Let us consider
the things which people commonly think they understand most dis-
tinctly of all; that is, the bodies which we touch and see. I do not mean
bodies in general – for general perceptions are apt to be somewhat
more confused – but one particular body. Let us take, for example,
this piece of wax.” (CSM II, 20; AT VII, 30) Not bodies in general, but
this particular body. Not pieces of wax in general, but this piece of wax,
says Descartes. The crucial assumption is that, at this stage of the in-
quiry, a perception of a general body, a general piece of wax, is apt to
be confusing, and so one ought not to rely on it. But this has implica-
tions for general perceptions generally. For note that Descartes speaks
of general perceptions generally as being confusing, and not just of

17 “What could be meant here,” says Compton, “is that not only the existence and na-
ture of our own minds (as thinking things) is ‘much truer and more certain’ than
the existence and nature of the piece of wax, but that the unity of our own minds
across time is even more certain than the unity of that piece of wax. For it is unity
across time, precisely, that is at stake in the case of the wax. And similarly, one could
hazard that it is what is at stake for the mind.
“If this were the argument, we would have something akin to Kant’s argument in
the transcendental deduction, namely, that in order for there to be ‘objects’ for us –
i.e., ‘unities’ of the manifold – there is presupposed the unity (and unifying activity)
of our transcendental ‘I’. And it is precisely, for Kant, the unity across time that is
centrally at issue. Descartes’ argument would be a ‘psychological’ or ‘existential’ ver-
sion of Kant’s ‘transcendental’ one. . . .
“I do not see how Descartes could do without a very strong sense of mental ‘sub-
stance,’ a change-independent substrate, intuitively given with the cogito. . . . ” First,
Descartes rarely, if ever, speaks about the identity of anything over an interval of time.
Second, were Descartes to follow Kant, he would have termed his view “metaphysi-
cal” – much to Kant’s chagrin – rather than “psychological” or “existential.” Third,
Compton, Baker, and Morris are agreed: Substance is an indispensable metaphysical
category in Descartes.
226 The Content of the Cogito

general perceptions about bodies being confusing. “When I reflect on

the cogito,” Descartes might have said, “my focus of attention is solely
on this I . It is not on thinkers and doubters generally, for general per-
ceptions, at this stage of the inquiry, are apt to be confusing. What I see
most evidently is that ‘I am’, ‘I exist’, is necessarily true whenever it is
put forward by me or conceived in my mind. (CSM II, 17; AT VII, 25)
There is no reference here to thinkers and doubters generally, nor
to what would happen to them if they put forward something before
themselves or conceived something in their minds.”18 If this is not an
unfair way of speaking on Descartes’ behalf, it is difficult to see how a
general statement, “Whatever thinks, exists” – the central premise in
the cogito-argument – can find support at this stage of the inquiry.

III. Ryle and the Elusive ‘I’

There are three and a half puzzling pages in Gilbert Ryle’s The Con-
cept of Mind (Chapter 6, “Self-Knowledge”) entitled “The Systematic
Elusiveness of ‘I’” (186–189). Ryle claims to explain therein why the
notion of ‘I’ is systematically elusive, and why there is a partial non-
parallelism between that notion and the notions of ‘you’ and ‘he’. My
interest in Ryle is limited: It aims only at showing that, given Descartes’
argument for the cogito, the ‘I’ is elusive in a significant way that Ryle
fails to notice. If my arguments are sound, it will follow that they also
defeat Ryle’s second claim about the partial nonparallelism between
the direct indexical word ‘I’, on the one hand, and indirect indexical
words ‘you’ and ‘he’, on the other. But I shall not labor to argue for
that separately here.
Let me outline Ryle’s central thesis by beginning with his exam-
ples.19 A singing-master criticizes the accents or notes of a pupil; he
does so by exaggerating each word the pupil sings. Had the pupil
sung slowly, the master could have parodied each word the pupil sang
before the next came to be sung. Then, humbly, the singing-master

18 Gaukroger offers some very interesting reflections on the nature of the self, which is
“a unified locus of subjectivity.” Consult, in particular, the first three of his five points
(G, 346–352).
19 There are at least twenty such examples in the last two sections of this chapter alone,
each example making roughly the same point. The examples I take are ones that
Ryle has discussed in detail.
 III Ryle and the Elusive ‘I’ 227

criticizes his own singing in a similar way; even more, he mimics with
exaggerations each word that he utters, including those that he utters
in parody of his own singing. Needless to add, the singing-master can
demonstrate a parody of a parody of a parody of a parody of . . . his
own singing. Two things are clear, says Ryle. First, the singing-master
can never get beyond the very earliest word of his song; second, at any
given moment he has uttered one note that has yet to be mimicked.
“He can, in principle, never catch more than the coat-tails of the object
of his pursuit, since a word cannot be a parody of itself.” (187).
Similarly – the second example – one may criticize a book in a review;
someone else may review that first review in turn and criticize it, but
the second review is not a criticism of itself. It can only be criticized by
a third review: “any review of any order could be published, though
at no stage would all the reviews have received critical notices” (187).
The last review, in any series of reviews, is, ex hypothesi, exempt from
review; ineluctably, there is always such a review.20
It is not hard to mimic how the argument would go for showing
the systematic elusiveness of ‘I’. Let us describe a situation in which a
person in a soliloquy (silent or verbal, public or private) is engaged in
the task of self-commentary.21 At one time, he makes several comments
about himself. He says to himself, say, that he was boorish to a friend.
A little later, that criticism is itself an object of appraisal: He thinks the
boorishness was in part playacting, because he (the friend) needed to
be taught a lesson (nor was there any other way to help him stiffen
his spine), and thus the person mitigates his self-condemnation. Still
later, this appraisal is itself appraised: He thinks that the animus with
which the act was performed revealed an old, harbored hurt, thus
recognizing the real cause of his action from the apparent one; and so
on. As the soliloquy proceeds, it leads the person to better understand
himself. But what a person knows (in this limited way) about himself can

20 Burman, in his conversations with Descartes, had asked him something astonishingly
similar: “But how can [the mind] be aware, since to be aware is itself to think? In
order to have the thought that you are aware, you must move on to another thought;
but if you do this, you can no longer be thinking of the thing you were thinking of a
moment ago. It follows that you cannot be aware that you are thinking, but only that
you were thinking.” (CSMK 315; AT V, 149)
21 Or self-suasion, self-dissuasion, self-reporting, self-exhortation, self-commiseration,
self-ridicule, self-admonition – all Rylean terms.
228 The Content of the Cogito

in principle be known by anyone else. Just imagine what a person says

of himself, then imagine the other person saying something similar.
Thus, “You were boorish,” “You only meant to teach him to take a clear
stand,” or “You wished to avenge an old, but grievous, insult.”
Now, at no point is the understanding of a person about himself
complete, says Ryle, because the vantage point from which the last crit-
ical appraisal was made is itself not appraised, examined. Of course, it
can in principle be examined, but only from the perspective of another
view, which will remain at that moment unexamined. “A higher or-
der action cannot be the action upon which it is performed. So my
commentary on my performance must always be silent about one
performance, namely itself, and this performance can be the target
only of another commentary. Self-commentary, self-ridicule, and self-
admonition are logically condemned to eternal penultimacy.” (186)
Each of us – no matter how morally perfect, intellectually brilliant, and
psychologically astute – will forever fall short of fulfilling, if only by one
step, the Socratic maxim, “know thyself.” “This, I think, explains the
feeling that my last year’s self, or my yesterday’s self, could in principle
be exhaustively described and accounted for, and that your past or
present self could be exhaustively described and accounted for by me,
but that my today’s self perpetually slips out of any hold of it that I
try to take” (187). One is thus logically prevented by this permanent
penultimacy from knowing oneself; it is not merely something very
difficult to perform or execute. This explains, thinks Ryle, not only
the systematic elusiveness of the ‘I’ but also the nonparallelism be-
tween ‘I’, on the one hand, and ‘you’ and ‘he’, on the other. Thus far,
Ryle.
Would – or should? – Descartes have been moved? Take the review
example. The book under review has content, and, in principle, a
reviewer can latch onto that content and criticize it. Likewise for the
second, third, fourth, . . . nth reviews. Each review has specific content,
and it is open to criticism. But the ‘I’, Descartes would have said, has no
content that can, in principle, be described. It is not the case that the
individual has caught a partial description, which then gets focused
on to reveal some additional descriptions missed in the first round of
descriptions, and then the third round gets even more, has greater
verisimilitude, and so on. Descartes would have claimed (and I take it
that David Lewis would, too) that the ‘I’ is remarkably different from
 III Ryle and the Elusive ‘I’ 229

the sorts of things Ryle has focused his attention on: namely, words,
notes, songs, reviews, and so on.
Put in a Rylean way, Descartes’ point would go something like this. I,
who am engaged in self-commentary, self-ridicule, or self-admonition,
encounter an ‘I’, my I, “this ‘I’ – that is, the soul by which I am what
I am –” (CSM I, 127; AT VI, 33) plus some description. This ‘I’ – “this
puzzling ‘I’ which cannot be pictured in the imagination” (CSM II, 20;
AT VII, 29) – a qualityless, descriptionless ‘I’, referred to by a thisness,22
is spewing out the description in an attempt at self-knowledge. That
description is open to examination, criticism, and commentary, which
in turn is subject to examination, criticism, and commentary, which
in turn . . . and so on. This is entirely susceptible to Rylean analysis.
What never gets captured in the net is the elusive ‘I’. The ‘I’ casts
the net, and the fish that is caught is a thought (which can be de-
scribed); never does the ‘I’ cast a net and catch itself. It cannot, in
principle, do that. “Should I, or should I not, put my knowing self
down on my list of the sorts of things that I can have knowledge of?
If I say ‘no’, it seems to reduce my knowing self to a theoretically
infertile mystery, yet if I say ‘yes’, it seems to reduce the fishing-net
to one of the fishes which it itself catches.” (178) To the “yes” part,
Descartes would have readily assented; to the “no” part, he would have
expressed his puzzlement thus: “It is only a theoretically infertile mys-
tery if you take away what you acknowledged in the ‘yes’ part: A net is
not a fish. The ‘I’ is simply not that which can be captured in any de-
scription beginning with the first description and ending with the last
description.”
Talking in terms of levels (see Figure 7.1), at each level there is
a description, if only a description of an earlier description; no two
descriptions are identical. Specifically, Reviewn is about (↑)Reviewn-1
which is about Reviewn-2 , and so on; at each level there is a review, if
only a review of an earlier review; no two reviews are identical. Like-
wise, Thoughtn is about Thoughtn-1 , which is about Thoughtn-2 , and
so on; at each level there is a thought, if only a thought about an

22 Says Ryle, “philosophers have speculated whether ‘I’ denotes a peculiar separate
substance and in what consists my indivisible and continuing identity” (177). It is
precisely this mistaken search that leads to “a family of enigmas” from which Ryle’s
theory is supposed to free us.
230 The Content of the Cogito

The Book I

Level0 Review0 I Thought0

Level1 Review1 I Thought1

Level2 Review2 I Thought2

Level3 Review3 I Thought3

. .
. . .
. . .
.
. .

Leveln-1 Reviewn-1 I Thoughtn-1

Leveln Reviewn I Thoughtn

The Elusive I
or
Why Isn’t A Book Elusive?

figure 7.1.

earlier thought; and no two thoughts are identical. Comparing the

two columns, one finds that the only constant at each level is the ‘I’,
which is the numerically same ‘I’ at any other level, indicating the
same ‘I’ perpetuating from one level to the next. The ‘I’ referring to
itself is indicated by . Only if we identify the ‘I’ with some descrip-
tion in a manner after Hume, as Ryle does, does Ryle have any force
against Descartes’ argument (but then Ryle’s view has some enigmas
of its own to face, as we shall see) and make the ‘I’ column in the
chart disappear. Descartes attempted to show that to make any such
 III Ryle and the Elusive ‘I’ 231

identification – even a partial identification – is to look in the wrong

place, to make an error about the content of that ‘I’.
Look at it this way. The ‘I’ and the book are remarkably different –
for one thing, no one speaks of the book as elusive; and so what Ryle
says about the book and its reviews cannot apply to the ‘I’ without
qualification. Consider: Each review of the book is a different and
distinct entity; it is assuredly not a continuous – indeed, not even a
discrete – part of the book, whereas the book provides the nucleus
that assures continuity over reviews about reviews about reviews . . . of
this book. By contrast, consider: Each December 31, a person reviews
his old self; and he does this, say, over a ten-year period. Each such
“review” is not only about himself but also a critical “review” of his
earlier “reviews” of himself. But these “reviews” – unlike the reviews of
the book – are not about distinct persons (each “review” representing
a different person), for no other than a good Rylean reason:

As if an ordinary person is really some sort of committee or team of persons,

all laced together inside one skin; as if the thinking and vetoing ‘I’ were one
person, and the greedy or lazy ‘I’ were another. But this sort of picture is
obviously of no use. . . . So the suggested reduction of a person to a team of
persons would merely multiply the number of persons without explaining how
it is that one and the same person can be both narrator and auditor, or both
vigilant and dreamy, both scorched and amazed at being scorched. (181)

But if there is no ‘I’ that assures continuity between these “reviews”

spread out over a ten-year period, what does?
Reexamine this crucial passage: “This, I think, explains the feeling
that my last year’s self, or my yesterday’s self, could in principle be
exhaustively described and accounted for, and that your past or present
self could be exhaustively described and accounted for by me, but that
my today’s self perpetually slips out of any hold of it that I try to take”
(187). Thus: I can know my past self in its entirety, but never my present
self in its entirety; on the other hand, I can know your past self in its
entirety and your present self in its entirety; and vice versa. At any given
moment, I am always partially inscrutable to myself; you are never, in
principle, inscrutable to me. To me, therefore, you are like a book;
unto myself I am no such thing.
The Rylean view has force only because it denies that ‘I’ is different
from the set of descriptions one might give of that ‘I’ or self; this is
232 The Content of the Cogito

precisely why I am able to know the whole of you without remainder but
fail in the attempt to know the whole of me. Why? Because any attempt
on my part to know that ‘I’ changes the very ‘I’ that I am attempting
to know. The knower and the known are never identical in my case.
For obvious reasons, too, this would render each ‘I’ at each level a dif-
ferent ‘I’. Because the ‘I’ at one level – say, level3 – investigating what
the ‘I’ is at level2 already has a thought that the ‘I’ at level2 does not
possess. If the time interval between these levels – between thoughts
and thoughts about thoughts – is shortened, the problem becomes ex-
tremely acute: We have a series of ‘I’s rapidly coming-into and going-
out-of existence. But there is simply no reason to think that the two ‘I’s,
in so short an interval of time, are numerically different and distinct
(while granting that the descriptions that the two ‘I’s spew out may
be remarkably so). This is Heraclitean flux with a vengeance. With-
out wishing to be a paradox monger, could ever an ‘I’ witness such a
flux?
If the foregoing argument is correct, then other enigmas follow.
First, there is no reason to think that I can capture in terms of a
description your puzzling ‘I’, any more than I can capture my own.
What is more, second, I know what it is for me to have a privileged
access to this ‘I’ that is mine, but I have no access to what you would
indexically designate as the ‘I’ that is yours. This should undermine
Ryle’s claim that

[n]o metaphysical Iron Curtain exists compelling us to be for ever absolute

strangers to one another, though ordinary circumstances, together with some
deliberate management, serve to maintain a reasonable aloofness. Similarly
no metaphysical looking-glass exists compelling us to be for ever completely
disclosed and explained to ourselves, though from the everyday conduct of
our sociable and unsociable lives we learn to be reasonably conversant with
ourselves. (173)

I have direct access to that ‘I’ of mine, to which you have none, and
vice versa; to this extent, if no more, we are absolute strangers.
Third, given the fallibility of first-person reporting, no less than of
third-person, Ryle concludes that self-consciousness does not have the
role that Cartesians attach to it. “Self-consciousness,” says Ryle, “must
not be described on the hallowed para-optical model, as a torch that
illuminates itself by beams of its own light reflected from a mirror in
 III Ryle and the Elusive ‘I’ 233

its own insides” (186). This is just mockery. Descartes allowed for the
possibility, even in the case where the cogito is established, that a thinker
may well be mistaken in his particular thought. Quite simply, the evil
genius can deceive me about what I am currently thinking. There is not
a solitary passage in Descartes suggesting that we are like metaphysical
looking glasses unto ourselves, forever completely disclosed and ex-
plained to ourselves. He would have insisted only that, given the proper
thought experiment, I cannot be mistaken in the belief that I exist
while I am thinking an admittedly fallible thought. Perhaps, then, self-
consciousness – as the word indicates, or as symbolizes – may be
required only to demonstrate the existence of the self, and nothing
more. That is, it is not required to serve as a surety, a guarantor, an
underwriter for the veracity of any thought that it is currently casting
its light on.
Fourth, the ‘I’, the thinker, has a property of self-consciousness not
possessed by its thoughts: I am aware of having a thought, my thought
is not aware of my having this thought or any other; it is just a thought.
There is nothing more to a book than the information encoded in its
inscriptions; the book itself has no capacity for self-reference; it cannot
behold its contents. One can, of course, imagine a book replete with
self-referential sentences and sentences that refer to other sentences;
it may even have a humorous self-referential title, such as the title of
Raymond M. Smullyan’s book,What is the Name of This Book? But one
cannot, per impossible, imagine a book that itself has any such property as
self-consciousness. On the other hand, the qualityless ‘I’ that Descartes
speaks of has just that property; and nothing else in the universe – save
a thinking thing – has that property.
The fifth and final argument: Ryle says, “For any performance of
any order, it is always possible that there should be performed a vari-
ety of higher order actions about it” (185). The emphasis is mine, for
I wish to indicate that it is not necessary that one perform a higher-
order action. Let us, then, suppose that a person, say Gilbert Ryle,
wishes to understand himself. He goes through reviews upon reviews
of himself (reviews of his own making about himself ), and finally he
comes to a halt at some stage. Let us call this the Ground State. Here
is the crucial step: Either it is possible that the Ground State is trans-
parent to Ryle, or it is not possible. To say that a state is transpar-
ent is to make an epistemic claim, but not a claim about epistemic
234 The Content of the Cogito

infallibility. It is a state in which a person, say Ryle, sees no epis-

temic need to evaluate, rate, rank, weigh, classify, pigeonhole, cod-
ify, catalog, measure, reflect upon, expound upon, comment upon,
define, explain, or criticize the state; he understands the state, he
thinks, perspicuously. Of course, he could be mistaken. On the other
hand, when in the cogito-state he is making the cogito-claim; he is in an
infallible state and knows himself to be in such a state, but he need
not be in the cogito-state to be in what he thinks to be a transparent
state.
Now consider the first alternative: If the Ground State is transparent
to Ryle, it would circumvent any need on Ryle’s part to move to a higher
stage, the better to know himself. As we saw, it is not necessary that Ryle
perform a higher-order action. Ryle could, in principle, know all of
the states of Ryle that in their totality constitute what can be described
about Ryle, including the Ground State, which is transparent to him,
and no state would remain unknown (ex hypothesi, there being no state
to know, in particular none from which alone the Ground State can be
known, evaluated, measured, and so on). If this is possible, Ryle has
failed to explain the elusiveness of ‘I’. It is false for him to say: “This, I
think, explains the feeling that my last year’s self, or my yesterday’s self,
could in principle be exhaustively described and accounted for, . . . but
that my today’s self perpetually slips out of any hold of it that I try to
take” (187).
No sooner is this said than the reader will instinctively revert to the
book-and-reviews case just presented and protest that there cannot
be, on Ryle’s view, such a transparent Ground State. Let us, then, con-
sider the second alternative. If there cannot be a transparent Ground
State, then Ryle is mistaken when he says that “your past or present
self could be exhaustively described and accounted for by me” (187).
And the reason is this. Consider the case in which it is not Ryle who
is trying to know Ryle, but I who am trying to know Ryle. Surely, in
order to exhaustively describe Ryle as Ryle says that I could – “your
past or present self could be exhaustively described and accounted
for by me” – I would have to be in some Provisionally Basic State, a
state in which I could describe, among his other states, Ryle’s own
Provisionally Basic State. Now, if that Provisionally Basic State of mine
is not transparent to me, I will never be epistemically sure whether
I have really understood Ryle, and I will push further to evaluate
 III Ryle and the Elusive ‘I’ 235

my own state of understanding of Ryle’s states; and each such state

of understanding of Ryle’s states being fallible, it will push me to
perform higher-order actions. But if I cannot be in a transparent
state, ex hypothesi, then surely I can no more know Ryle than he can
know himself. There is more symmetry between ‘I’ and ‘he’ than Ryle
acknowledges.
In conclusion: Either I can know myself, rendering the ‘I’ nonelu-
sive in cases where Ryle himself thinks that it is and ought to be elusive,
or I cannot know myself, for the same reasons that you cannot know
me either, thus making the ‘I’ elusive even in those cases where Ryle
thinks that it is not. The Rylean way ineluctably keeps us from seeking
the content of the ‘I’ in the right place.
 8

Memory, Explanation, and Will

The aim stated in the last chapter is pursued further in this final chap-
ter. That aim is to demonstrate how interpretations of various philo-
sophical theories of Descartes cohere with the core interpretation that
the cogito is not an argument. Three additional such philosophical
theories of Descartes are examined here, theories that have no overt
connection with one another – namely, his theories about memory,
logic and explanation, and will. I trust that their mutual indepen-
dence (something akin to independent testability) will strengthen my
interpretation of the cogito.
In section I, Descartes’ distinction between two types of memory is
drawn; this distinction is then examined, and it is shown how it lends
added support to the core argument. Now, Descartes had interesting
things to say about the nature and logic of discovery and explanation,
and he prided himself on differing from the traditional study of logic.
When we take this aspect of Descartes’ thought into consideration –
section II is designed for that purpose – it further strengthens the
central thesis. Finally, in section III, I analyze Descartes’ theory of
the will – not all aspects of it, but just that aspect that contributes to
the central thesis of this book.

I. The Role of Memory

There are two types of memory, according to Descartes. I shall call them
physical memory and intellectual memory, respectively. In the terminology

236
 I The Role of Memory 237

of Descartes’ posthumously published Treatise of Man, “the seat of mem-

ory” lies in the internal part of the brain. After leaving a gland, the
spirits pass through the tubes and into the pores of the filaments that
compose the solid part of the brain. Depending on the force of the
spirits and the degree of openness of the tubes, the spirits will expand,
bend, and rearrange the filaments they encounter. At first, the pas-
sage of the spirits will be difficult; but with time, intensity, frequency,
and repetition of the encounters, the passage will become increasingly
easier. They will then leave a pattern – an imprint, a trace, a crease – in
the brain that cannot be easily erased and that can easily be activated
without the presence of the relevant corresponding object. “And it
is in this that memory consists.”1 Now, oddly enough, Descartes thinks
that memory is located not only in the brain, but outside of it as well:
“[W]hat people call Local Memory is outside of us.” 2 For example, a
lute player has a part of his memory in his hands; some of our recol-
lection of a book is not in our heads but “on the paper of the copy
which we have read” (CSMK, 146; AT II, 48). This is physical memory.
“But,” says Descartes, “besides this memory, which depends on the
body, I believe there is also another one, entirely intellectual, which
depends on the soul alone”3 (CSMK, 146; AT II, 48). This is intellectual
memory.
Descartes often bemoans the weakness and unreliability of memory.
If the cogito were an argument, then the mind would have to traverse
the various premises to the conclusion, and see and remember the several
premises and the conclusion and the links between various proposi-
tions in the argument. Let me quote at length:
[D]eduction sometimes requires such a long chain of inferences that when we
arrive at such a truth it is not easy to recall the entire route which led us to it.

1 Descartes, Treatise of Man, 87–88. Also see CSMK, 143; AT III, 19–20; and CSMK, 233;
AT IV, 114.
2 Descartes, Treatise of Man, 71.
3 Curiously, Descartes maintains in a letter, dated June 4, 1648 probably written to
Arnuald that intellectual memory “was not in use in the womb” (CSMK, 354–355;
AT V, 193). I should point out that the original translation in Kenny is much stronger;
it reads: “intellectual memory, which was not possible in the womb.” But the infant in
the womb has thoughts, and so it has a soul. Intellectual memory, we were just told,
depends on the soul alone, and the soul has all the innate ideas even then. So it would
follow that something more would be required for intellectual memory in the womb
to commence operating. What, then, is it? Is one to infer that the infant in the womb
cannot engage in the cogito? See also CSMK, 356–357; AT V, 220–221.
238 Memory, Explanation, and Will

That is why we say that a continuous movement of thought is needed to make

good any weakness of memory. If, for example, by way of separate operations, I
have come to know first what the relation between magnitudes A and B is, and
then between B and C, and between C and D, and finally between D and E, that
does not entail my seeing what the relation is between A and E; and I cannot
grasp what the relation is just from those I already know, unless I recall all of
them. So I shall run through them several times in a continuous movement of
the imagination, simultaneously intuiting one relation and passing on to the
next, until I have learnt to pass from the first to the last so swiftly that memory
is left with practically no role to play, and I seem to intuit the whole thing at
once. In this way our memory is relieved, the sluggishness of our intelligence
redressed, and its capacity in some way enlarged. (CSM I, 25; AT X, 387–388)

The central question is, which, (if either) of the two memories –
physical memory or intellectual memory (or both) – can the thinker
rely on while he is in the process of getting to the cogito through the
method of doubt? Self-evidently, it cannot be physical memory, either
internal or external, since this kind of memory presupposes the existence
of physical objects, such as the brain of the thinker, the hands of a lute
player, and the pages of a book. If the thinker doubts the existence of
physical objects, as he does in his hyperbolic doubt, then he has little
reason to believe in the existence of physical memory, and even less
to rely on its alleged dicta.4
But intellectual memory cannot be relied upon either, for it pre-
supposes the existence of the soul, which is only established simul-
taneously with, or subsequent to, the cogito. So if the cogito were an
argument, the thinker at this stage in his inquiry would have no reason
to rely on intellectual memory.5 Therefore, the thinker cannot rely on

4 “But memory is no different from imagination – at least the memory which is corporeal
and similar to the one which animals possess. So we can conclude with certainty that
when the intellect is concerned with matters in which there is nothing corporeal or
similar to the corporeal, it cannot receive any help from those faculties.” (CSM I,
43; AT X, 416) Imagination is notoriously unreliable, therefore physical memory is
unreliable.
5 Nor will the following defense do: Intellectual memory is a function of conscious-
ness, and remains so whether one answers the ontological question of the ground
of memory. And as a function of consciousness, it is certainly available to be used by
the mediator in bringing together the various steps of an argument. The reliability of
memory is an epistemological question, and justifying its ontological ground in the
existence of the soul is irrelevant to deciding whether it is reliable and available to be
used by consciousness in its reasonings. There are two responses: First and foremost,
intellectual memory is notoriously unreliable, and so Descartes questioned relying on
 I The Role of Memory 239

memory, and hence cannot judge even the validity of the argument,
since that would presuppose the memory of the premises and the rules
of inference that have been invoked. I conclude that intuition, and not
deduction, is the only thing that Descartes can rely on at this stage of
the inquiry. Consequently, the cogito is a simple proposition and not
an argument.
“This overlooks the fact,” someone might protest, “that Descartes, in
the passage just cited, does in fact speak of the possibility of a deduction
being intuited, if one went over it several times. There is an analogy that
one can draw from vision. When I look out from my study window, I can
see, at a single glance, the white mailbox, the road, the lawn, the red-
brick house across the street, the lamppost, and the massive oak tree. I
do not look, as a matter fact, at each of these objects one at a time and
then put the whole thing together. Rather, I take in the whole scene at
once. Similarly, if by repeated practice and learning I could take in the
whole deduction at once, and not in stages, I would not have to look
at one proposition at a time, at each rule of inference at a time, and
then slowly and carefully put the whole argument together. I could
take in the entire cogito -argument at a single intellectual glance, in
what Descartes calls intuition. Thus, even if memory is unreliable, it is
quite dispensable. The cogito could be an argument, and the argument
could be intuited.” So someone might say.
There are several responses. First, the role of memory cannot be got
rid of in principle. One can reduce its role, but one cannot eliminate
it. As Descartes himself says, only after “I have learnt to pass from the
first to the last . . . swiftly,” is “memory . . . left with practically no role to
play.” But practically no role to play is not the same thing as no role
to play. There is an implied lemma in the foregoing: Memory has no
role to play in an act of intuition.
Second, the analogy with ordinary vision would not sit very well with
Descartes:
We can best learn how mental intuition is to be employed by comparing it with
ordinary vision. If one tries to look at many objects at one glance, one sees
none of them distinctly. Likewise, if one is inclined to attend to many things

it even in the opening paragraph of the Second Meditation. Second – and admittedly
this is bold – if the fundamental root (the soul) of the thing (intellectual memory) can
be held suspect, or if its nonexistence is seen as a distinct possibility, that should cast
doubt even on the existence of that thing, with whatever nature it is alleged to possess.
240 Memory, Explanation, and Will

at the same time in a single act of thought, one does so with a confused mind.
Yet craftsmen who engage in delicate operations, and are used to fixing their
eyes on a single point, acquire through practice the ability to make perfect
distinctions between things, however minute and delicate. The same is true
of those who never let their thinking be distracted by many different objects
at the same time, but always devote their whole attention to the simplest and
easiest of matters: they become perspicacious. (CSM I, 33; AT X, 400–401)

Descartes is maintaining that one ought not to “be distracted by

many different objects at the same time, but always devote [one’s]
whole attention to the simplest and easiest of matters.” This advice is
not just for craftsmen, but for knowledge seekers as well. Correspond-
ingly, Descartes would maintain that one ought not to be distracted
by many different propositions at the same time, but instead should
always devote one’s whole attention to the simplest and easiest propo-
sition. At least, one should do this at the beginning. “[T]wo things
are required,” says Descartes, “for mental intuition: first, the proposi-
tion intuited must be clear and distinct; second, the whole proposition
must be understood all at once, and not bit by bit” (CSM I, 37; AT X,
407). But if one were attempting to intuit the whole argument, one
would at some point in the beginning inevitably consider it bit by bit,
proposition by proposition, in order to ascertain the argument’s valid-
ity and soundness.
But, third, there is a more interesting reason why the analogy fails.
When I look out of my study window and see the various objects, they
do not depend on one another. If the oak tree were cut down, the
house would still remain; if the house were completely remodeled,
the mailbox would still remain; if the mailbox were painted brown,
this would have no effect on the red-brick house across the street.
But the propositions in the cogito -argument are essentially linked with
respect to their content, structure, and truth-value. If any of these latter
characteristics of the premises were to change, the argument might no
longer be valid or sound.

II. Discovery, Explanation, and the New Logic

This section is not simply a very terse note on the history of logic.
It is also meant to show the implications of Descartes’ view of logic,
explanation, and discovery for the cogito.
 II Discovery, Explanation, and the New Logic 241

Descartes outlined a new logic; or, at least, he believed that he

had one. Antoine Arnauld and Pierre Nicole, the joint authors of
La Logique, ou L’Art de Penser, famously known as The Port-Royal
Logic, agreed. The old scholastic logic concentrated on a theory of
communication of settled knowledge. It was designed to explain known
truths. The new logic aimed at providing a means for guiding reason
in the acquisition of knowledge. It was designed to discover new truths.
The cogito, construed as an argument, is patterned after the old way;
the cogito, construed as a nonargument, is in harmony with Descartes’
new way. Or so I shall argue.
Pierre de la Ramee, known as Peter Ramus, proposed in his master’s
inaugural thesis in 1536 a bold thesis: Quaecumque ab Aristotle Dicta
Essent. Commentitia Esse. This meant: “Whatever Aristotle Has Said Is
a Fabrication.” Logic needed to be refurbished. First, it was to be a
logic of learned disputation; that is, logic was to be devised to assist in
disputes, on the assumption that such disputation is a better means of
discovering the truth than anything else. Second, logic was to be in-
strumental in a theory of communication. “The logic of the scholastics
and the Ramists had been formulated as an instrument for the trans-
fer of knowledge from expert to expert.”6 But in communication, in
transferring the old wisdom to the next generation, order and system-
atization are of great importance; and so the Ramists proposed, in
proper scholastic fashion, the order to be followed: the most general
statement should come first, the next-most-general statement second,
and so on. Third and finally, logic was for a theory of speculative sci-
ences, and not for mere practical things.
Descartes was vehemently opposed to this. He thought that the old
logic was perhaps a satisfactory tool to explain what was already known
and established, but that it was useless for discovering new truths.7

6 Wilbur Samuel Howell, Logic and Rhetoric in England, 1500–1700, 347.

7 Thus: “I observed with regard to logic that syllogisms and most of its other techniques
are of less use for learning things than for explaining to others the things one already
knows or even, as in the art of Lully, for speaking without judgement about matters
of which one is ignorant” (CSM I, 119; AT VII, 17). And: “I do not mean the logic of
the Schools, for this is strictly speaking nothing but a dialectic which teaches ways
of expounding to others what one already knows or even of holding forth without
judgement about things one does not know. Such logic corrupts good sense rather
than increasing it. I mean instead the kind of logic which teaches us to direct our
reason with a view to discovering the truths of which we are ignorant.” (CSM I, 186;
AT IXB, 13–14)
242 Memory, Explanation, and Will

Evidently, then, Descartes was explicitly opposed to the syllogism as

a tool of discovery. Syllogisms may enable one, at best, to explain to
others what one already knows, or, at worst, to hold forth on matters
about which one is ignorant. Such a logic, and its techniques, corrupts
the mind. It does not guide it in the direction of truth. Descartes
subtitled his Discourse on the Method as follows: of rightly conducting one’s
reason and seeking the truth in the sciences. He would never have subtitled
it, as did Ralph Lever his 1573 treatise on logic, The Arte of Reason,
as: rightly termed, Witcraft, teaching a perfect way to argue and dispute. He
was not interested in arguing or disputing, in communicating settled
knowledge or explaining it; he was interested in discovering what had
hitherto been unknown. That called for a fresh logic.8
Descartes’ logic was a logic of discovery, and the logic of the scholas-
tics was the logic of explanation. Descartes’ logic was meant to be an
aid in making inquiries into nature, to discovering in a systematic way
the truth about things. The old logic was meant to explain what had
already been discovered. Essentially, the distinction between the order
of discovery and the order of explanation is the distinction between
epistemology and ontology. The first is about what and how we can
know and, in Descartes’ case, what and how we can know with cer-
tainty; the second is about what is the case regardless of what and how
we know. “[W]hen we consider things in the order that corresponds
to our knowledge of them, our view of them must be different from
what it would be if we were speaking of them in accordance with how
they exist in reality” (CSM I, 44; AT X, 418).
If we had a complete system of knowledge that reflected the order
in which we came to know the things in the universe – the order of
discovery (ordo cognoscendi) – and we set it beside a complete system of
knowledge that reflected the causal order of things in the universe –
the order of explanation (ordo ascendendi) – then each complete sys-
tem of knowledge would appear as a tree that is the inverse (loosely
speaking) of the other.9 (See Figure 8.1.)

8 See Chapter 1, sections III and IV.

9 To Morin on July 13, 1638, Descartes wrote: “You say also that there is a vicious circle in
proving effects from a cause, and then proving the cause by the same effects. I agree:
but I do not agree that it is circular to explain effects by a cause, and then prove the
cause by the effects; because there is a big difference between proving and explaining.”
(CSMK, 106; AT II, 197–198)
 II Discovery, Explanation, and the New Logic 243

Metaphysical level-1 M1

Metaphysical level-2 M2 M3

Metaphysical level-3 M4 M5 M6
.
.
.

Metaphysical level-n Mn-k M n-j Mn-i ....... Mn

Order Of Explanation

Discovery level-n Dn ........ Dn-i Dn-j Dn-k

.
.
.

Discovery level-3 D6 D5 D4

Discovery level-2 D3 D2

Discovery level-1 D1

Order Of Discovery

figure 8.1.

To explain. If we had a complete system of knowledge that reflected

the order in which we came to know the things in the universe (ordo
cognoscendi), the cardinal proposition would be the cogito. (This would
be D1 in the order of discovery.) From this proposition, propositions
about other things, such as God and the external world, would be
244 Memory, Explanation, and Will

known. (It would be very useful to compare the tree of the order of
discovery to Descartes’ Tree of Philosophy.)10 If we had a complete
system of knowledge that reflected the causal order of things in the
universe (ordo ascendendi), the cardinal proposition would be that God
exists (M1 in the order of explanation), from which the truth about
other things in the universe would follow. It does not follow, of course,
that what comes first in one system comes last in the other, and so on.
For example, the proposition that God exists does not come first in
the order of knowledge; it comes second. However, in the causal order
that proposition comes first, not next-to-last.
Eudoxus, playing Descartes in The Search for Truth, has just succeeded
in demonstrating to Polyander the truth of the cogito. He wants to move
to the next step, and he says to Polyander, “But what are you – you
who have doubts about everything but cannot doubt that you yourself
exist?” Polyander replies, “The answer to that is not at all difficult, . . .
I shall say I am a man” (CSM II, 410; AT X, 515). I quote at length
Eudoxus chiding Polyander:

You are not paying attention to my question, and the reply you give me, how-
ever simple it may seem to you, would plunge you into very difficult and
complicated problems, were I to press you even a little. If, for example, I were
to ask even Epistemon himself what a man is, and he gave the stock reply of
the scholastics, that a man is a “rational animal,” and if, in order to explain
these two terms (which are just as obscure as the former), he were to take us
further, through all the levels which are called “metaphysical,” we should be
dragged into a maze from which it would be impossible to escape. For two
other questions arise from this one. First, what is an animal ? Second, what is
rational ? If, in order to explain what an animal is, he were to reply that it is
a “living and sentient being,” that a living being is an “animate body,” and
that a body is a “corporeal substance,” you see immediately that the questions,
like the branches of a family tree, would rapidly increase and multiply. Quite
clearly, the results of all these admirable questions would be pure verbiage,
which would elucidate nothing and leave us in our original state of ignorance.
(CSM II, 410; AT X, 515–516)

The scholastic approach is no good: It sprouts endless definitions.

In defining one term, we use two others; in defining those two, we
use four more; in defining those four, and so on, down to a plethora

10 See Chapter 1, section III.

 II Discovery, Explanation, and the New Logic 245

of definitions. Thus, Descartes gives by way of illustration what would

happen if one attempted to define one of the terms, namely, animal.
But the same problem would arise if we attempted to define the other
term, namely, rational. Indeed, here we may well have a conflict of
definitions. For example, when Ramus claimed that a person was ra-
tional, he meant that it was his innate capacity to reason syllogistically;
Descartes thought otherwise; Aristotle might have had still a third view.
Ramus’ definition of rational would then lead us to define syllogism,
among other things. The result of all this, says Descartes, “would be
pure verbiage, which would elucidate nothing and leave us in our
original state of ignorance.”11
Epistemon, the scholastic, is quite offended:

I am sorry you despise the tree of Porphyry . . . which has long been universally
accepted in the Schools. To this day no better or more appropriate way has
been found for explaining what we are than displaying all the levels which
make up our whole nature, for in this way, by passing up and down through all
these levels, we can learn what our nature has in common with the natures of
all the other things, and in what respects it differs from them. And this is the
highest point to which our knowledge can reach. (CSM II, 410 –411; AT X, 516)

Porphyry of Tyre (third century a.d.), successor to Plotinus, had

proposed a method of classification. Each object or thing was to be
explained in terms of a genus and a species. When the whole tree, a
system of knowledge, was complete, there would be the highest genus
(summum genus) and the lowest species (species infima). Below the species
infima would be individuals. To explain something is to appropriately
place it on one of these metaphysical levels: In short, it is to define it.
“[I]n order to know what these are, we have to rack our brains trying to
find the ‘proximate genus’ and the ‘essential differentia’ which go to
make up their true definition” (CSM II, 417; AT X, 523). And retracing
those levels, we discover what we share in common with others and
wherein they differ from us. Consider, for example, the question, what
is a thinking substance? Bourdin replies with a diagram (Figure 8.2)
of a partial tree of Porphyry (CSM II, 344; AT VII, 506).

11 Rudolph Agricola, writes Lisa Jardine, with his applied dialectic handbook, “exclu-
sively concerned with practical argument strategies at the expense of the study of
formal inference, . . . prepared the way for the most notorious of the advocates of
‘loose’ reasoning, Petrus Ramus.” (“Humanistic Logic,” 184)
246 Memory, Explanation, and Will

Thinking Substance

Corporeal (possessing Incorporeal (not possessing

and using a body) or using a body)

Extended and Unextended and God Angel

divisible indivisible

Soul of Soul of Mind of Mind of

a horse a dog Socrates Plato

A Partial Tree Of Porphyry

Bourdin’s Scheme

figure 8.2.

What Epistemon was recommending – and what Descartes, in the

person of Eudoxus, was vigorously opposing – was that Polyander
should reply in the manner of the scholastics to Descartes’ question,
“But what are you – you who have doubts about everything but cannot
doubt that you yourself exist?” Part of Eudoxus’ objection is to the
metaphysical levels.

Indeed, whatever Epistemon may say, it strikes me that there is a great deal
of obscurity in these metaphysical levels. If, for example, we say that a body
is a “corporeal substance,” without making clear what a corporeal substance
is, these two words do not tell us any more than does the word “body.” In the
same way, if we assert that to be living is to be an “animate body,” without first
explaining what a body is and what it is to be animate – and likewise with all
the other metaphysical levels – we are uttering words and putting them as it
were in a certain order, but we are not really saying anything. (CSM II, 411;
AT X, 517)

But I do not see the scholastic logic and Descartes’ new logic as
being incompatible. The new logic takes the investigator from one
discovery level to the next in a way in which the old logic could not.
But once a truth has been discovered, one has to explain that truth.
 II Discovery, Explanation, and the New Logic 247

The explanation of a truth has to follow a certain pattern: the most

general statement coming first, the next-most-general statement com-
ing second, and so on. It is here that the old logic is used. Such a
logic would presumably take us from one metaphysical level to the
next. This is precisely what Descartes must have meant, since he has
Eudoxus say: “I have never had any intention of condemning the method
of explanation ordinarily employed in the Schools” (CSM II, 411; AT X,
516; my emphasis).
Now reconsider the following cogito-argument:12

Metaphysical level-4 [1] Whatever object has a property, then

that object exists.
Metaphysical level-3 [2] Whatever substance in which the
attribute of thought inheres, that
substance exists.
Metaphysical level-2 [3] I am a substance in which the attribute
of thought inheres.
Metaphysical level-1 [4] ∴ I, a substance, exist.

This is an argument following the order of explanation. It is clearly

scholastic, in the sense that as a purported explanation it begins with
the most general claim and then descends to less and less general
claims. Metaphysical level-4 is of the highest generality: It uses terms
like ‘object’ and ‘property’, which can take under their umbrella, or
subsume, a host of distinct metaphysical categories. For example, in
metaphysical level-3, ‘object’ and ‘property’ are replaced by the more
specific metaphysical categories ‘substance’ and ‘attribute’, respec-
tively; and, the general relation ‘has’ has been replaced by the more
specific metaphysical relation ‘inherence’. In metaphysical level-2, ‘I’
is substituted for ‘substance’, and ‘attribute of thought’ for ‘attribute’.
‘Exists’ remains univocal throughout.
The foregoing can also be cast in terms of ‘proximate genus’
and ‘essential differentia’. A partial tree of Porphyry might look like
Figure 8.3.
How would Descartes have responded? No differently, I sus-
pect, from the way he responded earlier, and it might have gone

12 It was outlined in Chapter 5, p. 150. One step is omitted here.

248 Memory, Explanation, and Will

Being

Item Property

Substance Non-substance Attribute Non-attribute

(event) (mode)

I Non-persons Thought Extension

A Partial Tree Of Porphyry

figure 8.3.

thus: “Whatever these modern philosophers may say, it strikes me

that there is a great deal of obscurity in the metaphysical levels in
this partial tree of Porphyry. If, for example, we say that whatever
has an attribute of thought, the subject of thought exists, without
making clear what, for example, ‘substance’, ‘attribute’, and ‘inher-
ence’ are, the proposition does not tell us much – and likewise with
all the other metaphysical levels. Thus we are uttering words and
putting them, as it were, in a certain order, but we are not really saying
anything.”
If Descartes were pursuing the old scholastic logic, his approach
in the Meditations on First Philosophy and the Discourse on the Method
could scarcely be more unsatisfactory. Descartes should have begun
with the highest truth, the widest generality. He should have begun
with God rather than himself, since God is the cause of everything, in-
cluding Descartes’ existence. The old logic, following Aristotle, traced
out the causal lineage. But in fact, Descartes begins with a particu-
lar, namely, I (Descartes) exist. He does not begin with the general
statement, “Whatever thinks, exists,” let alone with the existence of
God.13

13 See Appendix C, which deals with Francis Bacon’s relationship to Descartes on some
of the issues treated in this section.
 III Will, Cogito, and the Purposes of God 249

III. Will, Cogito, and the Purposes of God

The will plays an important role in Descartes’ epistemology, notwith-
standing the fact that what he said on the nature and function of the
will is disproportionately little.14 It is fundamental to his theory of
error. I shall argue that it is also fundamental to the cogito. What needs
to be argued is how Descartes can be certain that, in the light of his the-
ory of error, he cannot in the cogito-state err. I shall eventually argue that
his theory of error is mistaken, or at best insufficient; but even if his the-
ory of error is inadequate, the cogito is consistent with Descartes’ larger
aims, and the latter would be frustrated if the cogito were an argument.

[W]hen I look more closely at myself and inquire into the nature of my errors
(for these are the only evidence of some imperfection in me), I notice that they
depend on two concurrent causes, namely on the faculty of knowledge which is
in me, and on the faculty of choice or freedom of the will; that is, they depend
on both the intellect and the will simultaneously.15 (CSM II, 39; AT VII, 56)

Now, both the faculty of knowledge or intellect ( facultas cognoscendi)

and the faculty of will ( facultas eligendi) are endowed by God. Indeed,
so much so that Descartes claims that he knows by experience that
his will is “not restricted in any way,”16 that his will is so ample that
“the possibility of a further increase” is beyond his understanding,
and finally, that it is in virtue of his will that he understands himself
“to bear in some way the image and likeness of God” (CSM II, 39–40;
AT VII, 56–57). But clearly this raises a problem. Since the existence
of God has been proved, and God is all-good and all-perfect, God, the
supreme Craftsman, cannot make anything defective. Whatever He
creates is perfect. But if Descartes is perfect, that is, endowed with a
faultless intellect and will, whence his error? This is the problem that
Descartes sets himself to solve in the Fourth Meditation.

14 “Our volitions,” said Descartes in The Passions of the Soul, “are of two sorts. One consists
of the actions of the soul which terminate in the soul itself, as when we will to love
God or, generally speaking, to apply our mind to some object which is not material.
The other consists of actions which terminate in our body, as when our merely willing
to walk has the consequence that our legs move and we walk.” (CSM I, 335; AT XI,
343) In what follows, I am primarily concerned with the first sort of will.
15 See Appendix E, section I.
16 See especially Descartes’ reply to Gassendi concerning both his (Descartes’) experi-
ence of freedom as well what anyone else can experience (CSM II, 259–260; AT VII,
377–378).
250 Memory, Explanation, and Will

The source of error is not to be found in the intellect, even though

Descartes’ intellect is finite rather than infinite. Finite does not auto-
matically mean defective; it only means that it is not as comprehensive
as God’s intellect. Imagine a student who is very astute in plane geome-
try. Set him any solvable problem in the plane geometry of Euclid, and
he can solve that problem. Then we would say, and correctly, that in the
domain of plane geometry his knowledge is as perfect as it is possible
for a human mind’s to be.17 This student is unable, however, to solve
problems in spherical or hyperbolic geometry. These domains lie out-
side his sphere of competence. By contrast, of course, God can solve
any genuine problem, in any domain of mathematics, discovered, dis-
coverable, or otherwise – and then some. Analogously, God has given
a person a finite intellect that is fashioned to be adequate for whatever
knowledge of the universe a person can in principle acquire. These do-
mains lie within his sphere of competence. But his intellect is neither
appropriate nor equal to the task of discovering any knowledge of the
final causes or the purposes of God about any aspect of the universe.18
That is God’s preserve. These domains of the universe are to him what
spherical and hyperbolic geometry are to the student. In short, man’s
intellect is perfect, but limited to appropriate domains or spheres;
God’s intellect is perfect, infinite, and not limited to this sphere or that.
But if my mind is limited to knowing only about n domains or as-
pects of the universe, could not God have created my mind so that
I could know about (n + 1) domains? And if He could have created
such a mind, then why could He not have created one that could have

17 On October 8, 1628, Isaac Beeckman made the following entry in his journal on
the occasion of Descartes’ visit: “He told me that insofar as arithmetic and geometry
were concerned, he had nothing more to discover, for in these branches during the
past nine years he had made as much progress as was possible for the human mind.
He gave me decisive proofs of this affirmation and promised to send me shortly
his Algebra, which he said was finished and by which not only had he arrived at a
perfect knowledge of geometry but also he claimed to embrace the whole of human
knowledge.” (V, 78)
18 “ . . . all the purposes of God are hidden from us. . . . We think of God as a sort of
superman, who thinks up such-and-such a scheme, and tries to realize it by such-and-
such a means. This is clearly quite unworthy of God. . . .” (CSMK, 341; AT V, 158) Just
what is unworthy of God? Is it unworthy of Him to have any schemes at all? Or is it
unworthy of Him to try to realize the scheme by particular means? Clearly, the first
alternative will not do: It would preempt God from having any purposes at all. But
what is intended by the second alternative? That God can accomplish His purposes
without employing any means? This, surely, needs some defense.
 III Will, Cogito, and the Purposes of God 251

understood (n + 2) domains, or (n + 3) domains, and so on?19 The

objection is not that He should have created my mind to be as good as
His, that He should have made my mind “immense, incomprehensible
and infinite” (CSM II, 39; AT VII, 55). But it is rather, why did He make
my mind so that I could comprehend only thus and so domains, and
no more?
To this, Descartes would have replied, “I cannot produce any rea-
son to prove that God ought to have given me a greater faculty of
knowledge than he did” (CSM II, 39; AT VII, 56). To produce such a
reason, one would have to look at the whole universe from a God’s-
eye view. One would have to do this because “what would perhaps
rightly appear very imperfect if it existed on its own is quite perfect
when its function as a part of the universe is considered” (CSM II,
39; AT VII, 55–56). This surely is not unreasonable. For a long time,
for instance, the valves of the heart (the tricuspid and bicuspid valves,
respectively) appeared to be insignificant and without function, until
William Harvey discovered the circulation of the blood; then all of a
sudden the significance and function of the valves became abundantly
clear. The valves were there to ensure that the blood flowed only in
one direction, from the auricles to the ventricles. To understand the
function of the cog, one has to understand the function of the wheel.
Similarly, Descartes might have argued, the human mind consid-
ered in itself may appear to be a pretty insignificant piece of – and
to have no deeper function in – the universe. If one wanted to know
why it has limited powers, having access only to finite domains, one
would have to have a larger picture of the universe. Then the reason
for its limited nature would become abundantly clear. But, alas, in the
case of the mind, discovering a larger picture is not like discovering
the circulation of the blood. Descartes argued, specifically against the
Aristoteleans, that even if we could know all the basic physical princi-
ples of the entire universe, we still would not know the final causes.
Only the final causes – the purposes of God – would explain what the
mind is for and why the mind is limited. But the purposes of God are

19 George Schlesinger raises a parallel objection in the context of answering the ques-
tion, Why did God create a world, including the evil in it? (and infers that there is
no such thing as the best of all possible worlds): see his ‘The Problem of Evil and
the Problem of Suffering’ and Religion and Scientific Method; see also Robert Nozick’s
discussion of Schlesinger in An Examined Life, 224–226.
252 Memory, Explanation, and Will

inscrutable; hence, the larger picture will forever escape us, and we
shall not be able to understand why the nature of the human mind is
limited.20
Now, if his intellect is not the culprit in what causes Descartes to
err, then it can only be his will. The will, strictly considered in itself
and as endowed by God, “is both extremely ample and also perfect of
its kind” (CSM II, 40; AT VII, 58). Indeed, it is so much so that it is
in his will that a person is most Godlike. The function of the will is
to affirm or deny, to pursue or avoid. Judgment, which is what can go
wrong – and Descartes is providing a theory of why this happens – is
dependent on both the intellect and the will in the following way. The
intellect proposes a proposition, and the will accepts or denies it.21

20 Gassendi had argued that Descartes had failed to notice the difference between the
private and public purposes of God, and that knowing the purposes of God that are
on “public display” would lead us to bestow great praises on God (CSM II, 215–216;
AT VII, 310). Descartes responded that Gassendi’s arguments are best construed
not as a defense of purposes – for all of God’s purposes “are equally hidden in
the inscrutable abyss of his wisdom” – but rather as a defense of efficient causes,
for example, those pertaining to the functioning of the various parts of plants and
animals. These efficient causes mortals can know. Then, Descartes goes on to say
something interesting: “In ethics, then, where we may often legitimately employ
conjectures, it may admittedly be pious on occasion to try to guess what purpose God
may have had in mind in his direction of the universe” (CSM II, 258; AT VII 374–375).
In August 1641, the same year in which the Meditations were published, Descartes
wrote to Hyperaspistes, “It is self-evident that we cannot know God’s purposes unless
God reveals them. From the human point of view adopted in ethics, it is true that
everything was made for the glory of God, in the sense that we must praise God for
all his works.” (CSMK, 195; AT III, 431)
This is a bit puzzling. If we can adopt a conjecture about God’s purpose – strictly
as a heuristic device – as we go about solving our ethical problem, might we not
adopt a similar conjecture (again, strictly as a heuristic device) as we go about solving
our scientific problem? In conversation with Burman, in April 1648, Descartes had
said that “the knowledge of a thing’s purpose never leads us to knowledge of the
thing itself” (CSMK, 341; AT V 158). If conjecturing a thing’s purpose – “Aristotle’s
greatest fault” – is inefficacious, misleading, or “futile,” what makes the adoption of
a similar conjecture in ethics less so? Would knowledge of a thing’s purpose lead us
to the knowledge about how we should morally deal with that thing? Indeed, would
it not be more dangerous to adopt such conjectures in ethics than in science? It is
clearly insufficient to say, as Descartes did to Burman, that ethical purposes are ones
that are “known through revelation” (CSMK, 341; AT V 158). If the only purposes of
God postulated in ethics were those revealed by Him, they would not be conjectures
any more. They would be God’s decrees. Therefore, if conjectures ought not to be
allowed in science, they should be made impermissible in ethics as well.
21 Some recent commentators have claimed that for Descartes, propositions in them-
selves are neither true nor false, and that therefore error cannot be attributed to
 III Will, Cogito, and the Purposes of God 253

Until the will affirms or denies a proposition, there is no judgment.

And it is a judgment – in contrast, say, to desire – which can be true or
false.22
“So what then is the source of my mistakes?” asks Descartes. He an-
swers, “[T]he scope of the will is wider than that of the intellect; but in-
stead of restricting it to the same limits, I extend its use to matters which

them; error is possible only when propositions are judged – affirmed, denied, and
so on (Losonsky, Enlightenment and Action From Descartes to Kant: Passionate Thought,
15–16). Presumably, the support for this view comes from the two paragraphs out of
the third meditation wherein Descartes says that he wants to classify his thoughts and
“ask which of them can properly be said to be the bearers of truth value” (CSM II,
25; AT VII 36–37). They cannot be ideas in themselves; neither can they be will or
emotions; and thus Descartes concludes, so goes the claim, that they can only be
judgments.
My interpretation differs from this. Descartes would claim that the truth-value that
a proposition possesses is an ontological matter – it possess a truth-value whether any-
one judges it or not; it becomes an epistemic matter – error becomes possible – when
and only when a proposition is judged. My reasons for suggesting this interpretation
are three. First, consider “images of things,” such as a goat or a chimera; it is only for
these cases, says Descartes, that “the term ‘idea’ is strictly appropriate.” And ideas,
so long as one does not refer them to anything else, “cannot strictly speaking be
false.” Then Descartes goes on to say, “for whether it is a goat or a chimera that I
am imagining, it is just as true that I imagine the former as the latter.” This would
be odd if propositions did not have truth-value in themselves: I imagine a goat, but
do not refer it to anything else; it is just an image, an idea. Now, I have not enter-
tained the proposition “I am imagining a goat,” let alone affirmed it. Nevertheless,
says Descartes, “I am imagining a goat” is true.
Second, Descartes is more concerned about how to avoid making a mistake in
judgment, an epistemic matter, than he is about claiming that propositions in them-
selves are neither true nor false, an ontological matter. Descartes says, “Thus the only
remaining thoughts where I must be on my guard against making a mistake are judg-
ments. And the chief and most common mistake which is to be found here consists
in my judging that the ideas which are in me resemble, or conform to, things located
outside me.” (CSM II, 26; AT VII 37) Consider the proposition, “The sun is several
times larger than the earth.” If I refrain from judging it, then I have not in principle
run any risk of making a mistake; but surely the proposition has a truth-value nev-
ertheless. Indeed, if I refrain from judging it, because I have not yet come up with
any reason or evidence that tilts me one way or another, the proposition must have
a truth-value – after all, I am trying to discover its truth-value. What Descartes wants
to do in this part of the Third Meditation is to demonstrate what kind of mistakes
I make when judging a proposition (I might hasten to judgment based on “natural
impulse,” “blind impulse” or “spontaneous impulse”) and how I might avoid acting
in a precipitous manner (by relying solely on “natural light” ).
Third and finally, God is the creator of eternal truths, such as 2 + 3 = 5, and surely
said proposition is true regardless of whether any seeker inquires after or judges its
truth.
22 See Appendix E, section II.
254 Memory, Explanation, and Will

I do not understand” (CSM II, 40; AT VII, 58). To explain: Descartes’

intellect offers for consideration by the will, say, the statement S, “The
earth goes around the sun.” Here, as anywhere else, “the intellect
should always precede the determination of the will” (CSM II, 41; AT
VII, 60; see also CSMK, 195; AT III, 432). If the will affirms that S is
true, but does not know it to be so, it goes beyond the limit set by the
intellect. That is, if S is true, Descartes knows it only accidentally. If S
is false, then surely Descartes does not know S.
What is it, precisely, for the will to go beyond the limits or to restrict
itself to these same limits? Descartes must mean something like this.
The intellect in offering a proposition to the will also defines the limits
of the proposition. These limits are set by all the other propositions
linked to the one currently entertained by the will. Thus, S, “The earth
goes around the sun,” is limited by the propositions connected to it.
To know S would be to know these other propositions (presumably,
propositions of astronomy) and its links or connections with them. The
will is precisely at that limit if it knows these other propositions and the
various links. By contrast, it is not exactly at that limit (is within it) if
the will affirms S, but in the absence of the knowledge of at least some
of these other propositions, or if it mistakenly believes that the links
are weaker than they in fact are. It is outside that limit if it mistakenly
enlarges the set of propositions to which it is linked, or mistakenly
strengthens the links, or if it affirms S when S is false.
If this analysis is correct, then ultimately there must be at least one
proposition whose limit is defined by itself. Otherwise, there would be
an infinite regress, and there would be no possibility of knowledge.
Consider, then, the alleged argument of the cogito:

[1] Whatever thinks, exists. Premise

[2] I think. Premise
[3] If I think, I exist. 1, Universal Instantiation
[4] ∴ I exist. 2,3, Modus Ponens

The first premise cannot be known to be true, for it is surely inductively

dependent on the cogito. At this point, for Descartes to affirm the first
premise is for him to extend the use of his will to matters that he does
not yet understand. Once again, to know that the argument is valid,
let alone that the argument is sound, the will has to affirm the rules of
inference of universal instantiation and modus ponens. If the will were to
 III Will, Cogito, and the Purposes of God 255

acknowledge the argument to be valid in the absence of the knowledge

of the rules of inference, then its acknowledgment would be correct,
but only accidentally so. On the other hand, if it correctly acknowl-
edges the argument to be valid by staying within proper limits, that is,
if it knows the other propositions to be true, then the cogito-argument
is hardly the centerpiece of his philosophy. For that, we would have to
look elsewhere. By proper contrast, the cogito claim, “I am thinking,
therefore I exist,” sets its own limit. The will, to acknowledge it, does
not have to look elsewhere. Indeed, this is precisely what Descartes
affirms.
The will is freest when it is powerfully and ineluctably inclined “in
one direction.” “[T]he more I incline,” he writes in the Fourth Med-
itation, “in one direction – either because I clearly understand that
reasons of truth and goodness point that way, or because of a divinely
produced disposition of my inmost thoughts – the freer is my choice.
Neither divine grace nor natural knowledge ever diminishes freedom;
on the contrary, they increase and strengthen it.” (CSM II, 40; AT VII,
57–58) Indifference, far from being a sign of freedom, is an indica-
tion of bondage. In such cases, the will “turns aside from what is true
and good, and this is the source of my error and sin” (CSM II, 41;
AT VII, 58).
It is here that Descartes explicitly returns to the cogito by way of an
illustration of the freedom of the will and its inclination to accept the
cogito.

I have been asking whether anything in the world exists, and I have realized
that from the very fact of my raising this question it follows quite evidently that
I exist. I could not but judge that something which I understood so clearly was
true; but this was not because I was compelled so to judge by any external force,
but because a great light in the intellect was followed by a great inclination in
the will, and thus the spontaneity and freedom of my belief was all the greater
in proportion to my lack of indifference. (CSM II, 41; AT VII, 58–59; see also
CSMK, 233–234; AT IV, 115–116)

One might argue that for all that Descartes has said thus far, the
cogito might just as well be an argument. The argument is so clear that
Descartes’ will is inclined in the direction of affirming it; he could not
but judge it to be sound, and in so affirming, his will exhibits freedom.
This, I think, would be hasty. Before he has proved the existence of God
256 Memory, Explanation, and Will

in the Third Meditation, Descartes says, “In this first item of knowledge
[the cogito] there is simply a clear and distinct perception of what I am
asserting; this would not be enough to make me certain of the truth of
the matter if it could ever turn out that something which I perceived
with such clarity and distinctness was false.” He concludes, as we saw,
with an extraordinarily important general rule: “So I now seem to be
able to lay it down as a general rule that whatever I perceive very clearly
and distinctly is true” (CSM II, 24; AT VII, 35).
Descartes’ ideal seeker ought to be armed not only by a perfect, if
finite, intellect, but also by a proper will – a will that will not exceed
its limits. But the will initially learns the limits that it must recognize
only at the foot of the cogito; it is there that it learns the insignia of
truth. It is in the state of the cogito that the will learns what it means to
be ineluctably inclined toward affirming a proposition, and what it is
to be free.23 That is the first instance. Only then does it provide itself
with a general rule and decide henceforth to be guided by that rule.
Take away the cogito, and the reason for accepting that rule disappears.
Take away the rule, and the entire enterprise past the cogito collapses.
For example, the proofs of the existence of God and material things
centrally depend on the general rule. The rule determines whether or
not any given proposition, or any idea, bears the appropriate insignia
allowing it to be entered in the knowledge log.
Our task, then, is to determine if it could be the cogito, construed
as an argument, that could serve as that first instance from which the
general rule arose. One might claim that the cogito argument was so
powerfully clear that the will could not help but assent to it. The ar-
gument ineluctably bent the will in that “one direction.” From this
experience, the ego first learned the veracity of the general rule, and
the will accepted it. But such cannot be the case, because the argu-
ment already presupposes the correct application of the general rule
at various places in the cogito-argument. The argument experience
cannot be the first experience from which the general rule arose.
In fact, we can generalize: No argument experience can be the first
instance in which the veracity of the general rule is acknowledged.

23 This is one of the key points of this section, and I greatly value Kenny’s support of it;
see also Kenny’s “Descartes on the Will,” especially 148–149.
 III Will, Cogito, and the Purposes of God 257

On the other hand, the truth of any subsequent idea or proposition

is in question if it fails to satisfy the general rule. Thus, to construe
Descartes’ cogito as an argument is to defeat Descartes’ strategy and
larger aims.
It is time to raise some questions about Descartes’ view of the will;
in particular, I shall raise three. My first objection will raise the issue
of whether Descartes’ solution of how and why humans err invokes
God’s inscrutable purposes, an invocation of the very purposes he
had condemned in physics. “I consider the customary search for final
causes to be totally useless in physics” (CSM II, 39; AT VII, 55). The
second objection will raise the question of whether, in the ultimate
analysis, Descartes has a solution at all. The third objection will raise
concerns about the precise relation between the intellect and the will,
and about which one has primacy over the other.
The first objection: Consider three clear cases. Case 1: God has an
infinite intellect and an infinite, perfect will. His will is supremely
benevolent. “God’s will is incomparably greater than mine, both in
virtue of the knowledge and power that accompany it and make it
more firm and efficacious, and also in virtue of its object, in that it
ranges over a greater number of items” (CSM II, 40; AT VII, 57).
Thus, He knows everything, takes no incorrect actions, performs all
good ones, and makes no mistakes of thought or deed. “In God willing
and knowing are a single thing” (CSMK, 24; AT I, 149); “in God seeing
and willing are one and the same thing” (CSMK, 235; AT IV,119). This
is the ideal that no one else can reach.
Case 2: A person has a finite, but perfect, intellect. His will is good
and is such that it never transgresses the limits set by the intellect. Let
us divide all propositions that the intellect offers the will into three
categories. First is the proposition that is its own limit; when it is un-
derstood, the will is powerfully inclined to affirm it as being true.
Second are the propositions of logic and mathematics; and third are
empirical propositions that can be affirmed or denied only with a cer-
tain degree of probability. The ideal seeker never denies propositions
in the first two categories, and affirms them whenever he encounters
them.
Descartes maintains that the will may encounter a proposition –
such as, “My essence is corporeal” – and neither affirm nor deny it
258 Memory, Explanation, and Will

because he does not as yet have sufficient reasons for doing so. Here
he becomes indifferent. As he puts it,

[M]y intellect has not yet come upon any persuasive reason in favour of one
alternative rather than the other. This obviously implies that I am indifferent
as to whether I should assert or deny either alternative, or indeed refrain
from making any judgment on the matter. What is more, this indifference
does not merely apply to cases where the intellect is wholly ignorant, but
extends in general to every case where the intellect does not have sufficiently
clear knowledge at the time when the will deliberates. For although probable
conjectures may pull me in one direction, the mere knowledge that they are
simply conjectures, and not certain and indubitable reasons, is itself quite
enough to push my assent the other way. (CSM II, 41; AT VII, 59)

The point is well taken. Here I am imagining that the ideal seeker as-
signs precisely that probability to the propositions in the third category
that should be assigned to them, neither more nor less. In short, the
ideal seeker will be pulled in the direction of the probable conjecture
exactly to the extent that he should be so pulled. This ideal seeker can
in principle know the entire knowable universe. At no stage in his inves-
tigations will he ever affirm a proposition that transgresses the limits.
Descartes himself envisages two ways in which God could have
created a person so that he did not err. “God could easily have brought
it about that without losing my freedom, and despite the limitations in
my knowledge, I should nonetheless never make a mistake.” “The first
way”: “He could, for example, have endowed my intellect with a clear
and distinct perception of everything about which I was ever likely to
deliberate; or he could simply have impressed it unforgettably on my
memory that I should never make a judgement about anything which
I did not clearly and distinctly understand” (CSM II, 42; AT VII, 61).
“The second way”: “I can avoid error in the second way, which depends
merely on my remembering to withhold judgement on any occasion
when the truth of the matter is not clear”24 (CSM II, 43; AT VII, 62).
Case 3: A person has a finite, but perfect, intellect, and a will.
The will, however, because it is ample, often transgresses the limits
and makes mistakes. These mistakes may reside in the will’s affirm-
ing (or denying) a proposition when there is not sufficient warrant

24 For lack of space, I am unable to analyze these two ways and to show how Descartes’
ways presuppose God’s inscrutable purposes.
 III Will, Cogito, and the Purposes of God 259

for such an affirmation (or denial). Or it may reside in the will’s

affirming a proposition with a higher degree of probability, when
only a smaller probability should be affirmed. The will may some-
times understate the probability with which a proposition can be af-
firmed. Perhaps in time it will correct its mistakes and move forward.
But it does make mistakes. This is simply and correctly the view of a
human.
The question that arises is this. Why did not God create a finite
person on the model of Case 2 rather than on the model of Case 3?
The model seems simple and consistent enough. What is more, the
model is in harmony with the view of God as the supremely perfect
Craftsman and of a person as finite. He is a perfect Craftsman, so
His artifact cannot be defective; hence, His artifact, the ideal seeker,
does not make mistakes. The artifact, being a mere artifact, need
not be infinite – which, as a matter of fact, he is not. Notice that
this does not question why God did not give a finite person more
knowledge – say, of (n + 1) domains – rather than what He in fact
gave him (knowledge of n domains). Granted that we require God’s
purposes or the God’s-eye view of final causes to answer that question,
the issue raised here is a different one. Assuming we can know only
n spheres – for reasons knowable only to God – why did God not make
a person along the model of Case 2? Whatever the limits on his knowl-
edge, a person could acquire it piecemeal, but without making any
mistakes.
One might say: If we had the God’s-eye view, then we would be able
to see clearly why Case 3 is better than Case 2. Lacking such insight or
knowledge, we have no grounds to complain. There is a difficulty in
this reply. If the valves of the heart appeared insignificant and func-
tionless before the discovery of the circulation of the blood, they do
perspicuously appear significant and useful after the discovery. But we
will hardly similarly conclude – for there is no analogy here – that
before having the God’s-eye view, we would see some of our epistemic
states as being mistakes, but after acquiring the God’s-eye view (hypo-
thetically speaking), we would see those epistemic states as being only
apparent, and not real, mistakes. Were we to so conclude, we would
effectively be pronouncing the death of epistemology. Consequently,
on Descartes’ view, it must be intrinsically important – something we
fail to see now – that we make mistakes, real and genuine mistakes,
260 Memory, Explanation, and Will

through the transgression of our will. And why making mistakes is

important can be seen only from the God’s-eye view.

But I cannot therefore deny that there may in some way be more perfection in
the universe as a whole because some of its parts are not immune from error,
while others are immune, than there would be if all parts were exactly alike.25
And I have no right to complain that the role God wished me to undertake in
the world is not the principal one or the most perfect of all. (CSM II, 42–43;
AT VII, 61)

This is an unsatisfactory reply. The critic’s complaint is not that in

Descartes’ solution to the problem of why humans err, God has not
assigned Descartes a role that is the most perfect of all. His complaint
rather is that Descartes’ solution to the problem of why humans err
vehemently assumes final causes or the inscrutable purposes of God –
namely, that a universe in which humans make mistakes is more perfect
than a universe in which they do not make mistakes, and that it is the
purpose of God to make the most perfect universe. An Aristotelian
might rebut Descartes by saying that if some explanations in meta-
physics are incomplete without resorting to final causes, so also are
some explanations in physics; on the other hand, if invoking final
causes in physics is “totally useless” (CSM II, 39; AT VII, 55), what
makes them any different in metaphysics? Indeed, it would still not
follow that a universe in which a person makes mistakes is more per-
fect than one in which he does not – at least not without some pretty
powerful arguments. But Descartes provides none. He blandly as-
sumes that the present universe, mistakes and all, accomplishes God’s
purposes.26
To turn to the second objection: In one sense, Descartes has no
theory of error at all. A theory of error must, in part, explain why it is
that the ego commits error. Well, it lets the will transgress the limits.
But why does the will transgress the limits? There are several possible
causes. The transgressions may be caused by the emotions – emotions

25 This is a slip: Descartes intends something much stronger than merely “if all parts
were exactly alike.” After all, if each part were fallible, then all parts would be exactly
alike; what Descartes intends, of course, is that each part is but alike in being immune
from error.
26 There is also this: If I am essentially prone to error, and not just accidentally so, it is
difficult to see how I can know and correct my mistakes.
 III Will, Cogito, and the Purposes of God 261

such as anxiety, jealousy, confidence, and despair. In The Passions of the

Soul, Descartes says,

We are prompted to desire the acquisition of a good or the avoidance of an

evil simply if we think it possible to acquire the good or avoid the evil. But
when we go beyond this and consider whether there is much or little prospect
of our getting what we desire, then whatever points to the former excites hope
in us, and whatever points to the latter excites anxiety (of which jealousy is
one variety). (CSM I, 350–351; AT XI, 375)

Consequently, one conjecture is that the will errs because the emo-
tions, such as jealousy and anxiety, divert the will onto the wrong path.
Or the will may transgress because it harbors false beliefs, such as, “The
sun goes around the earth.” Holding onto such beliefs causes the will
to err.
Some such reply would restore Descartes’ view against the objection
that asks why God did not model a person along the lines of Case 2.
Descartes could argue that both the intellect and the will, when un-
hampered, function faultlessly in the limited domains. What causes the
will to make the mistakes is extraneous causes, such as bad emotions
and false beliefs. Understandably, Descartes did not take this option,
since it only shifts the problem to accounting for the source of the
false beliefs and the assenting to propositions under the influence of
bad emotions. Hence the purging of all prior beliefs.
The ideal seeker engages in methodical doubt; he doubts every-
thing under the sun. He doubts all of his beliefs, no matter how well
settled and deeply entrenched. He rejects as false his cosmology and
geography, his history and language, his morals and theology, his fa-
bles and mathematics, his medicine and jurisprudence – or so he is
made to do in Discourse on the Method. “[I]t is good,” says Descartes, “to
have examined all these subjects, even those full of superstition and
falsehood, in order to know their true value and guard against being
deceived by them” (CSM I, 113; AT VI, 6). What surprises me is that
he does not take the same stand against his emotions when engaging
in his methodical doubt. If the beliefs can deceive, the passions or
the emotions can deceive no less. No one was more aware of that than
Descartes. These passions or emotions may incline us toward affirming
(or denying) a proposition that we should deny (affirm), or at least
toward abstaining from affirming (denying) it. Thus, an ideal seeker
262 Memory, Explanation, and Will

should be purged not only of all false beliefs, but also of all bad emo-
tions that would turn the will “aside from what is true and good.” The
ideal seeker,before he starts his inquiry, now has an intellect purged of
all false beliefs and a will free from the influence of all bad emotions.
Let us call this the pristine state of the ideal seeker. And in this pristine
state he collects and stores true propositions, as and when they are
recognized. The first to be recognized is the cogito, then the general
rule, then the proof of the existence of God, and so on.
Now, in such a pristine state how can the will make mistakes? It seems
to me that it cannot. Yet Descartes maintains that the will can make
mistakes even in this state, since it is ample and perfect and so may
well transgress the limits. But what is the cause of such transgressions?
There is no answer; the will just simply plumps itself beyond the limit.
It just simply does so, and there is nothing else that can be said.27
What Descartes does say is puzzling: “In this incorrect use of free will
may be found the privation which constitutes the essence of error. The
privation, I say, lies in the operation of the will in so far as it proceeds
from me, but not in the faculty of will which I received from God, nor
even in its operation, in so far as it depends on him.” (CSM II, 41;
AT VII, 60)
Descartes seems to come very close to maintaining with respect
to the will what he has said about the intellect. The finite intellect
is a God-given instrument and cannot be defective in the domain of
inquiry for which it was intended. The will is a God-given instrument,
too, and cannot be erroneous in any way, either. The defect lies in
the operator of the will, not in the will itself. So, presumably, there is
a will* that operates the will. (If you find this paradoxical, replace
will* by ‘operator.’) When the first automobiles had to be started, one
had to crank them with a handle. Nothing need be wrong with the
car, but the handle or the cranking could be wrong. If this reading
is correct, it simply shifts the problem one step backward. What is
the reason for the will* operating sometimes without mistakes, and
sometimes mistakenly? There seems to be no apparent reason for any
of its actions. Thus, I maintain that when we properly understand the
ideal seeker, Descartes’ theory of error does not seem to amount to

27 There is clearly a tension in Descartes’ theory. As will be noted, Descartes also wants
to vehemently deny that the will can transgress when in the pristine state.
 III Will, Cogito, and the Purposes of God 263

much. For he really offers no explanation of the causes that incline

the will away from the true and straight path.
The will of a person endowed by God must be essentially good.
Otherwise, even in this pure case, there is no guarantee that the will
will not make mistakes. Indeed, if it is evil it will not be able to help
making mistakes,
The third and final objection pertains to the relation between the
intellect and the will. Which one has supremacy? The precise relation
between the intellect and the will is unclear. On the one hand, it seems
that both are autonomous; on the other hand, only the intellect is
autonomous. Thus the primacy goes to the intellect – morally speaking,
as Descartes would say. The intellect offers the will propositions for the
will to affirm or deny. The will seems to be autonomous in the sense
that it is free to affirm or deny and is not constrained by the intellect.
But the issue is not that simple. For instance, Descartes says that when
in the state of the cogito, he affirmed its truth, he was not “compelled
so to judge by any external force, but because a great light in the
intellect was followed by a great inclination in the will” (CSM II, 41;
AT VII, 59).
Thomas Hobbes complained that the

phrase “a great light in the intellect” is metaphorical, and so has no force in

the argument. Moreover, anyone who is free from doubt claims he has such
“great light” and has no less strong a propensity of the will to affirm what
he has no doubt about than someone who possesses real knowledge. Hence
this “light” can explain why someone obstinately defends or holds on to a
given opinion, but it cannot explain his knowledge of its truth. (CSM II, 134;
AT VII, 191–192)

For Hobbes, knowing, believing, and giving assent are independent of

the will. When a valid argument is presented, or something credible
is reported, we often believe against our will. Affirmation and denial,
says Hobbes, are acts of will. He then continues, “but it does not follow
that our inner assent depends on the will” (CSM II, 134; AT VII, 192).
What is puzzling in Hobbes is how affirmation is dependent on the
will, and yet our so-called inner assent is not dependent on it. What
is it dependent on? It cannot be the intellect, since Hobbes seems to
agree with Descartes that the role of the intellect does not extend to
affirming or denying propositions.
264 Memory, Explanation, and Will

The great light in the intellect forces the will to follow ineluctably.
What is this great light? Is it that the proposition is true? No, because
a proposition is true if it corresponds with the state of affairs, and that
says nothing about a natural light. Furthermore, the function of the
intellect is not to judge propositions to be true or false, it is simply to
offer propositions to the will to affirm or deny. When “a great light
in the intellect [is] followed by a great inclination in the will,” what
connects the intellect and the will? Is it a causal connection, such that
when the intellect exhibits a great light it necessarily causes the will to
affirm the proposition in question? But then the will can scarcely be
called free. If it is not a causal connection, then what connection is
it, such that the will cannot help assenting to the proposition in that
state? Perhaps the will is taught by nature, in the language of the Third
and Sixth Meditations, to assent; yet nature has not designed a causal
nexus – it has designed something else – between the intellect and the
will. However, is this something else any less determinative?28
Is it possible – just possible – for the will to deny the cogito while
in the cogito-state? If it is possible, then someone can consistently deny
that the cogito is certain. Not to deny its certainty would be for him to
transgress the limits. But here is what Descartes says in a letter, probably
to Mesland, written on February 9, 1645:

[W]hen a very evident reason moves us in one direction, although morally

speaking we can hardly move in a contrary direction, absolutely we can. For it
is always open to us to hold back from pursuing a clearly known good, or from
admitting a clearly perceived truth, provided we consider it a good thing to
demonstrate the freedom of our will by so doing.29 (CSMK, 245; AT IV, 173)

Judged by this passage, it is difficult to tell whether we can move in a

contrary direction. In one sense we can, provided we wish to demon-
strate the freedom of the will. But the purpose is not to deny a truth
we clearly perceive; it is something else, namely, to express and exalt
the freedom of the will. In other words, we can vigorously shake our
heads and loudly mouth sentences to the contrary – in order to exhibit

28 See Appendix E, section III.

29 See Kenny, “Descartes on the Will,” especially 151–159, for a brilliant analysis of this
letter and a refutation of M. Alquié’s thesis that the letter demonstrates that Descartes
“allows the possibility of rejecting the evident and the good in full knowledge of the
cause.”
 III Will, Cogito, and the Purposes of God 265

our freedom – but we would be utterly incapable of a genuine dissent

within ourselves to a truth we clearly perceive.
Finally, there is an interesting way to show the primacy of the in-
tellect over the will and the sheer beauty of the cogito. In the cogito-
state, the intellect – provider of “a great light” – elicits a response
from the will. In that state, the will must assent; not to do so would
be a mark of its bondage or unfreedom – “the more I incline in one
direction . . . because I clearly understand the reasons of truth . . . the
freer is my choice” (CSM II, 40; AT VII, 57–58). Generally speak-
ing, the will can err by overreaching or by understating. In the case
at hand, the will can err if it claims more or less than what is war-
ranted by the cogito-statement. It understates it if it claims less than
what is warranted by the cogito-statement. But what is it for the will
to understate in this case? Perhaps one has such a timid will, a will
so lacking in trust and confidence in itself, that it always hedges:
It gives assent to watered-down claims such as, “Probably, I exist.”
On the other hand, what is it for the will to overreach in the cogito
state? It is very hard to imagine. Perhaps something like, “I always
existed.”30
These cases of understating and overstating while in the cogito-state
are not sustainable alternatives. As argued earlier, the will is not ab-
solutely and unconditionally free; the alleged freedom to deny any-
thing under any circumstances is not freedom but bondage. For ex-
ample, it cannot deny what it perceives clearly and distinctly; it can
only pretend to deny it in order to exhibit its freedom. Let us look
more closely at the case of the timid will. Suppose that while in the
cogito-state, that will affirms only the proposition, “Probably, I exist.”
But it must have some countervailing evidence in order to rationally
make only a watered-down claim; if there is no such evidence, –
“where the intellect does not have sufficiently clear knowledge at
the time when the will deliberates”(CSM II, 41; AT VII, 59) – the as-
sertion “Probably, I exist” is irrational. One cannot argue that it is

30 Replying to Pierre Gassendi’s “Fifth Set of Objections,” Descartes wrote: “You here
ask me to say briefly whether the will can extend to anything that escapes the intellect.
The answer is that this occurs whenever we happen to go wrong. . . . Again, we never
understand anything in a bad fashion; when we are said to ‘understand in a bad
fashion,’ all that happens is that we judge that our understanding is more extensive
than it in fact is.” (CSM II, 259; AT VII, 376–377)
266 Memory, Explanation, and Will

the nature of a timid will only to assent diffidently to a circumspect

proposition.
Let us recall at the end what we said in the beginning about
Descartes on the ideal seeker. Among other things, such a seeker must
have a quick wit, ample and prompt memory, and the strongest ability
to reason; he must be skilled at ordering his thoughts, troubled by no
cares or passions, and capable of seeing clearly into his own actions; he
must not be precipitate in his judgments. But the timid will is prey to
the very cares and passions that Descartes had warned against. In short,
the timid will is not “of a fairly robust intellect” (CSM II, 320; AT VII,
475). It follows, therefore, that in the very act of either understating
or overreaching, the will will inevitably be compelled to acknowledge
the truth of the cogito.31 Why? “Because,” as Descartes says, “a great
light in the intellect [is] followed by a great inclination in the will”
(CSM II, 41; AT VII, 59). This is just as well, because otherwise I do
not see how the cogito could serve as the foundation. Descartes’ entire
structure would collapse.32

31 Here is a firm denial: “Since I understand falsity to be merely a privation of the truth,
I am convinced that there would be a total contradiction involved in the intellect’s
apprehending falsity under the guise of truth; but this would have to be the case if
the intellect were ever to determine the will to embrace what is false” (CSM II, 260;
AT VII, 378). One might argue that this passage prohibits the intellect only from
determining as true what is in fact false, but that it does not prohibit the timid will
from, say, determining a watered-down proposition to be true. Admittedly so, but
an interesting defense of Descartes can be set up. If there were no countervailing
evidence that would justify the assertion of a watered-down proposition, then a timid
will no less than an ideal seeker would be in a bit of quandary: He would not know
which proposition to assert, the proposition or the watered-down proposition! He
would feel indifference as to which path he should take, and “the indifference [he]
feels[s] when there is no reason pushing [him] in one direction rather than another
is the lowest grade of freedom” (CSM II, 40; AT VII, 58). See also the following
note.
32 Nor would an objection based on the following do: “Unlike Descartes, [recent com-
mentators] do not take seriously enough the cognitive lives of those who suffer from
doubt and anxiety about their own capacity to fix their beliefs (even if it is to fix on
withholding judgment on some topic). . . . Cases of cognitive irresolution that fit the
dysfunctional pattern Descartes describes [inThe Passions of the Soul under the head-
ing “Irresolution”(CSM I, 390–391; AT XI, 459–460)] are ones in which noncognitive
factors undermine cognitive processes. Excessive fear of making mistakes can keep
one from settling on a belief in certain circumstances even when there are no com-
peting hypotheses and all the evidence – which is good even if it is not clear and
distinct – supports a proposition.” (Losonsky, Enlightenment and Action From Descartes
to Kant: Passionate Thought, 16, 18) Consequently, the objection would run, anyone
 III Will, Cogito, and the Purposes of God 267

This has been a long book, not only in pages; allow me, then, to con-
fine its conclusion to a couple of paragraphs. The heart of this book
lies in the thesis that Descartes’ cogito is not an argument, but an intu-
ition. If it were an argument, it would have to be either an immediate
inference or a mediate inference. If an immediate inference, it would
have to begin with a proposition known to be certain, from which an
inference is drawn to another proposition, the latter also known to be
certain. But if the former proposition is already known to be certain
before any inference is drawn, the ideal seeker has already arrived at
the Archimedean point, and he need look no further. Clearly, then,
the cogito cannot be an immediate inference. But the cogito cannot be
a mediate inference (decidedly a dogma in Cartesian studies) either.
If it is a mediate inference, then the inference would have to be gov-
erned by the rules of inference or the principles of logic. However,
the principles of logic, no less than the statements of mathematics,
are subject to doubt, courtesy of Descartes’ demon. There is no histor-
ical evidence, and even less philosophical argument in Descartes, to
suggest that the principles of logic are immune from doubt when the
statements of mathematics are not so exempt. What is more, Descartes’
view would have to face a new Cartesian Circle, more devastating than
the old one: Prior to the proof of the existence of God, no principles
of logic can be certified to be true or reliable. Without true or reliable

who is cognitively irresolute could deny the truth of the cogito because even in the
cogito -state he suffers from doubt and anxiety about his own capacity to fix his beliefs.
Lacking that capacity, he either understates his belief or anxiously overstates it.
First, to someone who is cognitively irresolute, Descartes’ advice is to make a
decision – rightly or wrongly – and to stick with it (CSM I, 390–391; AT XI, 460).
Notice that Descartes is offering this advice as a remedy for someone who is consti-
tutionally indecisive, hardly an epistemic criterion. Descartes urges this remedy in
the hope of curing the individual of his indecisiveness, not to suggest that whatever
decision such an individual made would be the right or rational decision. At best, it is
advice as to what the individual should do if he desires to embark upon the journey
of becoming an ideal knowledge seeker. Far from being an ideal seeker, such an
individual cannot be in the pristine state (described earlier). As Descartes says, “an
excess of irresolution results from too great a desire to do well and from a weakness
of the intellect, which contains only a lot of confused notions, and none that are
clear and distinct” (CSM I, 390–391; AT XI, 460). Not being in the pristine state –
a state in which there are no confused notions, as what notions do inform the state
are clear and distinct – he who is cognitively irresolute cannot be a counterexample
to Descartes’ claim that no one in the cogito state can deny the truth of the cogito.
268 Memory, Explanation, and Will

principles of logic, Descartes could not execute the proof of the ex-
istence of God. There is also this: If the cogito were an argument, the
ideal seeker would perforce have to rely on memory as he moved from
the premises to the conclusion. Now, Descartes drew a distinction be-
tween two types of memory: physical memory and intellectual memory.
There is no question about relying on physical memory, since at this
stage in the Meditations on First Philosophy Descartes is unsure whether
he even has a body. Nor will intellectual memory do: Memory is unre-
liable, says Descartes, immediately after the First Meditation. Ergo, the
cogito cannot be an argument, since it rests on memory.
On the constructive side, the cogito is an intuition; in order to see
that, one has to conduct a thought experiment. One has to engage
in the process of doubt; and in that process, one must notice oneself.
What one will notice when in the utmost state of doubt engineered
by a most powerful, malicious demon is that one nevertheless exists.
Whatever the breadth or depth of the doubt, at least one proposition
escapes from it in virtue of the very process of engaging in doubt. This
also enables us to understand the content of the cogito and why (pace
Ryle) the ‘I’ is so elusive. Finally, Descartes’ theory of the will makes
it impossible for an ideal seeker to deny, while in the cogito-state, the
truth of the proposition of the cogito, while allowing the ideal seeker to
doubt even the validity, let alone the soundness, of the cogito-argument.
Undoubtedly, then, Descartes’ cogito cannot be other than an intuition.
 Appendix A

Comments on Jeffrey Tlumak’s “Certainty and

Cartesian Method”

Jeffrey Tlumak, in “Certainty and Cartesian Method,”1 argues that

three important concepts, namely, irrevisability, maximal warrant, and
unmistakability, which are often read as entailing one another, are
nevertheless quite different and distinct concepts. I shall argue that,
far from being independent, they are closely interlinked, and that if
Tlumak were right these concepts could have a devastating effect on
Descartes’ cogito. These concepts, together with the claim about their
independence, are also utilized by Tlumak to show how to escape from
the Circle; but having supplied the groundwork, I shall leave that as
an exercise for the reader.
Let us begin with Tlumak’s definitions:

(I) h is irrevisable for S at t if and only if (i) on the basis of some (possibly
empty) set of evidential propositions, e , S is justified in believing h at t;
and (ii) there is no t’ and possible e ’ such that t’ is later than t, e is a
subset of e ’, and e ’ fails to justify S in believing h at t’, where not-h is not
a member of e ’ (T, 46).
(M) h is maximally warranted for S if and only if S is warranted in believing
h at t, and it is inconceivable that there is an h’, a t’ (where there are
no restrictions on the choice of t’), an S ’ (which may or may not be
restricted to S ) and an e ’, such that e ’ confers greater warrant on h’ for
S ’, at t’ than h has for S at t (T, 50).
(U) [C]ertainties are beliefs which cannot be mistaken (T, 45).

1 Hereinafter T.

269
270 Appendix A

At the outset, Tlumak lists six conditions for adequacy of analysis of

the concept of certainty. He is careful to say that these are not intended
to be jointly exhaustive conditions; however, any analysis that does not
square with them should not be regarded as adequate. It is not my pur-
pose to offer, and analyze, that list here; it would take me far outside the
scope of this book. I simply wish to grant these conditions to Tlumak
and to raise an objection or two. Tlumak says: “(M) fulfills all adequacy
criteria. So unless it turns out to be too strict, ruling out indubitable
knowledge of the superstructure of Descartes’ reconstructed science,
(M) seems a worthy contender.” (T, 50) He proclaims his real concern
to be the following: “I want to show that when viewed abstractly, and
not in the context of other Cartesian doctrines, irrevisability, maximal
warrant, and unmistakability are logically distinct notions” (T, 51). I
will argue that these are not logically distinct notions, and that if they
were distinct and were imported into the Cartesian system, they would
wreak havoc on it, especially on the cogito.
Tlumak claims that maximal warrant entails irrevisability, but the
converse is not true. Let us grant the former. For the latter claim, he
offer this argument:

For suppose that no evidence over and above that which supports h for S at
t could conceivably overturn h. Even if this required that h have maximally
good evidence bearing on it . . . other propositions of other kinds, for which
other kinds of evidence are relevant, could conceivably have still more impres-
sive credentials. Constant rational acceptance of h is compatible with greener
pastures in an entirely different ball park. (T, 52)

Let us suppose that there are two domains of inquiry, D 1 , and D 2 ; h 1

is entertained in D 1 , and h 2 in D 2 . One way of understanding Tlumak
is to suppose that the maximal warrant that any hypothesis in D 1 can
have is .8, whereas the maximal warrant of a hypothesis in D 2 can rise
as high as .9. Thus, constant rational acceptance of h 1 is compatible
with h 2 having still more impressive credentials. But this will not do:
Assigning .8 shows we are not confident that no new evidence will
overturn h; we suppose that there is a .2 probability of new evidence
surfacing that will overturn h 1 . On the contrary, we are assured “that
no evidence over and above that which supports h for S at t could
conceivably overturn h” (my emphases). In that case, assigning any
probability less than 1 is utterly inexplicable, inexplicably cautious.
 Jeffrey Thumak’s “Certainty and Cartesian Method” 271

On the other hand, if both hypotheses in their respective domains

enjoy a probability of 1 (maximal warrant), it is difficult to tell which
enjoys the greener pasture, or why.
The relationship between maximal warrant and irrevisability thus
understood would have a devastating effect on the cogito. The doubter
in the Second Meditation, having arrived at the cogito, would be left
in a quandary: “Well,” he might say, “I have produced maximal evi-
dence for the cogito and it is irrevisable; but that is not enough. For
all I know, there are other propositions in other domains with greater
warrant than the cogito. Why, then, is it rational for me, in view of such
a possibility, to stick with the cogito as my starting point – especially
when I claim that there is no certainty with greater warrant than the
cogito?”
Consider next Tlumak’s argument in support of the claim that un-
mistakability does not entail irrevisability (T, 52). Here is what he
says: “All necessary truths are unmistakable. Some necessary truths
are revisable. One can readily imagine a persuasive argument com-
ing along warranting suspension of judgment on a necessary truth.
Therefore, some revisable truths are unmistakable, so that unmis-
takability does not entail irrevisability.” (T, 52) The doubter in the
Second Meditation, having arrived at the cogito, might think: “Well,
I think the cogito is a necessary truth; but that is not enough. Some
necessary truths are revisable because ‘a persuasive argument [might
come] along warranting suspension of judgment on a necessary truth’.
Consequently, for all I know, the cogito may be just such a neces-
sary, but revisable, truth.” Is it not far more plausible to think that
when Tlumak says, “Some necessary truths are revisable,” he sim-
ply means, “What appear to be necessary truths (but are not) are
revisable”?
Tlumak might respond thus:

The argument rests on the assumption that for any proposition whatsoever,
it is possible warrantedly to withhold it if it is false, and possible warrantedly
to believe it if it is true. But this assumption falsely rules out the possibility of
unconfirmable truths [as well as the possibility of unfalsifiable falsehoods?].
Generalizing, since truth conditions, justification conditions, and belief con-
ditions are not invariably correlated, one may say that irrevisability (a concept
of privileged evidentiary status) does not entail unmistakability (a concept of
privileged doxastic status). (T, 52–53)
272 Appendix A

The doubter in the Second Meditation, having arrived at the cogito,

might think: “Well, I believe the cogito to be irrevisable; but that is not
enough. I might well be guilty of falsely ruling out the possibility of
it being an unfalsifiable falsehood. Consequently, for all I know, since
truth conditions, justification conditions, and belief conditions are
not invariably correlated, I may not have privileged evidentiary status
for the cogito, merely privileged doxastic status. In short, I had better
abandon claims to knowledge – even in the case of the cogito.”
There is also this concern: Irrevisability, Tlumak claims, is an epis-
temic concept (a concept of privileged evidentiary status), in contrast
to mistakability, which he says is a doxastic concept. If this were true,
one could see how the former does not entail the latter. But it is dif-
ficult to see irrevisability, as defined, as an epistemic concept: It uses
the notion of ‘justified in believing h ’, not the notion of ‘knows h’.
The latter entails that h is true, the former does not. But suppose that
irrevisability were an epistemic notion. Then Tlumak’s argument in
favor of the claim that irrevisability does not entail maximal warrant
collapses. For if h is irrevisable, and hence true, there could not be any
fresh evidence, e ’, to justify not-h, or to withhold assent from h; h would
be maximally warranted.
“Of the three” notions, says Tlumak, “perhaps unmistakability fares
least well, largely because it lacks an evidentiary component relativiz-
ing certainty to persons in epistemic situations” (T, 53). As it stands,
this is true. However, there is an interesting exit: We can analyze the
notion of certainty employed by [U] in terms of [M]. We have already
seen that [M] entails [I]; consequently, [U] would entail [I] as well.
[M] and [I], according to Tlumak, are epistemic concepts; the whole
purpose of offering [M] and [I] is that any analysis of metaphysical cer-
tainty must satisfy them; perhaps other things might be required in the
ultimate analysis. But this will at least insure that [U] has “evidentiary
component relativizing certainty to persons in epistemic situations,”
and that it need not, therefore, fare least well.
 Appendix B

Comments on Robert Nozick’s “Fiction”

In “Fiction,”1 Robert Nozick suggests the possibility that he, Nozick, is

a fictional character and so is everyone else, except for the one who is
the author of the piece. My aim here is not to argue with “Fiction” as a
whole, but only with aspects of it that are central to Descartes. Nozick
says:

“What!” you say, “we don’t really undergo suffering? Why it’s as real to us as
Oedipus’ is to him.” Precisely as real. “But can’t you prove that you really exist?”
If Shakespeare had Hamlet say “I think therefore I am,” would that prove to
us that Hamlet exists? Should it prove that to Hamlet? And if so what is such a
proof worth? Could not any proof be written into a work of fiction and be pre-
sented by one of the characters, perhaps one named “Descartes”? (Such a char-
acter should worry less that he’s dreaming, more that he’s dreamed.) (N, 462)

Nozick starts with the assumption that Hamlet is a fictional char-

acter. Then, of course, no matter what speech Shakespeare puts into
Hamlet’s mouth, including any proof of Hamlet’s existence, it is not
going to prove that Hamlet exists. Ex hypothesi, it could not prove to
Hamlet that he exists. By analogy, then – so Nozick seems to suggest –
if we, like Hamlet, are fictional, no proof of our existence that we can
mouth will confirm us in our existence. Recall how in former years
philosophers used to prove the existence of “other minds.” When we
are in a certain mental state (say, pain), these philosophers said, we

1 Hereinafter N.

273
274 Appendix B

exhibit a typical physical reaction (say, wincing). Our mental state is

unobservable to the rest of the world; not so our physical reaction, but
the two are causally connected. One can thus infer the cause (mental
state) from the effect (physical reaction). By analogy, then, we infer
that someone is in pain (mental state) when we observe that person
wincing (a typical physical reaction). Nozick presents the flip side of
such an argument. Hamlet is fictional, and could “prove” his existence
by any philosophical proof that could be put into his mouth; but such
a proof is worthless. Ergo, our proof – the cogito – of our existence may
be worth just as much.
Will this do against Descartes? Let us see. We do engage in thinking,
and this does not require a formal proof so much as being engaged
in thinking. If there is a “proof” of our existence, it comes later. We
know that Hamlet cannot engage in thinking – that is what being
fictional means. It is not enough that Nozick merely stipulates that
Hamlet thinks or is engaged in thinking. Nozick might retort, “Aren’t
you just begging the question? You are assuming that you are real and
Hamlet fictional.” Descartes might rejoin that Nozick just assumes that
Hamlet is fictional; Nozick must first distinguish between a fictional
character and a real character. Descartes would distinguish them (so
long as we were talking only about characters) as those that lack the
capacity to think (fictional) and those that posses it (nonfictional).
One might go a step further and say: If Nozick does not draw an initial
distinction, his conclusion can actually go in the other direction. It is
not that both Hamlet and I are fictional – we are both real; and that
reality is proved by our both being engaged in the cogito.
Nozick’s response might simply be that he need only demonstrate
the possibility of the cogito failing to demonstrate the existence of the
‘I’ engaged in thinking, not deny the possibility that what we had
hitherto thought to be fictional characters are not really fictional. Such
a possibility injects doubt into the heart of the Cartesian enterprise.
I construct the Cartesian response as follows: When Nozick says, “If
Shakespeare had Hamlet say ‘I think therefore I am,’ would that prove
to us that Hamlet exists?” (N, 462, my emphasis), Descartes would
deny that it proves any such thing to us, any more than it would prove
to us that the mechanical robot (lacking the property to think) exists
if it says, “I am thinking, therefore I am.” The cogito is not, nor is
it intended to be, an abstract, detached, formal proof; it requires a
 Comments on Robert Nozick’s “Fiction” 275

thinking thing as a starting point – a thinking thing that could say to

another (perhaps to its author), or even to itself: “Attempt to persuade
me that I am fictional: if you fail, then of course I am not fictional; on
the other hand, if you succeed in persuading me, it would be a pyrrhic
victory: for you have convinced me. Therefore, I am not fictional.”
Let us say, thinkingF is the thinking of a fictional entity, and
thinkingR is the thinking of a real entity. The first step is to deny
that the two types of thinking are the same. ThinkingF cannot, from
a Cartesian perspective, conduct the cogito thought experiment; only
thinkingR can do so. An author might create a fictional character and
make that fictional character say that he (the fictional character) is un-
dergoing the cogito thought experiment, but in reality he is doing no
such thing; it is merely the author’s projection. ThinkingR is required
for such a projection.2
Does anything that is a thinkingR thing even understand what it is
to be a thinkingF thing? This is not analogous to the question, what is
it like to be a bat? That bats have experiences is assumed; what is in
question is the accessibility of such experiences. The question before
us is a shade deeper. There is the question, what is it like for us to have
experiences of fictional entities? But before that is the question, what
is it like for fictional entities to have fictional experiences? And this
question follows on the heels of the primary question, can fictional en-
tities have experiences? If they do, are these experiences like those of
humans, or not? Are we imprisoned in our own mode of thinkingR in a
way in which thinkingF things are not imprisoned in their mode? How
then do they distinguish between the two modes of thinking? Descartes
would aver that anyone who can persuade himself that he is a thinkingF
thing has deluded himself; in the very act of persuading himself that
he is a thinkingF thing, he has demonstrated that he is a thinkingR
thing. What is quite lacking in Nozick’s proposal is an account of the
genuine difference – or an account of why we are mistaken in claiming
that there is a difference – between thinkingF and thinkingR , such that
“I am thinkingR , therefore I exist” is acceptable, but “I am thinkingF ,
therefore I exist” is not. What is acceptable, at best, is only the innocu-
ous “I am thinkingF, therefore I existF ” (where exist F stands for exists
in a fictional world – a roundabout way of saying nonexistent).

2 Or is it? See the arguments following the next paragraph.

276 Appendix B

Finally, let us suppose the author of the fictional world to which

Hamlet belongs is S. Could S be fictional? Yes, he could be, says Nozick,
provided that he had been created in turn by another author, S ’; S ’
could be fictional, provided that he had been created in turn by author,
S ’’, and so on. “Must there be a top-floor somewhere, a world that
itself is not created in someone else’s fiction. Or can the hierarchy go
on infinitely?” (N, 463) Clearly, such a hierarchy of fictional worlds
could not go on indefinitely. Why? To begin with, notice that this is
not analogous to asking if there is a first cause when there is a series
of causes, each cause in the series being real and producing the next
cause. The issue is not about the reality of the first cause; the question
is only if there is an ultimate, first, noncaused cause. Can there be a
series of fictional causes, each cause in the series being produced by an
earlier cause? Can there be an ultimate, first, fictional cause, one not
produced by an existing, real entity? Or, at the other extreme, can a
fictional cause produce a real (that is, nonfictional) effect? Why not,
given that Nozick seems to place no plausible restriction on what is to
serve as a cause?
Let us say that there are three types of worlds – the actual world,
possible worlds, and fictional worlds;3 and there is the possibility
that there is nothing. Why is there something rather than noth-
ing? If there were only fictional worlds, it would prompt a similar
question: Why are there fictional worlds rather than nothing? An
answer to the latter question would be even more puzzling than
Nozick’s own puzzling answer to the former question.4 If there were
at least one real world with a thinker in it, the existence of the fic-
tional world could easily be accounted for in terms of the thoughts
of that thinker. If not, how could a fictional world spring up from
nothing?5

3 For my purposes, an intuitive distinction between possible worlds and fictional worlds,
allowing for overlap, is more than enough.
4 Nozick attempts to answer this question in the intricate Chapter 2, “Why Is There
Something rather than Nothing?” in his Philosophical Explanations. For my arguments
against Nozick’s theory, see “Something, Nothing, and Explanation.”
5 This might raise some interesting questions about the type and nature of causation,
because Nozick entertains not only the possibility that something could come from
nothing, but also the possibility “that a character of one world creates another fictional
world wherein a character creates the first world” (N, 463). In the latter case, which,
if any, came first?
 Comments on Robert Nozick’s “Fiction” 277

Descartes says, in the paragraph containing the cogito claim in Med-

itations on First Philosophy, “Is there not a God, or whatever I may call
him, who puts into me [“puts into my mind” – French version] the
thoughts I am now having? But why do I think this, since I myself may
perhaps be the author of these thoughts?” (CSM II, 16; AT VII, 24)
So, if I were the only existing being in the universe – if there were
no one but me to author my thoughts – the very thought that I might
be a fictional entity, let alone the thought that I am a fictional entity,
would assure me of my reality. No matter how fictional, how varied
the universe is, no matter what its genre – “tragedy, farce, or theater-
of-the-absurd” (N, 463) – the evil genius, or Nozick, cannot persuade
me, if indeed it is I who am persuaded, that I am fictional.6 This is not
something that Hamlet can do or claim, surely.

6 “Our author, we know, is outside our realm, yet he may not be free of our problems.
Does he wonder too whether he is a character in a work of fiction, whether his writing
our universe is a play within a play? Does he have me write this work and especially this
very paragraph in order to express his own concerns?” (N, 463) It is the burden of the
present argument to demonstrate that the concerns of such an author are ill-founded.
 Appendix C

Cogito and the Port-Royal Logic

The principal author of Logic or The Art of Thinking,1 famously known

as the Port-Royal Logic, was Antoine Arnauld, although there is some in-
dication that parts of it may have been written by Pierre Nicole (1625–
1695) as well as by the Messieurs of Port Royal. Arnauld got his hands
on the original manuscript of Descartes, Rules for the Direction of the
Mind (also found in the remnant possessions of Leibniz) – long be-
fore it was published – through the good graces of one of Descartes’
ardent supporters, Claude Clerselier. Consequently, when he came to
compose Port-Royal Logic, Arnauld was heavily influenced by Descartes’
work, which Arnauld duly and explicitly acknowledged.2 Incidentally,
Port-Royal Logic was written for a young nobleman whose father, Duc du
Luynes, had translated Descartes’ Meditationes de Prima Philosophiae into
French, a translation that Descartes had approved. It is surely, then,
worth considering whether the authors of Port-Royal Logic would have
construed the cogito as an argument or as an intuition. This will have
to be a historical conjecture, since the Port-Royal Logic does not treat
this issue explicitly.

1 Hereinafter AN.
2 Buroker, in her Introduction to Logic or The Art of Thinking, writes: “Rene Descartes
was the true philosophical father of the Port-Royal Logic. . . . Arnauld and Nicole whole-
heartedly embraced Descartes’ rationalism. In fact, the Port-Royal theory of knowl-
edge is taken almost verbatim from Descartes.” (AN, xx) For Descartes’ huge influence
on Port-Royal Logic, see also AN, xx–xxiii. I am indebted as well to the Introduction in
the Port-Royal Logic edited by James Dickoff, Patricia James, and Charles W. Hendel.

278
 Cogito and the Port-Royal Logic 279

In “Fourth Part of the Logic” of Port-Royal Logic, Chapter 6 begins

with a Rule, thus: “Rules concerning axioms, that is, propositions which
are clear and evident in themselves . . .” (AN, 246). There are two car-
dinal rules. First is the rule that no axioms require demonstration, that
is, proof, although some axioms may require explanation. The rule is
stated thus:

In order to show clearly that an attribute applies to a subject – for example, to see that
being larger than its part applies to the whole – whenever we need only to consider the
two ideas of the subject and the attribute with moderate attention, so that we cannot do
it without recognizing that the idea of the attribute is truly included in the idea of the
subject, then we have the right to take this proposition for an axiom, not needing to be
demonstrated. This is because it has in itself all the evidence that could be given in a
demonstration, for a demonstration could do nothing more than show that this attribute
applies to the subject by using a third idea to show this connection, which is already seen
without the aid of a third idea. (AN, 248)

The second is the rule that what is not an axiom must be demonstrated.
It is stated as follows:

When the mere consideration of the ideas of the subject and attribute is not enough to
show clearly that the attribute applies to the subject, the proposition that affirms it should
not be taken for an axiom. (AN, 249)

The text immediately goes on to say: “But it must be demonstrated, by

using some other ideas to show the connection . . .” (AN, 249).
Prior to the statement of the two rules, Port-Royal Logic illustrates
and comments on what is an axiom as follows:

We should not fear that there are people who seriously agree with the con-
clusion that we do not know what is true or false about anything in itself. For
some things are so simple and evident – such as: “I think, therefore I am”; “the
whole is greater than its part” – that it is impossible to doubt seriously whether
they are in themselves the way we conceive them to be. The reason is that we
could not doubt them without thinking of them, and we could not think of
them without believing them to be true, and consequently we could not doubt
them.3 (AN, 248)

It concludes: “When we only need to consider the idea, the proposition

can be taken for an axiom, especially if this consideration requires

3 See also AN, 228–229, where Port-Royal Logic briefly refers to the cogito against the
background of the skepticism of the New Academicians and the Pyrrhonists.
280 Appendix C

only the moderate attention of which all ordinary minds are capable”
(AN, 248). Port-Royal Logic takes the cogito to be a crystal clear example
of an axiom, not a demonstration. It is in no need of a proof, although
it might be in need of an explanation (AN, 249). This is heavy testimony,
in my view, that Arnauld – a preeminent metaphysician and theologian,
writing a book on logic and epistemology, conversant with Descartes’
views, in close contact with Descartes and the Cartesians – should have
held the cogito not to be an argument.4 Surely, this should take us a
step toward settling the exegetical issue, even if it leaves open the
philosophical question.
“If I want to know,” says the Port-Royal Logic,

whether the soul is spiritual, suppose that, not getting to the bottom of it
right away, I choose the idea of thinking to clarify the matter. It is clear that
it will be useless to compare the thinking with the soul if I do not conceive
any connection between thinking and the attribute spiritual by which to judge
whether it does or does not apply to the soul. I can easily say, for example, that
the soul thinks, but I could not infer from this, therefore it is spiritual, unless I
conceive some connection between the terms “to think” and “spiritual.” (AN,
135–136)

Quite clearly, the last sentence is attempting to supply the missing

premise without which the argument would be invalid. Confronted
with the obviously parallel argument in which I wish to know whether
“I exist,” is true, Port-Royal Logic would not regard “I think” as a suffi-
cient premise (as Kenny contends) from which to extract “I exist.” It
would enjoin supplying the missing premise (without which it would
deem the argument invalid).
A careful, detailed study would have to address at least the follow-
ing issues. First, according to the Port-Royal Logic, how is a judgment,
especially a simple judgment – in particular, the cogito – to be formed?
Second, there is a rather serious objection against those who, like
Kenny, maintain that the cogito is an immediate inference. Conse-
quently, it will need to be asked, what immediate inferences did
Port-Royal Logic recognize? It seems to have recognized only the stan-
dard Aristotelian ones. For example, in the “Second Part of the Logic,”

4 “The authors [Arnauld and Nicole] also cite Descartes’ famous cogito argument [sic]
(‘I think, therefore I am’) as the primary example of knowledge by intellectual
intuition.” Buroker, “Introduction,” xxi.
 Cogito and the Port-Royal Logic 281

Chapters 17–19 (AN, 129–134) deal with conversion; therein seven

axioms and three rules are stated to justify the conversions from A
(by limitation), E, and I, propositions, none being converted from
O. Skipping the stock difficulty of how these axioms and rules are
to be justified at the end of the First Meditation, let us ask which, if
any, immediate inference in the guise of conversion would be rele-
vant here. It turns out that none are relevant. E and I clearly will not
do; A can be converted only by limitation, which means that A would
have to be known to be true before I can be demonstrated to be true.
At the very least, then, three things will need to be shown: (1) that
the cogito is an immediate inference, (2) the type of immediate infer-
ence it is, and (3) how the axiom or rule justifying the cogito as an
immediate inference is in turn justified from within the framework of
Port-Royal Logic in such a way that it does not run afoul of the First
Meditation.
Third, Port-Royal Logic’s account of conditional statements has trou-
blesome implications for their use in the cogito-argument, either as a
premise or as a rule of inference. Conditional statements in a deduc-
tion or demonstration, Descartes would have contended, have to be
necessary truths. If so, these conditional statements in a deduction
would themselves raise the question of how they are to be justified.
Port-Royal Logic’s account of a conditional statement (AN, 99–101), re-
flecting the Cartesian account, is vastly different from ours. For one
thing, in ours it is a material implication; in Port-Royal Logic’s account,
it is an inference. Port-Royal Logic says,

Conditionals are propositions made up of two parts connected by the condi-

tion if. The first part, which contains the condition, is called the antecedent,
and the other part is called the consequent: “if the soul is spiritual” – this is
the antecedent – “it is immortal” – this is the consequent.
This inference is sometimes mediate and sometimes immediate. It is only
mediate when there is nothing in the terms of either part that links them, as
for instance:

If the earth is immovable, the sun revolves.

If God is just, the wicked will be punished.

These inferences are quite valid, but they are not immediate because the
two parts, lacking a common term, are connected only by what one has in
mind, which is not expressed. . . .
282 Appendix C

To determine the truth of [immediate inferences] we consider only the

truth of the inference. For even if both parts are false, if the inference from
one to the other is valid, the proposition insofar as it is conditional is true.

If a creature’s will can obstruct the absolute will of God, God is not omnipotent.
(AN, 99–100)

It is quite evident that it would not be enough to maintain that

the cogito is an immediate inference. “I exist” is not logically linked to
“I think,” because there is nothing in the terms of either part that links
them, and hence it is prima facie only a mediate inference. Construed
as a conditional, it could still be true, while the antecedent is false;
therefore, the truth of the antecedent would have to be justified in-
dependently. But if the truth of the antecedent is justified first and
independently, Descartes need look no further for the first truth. If
the cogito is justified on the basis of a conditional rule, we would run
into the same problem.
Fourth, how precisely should the relationship between the cogito and
the two rules governing axioms be explicated? Fifth, Port-Royal Logic
recognizes principally two sorts of truths. The first sort of truth relates
to the nature of things and their immutable essences. These statements
are deemed necessary truths, if true. The second sort of truth relates
to existing things, especially contingent events, both past and future
contingent events (AN, 263). Among the questions to be answered
are: Does the cogito-statement obviously fit into one of the two types of
statements? If so, how would Port-Royal Logic have classified it?
Finally, Port-Royal Logic says that

even if it is necessary to have the subject matter in order to organize it, it is

not, however, necessary to learn how to find the subject matter before learning
how to organize it. In order to learn how to organize the subject matter, it is
sufficient to have some general subjects to use as examples. . . . It is true, then,
that we must have some subject matter in order to apply the rules of arguments
to it, but it is false that this matter has to be found by the method of topics.
(AN, 181)

Port-Royal Logic is here quarreling with Peter Ramus, who held ex-
actly the opposite view. Aristotle and the schoolmen discussed the
topics – the logic of discovery or invention – after enunciating the rules
for valid reasoning – the logic of justification or explanation. Ramus
contended that we need to discover thecontent of our argument first,
 Cogito and the Port-Royal Logic 283

before we can put its premises and conclusion in proper order. If the
cogito is an argument, shouldn’t Descartes go along at least part of the
way with Ramus? For without knowing or discovering the content of
the cogito, I could not determine its form; and if I could not determine
its form – because whatever rules determine the form of a statement
or an argument may be open to doubt in the First Meditation – I could
not determine its validity. If it is doubted that in order to find the con-
tent of the cogito, the method of topics (or, for that matter, any rule
or method) is necessary, then the enunciation of the rules of valid
reasoning ought to come only after the content of the cogito has been
established.
 Appendix D

Bacon and Descartes

Descartes is remarkably united with Sir Francis Bacon in his logic

of discovery. They both doubted and rejected the ancient systems of
knowledge and the unexamined evidence of the senses. The human
mind, said Bacon in The New Organon,1 is beset by idols of four kinds:
Idols of the Tribe, Idols of the Cave, Idols of the Marketplace, and
Idols of the Theater. The Idols of the Tribe represent general human
failings in virtue of humanity’s mental make-up; Idols of the Cave rep-
resent failure owing to the peculiarities of the individual; Idols of the
Marketplace reflect the poor choice of language that obstructs the
understanding; and finally, Idols of the Theater represent the various
outmoded ancient and modern philosophical systems. The mind had
to be purged of these idols in order that true inquiry could begin.
But one also has to cast out the old logic:

As the sciences which we now have do not help us in finding out new works, so
neither does the logic which we now have help us in finding out new sciences.
(B, 41)

Again:
The logic now in use serves rather to fix and give stability to the errors which
have their foundation in commonly received notions than to help the search
after truth. So it does more harm than good. (B, 41)

1 Hereinafter B.

284
 Bacon and Descartes 285

So learned disputations had to be replaced by experiments. Ex-

periments must be performed that are of “no use in themselves but
simply serve to discover causes and axioms” (B, 96). These experi-
ments are called Experimenta lucifera, experiments of light, which are
to be distinguished from Experimenta fructifera, experiments of fruit.
The experiments of light will give us a collection of the first particu-
lars; however, “the understanding must not . . . be allowed to jump and
fly from particulars to axioms” (B 98). It should move cautiously and
carefully.

But then, and then only, may we hope well of the sciences when in a just scale
of ascent, and by successive steps not interrupted or broken, we rise from
particulars to lesser axioms, and then to middle axioms, one above the other;
and last of all the most general. (B,98)

Both Bacon and Descartes wanted to create a new system of knowl-

edge on well-founded truths: Bacon began with simple, careful obser-
vations and experiments, Descartes with simple, powerful intuitions.
The generalizations and higher generalizations came later, emphati-
cally not before. What role did the syllogism play in Bacon’s logic of
discovery? Not much.

The syllogism is not applied to the first principles of science, and is applied
in vain to intermediate axioms, being no match for the subtlety of nature. It
commands assent therefore to the proposition, but does not take hold of the
thing. (B, Aphorism 13, p.41)

The syllogism consists of propositions, propositions consist of words, words

are symbols of notions. Therefore if the notions themselves (which is the root
of the matter) are confused and over hastily abstracted from the facts, there
can be no firmness in the superstructure. Our only hope therefore lies in a
true induction. (B, Aphorism 14, p. 41)

What we find in Bacon is remarkably parallel to what we find in

Descartes. The first principle – in Descartes’ terminology, the first
item of knowledge – is not to be founded on a syllogism. To “take
hold of the thing,” one needs to directly experience it. To rely on a
syllogism is ultimately to rely on words, and words are often confus-
ing, whence the Idols of the Marketplace. On such a foundation one
cannot erect a superstructure of knowledge. Bacon concluded that
the way to search for and discover truth is to “derive axioms from the
286 Appendix D

senses and particulars, rising by a gradual and unbroken ascent, so

that it arrives at the most general axioms last of all. This is the true way,
but as yet untried.” (B, 43) Descartes concluded likewise: “It is certain
that if we are to discover the truth we must always begin with particu-
lar notions in order to arrive at general ones later on” (CSM II, 271;
AT VII, 206). There is simply too much similarity in the structure of
their respective enterprises. Perhaps, then, these are powerful histor-
ical reasons to conclude that the cogito was not intended by Descartes
to be an argument. If it were, then it would not be the first item of
knowledge.
 Appendix E

Comments on Anthony Kenny’s “Descartes

on the Will”

Kenny’s paper “Descartes on the Will”1 is the locus classicus on

Descartes’ theory of will, error, and judgment. Since much of what
Kenny says in his paper has no direct bearing on the core thesis of this
book, I confine myself to making an observation or two on aspects of
his paper that bear more directly on Chapter 8.

I
There is first a purely historical issue. Kenny is keen to establish that
Descartes’ theory of the will in the Regulae – published only posthu-
mously, composed as early as 1628, if not earlier – is sharply different
from the one he offered in later works such as the Principles of Philoso-
phy and Meditations on First Philosophy. Consequently, when Leslie Beck,
in his The Method of Descartes (17), cites a passage from the Regulae that
smacks of what Descartes was alleged to have claimed only later, Kenny
responds thus:

When he wrote the Regulae, Descartes still held the orthodox Thomist
view. . . . There is, it is true, one passage which Leslie Beck sees as presup-
posing the later view “that judgement whether in its pure or practical use is an
assent or dissent, an act of will.” But the passage is most naturally interpreted
as applying only to practical matters. Descartes exhorts us, in studying, “to
think solely of increasing the natural light of reason, not with a view to solving

1 Hereinafter KW.

287
288 Appendix E

this or that scholastic problem but in order that in all the happenings of our
life, our intellect may show our will what alternative to choose.” [CSM I, 10;
AT X, 361] In this passage it is choice, and not judgement, not even practical
judgement, which is attributed to the will. (KW, 135)

I shall argue that Beck’s view is defensible. The passage that Beck
cites comes from the very first rule of the Regulae. First, while it is true
that Descartes speaks of choice, he explicitly speaks of judgment as well
in this rule, and judgment not only in the practical context, but also in
the pure – arguably theoretical (that is, scientific or metaphysical) –
context. Indeed, the theoretical context is Descartes’ predominant
concern in this rule. Descartes wants men to pursue several sciences,
not just one or two, on the grounds that this is the only way to attain the
most “respectable and commendable ends” – “good sense,” “universal
wisdom” – namely, “contemplating the truth”: “For every other science
is to be valued not so much for its own sake as for its contribution to
universal wisdom.” (I am slightly overreaching when I speak not of a
but of the most commendable aim; I have in mind the final paragraph
of the Third Meditation, which dovetails nicely with what Descartes
refers to as the aim that produces happiness.)
Second, Descartes wants to enable us to make judgements “in all
the happenings of our life” (my emphasis).
Third, I added the emphasis to indicate that the context of the
first rule is of cardinal importance. Its caption reads: “The aim of our
studies should be to direct the mind with a view to forming true and sound
judgements about whatever comes before it” (CSM I, 9; AT X, 359). In this
first rule, Descartes is bent on persuading us that people often reason
erroneously about scientific studies because they rely on a mistaken
analogy. In practical matters, they say, a person cannot do everything:
“[O]ne man cannot turn his hand to both farming and harp-playing.”
From this, says Descartes, they incorrectly conclude that “the same
must be true of the sciences as well.” Hence, they advocate studying the
sciences separately; presumably thinking along the lines of one man,
one science. But the practical arts require some “bodily aptitude and
practice,” and so prove restrictive; whereas knowledge is acquired by
the mind: “[T]here is no need to impose any restriction on our mental
powers; for the knowledge of one truth does not, like skill in one art,
hinder us from discovering another” (CSM I, 9; AT X, 359–360).
 Comments on Anthony Kenny’s “Descartes on the Will” 289

Indeed, the passage that Kenny cites from the first rule in defense
of his interpretation, namely, that Descartes was primarily concerned
with choice in practical matters, continues thus: “He will soon be sur-
prised to find that he has made far greater progress than those who
devote themselves to particular studies, and that he has achieved not
only everything that the specialists aim at but also goals far beyond any
they can hope to reach” (CSM I, 10; AT X, 361). ‘He’ refers to some-
one who seriously wishes to investigate the truth, not, or not only, to
someone interested in making practical choices. ‘Particular studies’
refers to particular scientific disciplines. Finally, in what immediately
precedes the passage that Kenny cites, Descartes says that “the sciences
are so closely interconnected that it is much easier to learn them all
together than to separate one from the other. If, therefore, someone
seriously wishes to investigate the truth of things, he ought not to
select one science in particular. . . . He should, rather, consider sim-
ply how to increase the natural light of his reason. . . .” The notion
of the natural light of reason is used far more in the context of sci-
ence and metaphysics, as in the Third Meditation, than in the practical
context.
I conclude: There is far greater continuity between Descartes’
earlier work and his later work than Kenny would have us believe.

II
“Now as far as ideas are concerned,” says Descartes,

provided they are considered solely in themselves and I do not refer them
to anything else, they cannot strictly speaking be false; for whether it is a
goat or a chimera that I am imagining, it is just as true that I imagine the
former as the latter. As for the will and emotions, here too one need not worry
about falsity; for even if the things which I may desire are wicked or even non-
existent, that does not make it any less true that I desire them. Thus the only
remaining thoughts where I must be on my guard against making a mistake
are judgements. (CSM II, 26; AT VII, 37)

Kenny finds this argument “strange.” “One could,” says he, “as well
argue that judgements in themselves could not be false, on the grounds
that whether what I judge is true or false, it is none the less true that
I judge” (KW, 145).
290 Appendix E

What is Descartes’ justification for lumping desires with judgments?

Kenny supplies an answer – an answer, he thinks, that Descartes should
have given. “If we take together all those mental activities which can
have rightness or wrongness ascribed to them, we will find that they
include all those activities which Descartes ascribed to the will and
exclude those which he ascribed to the intellect” (KW, 144). However,
this is open to the objection that “[t]he right, we might say grandly, is
a genus of which the species are the true and the good; and Descartes’
classification emphasizes the unity of the genus at the cost of ignoring
the diversity of the species” (KW, 144). But, says Kenny, according to
Descartes a mind properly used is infallible, and hence every erroneous
judgment is a moral failure. “What theologian or philosopher,” asks
Descartes – Kenny’s translation is quite different from CSM II, 105 – “or
indeed what rational man has ever denied that we are in less danger of
error the more clearly we understand something before assenting to it,
and that it is a sin to make a judgment before the case is known?” (KW,
144–145) Furthermore, the truth-value of a judgment is dependent on
what is asserted, and not on the assertion. What is asserted (phrastic) is
on a par with the content of the desire; the assertion is on a par with
the desire (neustic).
This raises a question or two. First, suppose that I judge or assert
the proposition, “The sun goes around the earth.” The phrastic “The
sun goes around the earth” is false; the judgment (containing the
assertion) is false, too. It is true that I have also sinned. It may also be
the case that I have made an inelegant, injudicious, implausible, even
funny, claim. But my sinning – or any of these other things – is parasitic
upon my making an erroneous judgment. With desires it is different.
Suppose I desire that everyone stop smoking. The phrastic “Everyone
stop smoking” – let us assume it to be so, without further argument –
is false; on the other hand, my desire may be right or wrong, cruel
or kind, serious or laughable, but not true or false – regardless of the
truth-value of the phrastic. Consequently, the neustic of a judgment
and the neustic of a desire yield essentially different outcomes.
Second, “The point Descartes has missed is what we may call – to
adapt an expression of J. L. Austin’s – ‘the onus of match’” 2 (KW, 145–
146). Consider the assertion of a simple declarative sentence, “The sun

2 J. L. Austin, “How to Talk,” in his Philosophical Papers, 190.

 Comments on Anthony Kenny’s “Descartes on the Will” 291

goes around the earth.” The judgment is false, because the state of
affairs that obtains is one in which it is the earth that goes around the
sun. What shall we blame? Shall we blame the assertion or the state
of affairs? Well, if it is a map of a street, the onus is on the map to
precisely depict the street as it actually is; if it is an architectural plan
of how the street should have been laid out in the first place, then
the onus is on the street to match the plan. “Descartes, in lumping
together affirmation and desire, negation and aversion, confounds
the different onus of match involved in the two different kinds of
assent and dissent. This, it seems to me, is the fundamental defect in
his theory of judgement as an act of the will.” (KW, 146) Thus, Kenny.
It is difficult to come up with an example of an assertion in which
the onus of match is on the state of affairs. Consequently, I am going to
assume, pending examples and arguments from Kenny, that in the case
of assertions the onus of match is always on the phrastic component
of the assertion, and the aim of the neustic is always to get that match
right. Now, consider the following. A pious Christian astronomer, in
the days after Galileo, says, “I desire that the sun goes around the
earth.” Is it obvious that the onus of match is on the state of affairs?
Is it obvious that the onus of match is on the astronomer to bring his
desires in line with physical reality? If so, it would obliterate the differ-
ence between an assertion and a desire. (Notice: Crazy desires are no
less desires; and in the case of this example, if the astronomer believes
that his theory of the solar system would be enormously simplified if
indeed the sun went around the earth, it is not even clear that the
desire is that crazy.) Or suppose that I said, “I desire that everyone stop
smoking.” Is the onus of match on me? Must I now think, act, and
do things in order to bring about that state of affairs? And if I said,
“I desire that extraterrestrial beings descend on earth,” would I
similarly have an onus to do things that would bring about that state
of affairs? What would I have to do? The point is this: Desires do not
always, or necessarily, come attached with onuses; judgments never fail
to have an onus attached to them.
Third and finally, Kenny considers only desires and judgments as
lumped together, but they are not the only ones put together in one
class. Will, emotions, and chimeras belong in that class, as well as any
ideas that “are considered solely in themselves and I do not refer them
to anything else.” It is not clear how Kenny’s analysis could apply to
292 Appendix E

all of them. For example, consider chimeras. Here Descartes is simply

entertaining an idea, not making a judgment. It is not a desire. Let
us even grant that there is a phrastic involved, for the sake of argu-
ment. What is its neustic? Nonetheless, it is not obvious that we should
speak of the idea of chimeras as being true, if accidentally the alleged
chimeras do exist. The argument is so much stronger if the idea is false.
Even if the notion of rightness ties together judgment and desire, it is
not evident how rightness could tie in ideas like chimeras as well.
Descartes’ theory of will and judgment is decidedly not adequate
for several reasons, not least of which are reasons that Kenny himself
has devised (see especially KW, 156–159). But on the present score,
Descartes is not guilty of Kenny’s charges.

III
Let us follow Kenny following David Hume on the nature of human
free will. There is the liberty of spontaneity and there is the liberty
of indifference (A Treatise of Human Nature, 262–265). The liberty of
spontaneity is defined in terms of wanting, and goes thus: “[W]e are
free in doing something if and only if we do it because we want to
do it.” The liberty of indifference is defined in terms of power: “[W]e
are free in doing something if and only if it is in our power not to
do it” (KW, 146). What follows is a masterly account of how, like most
philosophers, Descartes attempted to combine the two forms of liberty
in his account of human free will. I have always found Hume’s account
unsatisfactory. The liberty of spontaneity dispenses with a crucial step,
and it does so without much justification. It is the step that answers
the question, “Are not the wants themselves determined?” If our wants
themselves are determined, that is, if we do not enjoy the liberty of
indifference with respect to our wants, then it is never in our power not
to want what we want. I grant that Descartes has a right to stipulate
how he is going to use the words ‘free’ and ‘voluntary’. “Descartes was
willing to call an act voluntary,” says Kenny, “if it was in accordance with
the agent’s desires, whether or not it was unavoidable; indeed, unlike
most scholastics, he was prepared to call an unavoidable, but welcome,
action ‘free’ as well as ‘voluntary’” (KW, 139). I contend only that he
has provided no philosophical justification for this. One need read no
further than Kant.
 Comments on Anthony Kenny’s “Descartes on the Will” 293

Descartes’ view about where the credit lies when we act correctly can
now be seen as equally puzzling. In Principles of Philosophy, Descartes
says, “[W]hen we embrace the truth, our doing so voluntarily is much
more to our credit than would be the case if we could not do other-
wise” (CSM I, 205; AT VIIIA,19). Surely, here Descartes is invoking the
liberty of spontaneity. Kenny comments on this passage as follows: “At
first sight this appears to mean that when we embrace truth clearly
seen, we are free not to embrace it; but from the sequel it is clear that
this is not so. Perhaps Descartes means that in such a case the credit
goes not to us but to the author of our nature, as the credit for the
precise operation of a machine goes to its maker.” (KW, 149, note 51)
I suspect that Kenny had in mind a passage that comes just six para-
graphs later: “[T]he minds of all of us have been so moulded by nature
that whenever we perceive something clearly, we spontaneously give
our assent to it and are quite unable to doubt its truth” (CSM I, 207; AT
VIIIA, 21; my emphasis). Undoubtedly, what we lack, according to this
passage, is the liberty of indifference. What ineluctably follows is that
we deserve credit about as much as the machine.
 Bibliography

Abbott, Edwin A. Francis Bacon: An Account of his Life and Works. London:
Macmillan, 1885.
Adams, Robert M. “Primitive Thisness and Primitive Identity.” The Journal of
Philosophy, Volume 76, No. 1 ( January 1979), 5–26.
Anderson, Alan, and Belnap, Nuel. “Enthymemes.” The Journal of Philosophy,
Volume 58 (1961), 713–723.
Arnauld, Antoine, and Nicole, Pierre. Logic or the Art of Thinking. Translated
and edited by Jill Vance Buroker. Cambridge: Cambridge University Press,
1996.
Ashworth, E. J. “Traditional Logic.” In Quentin Skinner, Eckhard Kessler, and
Jill Kraye, editors. The Cambridge History of Renaissance Philosophy. New York:
Cambridge University Press, 1988, pp. 143–172.
Augustine. On the Free Choice of the Will. Translated with an introduction by
Thomas Williams. Indianapolis: Hackett, 1993.
Austin, J. L. “How to Talk.” Philosophical Papers. New York: Oxford University
Press, 1961,
Ayer, Alfred J. “I Think, Therefore I Am.” In Willis Doney, editor. Descartes: A
Collection of Critical Essays. London: Macmillan, 1968, pp. 80–87.
Ayer, Alfred J. Language, Truth, and Logic. London: Victor Gollancz,
1964.
Bacon, Francis. The New Organon and Related Writings. New York: Bobbs-Merrill,
1960.
Baker, Gordon, and Morris, Katherine J. Descartes’ Dualism. New York:
Routledge, 1996.
Beck, L. J. The Method of Descartes: A Study of the Regulae. Oxford: Oxford
University Press, 1970.
Belaval, Y. Leibniz: Critique de Descartes. Paris: 1960.
Black, Max. A Companion to Wittgenstein’s Tractatus. Ithaca, N.Y.: Cornell
University Press, 1970.

294
 Bibliography 295

Borges, Jorge Louis. “Borges and I.” In Douglas R. Hofstadter and Daniel C.
Dennett, editors. The Mind’s I: Fantasies and Reflections on Self and Soul. New
York: Bantam, 1982, pp. 19–20.
“The Circular Ruins.” In Douglas R. Hofstadter and Daniel C. Dennett,
editors. The Mind’s I: Fantasies and Reflections on Self and Soul. New York:
Bantam, 1982, pp. 344–348.
Buchdahl, Gerd. Metaphysics and the Philosophy of Science. The Classical Origins:
Descartes to Kant. Cambridge, Mass.: MIT Press, 1969.
Buroker, Jill Vance. “Introduction.” In Antoine Arnauld and Pierre Nicole,
Logic or The Art of Thinking. Translated and edited by Jill Vance Buroker.
Cambridge: Cambridge University Press, 1996, pp. ix–xxvi.
Chadwick, Henry. Augustine. New York: Oxford University Press, 1986.
Curley, Edwin M. “Certainty: Psychological, Moral, and Metaphysical.” In
Stephen Voss, editor. Essays on the Philosophy and Science of Rene Descartes.
New York: Oxford University Press, 1993, pp. 11–30.
Descartes against the Skeptics. Cambridge, Mass.: Harvard University Press,
1978.
Descartes, Rene. The Philosophical Writings of Descartes, Volume 3, The Cor-
respondence. Translated by John Cottingham, Robert Stoothoff, Dugald
Murdoch, and Anthony Kenny. Cambridge: Cambridge University Press,
1991.
The Philosophical Writings of Descartes, Volumes 1 and 2. Translated by
John Cottingham, Robert Stoothoff, and Dugald Murdoch. Cambridge:
Cambridge University Press, 1984.
Descartes’ Conversation with Burman. Translated with an introduction and
commentary by John Cottingham. Oxford: Oxford University Press, 1976.
Treatise of Man. Translation and commentary by Thomas Steele Hall.
Cambridge: Harvard University Press, 1972.
Dickoff, James, James, Patricia, and Hendel, Charles W. “Introduction.” In
Antoine Arnauld and Pierre Nicole, Logic or The Art of Thinking: Port-Royal
Logic. Translated and edited by James Dickoff, Patricia James, and Charles
W. Hendel. New York: Bobbs-Merrill, 1964, pp. xxvi–xli.
Ellis, Brian David. Rational Belief Systems. Lanham, Md.: Rowman and Littlefield,
1979.
Faust, David. The Limits of Scientific Reasoning. Minneapolis: University of
Minnesota Press, 1984.
Feldman, Fred. “Epistemic Appraisal and the Cartesian Circle.” Philosophical
Studies, Volume 27 (1975), 37–55.
Frankfurt, Harry G. Demons, Dreamers, and Madmen: The Defense of Reason in
Descartes’ Meditations. Indianapolis, Ind.: Bobbs-Merril, 1970.
Gaukroger, Stephen. Descartes: An Intellectual Biography. New York: Oxford
University Press, 1995.
“The Sources of Descartes’s Procedure of Deductive Demonstration in
Metaphysics and Natural Philosophy.” In John Cottingham, editor. Reason,
Will, and Sensation: Studies in Descartes’s Metaphysics. New York: Oxford
University Press, 1994, pp. 47–60.
296 Bibliography

Cartesian Logic: An Essay on Descartes’s Conception of Inference. New York: Oxford

University Press, 1989.
Geach, Peter T. Logic Matters. Berkeley: University of California Press, 1980.
Gewirth, Alan. “The Cartesian Circle.” Philosophical Review, Volume 50 (1941),
368–395.
Haack, Susan. “The Justification of Deduction.” Mind, Volume 85 (1976),
112–119.
Hacking, Ian. “Leibniz and Descartes: Proof and Eternal Truths,” Proceedings
of the British Academy, Volume 59 (1973), 1–16.
Haldane, Elizabeth S. Descartes: His Life and Times. London: John Murray, 1905.
Hintikka, Jaakko. “Cogito, Ergo Sum: Inference or Performance?” In Willis
Doney, editor. Descartes: A Collection of Critical Essays. London: Macmillan,
1968, pp. 108–139.
Howell, Wilbur S. Logic and Rhetoric in England, 1500–1700. Princeton, N.J.:
Princeton University Press, 1956.
Hume, David. A Treatise of Human Nature. David Fate Norton and Mary J.
Norton, editors. New York: Oxford University Press, 2000.
Jardine, Lisa. “Humanistic Logic.” In Quentin Skinner, Eckhard Kessler, and
Jill Kraye, editors.The Cambridge History of Renaissance Philosophy. New York:
Cambridge University Press, 1988, pp. 173–198.
Joy, Lynn Sumida. Gassendi the Atomist: Advocate of History in an Age of Science.
New York: Cambridge University Press, 1989.
Kant, Immanuel. Critique of Pure Reason. Translated and edited by Paul
Guyer and Allen W. Wood. New York: Cambridge University Press,
1998.
Kenny, Anthony. “Descartes on the Will.” In John Cottingham, editor. Descartes:
Oxford Readings in Philosophy. New York: Oxford University Press, 1998,
pp. 132–159.
Descartes: A Study of His Philosophy. New York: Random House, 1968.
Kline, Morris. Mathematics: The Loss of Certainty. New York: Oxford University
Press, 1980.
Kneale, William, and Kneale, Martha. The Development of Logic. Oxford:
Clarendon Press, 1964.
Leibniz, Gottfried W. Die Philosophischen Schriften von G. W. Leibniz. Edited by
C. I. Gerhardt. Hildesheim: Georg Olms Verlagsbuchhandlung, 1960.
Lewis, David. On the Plurality of Worlds. New York: Oxford University Press, 1986.
Philosophical Papers, Volume 1. New York: Oxford University Press, 1983.
Locke, John. An Essay Concerning Human Understanding, Volume 2. Annotated
by Alexander Campbell Fraser. New York: Dover, 1959.
Lohr, Charles H. “The Sixteenth-Century Transformation of the Aristotelian
Division of the Speculative Sciences.” In D. R. Kelley and R. H. Popkin, ed-
itors. The Shape of Knowledge from the Renaissance to the English Enlightenment.
Norwell, Mass.: Kluwer, 1991, pp. 49–58.
“Jesuit Aristotelianism and Sixteenth Century Metaphysics.” In G. Fletcher
and M. B. Scheute, editors. Paradosis. New York: Fordham University Press,
1976, pp. 203–220.
 Bibliography 297

Losonsky, Michael. Enlightenment and Action From Descartes to Kant: Passionate

Thought. New York: Cambridge University Press, 2001.
Menn, Stephen. Descartes and Augustine. New York: Cambridge University Press,
1998.
Montaigne, Michael de. An Apology for Raymond Sebond. Translated and edited
by M. A. Screech. New York: Penguin, 1987.
Nozick, Robert. An Examined Life. New York: Simon and Schuster, 1989.
“Fiction.” In Douglas R. Hofstadter and Daniel C. Dennett, editors. The
Mind’s I: Fantasies and Reflections on Self and Soul. New York: Bantam, 1982,
pp. 461–464.
Philosophical Explanations. Cambridge, Mass.: Harvard University Press, 1981.
O’Neill, Onora. Constructions of Reason: Explorations of Kant’s Practical Philosophy.
New York: Cambridge University Press, 1989.
Perry, John. “Frege on Demonstratives.” Philosophical Review, Volume 86
(1977), 474–497.
Popkin, Richard H. The History of Scepticism: From Erasmus to Descartes. Revised
edition. New York: Harper Torchbooks, 1964.
Popper, Karl. Conjectures and Refutations: The Growth of Scientific Knowledge. New
York: Harper Torchbooks, 1965.
Prichard, H. A. “Descartes’s Meditations.” In Willis Doney, editor. Descartes: A
Collection of Critical Essays. London: Macmillan, 1968, pp. 140–168.
Quine, W. V. O. Methods of Logic. Revised edition. New York: Holt, Reinhart
and Winston, 1959.
Rodis-Lewis, Genevieve. Descartes: His Life and Thought. Translated by Jane
Marie Todd. Ithaca, N.Y.: Cornell University Press, 1999.
Rorty, Amelie O., editor. Essays on Descartes’ Meditations. Berkeley: University
of California Press, 1986.
Ryle, Gilbert. The Concept of Mind. New York: Penguin, 1966.
Sarkar, Husain. “Something, Nothing, and Explanation.” Southwest Philosophy
Review, Volume 9, No. 1 ( January 1993), 151–161.
Schlesinger, George. Religion and Scientific Method. Dordrecht: D. Reidl, 1977.
“The Problem of Evil and the Problem of Suffering.” American Philosophical
Quarterly, Volume 1 (1964), 244–247.
Sextus Empiricus. Outlines of Pyrrohnism. The Loeb Classical Library. Translated
by Rev. R. G. Bury. New York: Putnam’s Sons, 1933–44.
Shakespeare, William. Macbeth. The Arden Shakespeare. Edited by Kenneth
Muir. London: Methuen, 1962.
Smiley, T. J. “Entailment and Deducibility.” Proceedings of the Aristotelian Society,
n.s., Volume 59 (1959), 233–254.
Smullyan, Raymond M. What Is the Name of This Book? New York: Prentice Hall,
1978.
Spinoza, Benedict de. The Chief Works of Benedict de Spinoza, Volume 2. New
York: Dover, 1951.
Stout, A. K. “The Basis of Knowledge in Descartes.” In Willis Doney,
editor. Descartes: A Collection of Critical Essays. London: Macmillan, 1968,
pp. 169–191.
298 Bibliography

Strawson, Peter. Introduction to Logical Theory. London: Methuen, 1967.

“Review of von Wright.” The Philosophical Quarterly, Volume 8 (1958),
372–376.
Taylor, Charles. Sources of the Self: The Making of the Modern Identity. Cambridge,
Mass.: Harvard University Press, 1989.
Tlumak, Jeffrey. “Certainty and Cartesian Method.” In Michael Hooker,
editor. Descartes: Critical and Interpretive Essays. Baltimore: Johns Hopkins
University Press, 1978, pp. 40–73.
van Cleve, James. “Foundationalism, Epistemic Principles, and the Cartesian
Circle.” In John Cottingham, editor. Descartes: Oxford Readings in Philosophy.
New York: Oxford University Press, 1998, pp. 101–131.
von Wright, Georg H. Logical Studies. London: Routledge and Kegan Paul,
1957.
Vrooman, Jack R. Rene Descartes: A Biography. New York: G. P. Putnam’s Sons,
1970.
Williams, Bernard. “Introduction.” In John Cottingham, editor and translator.
Meditations on First Philosophy by Rene Descartes. With Selections from Objections
and Replies. New York: Cambridge University Press, 1986, pp. vii–xvii.
Descartes: The Project of Pure Inquiry. New York: Penguin, 1978.
“The Certainty of the Cogito.” In Willis Doney, editor. Descartes: A Collection
of Critical Essays. London: Macmillan, 1968, pp. 88–107.
Wilson, Margaret D. Descartes. New York: Routledge and Kegan Paul, 1986.
Wittgenstein, Ludwig. On Certainty. New York: Harper and Row, 1969.
Tractatus Logico-Philosophicus. London: Routledge and Kegan Paul, 1961.
 Name Index

Abbot, Edwin A., 16 Blake, William, 109

Adam, 54 Borges, Jorge Louis, 171n28
Adams, Charles, xi Bourdin, Pierre, 57, 111n12, 114–15n15,
Adams, Robert Merrihew, 216, 218 191–2, 245
Agricola, Rudolph, 245n11 Brahe, Tycho, 124
Agrippa, Henricus Cornelius, 133–4 Buchdahl, Gerd, 21n20
Alquié, Ferdinand, xi, 264n29 Burman, Frans, 142, 198n22, 227n20,
Ambrose (Bishop of Milan), 59 252n20
Anderson, Alan, 159, 160n22 Buroker, Jill Vance, 44n3, 176, 278n2,
Apollonius of Perga, 132, 132n28, 280n4
202n26, 208
Aquinas, Thomas, 134, 216
Caterus, Johannes, 110n10
Archimedes of Syracuse, xii, 132, 183,
Chandoux, N. de, 1–2, 28
208
Charron, Pierre, 23n21
Aristotle, 7, 12, 62n4, 133, 134, 245, 248
Christina (Queen of Sweden), 7n9
Arnauld, Antoine, 83–4, 85, 86n20, 109,
Cicero, Marcus Tullius, 203
116, 116n18, 117, 130n25, 181n6, 191,
Clerselier, Claude, 16, 115n16, 278
241, 278, 278n2, 280, 280n4
Cleve, James van, 32n28, 46–7n5, 102,
Ashworth, Earline Jennifer, 189n11
127–30
Aubrey, John, 18n17
Compton, John, xiv, 58–9n2, 221–2n12,
Augustine (Bishop of Hippo), 58, 85
223n15, 225n17
Austin, John Langshaw, 290
Copernicus, Nicholas, 124
Ayer, Alfred Jules, 138, 151, 152, 217–8
Crace, Jim, xiv
Curley, Edwin M., xii, 4n5, 51n7, 89,
Bacon, Francis, 1, 14n14, 16n16, 284–6 101, 102, 109–26, 127n24, 152, 154–61,
Bagni, Guido, 1 167
Baillet, Adrien, 2n3
Baker, Gordon, xii, xiv, 56, 142n4, 151–2,
201n25, 215–17, 221–2n12, 225n17 Democritus, 73
Beck, Leslie J., 21n20, 287, 288 Dickoff, James, 278n2
Beeckman, Isaac, 250n17 Dolz, Johannes, 189n11
Belaval, Yvon, 161–2n23 Duboux, Emile, xi
Belnap, Nuel, 159, 160n22
Berulle, Pierre de, 1, 2 Ellis, Brian, 109n8
Black, Max, 199n23 Epicurus, 73

299
300 Name Index

Euclid, 18n17, 131 Kneale, William, 18n17

Eve, 54 Kuhn, Thomas Stanley, 152

Faust, David, 109n9 Lambert, Karl, 167

Feldman, Fred, 127, 129 Lambert of Auxerre, 132
Fermat, Pierre de, 2n2, 143–4n6 Languis, Wilhelm, 192
Fonesca, Pedro, 134 Leibniz, Gottfried Wilhelm von, 21, 38n1,
Fraassen, Bas C. van, 167 141, 145, 147, 161–2n23, 278
Frankfurt, Harry, 54–7, 113, 118, 153, Leo X, 134
154, 166n25, 207 Lever, Ralph, 242
Frege, Gottlob, 56 Lewis, David, 73, 219–20, 221–2n12, 228
Lichtenberg, Georg, 151, 152
Galileo, Galilei, 13–14 Locke, John, 7, 141, 145, 216
Gassendi, Pierre, 1, 26, 71–2n9, 84, 149, Lohr, Charles H., 134n30
213, 214, 217, 249n16, 252n20, Losonsky, Michael, 9n12, 252–3n21,
265n30 266–7n32
Gaukroger, Stephen, xii, 62n4, 102, Lull (Lully), Raymond, 189, 241n7
110n10, 130–7, 161–2n23, 196n17,
198n20, 198n21, 199–200n24, Machiavelli, Niccolo, xv, 27
204n29, 226n18 Maurice of Nassau, 104n4
Geach, Peter Thomas, 159–60, 160n22 Meibornium, Marcum, 192
Gewirth, Alan, 127, 129 Meinong, Alexis, 153
Gillot, Jean 7n9 Menn, Stephen, 58–9n2
Mersenne, Marin, 1, 2n2, 5n7, 5n8, 13,
Haack, Susan, 105n6 14, 62–3n6, 77, 94, 116, 116n18,
Hacking, Ian, 161–2n23 130n25, 133, 143–4n6, 190, 192,
Hahn, Russell, xv 212–13, 222
Haldane, Elizabeth S., 2n3 Mesland, Denis, 190, 211, 211n1, 264
Harvey, William, 17–18, 251 Mill, John Stuart, 133
Hendel, Charles W., 278n2 Montaigne, Michele de, xi, 4, 8n11, 73,
Hilbert, David, 113n13 101, 102–4
Hintikka, Jakko, 148–50, 152, 167 Moore, George Edward, 171
Hobbes, Thomas, 1, 18, 142n4, 263 Moore, Terrence, xv
Howell, Wilbur Samuel, 241n6 Morin, Jean-Baptiste, 242n9
Hume, David, 103n2, 127, 171, 212, Morris, Katherine J., xiv, 56, 142n4,
212n2, 230, 292 151–2, 201n25, 215–17, 221–2n12,
Husserl, Edmund, xiv, 221–2n12 225n17
Huygens, Constantijn, 132n28, 202n26
Hyperaspistes (pseudonym), 49n6, 71n9, Newton, Issac, 21, 48
84 Nicole, Pierre, 241, 278, 278n2, 280n4
Nozick, Robert, 173n29, 251n19, 273–7
James, Patricia, 278n2 O’Neill, Onora, 25n22, 40n2
Jardine, Lisa, 205n32, 206n33, 245n11 Ovid, xi
Joy, Lynn Sumida, 26n23
Pappus of Alexandria, 22, 131, 132, 208
Kant, Immanuel, 25n22, 40–1, 225n17, Perry, John, 219, 221–2n12
292 Peter of Spain, 132
Kaplan, David, 167 Plotinus, 245
Kemp Smith, Norman, xii Pollot, Alphonse, 166, 213
Kenny, Anthony, xii, xiii, 38n1, 86n21, 90, Polyander, 8, 8n11, 10, 39, 61, 62,
127–8, 141, 143n5, 143–4n6, 150, 152, 160n22, 167, 244, 246
153–4, 158n17, 205n31, 256n23, Popkin, Richard H., 58, 192n13
264n29, 287–93 Popper, Karl Raimund, 14n14, 53–4
Kline, Morris, 113n13 Porphyry, 85, 88, 245, 247, 248
Kneale, Martha, 18n17 Prichard, H. A., 46n5
 Name Index 301

Ptolemy (Claudius Ptolemaeus), 48, 124 Smullyan, Raymond, 233

Pythagoras of Samos, 18n17 Socrates, 8, 93
Solon, 190
Quine, Willard van Orman, 141 Spade, Paul Vincent, 189n11
Quintilian, 62n4 Spinoza, Baruch (Benedict de), 29n24
Stout, A. K., 203n27
Ramus, Petrus (Peter, Pierre de la Strawson, Peter F., 160n22, 162, 164
Ramee), 7, 241, 245, 245n11, 282–3 Suarez, Francisco, 134
Regius, Henricus (Henri de Roy), 85n20
Rembrandtsz, Dirk, 7n9 Tannery, Paul, xi
Reneri, Henricus (Henri Regnier), 166, Tantalus, 37
213 Taylor, Charles, 154n13
Revius, Jacobus, 31n27 Tepelli, xi
Richter, Truitt, 57n8 Tlumak, Jeffrey, xiv–v, 32n28, 75n11,
Rodis-Lewis, Genevieve, 2n3, 4n4, 7n9, 125n23, 269–72
18, 23n21, 31n26, 31n27, 84n18, Toletus, Francisco, 134
104n4, 104n5, 191n12
Rousseau, Jean Jacques, xi Valla, Lorenzo, 206n33
Russell, Bertrand, 151, 152, 167, 199n23 Ville-Bressieu, Etienne de, 1
Ryle, Gilbert, 46n5, 226–35 Voetius, Gisbert, 104n4

Sarkar, Casim Ali, xv Watson, Richard A., xiv

Sarkar, Husain, 276n4 Williams, Bernard, xii, 144, 151, 154n13,
Schlesinger, George, 251n19 158–9n18, 162–8, 222n14
Schulter, Henry, 7n9 Wilson, Catherine, xv, 11n3, 78n15,
Sebond, Raymond, 8n11, 104n3 161–2n23
Sextus Empiricus, 4, 192, 203, 204–6 Wilson, Margaret Dauler, xii, 55–7,
Shakespeare, William, 107n7, 149, 273, 145–6, 222n14
274 Wittgenstein, Ludwig, 46n5, 199n23
Silhon, Jean de, 111n12, 187–8 Wordsworth, William, 109
Smiley, Timothy J., 160n22 Wright, Georg H. von, 160n22
 Subject Index

Aristotelianism, 18 199–200n24, 207, 209–10, 226, 239,

atheism, 104n4 240, 247, 254–7
and logic, 126, 135 and bootstrap approach, 83
and mathematics, 121 as the central claim, 138–9
axiom(s), 23, 279–80 and clear and distinct ideas, 62–3n6,
and generality, 22, 285–6 98–100, 225, 265–7
of metaphysics (philosophy), 17, 19, and conditional statements, 281–2
20, 22, 148, 181 content of, 214–26
of physics, 14, 17, 19, 22 as an empirical necessary truth, 73
and simplicity, 21–2 as an enthymematic argument, 140,
142
Bourdin’s scheme 245–6 as an exercise, 208
and the extraordinary-doubt model,
Cartesian Circle 68–9
new, 102, 109–13, 137 and falsificationism, 78–9
old, 102, 109, 116–17, 126–30, 130n25, and fictional (possible) worlds, 70–2,
137 194, 275–7
causal principle (ex nihilo nihil fit), 52, as a first-order mental process, 75–6
150 as a first-order mental state, 74, 77,
certainty, 24, 46–7n5, 54, 62, 66n7, 76, 79–80, 128
80, 129–30, 233n15 as the first truth (axiom), 91, 143,
metaphysical, 122, 127, 129, 178 145, 183, 186, 203, 208, 279–80,
moral, 122, 178 282
practical, 127 and the general rule, 93, 114–15n15,
psychological, 12, 122 161–2n23, 193, 224
subjective, 155 by induction from one’s own case, 144,
clear and distinct idea(s), 15, 18, 23, 24, 146–8, 192
61, 62n4, 92–3, 94–5, 96–100, 121–2, and the infinite regress, 94–5
143–4n6, 127–30, 193, 223–5, 240, and the mechanical robot, 274
256 as a mental process, 69–70, 75
and will, 265–7 as a mental state, 69–70, 72, 75, 83,
cogito, 67 217, 225
and absolute truth, 96–8, 147–8 not an argument, 181, 193,
as an argument, 110, 112, 114, 138–40, 199–200n24, 203–7, 209–14, 280
141, 156, 162–3, 177, 195, 197, not a proof, 154–6

302
 Subject Index 303

and the ontological argument, 141, existential inconsistency, 168, 169, 170,
153–4 172, 174
as an ontological claim, 91 existential presupposition, 149, 157–9,
and particular existence, 64–5, 78 159–60, 173
as the perfect solution, 80–2 existential tautology, 174
as performance, 169, 170 experiments, 18, 35
as a second-order mental state, 74, of fruit (Experimenta fructifera), 285
76–7, 79–80 of light (Experimenta lucifera), 285
and the six criteria, 82 extraordinary-doubt model, 46n4, 47, 66,
as a syllogism, 142–8, 156–7, 68–9, 74–5, 104n4, 155, 166–7n25,
158–9n18 238
and theory of error, 265–6 and the deterministic case, 48
and the vicious circle, 94–5 and the evil demon, 52–3, 68
and the wax, 210–12, 224–5 and the mathematical case, 48–9, 178
and the will, 264–6 and the probabilistic case, 47–8
conscientia, 216–17
Counter-Reformation, 1 faculty
Curley’s Central Contention, 110–13 of knowledge and will, 263–5
of knowledge or intellect ( facultas
deduction, 141, 148, 176, 177, 179–81, cognoscendi), 249, 250, 251, 256, 262
185, 192, 198, 199–200n24, 200–1, of will (facultas eligendi), 249
203n27, 205n31, 220, 237–8, 281 faith, 8, 37–8
dependence, 16–20 of Descartes, 31n27
Descartes’ dream, 16n16 and Eucharist, 83–4
doubt, 44–5, 69, 74, 86, 88, 90, 109, 140, and reason, 58–9, 58–9n2, 104, 104n2,
155, 189 108
and the causal account of principles, Fates, 190
55–7 final causes (see also God, purposes of),
degrees of, 41–2, 90–1 250
and disbelief, 46n4 in physics, 257–60
and the dream argument, 51
hyperbolic, see extraordinary-doubt general rule, 26–7, 92, 198–9n22, 224–5,
model 256–7
and logic, 57, 166–7, 178, 187, 189, generality, 22
191–2, 194 God
and mathematics, 49, 70n8, 166–7, as Craftsman, 11, 249, 259
178, 192, 194 and demonstrative knowledge, 187–8
notion of, 77 existence of, 38, 88, 111–6, 116–8, 125,
and principles, 50, 54–5 153, 194
and reason, 105–6 God-argument, 112, 114
and the senses, 50–1 intellect of, 250, 257
and simple and general things, 52 as lawgiver, 190
and the syllogism, 188–9, 189n11 omnipotence of, 53, 190–1
private purposes of, 252n20
epistemic problem, 80 public purposes of, 252n20
Epistemon, 10, 140, 244, 245, 246 purposes of, 250, 250n18, 251–2, 257,
essences, 14n14 258n24, 259–60
eternal truth(s) (common notion[s]) will of, 257, 282
144, 156–7, 158, 161–2n23, 163, Ground State, 233–5
166–7, 186, 199–200n24, 252–3n21
creation of, 190–1, 194 hyperbolic doubt, see
Eucharist, 83–6, 224n16 extraordinary-doubt model
Eudoxus, 8, 10, 140, 244, 246, 247
evil demon (genius), 49n6, 52–3, 57, 68, ‘I’, 67, 77, 84, 85–6, 87, 91, 153, 207–8,
78, 103, 111, 120, 145, 146, 177–8, 212, 214–16, 217–18, 220–1, 225n17,
187, 192, 194, 213, 277 229
304 Subject Index

‘I’ (cont.) scholastic, 22, 24, 178, 241–2, 246–8,

and the Heraclitean flux, 232 284
identity of the, 222, 222n13, 223n15, Stoic, 204–7
224, 230
and the Socratic maxim, 228 mathematics, 24
as systematically elusive, 226–35 maximal warrant, 269–72
ideal seeker(s), 24, 62, 63, 82, 106, 146, mechanics, 16–20
152, 167–8, 207, 256, 257–8, 259, medicine, 16–20
266 memory, 177, 180–1, 182, 194–5, 198,
and natural class, 9 201n25, 202n26, 203n27, 205n31,
and the original state of ignorance, 5 210
and the pristine state, 261–2, 266–7n32 and the infant, 224
problem of the, 3–4 intellectual, 236–7, 237n3, 238–9,
and the stage of determining the right 238–9n5
qualities, 6–9 local, 237
and the stage of preparation, 9–10, 26 physical, 236–7, 238
and the stage of sifting 5–6 unreliability of, 237–8, 238–9
and theory of error 260–3 mental process, 42, 69–70
identity of indiscernibles, 221 first-order, 43, 75–6, 107–8
Idols (of the Cave, Marketplace, Theater, second-order, 43
Tribe), 284, 285 mental state, 42, 69–70
imagination, 238n4 first-order, 43, 68, 74
impulse (blind, natural, spontaneous), second-order, 43, 68, 76
108, 122, 253n21 metaphysics, 16–20
Index, 23n21 method, 2, 12
indubitability, 12, 44, 77 of analysis, 13–16, 22, 135–6, 200–1
descriptive, 113–14 and bootstrap approach, 30
normative, 113–14, 154 of disputationes, 134
inference(s) four rules of, 13–16
immediate, 143–4n6, 280–2 historical, 26–7, 28, 29
mediate, 280–2 humanist-Aristotelian, 132–3
intuition, 80, 179–81, 182, 185, 192, of Kant, 25n22
195–6, 198, 199n23, 199–200n24, of Lull, 133–4
200–1, 203n27, 239–40 mathematical, 23, 24, 26, 28, 29
and memory, 239 and the moral code, 30–2
primary, 199, 201–2, 202n26 and morals, 16–20, 25, 30–2
secondary, 199, 201–2, 202n26 paradox of, 23
irrevisability, 269–72 and practice, 21
and probability, 28
Jesuits, 7n9 of synthesis, 13, 14–15, 22, 135–6,
judgements, 215, 243 200–1
Middle Ages, 29n25
Kenny dilemma, 154 modal principle, the, 159–60
moral code, 30–2
language-game, 151–2
Lateran Council, 134, 135 naive realist, 207
logic natural light, 11, 30, 122, 253n21, 289
Aristotelian, 178, 204–7 Neanderthal man, 54
of Bacon, 285 neustic, 290–1
of Descartes, 246 New Academicians, 279n3
and discovery, 241–3, 241n7
and explanation, 241–3, 241n7, ontological argument, 125–6, 141,
247 153–4
history of, 240–8, 278–83, 278–83, onus of match, 290–1
284–6 order of discovery (ordo ascendendi), 242,
justification of, 105 243
 Subject Index 305

order of explanation (ordo cognoscendi), Stoic conditional argument, 204–7

242, 243 Stoics, 62n4, 161
ordinary-doubt model, 68 substance, 83–6
Sulmo principle, xi, 176
paradox, 23, 83, 112 summum genus (the highest genus), 245
phrastic, 290–1 syllogism, 133, 242, 245, 285, 191–2
physics, 16, 17, 18–20 Aristotelian, 12, 42–3, 157, 158,
Port Royal, 278 and explanation, 242
presupposition, the relation of, 162–3 probable, 141
primitive suchness, 214, 215, 218, 220–1, as verbiage, 245
221n12
primitive thisness, 214, 218, 221n12 thinker, the, 211, 215
principles (rules) of inference (logic), thinkingF , 275–6
110–13, 116, 145, 146, 148, 152, thinkingR , 275–6
156–8, 161, 178, 187, 193, 195, thought, 84, 184
199–200n24, 203–7, 213–14, 239 and language, 44n3
probability, 1–2, 28, 143–4n6 tree of philosophy, 17, 20, 199n23, 244
problems tree of Porphyry, 245, 246, 247–8
genuine, 37 truth, 9, 20, 81, 92, 95–6, 118
imperfect, 36, 38, 81 absolute, 10–11, 95, 97
imperfectly understood, 36, 38 existential, 145
perfect, 35, 38, 80–2 indefinability of, 77
perfectly understood, 36, 38, 80–2 of language, 166–7
philosophical, 38 of logic, 166–7n25
progress in philosophy, 151–2 of a mathematical statement, 4,
proof 166–7n25, 145
notion of, 161, 161–2n23 and natural light, 11
subjective conception of, 118 phenomenal, 10–2
proposition and proof, 161–2n23
assent-compelling, 118–21 rational, 10–2
atomic, 199n23 relative, 95, 97–8
compound, 199n23 Turk, the, 31–2
general, 64–5, 140–1, 147–8, 192–3
particular, 64–5, 140–1, 147–8, 192–3 unity
truth-value of, 252–3n21 of composition, 223
Provisionally Basic State, 234–5 of the ‘I’, 225n17
Pyrrhonism, 192 of method, 21
Pyrrhonists, 279n3 of nature, 223
of our minds, 225n17
quantification theory, 148–9 of the piece of wax, 225n17
of science, 16, 21, 289
reason, 5, 7–8, 11, 25n22, 37–8, 69, unmistakability, 269–2
102–9, 135, 137, 145
private 40–1 wax, 98–9, 210–12, 224, 225
public 40–1 will, 9, 45, 249, 249n14, 252
and religion, 58–9, 101, 104–5 assent of the, 263–4
riddles, 36 and doubt, 45, 66
and the essence of error, 262
self-ascription, 214, 219–20 freedom of the, 255
self-consciousness, 230, 232–3 and liberty of indifference, 292–3
simplicity, 15, 21, 63–5 and liberty of spontaneity, 292–3
soul, 211, 213 limits of the, 254–5, 260, 265n30
species infima (the lowest species), 245 operator of the, 262
statement, 169–70 and theory of error, 260–3