K 8.75%.
i. 3‘ this; »

Maw—WA”... is .
..1.m.o :I(naas.s.
a 1'5»!
...

s 5" .0“me
.Wriwmmtx?‘

. a
..i
...Wu.rd..
it LMW

1.3.} x?
”an huh... 442‘s“. “haw-i...
3.51. e x

 

 

_ :8... 5.11.“! ...........

 

 

.nchzvsI'. v.
.x I Art

£04 5%. Lu\..

 

I'll!

 

 

l LIBRARY
2001 Michigan State
University

This is to certify that the
dissertation entitled

THE INDISPENSABILITY OF METAPHYSICAL REALISM

presented by

John D. Mariana

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Philosophy

 

 

A

Major Professor’s Signature

/ ' "(

Date

MSU is an afﬁrmative-action, equal-opportunity employer

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K‘IPij/ACITAPFGS/ClRC/Daieoue Indd

THE INDISPENSABILITY OF METAPHYSICAL REALISM
By

John D. Mariana

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
DEPARTMENT OF PHILOSOPHY

2008

ABSTRACT
THE INDISPENSABILITTY OF METAPHYSICAL REALISM
By
John D. Mariana
The past hundred years or so have witnessed increased philosophical attention
devoted to elaborating and defending competing versions of empiricism, naturalism and
pragmatism, and a corresponding decrease in philosophical conﬁdence in rationalism and
traditional metaphysical realism. There are, to be sure, a number of reasons for this, but
the discovery of alterative logics and alternative geometries, coupled with several
dramatic nineteenth and twentieth century revolutions in scientiﬁc knowledge, have
perhaps done the most to erode belief in the possibility of genuine a priori knowledge
about a determinately structured reality. This has prompted a number of philosophers,
beginning at least with the logical positivists, to insist that traditional metaphysical
questions about the structure of reality in itself are unanswerable (or cannot even be
properly asked) because they are either meaningless or unintelligible, or involve terms
that lack any knowable referent. In short, their claim is that traditional metaphysical
questions cannot be answered because they surpass the bounds of human understanding.
This claim is, of course, quite in line with empiricist epistemological
commitments generally, but, as an apparently universal claim about the scope and limits
of human knowledge, it immediately prompts the following question: what could an
empiricist possibly appeal to in support of such a statement? It should (I hope) be
immediately apparent that there are only two options here. The ﬁrst would of course be

to argue that traditional metaphysical questions are unintelligible or meaningless by

offering an empirical account of language acquisition, but this would obviously involve
begging the question against the metaphysician.

The only other option, however, would be to turn to a universal theory of meaning
that would enable us to say precisely why metaphysical questions could not possibly be
meaningful. But if this theory is not to be empirically supported, how could it possibly
claim to establish the bounds of meaningfulness for all possible languages without itself
requiring the support of a realist metaphysics, coupled with a rationalist epistemology
that accounts for our a priori awareness of this synthetic universal truth? In other words,
the conditions of the success of the empiricist’s argument require its failure.

So why not just abandon the original attempt to rule out metaphysical realism,
along with any attempt to produce a universal theory of meaning? This is out of the
question, because there can be no rational argument for doing so. Any argument intended
to support the conclusion that we should abandon all attempts to construct a successful
universal theory of meaning would eliminate whatever rational basis we might have had
for thinking that the premises of the argument itself are themselves meaningful.

If what I have said here is correct, then we have the following results. First, there
can be no successful argument against the very possibility of a priori knowledge of a
determinately structured reality. And secondly, since we can have no rational basis for
abandoning the attempt to construct a successful universal theory of meaning, our
attempts to develop such a theory must provide an explanation of how we can have a
priori knowledge, since any successful theory of meaning must assume that such

knowledge is possible. Thus, we cannot possibly dispense with metaphysical realism.

Copyright by
JOHN D. MARIANA
2008

To PLA T0
who got me started
and
To BETH

who keeps me going

ACKNOWLEDGEMENTS

I would like, ﬁrst of all, to thank my doctoral committee — most especially Debra
Nails, for her unfaltering encouragement, her support, her advice, her advocacy on my
behalf, her unstinting appreciation of my work, her forgiveness and understanding, and
her good humor and aplomb in the face of my repeated and ﬂagrant disregard for proper
formatting. Thanks also to Rich Hall and Win Wilkinson, who have now seen me
through not one but two advanced degrees (troupers, both of them), from whose example
I have learned more about philosophical method and conceptual clarity than I imagined
there was to know; and many, many thanks to Matt McKeon for his frustratingly
penetrating and thoughtful criticisms, my confrontations with which have inevitably
improved both my thinking and my understanding. I would also like to thank Carol
Slater (of Alma College) for consenting to serve as the ﬁfth (and external) member of my
committee, and for carrying forward the ﬁne practice of critical interdisciplinary review.

I would also like to thank Joe Hanna for teaching me the difference between
objectivity and mind-independence — a nice distinction the full implications of which, I
think, are still too little appreciated — and for serving (to some extent unwittingly) as the
catalyst for early versions of many of the arguments contained in the present work. I owe
a great deal to all of these philosophers, though of course any mistakes I have made are
entirely my own, and not theirs (and in most cases are things they warned me about
repeatedly). I also want to thank Mike Reno, my dear friend, long-time colleague and
most acidic critic, whose assiduousness in argument, intellectual honesty, and personal
insight have nourished me deeply, and from whose input, on all manner of philosophical

topics, I have beneﬁted enormously. Thanks also to Sarah Ward, without whose

vi

assistance on numerous occasions I could not possibly have advanced through my
graduate program smoothly and successfully.

My greatest debt is owed, however, to my love and my lifemate, Beth, who put up
with more from me in the course of the past eight years than should be asked of anyone.

That she forgives me is a testament to her resilience, her strength, and her commitment.

vii

PREFACE

It would seem to me most natural to note, by way of opening remarks, the extent
to which contemporary philosophy has come to be dominated by empiricism and
pragmatism. My impression is that rationalism and metaphysical realism are minority
views as things stand, and that any defense of either one must ﬁrst address the prevailing
consensus that these positions have been decisively exposed as untenable, before such a
defense can even properly get down to business. Other authors, such as Jerrold J. Katz
and Laurence BonJ our, who have recently offered defenses of both rationalism and
traditional realism have, however, opened their works with laments over the current
dominion of empiricism, and these laments have been met with bemused replies in the
literature. I must, then, address myself ﬁrst to these replies.

Katz remarks, in the Introduction to his Realistic Rationalism, that “[I]n the
course of this century, a large segment of Anglo-American philosophy was persuaded to
abandon the traditional conception of philosophy on which it is an a priori inquiry into

”I

the most general facts about reality. He claims that this conception of philosophy has
been replaced by “one or another of two naturalist conceptions of philosophy: philosophy
as therapy designed to cure the linguistic illness of which philosophy itself is the cause,
and philosophy as an a posteriori discipline within natural science.”2 Indeed, it is Katz’s
view that the Wittgensteinian and Quinean approaches to philosophy are now taken to be

the only serious contenders on the contemporary scene. He argues that once “the broad

notion of analyticity, which served as the logical empiricist’s basis for a priori

 

' Katz 2000, p. xi.
2 Ibid.

viii

knowledge, disappears” in the wake of Quinean and Wittgensteinian holism, what
remains of rationalism goes with it.3 Thus, according to Katz, “the only issue left
unresolved by the linguistic turn, the issue that today divides Anglo-American
philosophy, is whether it is Wittgenstein’s ontological naturalism with its positivistic and
therapeutic emphasis or Quine’s epistemic naturalism with its scientistic and pragmatic
emphasis that goes in its place.”4
Alex Oliver, in his review of Realistic Rationalism, is nonplused. “This is not
how it feels on this side of the Atlantic,” he coolly reports, “but then ‘Anglo-American’,
like ‘Continental’, is more a term of sociology than geography.”5 Could this be the same
Alex Oliver whose quest has been to deﬂate plural noun phrases by ﬁnding an
appropriate formal notation through which to capture their logical function? But never
mind. Not being on the other side of the Atlantic, I can’t say how things feel there. There
has been, after all, a great deal of philosophical work underway “on the Continent,” to
which Britain is closer (at least geographically), which is not so thoroughly preoccupied
with Wittgenstein and Quine: existentialism, phenomenology, critical theory, and
hermeneutics, for instance. Anglo-American philosophers have certainly taken notice of,
and pursued, these sorts of philosophical projects, and it was certainly a gross
overstatement on Katz’s part to claim that the battle between Quinean and
Wittgensteinian holism is the issue that today divides Anglo-American philosophy.

On the other hand, it is worth noting that in his survey and retrospective of

twentieth century analytic philosophy, Avrom Stroll names Wittgenstein as the greatest

 

3 Ibid., p. 184.
‘ Ibid.
5 Oliver 2000, p. 1 1 1.

ix

philosopher of the century.6 Indeed, Stroll considers Wittgenstein to be perhaps “the

most original philosopher since Kant.”7

He also gives exceedingly high marks to Quine,
however, noting that other twentieth century philosophers, such as Austin, “fail the
criterion of scope” - that is, they are not sufficiently wide-ranging in their thinking and

their work — and that we are thus “left with Quine and Wittgenstein.”8

But while Quine is
a brilliant philosopher, he also ultimately fails the criterion of scope.9 For Stroll’s money,
then, Wittgenstein carries the day as the most inﬂuential and wide-ranging philosophical
ﬁgure of the twentieth century, despite the fact that Wittgenstein produced no ethics, no
metaphysics (of course), no philosophy of science, no aesthetics, no political theory, no
philosophy of law or of rights, no serious social theory to speak of, and, depending upon
whom one asks, no philosophical theory at all (only therapy). What about Bertrand
Russell, who is sometimes identiﬁed as the founder of analytic philosophy? Stroll tells us
that we have already seen the “etiolation of the reputations of Russell, Moore, Carnap,

and Ryle.”lo

Are we so soon living in the time of the posterity of the twentieth century,
despite the fact that every living philosopher was born before the year 2000? When
current philosophical preoccupations die away, can we be so sure that someone like
Russell will not be the one who is remembered? What about Heiddeger? Stroll does not

even include him in the ranks of analytic philosophers, although his Being and Time is a

deeply analytic work if it is anything.ll But then, “analytic,” like “continental,” is more a

 

6 Stroll 2000, p. 253.

7 Ibid., p. 251.

‘ Ibid.

9 Ibid., p. 252.

'° Ibid., pp. 250-251.

” Heidegger (1962, p. l) announces, on the ﬁrst page of Being and Time, that “it is ﬁtting that we should
raise anew the question of the meaning of Being,” inasmuch as we do not “in our time have an answer to the

term of sociology than philosophy, and “Continental” ﬁgures like Heidegger do not even
appear on analytic radar screens.

Oliver also complains that Katz’s history of Anglo-American philosophy since the
linguistic turn is “very broad brush.”12 He characterizes Katz’s account as “history with

villains?”

Anyone who does not ﬁnd dismissals of rationalism and metaphysical realism
frequent in contemporary philosophical literature, however, does not read contemporary
philosophical literature. Such dismissals might appear infrequent only because the
dismissive attitude they reﬂect has come to be taken so much for granted that such
dismissals have become routine to the point of being banal — often being expressed in
exactly the same, or nearly the same terms. One ﬁnds it commonplace for philosophers
who are neither metaphysicians nor epistemologists to report that there is agreement that
the traditional conception of philosophy as a metaphysical and epistemological enterprise
can no longer seriously be sustained, that awareness of reality as such, or of a mind-
independent world, has been recognized to be impossible, and that the standpoint which
would be necessary for such awareness is unavailable to human knowers. Kai Nielsen
claims that we should “set aside the ontotheological tradition of philosophy — the grand
old tradition of philosophy,” which consists of “the foundationalist epistemological

tradition” and “the grand metaphysical tradition,” which Rorty has shown to be

“moonshine.”l4 Susan Bordo, in her critical review of Cartesian philosophy, The Flight

 

question of what we really mean by the word ‘being’.” If that is not a call for an analysis, then I don’t know
what is.

'2 Oliver 2000, p. 1 l 1.

'3 Ibid.

” Nielsen , pp. 3, II, and 33. The unapologetically contemptuous tone of Nielsen’s language cannot be
missed, even in these few quotations, which forces one to conclude either that Nielsen thinks there are no
adherents of the tradition leﬁ who might be offended by his tone, or that he is an unpleasant person. Be that
as may, I should only like to note, in conjunction with Nielsen’s use of the term “ontotheological” as a

xi

to Objectivity, objects to what she calls the “anticultural attitude” of traditional
philosophy. This attitude, she claims, “has been essential to the discipline’s conception
of itself as ‘cultural overseer,’ able to transcend history and discover an ultimate ‘neutral
framework’ within which to situate other human endeavors or describe reality.”ls She
notes, however, that philosophy “has been forced to recognize that its ‘enduring’ issues
and ‘timeless’ concerns are the products of very particular cultural circumstances.”16 Iris
Marion Young and Maria Lugones criticize the vaunted impartiality and objectivity of
reason on the grounds that impartial and objective reason “leaves behind the particular
perspectives from which it begins, and reconstructs them as mere appearances as opposed
to the reality that objective reason apprehends,” thus ignoring, trivializing, and obscuring
differences in its quest for unity and for a privileged perspective on the nature of reality.17
Young regards such a standpoint as simply impossible, since “the experience of these
[‘mere’] appearances, however, is itself part of reality,” and Lugones claims that the
impartial and objective reasoner is “a ﬁction of his own imagination,” “shot through and
though” with “paradoxical incoherence.”18 Konstantin Kolenda argues that we should
refrain ﬁ'om teaching philosophy in what he calls “the realist mode,” and should bring our
recognition of the demise of the metaphysical and epistemological tradition of philosophy
to the classroom: “The recommendation proceeds from the assumption that a historicist

approach, combined with the acknowledgment of the centrality of language, would put

 

catch-all descriptive term for traditional metaphysics and epistemology, that adherence to traditional realism
is sometimes diagnosed by anti-realists as arising out of a desire for the universe to have a God who could
underwrite human knowledge a la Descartes.

'5 Bordo 1987, p. 3.

'5 Ibid.

"’ Young 1990, p. 102.

xii

the teaching of philosophy on a sounder track.“9 Kolenda insists that “[b]y letting go of
the (mistaken) idea that philosophy has access to special facts and concepts indigenous to
its own domain, philosophers, both as practitioners and as teachers, are likely to play a
more useful role in the intellectual life of our culture.”20
Now where would these philosophers have gotten these ideas? It would not be
sensible to enter into a detailed account of the historical background of the state of
contemporary philosophy in this preface; historiography of this sort is too dicey, too
controversial, and too likely to produce the impression that what follows the historical
account hinges on its accuracy. I will need, of course, to discuss what I take to be the
historical and philosophical basis for the rejection of rationalism and traditional realism,
but I believe that this basis can be elaborated in terms that are quite generally
recognizable, and do not require a great deal of storytelling. I also believe that one of the
faults of recent attempts to defend rationalism and metaphysical realism is that they have
not addressed themselves to the broadest and historically deepest motives for rejecting
these positions, and so have not offered a challenge to the empiricist and pragmatist
traditions at their roots: namely, the supposed elimination of transcendental metaphysics

and the fundamental failure of the rationalist project. Although both Ayer and positivism

are now gone, the legacy of “The Elimination of Metaphysics” lies not in its relationship

 

'8 Ibid.; and Lugones 1994, p. 467. Bordo’s, Young’s, and Lugones’ criticisms all turn, to some extent, on
another fairly common diagnosis of adherence to traditional realism (and rationalism) as arising from a
desire for social control and domination.

'9 Kolenda 1992, p. 121, italics mine.

2° Ibid. It’s hard to see how letting go of the idea that philosophy has any speciﬁc or special subject matter
is going to enable philosophy to play a more useful role in the intellectual life of our culture, or, for that
matter, a more useful role in the intellectual life of the university. Indeed, it’s hard to see how letting go of
this idea is even consistent with philosophers’ insistence that we deserve a portion of the university’s
resources. I submit that it is not an accident that while enrollments in “core” philosophy courses are falling,

xiii

to the promotion of positivistic philosophy, but in its crystallization of a movement in
philosophy away from pure reason which is still underway. This movement is animated
by the conﬁdent belief that metaphysical realism is untenable, that foundationalist
rationalism is a lost cause, and that philosophy can and must do without them, and be
done without them. I believe, however, that the attempts that have been made in the past
century to philosophize without metaphysical realism have failed in their central ambition
— they have appeared to repudiate metaphysical realism only by covertly appealing to
metaphysicaily realist postulates of their own. If I am right about this, and about the
reasons I shall offer as to why these attempts have failed, it will remain only to examine
and undermine the motives which caused philosophers to attempt this elimination of
metaphysics in the ﬁrst place, and to show not only how rationalism and metaphysical
realism might be renewed, but also why such projects remain indispensable to

philosophy.

 

enrollments in “Critical Thinking” are still burgeoning. What else, these days, are philosophy teachers good
for? Seems to me that we’ve historicized ourselves right out of business.

xiv

TABLE OF CONTENTS

PART I: THE ROOT OF THE PROBLEM

CHAPTER 1

REPRESENTATIONALISM, THE RISE OF EMPIRICISM, AND THE REJECTION
OF REALIST METAPHYSICS ............................................................... 1

PART II: POSITIVISM

CHAPTER 2
AYER AND CARNAP: THE ELIMINATION OF METAPHYSICS ....................... 15

PART III: NEO-PRAGMATISM

CHAPTER 3
QUINE’S REJECTION OF THE A PRIOR] ..................................................... 35

CHAPTER 4
PUTNAM ON DIRECT REFERENCE AND CONCEPTUAL CONTENT ................ 55

CHAPTER 5
RORTY: THE END OF REPRESENTATIONALISM AND THE ELIMINATION OF
EPISTEMOLOGY ............................................................................... 79

PART IV: CONTEMPORARY RATIONALISM, A PARTIAL SURVEY

CHAPTER 6

JERROLD J. KATZ’S REALISTIC RATIONALISM ......................................... 99

CHAPTER 7

LAURENCE BONJOUR’S DEFENSE OF PURE REASON ................................ 127

CHAPTER 8

MOVING BEYOND REPRESENTATIONALISM: SKETCH OF A PROGRESSIVE
RATIONALISM ............................................................................... 150

WORKS CITED ................................................................................... 171

XV

CHAPTER 1

REPRESENTATIONALISM, THE RISE OF EMPIRICISM, AND THE
REJECTION OF REALIST METAPHYSICS

It is well known that Descartes disapproved of the methods, principles, and
assumptions of the “Schoolmen,” the Catholic Scholastic academics who still controlled
much of higher learning in seventeenth-century Europe. It is also well known that
Descartes established foundationalism as an epistemological position because he believed
that to secure our entire system of beliefs we must begin with indubitable propositions. It
is less well appreciated, however, just how and why these two facts are related (although
much has been written on the subject), and it is rarely, if ever, observed that what
Descartes took himself to be doing was reforming philosophy speciﬁcally through the
elimination of speculative metaphysics. He was appalled most by what he perceived to
be the unruly and brazen insecurity of Scholastic metaphysical postulates. He had thus
become convinced that if he “wanted to establish anything ﬁrm and lasting in the
sciences,” he would have to “to begin again from the ﬁrst foundations.”l His goal was to
identify a method that would distinguish legitimate from illegitimate bases for belief, and
that would prepare legitimately grounded beliefs for certiﬁcation, thus establishing with
certainty our knowledge of the world. In short, he was hoping to provide philosophy as a
whole with the sort of security enjoyed by the most rigorous logical proofs. In aspiring to
the successful completion of such a project —- the reformation of philosophy through the

elimination of speculative metaphysics and the development of a secure metaphysics -—

 

' Descartes 1979, p. 13.

Descartes established himself, in fact, as the ur--positivist.2 The speculative (that is,
unsecured), largely Aristotelian metaphysics of the Schoolmen was to be replaced with a
secure and certiﬁable metaphysics that would provide an account of both the possibility
and the actuality of our knowledge of the world outside our minds.

Descartes’ metaphysics was a version of representationalism that borrowed from
the Scholastic metaphysics he intended to eliminate.3 His view, in brief, was that our
awareness of objects in the world was a result of the objects giving rise to impressions or
representations of themselves in our minds through the medium of our senses. The
representational features of an idea are dictated by the actual features of the object it
represents; our ideas are in this sense literally informed by their objects. For Descartes,
of course, what guarantees that our senses can and do carry out this task successfully is
that our senses, along with the rest of our corporeal and non-corporeal faculties, were
created by a non-deceitful, benevolent God. Descartes believed that the existence and
benevolence of God could be established with certainty, which would, in turn, establish

with certainty the accuracy (or, at least, the capacity for accuracy) of our senses. A

 

2 As Moritz Schlick says, in “Positivism and Realismz” (1959, p. 83) “If every view is to be labeled
positivist, which denies the possibility of metaphysics, then nothing can be said against it as a mere
deﬁnition, and in this sense I would have to declare myself a strict positivist.” It is in this sense that
Descartes founded, at once, both the modern rationalist and the modern empiricist traditions. He meant to
declare what he regarded as speculative metaphysics to be impossible. Schlick continues: “[positivists] are
always talking of the ‘given,’ and state their basic principle mostly by saying that, like the scientist, the
philosopher must abide throughout in the given.” Descartes would have agreed; any disagreement lies only
in what can be said to be taken to be given, as Schlick goes on to point out.

3 Note the Aristotelian (even Platonic) ﬂavor of the notion of an idea’s being informed by the object it
represents. The notion seems to be this: the form of the object is transmitted to the mind so that the mental
item — the idea — that represents (or depicts, or reﬂects) the object captures or ﬁgures its form. It is not
fully worked out by Descartes in the Meditations just how ideas represent their objects (whether ideas are
literally little pictures of their objects, or merely serve to register their formal elements and the relations
between them, or what), and there is, in fact, some disagreement about how, exactly, Descartes conceived
of the nature of representation (see Anthony Kenny 1968, chapter 5). To be sure, the conceptual
underpinnings of this theory of mental representation are not strictly Aristotelian (or Platonic), but were
worked out to some degree by Scholastic and Medieval philosophers long before Descartes put them to this
use. The notion of form that serves as the basis of his theory can, however, be traced back to the ancients.

benevolent God would, after all, not have equipped us with senses that are wrong all of
the time, so they must be capable of providing us with accurate information about the
world and its objects, and must actually do so at least some of the time. Descartes relied
upon his proofs for the existence of God to certify his representationalist schema, and
these proofs would themselves constitute a self-certifying secure metaphysics, as against
the speculative metaphysics of the Schoolmen, which, though dependent upon God, did
not rest upon indubitable foundations, and so remained unmethodical, unveriﬁable, and
uncertain (and, thus, could not constitute knowledge).

Representationalism, as codiﬁed by Descartes, exerted a powerful inﬂuence over
the minds of Western philosophers, and continues to do so to this day. It offers such an
intuitively agreeable metaphysical picture of the possibility of our knowledge of things
that its pull seems to be nearly irresistible, as we shall soon see. What makes its
popularity so remarkable is that it has survived despite being completely and soundly
repudiated not even one hundred years after the publication of Descartes’ Meditations.
By that point, arguments to the effect that it is impossible to establish the existence of
anything (much less God) by deductive proof were well known, and Hume was quick to
point out that without God, representationalism had nothing and no one to vouch for it.
The point is essentially the same one that Hilary Putnam, Richard Rorty and others have
become so fond of making: there is no way for us to stand outside our own system of
ideas (or conceptual scheme, or language, or whatever), and see that our ideas and their

objects are connected to each other in the way representationalism says they are.4 Any

 

4 Putnam and Rorty seem to treat this as a refutation of metaphysical realism and rationalism simpliciter,
though of course it is not, as should be fairly obvious. It would, perhaps, serve as such a refutation if
representationalism were the only plausible version of rationalism available, or the only epistemological

attempt to do so (if such a thing could even be imagined) would just involve more
concepts or ideas, whose relation to the world and its objects would be in the same boat
as the rest of our system.

We can see why Descartes felt he needed to rely on his proofs of the existence of
God to certify his representationalism — what besides God possibly could certify such an
epistemology? If our epistemology calls for our awareness, and ultimately our
knowledge, of the world to involve the world’s having a certain sort of effect upon our
minds through our senses, establishing that this epistemology is correct requires that we
can conﬁrm that this relationship holds. To do so, we would either have to be able to
observe that it holds (which, of course, we cannot), appeal to someone or something that
could make such an observation or could be a priori aware that the relationship holds, or
appeal to a priori considerations of our own. In the absence of deductive proofs of the
existence of God, our own a priori considerations are all that we have left, and Hume
argued that no such considerations were available. Thus, there seems to be no way to
establish that representationalism is an accurate picture of the relation between the mind
and the world, and, what is worse, this vitiates it as an epistemological position, since this
relation is itself part of the world, and so should be capable of accurate representation.5 It
seems, however, to be simply impossible to represent the representation-relation (or, at

least, to know that we have done so properly), since we may ask of any representation of

 

basis for metaphysical realism. I believe that, in fact, this is precisely what Putnam and Rorty suppose, but
in this I think they are mistaken. I will return to this point in Chapters Four and Five.

5 It could be said that we ought to accept representationalism on inductive grounds, as the best explanation
of our experience of the world as consisting of medium-sized discrete physical objects. This is, in fact, the
position of scientiﬁc realism: namely, that while representationalism cannot be established, it can still be
maintained abductively. Of course, quantum physics puts a great deal of pressure on the idea that there
even are medium-sized physical objects, so no na'l‘ve construal of representationalism seems plausible.

Still, some sort of more sophisticated form of representationalism could perhaps be advanced. I shall argue
in Chapter Five, however, that Rorty has revealed representationalism to be untenable in any form.

the relation itself how we know that this representation accurately represents its object,
which leads to an inﬁnite regress.6

Kant could not, however, accept the skeptical Humean conclusion that we simply
cannot say anything about the source(s) of our impressions and, ultimately, our ideas. He
could see that Hume was right about Cartesian representationalism, and about
representationalism in general, but was dissatisﬁed with where Hume left things. Kant
felt that, by recognizing the inherent untenability of representationalism, Hume had very
nearly brought philosophy to the revolution in metaphysics that Descartes had himself
envisioned, but had stopped just short, and had, in effect, repudiated metaphysics
altogether.7 This, to Kant, was unacceptable: Hume “ran his ship ashore, for safety’s
sake, landing on skepticism, there to let it lie and rot,” whereas he “certainly struck a

spark from which light might been obtained, had it caught some inﬂammable substance

 

6 The only apparent way out of this problem would seem to be to show that the representation of the
representation-relation stands in the same relation to the relation itself as is depicted in the representation of
this relation. Unless we could do so on a priori grounds, however, there would seem to be no way to do
this. This, I believe, is the reason why Putnam and Rorty maintain that any attempt to do so would just
involve more (intratheoretical) concepts, and thus launch us upon an inﬁnite regress: they maintain that
none of our concepts or ideas are “privileged” over against any other, which is just as much as to say that
there are no a priori concepts or ideas. Naturally, given such an empiricistic assumption, it would be
impossible, using the resources of our conceptual scheme, to raise ourselves up out of our conceptual
scheme so as to have a look at the whole thing, in addition to its “ﬁt” with reality itself. Any time we ask
the question “How do our concepts hook on to reality?” we can answer only by appealing to more concepts,
about which we can always ask “how do they hook on to reality?” and so on, and so on.

Even if we do not go along with such a general rejection of a priori reasoning, however, there is
good reason to think that there is no way to support representationalism on an a priori basis, and, thus, no
way to support it at all. I shall argue, in Chapter Five, that the best way out of this problem is to reject
representationalism altogether in favor of some other version of metaphysical realism.

7 In a passage that has since become famous, Hume advises us as follows: “if we take in our hand any
volume; of divinity or school metaphysics, for instance; let us ask, Does it contain any abstract reasoning
concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of
fact and existence? No. Commit it then to the flames: For it can contain nothing but sophistry and
illusion” (Hume 1993, p. 114). Though Hume was no doubt being glib, this seems to be a case of what
Putnam refers to as the tendency of philosophers who lose sight of the transcendence of reason to slip into
cultural imperialism (see Putnam, “Why Reason Can’t Be Naturalized” (I983e), p. 235 and pp. 238-240).
Positivists did not tend to be any more charitable than Hume, though they never, to my knowledge,
advocated book burning.

and had its smouldering ﬁre been carefully nursed and developed.”8 Kant recognized that
what Hume had done was to demonstrate that Cartesian “secure” metaphysics was, in
fact, no more secure than Scholastic metaphysics, and therefore could only be
speculative, which is to say worthless. He believed, however, that this put him in a
position to establish, once and for all, a truly secure metaphysics, to succeed where
Descartes had failed — to bring about a reformation of philosophy by ridding it of
speculative metaphysics, and by ﬁnally founding metaphysics as a science. In this, Kant
was carrying on the positivistic tradition handed down from Descartes.

Descartes had, in effect, proposed a new standard for every ﬂedgling
epistemology to meet, and, until fairly recently,9 this standard had been acknowledged by
Western philosophers as naturally and properly suited to the epistemological enterprise in
general. Descartes’ standard is: epistemology, as the science of human knowledge, must
be secure — that is, it must derive its force ultimately from premises that are logically
rigorous and rationally indubitable — otherwise, it can have no authority. Descartes
certainly was not the ﬁrst epistemologist, or the ﬁrst philosopher, to propose
epistemological standards (either for what was required for a single belief to count as
knowledge, or for what was required for an epistemological theory to count as minimally

acceptable), but he is regarded as the originator of modern epistemology because it was

 

3 Kant 2001, pp. 6 and 2.

9 This standard has largely been abandoned, or is in the process of being abandoned, across all or nearly all
philosophical disciplines. There is now widespread conviction that the search for transhistorically and
transculturally certain foundations was naive and hopeless from the outset. Of course, it is almost not
worth pointing out that this position — in insisting upon the hopelessness of discovering secure
metaphysical foundations for our epistemology - requires heavy metaphysical assumptions of its own, to
the effect that there are no transhistorical or transcultural truths, or none of which we can be aware. In any
event, this conviction seems to me generally to be grounded either (1) on some version or other of political
or ethical pluralism, in which case it is simply a non sequitur (since I take it for granted that no
epistemological consequences follow from our political or ethical commitments as such), or (2) on neo-
pragmatism, in which case it is mistaken, as I shall argue in Chapters Three, Four and Five.

Descartes who, in implicitly distinguishing secure from speculative metaphysics,
launched the modern epistemological project of discovering a secure metaphysical basis
for our epistemology — that is, an irrefutable metaphysical account of what makes it
possible for us to have any knowledge of the world outside our minds, and of how we
acquire such knowledge. This is precisely what is now meant, thanks to Kant, by a
“critical” project, or a “critique”: namely, an attempt to provide an account of what
makes something possible, in order to pave the way to explaining how and why it is
actual. Kant was attempting to meet Descartes’ standard, but knew that Descartes’
metaphysics could not do so. Consequently, he concluded that metaphysics needed to be
reformed once again.

Kant began, as he himself notes, by applying Hume’s conclusion about causation
to our other ideas that are supposed to be a priori, such as space, time, and number, and
found that in each case they are such as to be incapable of arising through empirical
observation alone, but are nevertheless indispensable to reasoning. Indeed, Kant claims,
what he discovered was that these a priori ideas are the forms taken (or taken on) by all
of our experience —— the very conditions of experience, or (Kant postulated) of our having
experience at all. This was the secure metaphysics Kant was looking for: metaphysics
was now to be conceived as a science of concepts —- the conceptual apparatus of pure
reason. It would be a priori because it was, as it were, built in to all experience simply in
virtue of its being experience at all; it would necessarily precede experience because it
formed the outer limits of experience for us as experiencing beings. It would be secure
because it was the only form in which we could have experiences, and was what

explained the possibility of our experience.

There are, however, three serious, and related, problems with Kant’s metaphysics.
Firstly, by reconstruing metaphysics as a science of concepts, Kant broke with the
traditional philosophical usage of “a priori” to refer to information we possess about the
world in advance of experience of it. For Kant, a priori truths were truths about the
structure of pure cognition, not truths about the world. On this view, a priori knowledge
becomes analytic knowledge about the necessary structure of experience and about the
propositions that express the facts about this structure, thus paving the way for the
reduction of traditional epistemology to theories of reference that was distinctive of the
Linguistic Turn. Secondly, his metaphysics involves the claim that the world comes to us
ﬁltered through our forms of sensible intuition, one consequence of which is that we
cannot know whether our concepts “hook on” to the world (for all the reasons given
above), or what the world is in fact like outside of our experience of it. This, however,
invites the objection that we really have no grounds for talking about “the world” as such,
or about “the world” as the cause of our experience, and, thus, no grounds for saying that
we can really say anything about “the world” itself at all. As a result, Kant’s attempted
rescue and reform of metaphysics turns out, like Descartes’, to collapse into speculation.
Thirdly, as should by now be clear, Kant’s epistemology was representationalism without
the representation-relation. Kant largely preserved the pluralistic metaphysical schema of
representationalism, and merely replaced representation with “ﬁltering”; our ideas “of”
objects, or “of” the world, capture how the world appears to us, as ﬁltered through our
sensory and cognitive apparatus; this is Kant’s famous transcendental idealism. Kant did
not so much replace representationalism as modify it. It was a breathtakingly ingenious

modiﬁcation, to be sure, but it was a modiﬁcation nonetheless.

A. J. Ayer, like Descartes and Kant before him, felt the need for reform, and
proposed again to reform philosophy by eliminating speculative metaphysics. Unlike
Descartes and Kant, however, rather than attempting to replace speculative metaphysics
with a secure metaphysics, Ayer intended to replace speculative metaphysics with
nothing at all. The epistemological work that metaphysics had formerly been called upon
to do — the explanation of the possibility of a priori knowledge — could now be done,
according to Ayer and other positivists, by the notion of analyticity, entirely within the
conﬁnes of (natural or artiﬁcial) language alone. What appeared to be a priori awareness
of certain necessary truths about the world (such as mathematical truths) could be
understood, and accounted for, as intralinguistic or intrasystemic deﬁnitions; they are
necessarily true because we do not allow them to be false10 — we would sooner reinterpret
our physical theories than allow experimental evidence to overturn an analytic
(mathematical or logical) statement.ll In other words, their being necessarily true is
strictly a matter of convention.

The revolution in philosophy that had begun with Descartes’ attempt to reform
metaphysics had ended with the empiricist revolution of the late nineteenth and early
twentieth centuries, and with the Linguistic Turn; in short, it had ended with the rejection
of metaphysics as a philosophical enterprise concerned with understanding the nature and

structure of the world as it is in itself, and with its expulsion from philosophy.12 Like

 

'0 And because they form a consistent system, do not contradict one another, are the most basic statements
of their kind, and so on.

“ Ayer 1952, pp. 76-77.

'2 Of course positivism is now deﬁrnct, and metaphysics has made something of a wary and defensive
return to philosophy, but metaphysics in the analytic mode is, by and large, not a resumption of traditional
metaphysics, as I shall argue in Chapter Eight, and only a handful of philosophers today carry on the
tradition. Nowadays, philosophers of all sorts either propose that we go on without traditional metaphysics,
or they try silently to put this advice into practice, by conducting themselves as if traditional metaphysical
considerations were superﬂuous. John Rawls has famously remarked that his theory of justice is not a

Descartes and Kant, however, positivists sought a method that would enable philosophers
to distinguish legitimate from illegitimate bases for belief, and that would prepare
legitimately grounded beliefs for certiﬁcation, thus establishing with certainty our
knowledge of the world. They sought a means of grounding our entire epistemological
enterprise, and they believed that they had identiﬁed the ground: by construing objects as
groups of sensory stimulations, and reports of these stimulations as observation
statements, the entirety of our empirical knowledge (which is to say, the entirety of our
knowledge) could then be grounded upon our sensory stimulations coupled with our
linguistic conventions. The only question that remains is how we are to distinguish
observation statements (or statements of fact) from analytic statements on the one hand,
and nonsense on the other; after all, some statements that looked like factual statements
or analytic statements were in fact neither, but were illegitimate metaphysical assertions.
Positivists had the answer to this too: in order for a statement to be meaningful, it had to
be veriﬁable, which is to say that it had to be conﬁrrnable or (in Quine’s phraseology)
inﬁrmable by experience.

The chief problem with the criterion of veriﬁcation, however, was that, in order

for it to count as part of the apparatus of positivistic epistemology, it would have to be

 

metaphysical theory but a political one, and he suggests that “the justiﬁcation of a conception of justice is a
practical social task rather than an epistemological or metaphysical problem.” He further adds that his
conception of justice as fairness “presents itself not as a conception of justice that is true, but one that can
serve as a basis of informed and willing political agreement between citizens viewed as free and equal
persons.” (See Rawls 1985, p. 224 (n.2), and p. 230.) Kai Neilsen has proposed “setting aside the
ontotheological tradition of philosophy, the grand old tradition of philosophy, ” and pursuing philosophy
“within the limits of wide reﬂective equilibrium alone.” (See Nielsen 1994, p. 3.) Nelson Goodman sees
himself “at odds with rationalism and empiricism alike, with materialism and idealism and dualism, with
essentialism and existentialism, with mechanism and vitalism, with mysticism and scientism, and with most
other ardent doctrines,” and claims that there is no way of saying what reality is like “apart from all frames
[of reference].” (See Goodman 1978, p. x and p. 3.) Bas C. van F raassen advocates a thoroughgoing
instrumentalism about scientiﬁc theories, and insists that such an account of science must “involve
throughout a resolute rejection of the demand for an explanation of the regularities in the observable course

10

construed as a statement either of a matter of fact or of an analytic truth, but the statement
was neither veriﬁable itself on the basis of observation or experience, nor true by
deﬁnition (analytic). What was ultimately worse for positivism than the unveriﬁability of
the veriﬁcation principle, however, was Quine’s discovery that the notion of analyticity
just could not be made to work within the conﬁnes of a strictly empiricist epistemology,
which meant that there could be no hope for differentiating analytic statements from
statements of matters of fact, and thus no hope for positivistic epistemology.

After the demise of positivism, “analytic” philosophy went off in two different
directions: neo-pragmatism, which has two branches — the internalist/holist camp
(inspired by Quine), and the linguistic analysis/philosophy-as-therapy camp (inspired by
Austin and Wittgenstein, respectively) —- and scientiﬁc realism. Although the positivists’
notion of analyticity had been repudiated, neo-pragmatic “analytic” philosophers went on
treating ideas as linguistic items, since this (supposedly) did not require adopting any
metaphysical assumptions or postulates concerning the relationship between the mind and
the world “in itself.” That is, neo-pragmatists remained committed to empiricism, in
spite of the defeat of positivism, because representationalism and metaphysics seemed to
have perished once and for all, with the last of the positivists. The subject matter of
philosophy has been taken by neo-pragmatists to be linguistic (and second-order - about
language or concepts, and not about the world) ever since. As Quine put the matter in
“Epistemology Naturalizedz” “Surely Hume would have used contextual deﬁnition

with pleasure if [he] had thought of it.”'3 The important thing to notice, for our purposes,

 

of nature, by means of truths involving a reality beyond what is actual and observable, as a demand which
plays no role in the scientiﬁc enterprise.” (See van Fraassen 1980, p. 203.) I could go on.

3 Quine 1985, p. 18. I tend to think, as Quine apparently does not, that it is quite signiﬁcant that Hume did
not think of it, and nor did any philosopher in the history of the discipline, principally because they

11

however, is that the neo-pragmatists’ rejection of metaphysics was a result of the
retention by the epistemological tradition of representationalism as the epistemology of
choice.

In spite of Hume and neo-pragmatism, scientiﬁc realists adopted
representationalism as their epistemology, but, rather than offering a secure metaphysics
to support it, they opted for an inductive stance: realism about the objects of science, and
representationalism as an account of our knowledge of these objects, provides us with the
best explanation of the success of the scientiﬁc enterprise. Consider the following from

Hilary Putnam:l4

It is beyond question that scientists use terms as if the associated criteria were not
necessary and sufficient conditions, but rather approximately correct characterizations of
some world of theory-independent entities, and that they talk as if later theories in a
mature science were, in general, better descriptions of the same‘entities that earlier
theories referred to. In my opinion, the hypothesis that this is right is the only hypothesis
that can account for the communicability of scientiﬁc results, the closure of acceptable

scientiﬁc theories under ﬁrst-order logic, and many other features of the scientiﬁc
method.[5

This position has provoked scads of replies, from Bas van Fraassen’s “adequacy”
theory[6 to Latour and Woolgar’s inscriptionism.l7 Of course, scientiﬁc realism has also

attracted the replies and scorn of other neo-pragmatists, most notably those neo-

 

virtually all shared the same conviction: that the subject matter of philosophy is the nature of reality.
Perhaps the most important consequence of the view that the subject matter of philosophy is linguistic is
the view that ideas and concepts are linguistic, and that, therefore, our experience of the world takes place
in and through language. This view is unquestionably false (see, for instance, Laurence BonJour’s
“Analytic Philosophy and the Nature of Thought,” (forthcoming), and Bryan Magee’s (I999) Confessions
of a Philosopher, chapter 5), but what should here be noted is that it relies, ultimately, on the defeat of
representationalism and the rejection of metaphysics.

'4 Prior to his conversion to internal realism.

'5 Putnam 1975, p. 237.

‘6 See van Faassen, 1980.

'7 See Latour and Woolgar, 1979.

12

pragmatists who formerly identiﬁed themselves as scientiﬁc realists but have since
jumped ship —— such as Putnam himself, and Nelson Goodman — which is, incidentally, a
common occurrence. Putnam’s replies tend to the global, and generally attack scientiﬁc
realism at the point of its fondest epistemological aspirations. Putnam has pointed out,
quite rightly, that the representation-relation scientiﬁc realists depend upon, and hope
will one day be “worked out,” can never be worked out from within the conﬁnes of an
empiricist programme, since any such “working out” of the representation-relation would
simply involve more concepts (more linguistic items) — in other words, boot-strapping.'8
Neo-pragmatists tend exasperatedly to insinuate that scientiﬁc realists, as heirs of
positivism, do not seem to have learned anything from its downfall.

All of this has left the traditional epistemological project of modern Western
philosophy with the following two choices, both empiricist: (1) either our epistemology
has, and can have, no possible foundation, and any talk of the “ﬁt” between our
conceptual scheme and reality itself is just hopeless pining for a “God’s eye view” that
we can never have; or (2) our best hope lies in the continued success of science, and of
intertheoretic reduction, which increases the likelihood that scientiﬁc realism will be
accepted as the best explanation of the success of science. The question is, why be an
empiricist? What apparently makes rationalism so implausible is that its ﬂagship
epistemology, representationalism, was demolished over two hundred years ago. This
does tend to make scientiﬁc realists look not only quaint, but a bit daﬁ. The rejection of
metaphysics, however, that began with Descartes’ rejection of speculative metaphysics
and ended with positivism’s repudiation of all metaphysics as speculative, depends upon

the prominence, and ultimately the failure, of representationalism. In the following

 

'8 Putnam 1983, pp. 17-18.
13

chapters, I shall argue that the relatively recent repudiation of metaphysics in general as a
philosophical discipline concerned with the nature of reality itself was and is unjustiﬁed
and, in any event, unsuccessful anyway, not simply because epistemologies other than
representationalism are possible, but because any epistemology that attempts to ground
itself so as to meet Descartes’ standard (or one like it) requires a corresponding
metaphysics, and because every epistemology must meet this sort of standard to count as
a minimally acceptable account of knowledge. I shall also argue that neo-pragmatic
arguments to the contrary are nothing to the point, because they themselves depend upon
the very rejection of metaphysics that I am calling into question. Looking back over the
history here traced, we can see that the rejection of metaphysics relies on the retention by
the epistemological tradition of the repersentationalist schema; what must go is
representationalism, not metaphysics. I propose a return to metaphysics. If we cannot
hope for the secure metaphysics that was the dream of Descartes and Kant, then we must
return to speculative metaphysics, if that is the best we can do, because, as I shall argue,

philosophy cannot get on without it.

14

CHAPTER 2

AYER AND CARNAP: THE ELIMINATION OF METAPHYSICS

The history of modern epistemology can, as we have seen, be viewed as a history
of repeated attempts at the reformation of philosophy through the elimination of
speculative metaphysics, beginning with Descartes’ Schoolmen and Scholasticism, and
his reformation project. Kant had the natural theologians and the Cartesians to contend
with, and British and German proto-positivists had the German idealists (post-Kantians —
Hegel, Schelling, F ichte and others).I The rejection of speculative metaphysics
undertaken both by positivists and neo-pragmatists, then, was not so wholly novel as it
may seem. Several successive waves of such rejection had preceded the empiricist
revolution of the late nineteenth century. The difference between Descartes and Kant, on
the one hand, and both positivists and neo-pragmatists, on the other, however, is that,
while Descartes repudiated the metaphysics of his day, and Kant repudiated the
metaphysics of his, both Descartes and Kant inaugurated new metaphysical enterprises in
an attempt to produce a secure metaphysics, which were then overthrown in their turns,
whereas positivists’ and neo-pragmatists’ stated intentions, though perhaps unfulﬁlled,
were to replace speculative metaphysics with nothing at all, other than a new philosophic
method. In this sense, at least, positivists and neo-pragmatists have been truly

revolutionary in their hopes, if not in the event.

 

' Thus, the history of modern philosophy has been a history of empiricist revolution, even if Descartes and
Kant considered themselves, and have since been considered by others to be, rationalists. Their versions of
rationalism have, as it were, been emissaries and harbingers of the revolution, by equating epistemology
(and rationalism) with the project of accounting for the role of the a priori in grounding our empirical
knowledge, and thus by progressively narrowing the scope of the a priori to ﬁt this account. (See, also,
BonJour 1999, pp. 20-26.)

15

Both positivists and neo-pragmatists have claimed that the utterances of
metaphysicians are meaningless. Both have claimed that the debates which have so
preoccupied metaphysicians in the past are pointless, irresolvable, and ultimately
misguided. They have claimed that metaphysics should be, must be, overcome,
eliminated, purged from the discipline of philosophy, reclassiﬁed as poetry or mysticism
(or both, or as something far worse and less beautiful and expansive than either), and
renounced, in order that we may get on with the serious business of philosophy. They
say these things in the spirit of one who has purchased a set of well-used but sturdy tools
of the various practical trades, and has found among the lot divining rods, phrenological
tables, astrological charts, and crystal balls. The point is simply to clear out the
superstitious and impractical clutter, and to eliminate the blunt and clumsy and simply
useless implements in favor of the tools that clearly work.

Despite the fact that both positivists and neo-pragamatists say quite similar things
about metaphysics and about metaphysicians, however, and despite the fact that they
seem to say them for at least some of the same reasons — at least according to the
ﬁgurative analogy I offered just now — neo-pragmatists and positivists have (in the main)
rather different ideas of just what constitutes the serious business of philosophy. Thus,
they have also importantly different reasons for rejecting metaphysics and calling for its
elimination or excision from philosophy. Positivists were, like Descartes and Kant,
hoping to discover the true method in philosophy, the ground of our knowledge of the
world, the philosophical threshing basket which will enable us, once and for all, to
separate the wheat from the chaff, and they believed that the work of explaining the

apparent necessity of some truths formerly assigned to metaphysics could now be

16

accomplished by logical tools alone. A good deal of what has passed for respectable
metaphysics at one time or another has also, of course, appeared to be hogwash by the
lights of the succeeding age; as I shall argue in this and the following chapters, however,
the positivists’ and neo-pragmatists’ wholesale rejection of metaphysics was
unsuccessful, unjustiﬁed, and untenable. As we will see, positivists are committed to a
metaphysical doctrine of their own, which positivism should prohibit, and this is a

consequence of the indispensability of metaphysics to any comprehensive epistemology.2

A.J. Ayer identiﬁes as his target the traditional conception of philosophy as a
ﬁrst-order discipline distinct from science — the conception that involves “the thesis that
philosophy affords us knowledge of a reality transcending the world of science and
common sense.”3 He begins his assault on metaphysics by noting that the problem with

transcendental, or what I am calling speculative, metaphysical claims,4 is that they would

 

2 An argument quite similar to the one developed in this chapter is developed independently by Michael
Williams in his “The Elimination of Metaphysics” (1986) though he arrives at somewhat different
conclusions. Clark Glymour also makes an argument to similar effect against Putnam, in his “Conceptual
Scheming” (I982), but the general line of argument, similar to the one I develop here, can be equally well
applied to Ayer’s position in Language, Truth and Logic or, for that matter, to Quine’s position in “The
Limits of Knowledge” (l976a).

3 Ayer 1952, p. 38. In other words, metaphysics, broadly construed.

’ According to Ayer, and to positivists generally, all metaphysical claims are speculative, because,
according to positivism, there can be no secure metaphysics — unless, of course, we accept the Kantian
psychologistic deﬂation of the content and work of metaphysics; that is, unless we take the development of
a theory of pure cognition, or later, of an ideal language, to be metaphysics. I suppose that 1 cannot really
object if someone chooses to use the word in this way, but, as I have already noted in Chapter One, such
usage is not consistent with the traditional philosophical usage of “metaphysics” to refer to the study of the
nature of reality itself beyond direct experience. Moreover, I would, in a Kripkean move, be willing to let
the positivists have the term “metaphysics” to refer to the development of an ideal language or whatever,
but I would coin the term “schmetaphysics” to refer to the study of the nature of reality beyond direct
experience, and argue that it should be clear that (l) metaphysics in the positivists’ sense is not
schmetaphysics, and (2) positivists do not (intentionally) do schmetaphysics at all.

17

appear to have no basis in experience. Ayer observes that we might ask the
metaphysicians for an account of the source of their premises, on the basis of which they

supports their metaphysical hypotheses:

Must [the metaphysician] not begin, as other men do, with the evidence of his senses?
And if so, what valid processes of reasoning can possibly lead him to the conception
of a transcendent reality? Surely from empirical premises nothing whatsoever
concerning the properties, or even the existence, of anything super-empirical can
legitimately be inferred.5

Lest any questions should already seem to have been begged against the metaphysician,
Ayer is prepared to acknowledge the claim on the part of the metaphysician to a priori
awareness that would provide the requisite support for her metaphysics. Thus, Ayer
concludes, “one cannot overthrow a system of transcendent metaphysics merely by
criticizing the way it comes into being. What is required is rather a criticism of the nature
of the actual statements which comprise it,” presumably, in order to show that they can
have no conceivable meaning.6 Ayer is prepared with just such a criticism: “Our charge
against the metaphysician is not that he attempts to employ the understanding in a ﬁeld
where it cannot proﬁtably venture, but that he produces sentences which fail to conform
to the conditions under which alone a sentence can be literally signiﬁcant.”7 What is the
criterion of literal signiﬁcance? The criterion of veriﬁability. This criterion states that a
sentence is literally meaningful if, and only if, the speaker of the sentence can identify
what observation or set of observations would lead her to accept the sentence, and what

observation or set of observations would lead her to reject it. Following the criterion, we

 

5 Ayer 1952, p. 33.
6 Ibid., p. 34.
7 Ibid., p. 35.

18

can see that a sentence such as “‘the Absolute enters into, but is itself incapable of,
evolution and progress’” is unveriﬁable, and thus meaningless.8
Sentences such as “unicoms are ﬁctitious,” however, might at ﬁrst appear to pose
a problem for the criterion, inasmuch as such a sentence seems to express a truth in a way
that the sentence involving “the Absolute” does not, since the conventions of
quantiﬁcational logic, which call for existential generalization over such sentences,
would require us to infer that (1) unicoms exist, and that (2) unicoms are a sort of
existent entity called “ﬁctions.” Clearly this is not the proper conclusion. Thus, what is
wanted is an analysis of the logic of the sentence that reduces the apparent referring term
“unicorn” and the apparent predicate “ﬁctitious” to circumlocutions which would enable
us to avoid conclusions (1) and (2) above, while preserving the conventions of
quantiﬁcational logic. Metaphysicians, Ayer suggests, are often bewitched into thinking
that there are such things as substances by the structure of language, in more or less the
way that a straightforward reading of “unicoms are ﬁctitious” according to the principles
of quantiﬁcational logic would yield conclusions (1) and (2). Philosophers who, Ayer
says, are “infected by the primitive assumption that to every name a single real entity
must correspond,” are misled by the further assumption that we must draw a sharp
distinction between a thing and its attributes.9 In calling the thing-in-itself a “substance,”
such philosophers are then forced to the illegitimate and unintended metaphysical
conclusion that there is such a thing as substance as a feature of reality in itself. Thus are

many philosophers ineluctably led to make metaphysical assertions without intending to

be metaphysicians (but only, presumably, by meaning to talk about certain sensible

 

‘ Ibid., p. 36.
9 Ibid., p. 42.

19

entities). There remains, then, the correction of these metaphysical errors which result
from seemingly innocuous assumptions about language and its use, and the correction of
these errors by way of the production of sentential or linguistic analyses will constitute an
important part of the business of philosophy, inasmuch as philosophy will be charged
with the reﬁnement of the ideal language of scientiﬁc discourse.

Now it is not my intention to defend such statements as “the Absolute enters into,
but is itself incapable of, evolution and progress.” However, we must be clear about what
warrants us in using the criterion of veriﬁability to rule it out as meaningless. Perhaps
the most famous objection to the criterion of veriﬁability - which I will henceforward
refer to as the principle of veriﬁcation or as the veriﬁcation principle — is that it is itself a
sentence which purports to be meaningful and which purports to set the standard for
meaningful utterance, but that it does not satisfy itself, since it is not empirically
veriﬁable.

It is not clear, however, that this is the case. The claim that a statement is
meaningﬁrl if, and only if, its speaker can identify those observations that would lead her
to accept the statement, and those observations that would lead her to reject it, can be
veriﬁed (if any statement can be veriﬁed at all) by investigating whether in fact speakers
treat statements as meaningful if, and only if, they can identify those observations that
would lead them to accept those statements, and those observations that would lead them
to reject them. The problem is not that the veriﬁcation principle is not itself veriﬁable, it
is that positivists did not intend the principle as itself an empirical generalization that
would stand or fall on the results of empirical research. They intended it to apply to all

utterance in advance of investigation into the practices of speakers. One of the

20

consequences of accepting the principle, however, is that the only statements that are
meaningful and can be known to be true in advance of any investigation are analytic, or
true in virtue of the meanings of their terms alone (i.e., tautological). The veriﬁcation
principle is not, however, analytic, so it must be an empirical generalization. Some
positivists were prepared to accept this consequence until it was realized'0 that no
statement can be veriﬁed — that is, established or conﬁrmed to be true — empirically;
statements can only be more or less well conﬁrmed by the sum total of past experience.
What this means, however, is that empiricism can only be more or less well supported by
experience, according to its own standards. It cannot be established by experience, or by
reason alone. Thus, the criterion would seem to be without support, merely a piece of
positivist dogma.

This provoked two different reactions among positivists: certain positivists
acceded to the charge of dogmatism, but attempted to mitigate the damage by claiming a
more limited scope for the principle; others refused to be called dogrnatists, and insisted
that the principle which establishes the criterion for meaningful utterance cannot itself be
held to this criterion. However, it is not altogether clear why the principle should be
accorded such a status. Even if it could be counted among the conventions of our
language, we would then be forced to regard it as analytic, which it clearly is not, and
which would render it silent on matters of fact, such as which statements are meaningful

and which are not, which would amount to suicide for positivism.H

 

'0 Perhaps it would be better to say “until it was remembered,” since Hume had already made the point.

” I take it for granted that the question of which statements are meaningful and which are meaningless is a
matter of fact — about the relationship between sentences (or statements or propositions or whatever) and
their objects, whatever they are — and cannot simply be “decided” by the adoption of an analytic principle
of sentence veriﬁcation or construction.

21

It is not that there is no way in which the truth of the statement could be borne
out by observation, or sets of observations. Rather, the principle is supposed to state the
criterion for the meaningfulness of statements, thus, the judgment that the principle itself
is meaningful or meaningless would already have to assume the principle (or some other
such principle) in the ﬁrst place. We would thus need either another principle (in which
case it is not clear why we even need the veriﬁcation principle), or a metaveriﬁcation
principle, to which to appeal in such cases, which would claim that “The sentence ‘A
sentence is meaningful if, and only if, the speaker of the sentence can identify what
observation or set of observations would lead her to accept the sentence, and what
observation or set of observations would lead her to reject it’ is meaningful if, and only if
...” and so on up. This, of course, would produce an inﬁnite regress of metaveriﬁcation
principles, which I can only assume would be vicious.

What is even more important to notice, however, about this admittedly shopwom
point, is that the veriﬁcation principle could not possibly apply to all utterance in
advance. Ayer observes that the criterion of veriﬁability will require some adjustment if
it is to allow for what seem to be perfectly reasonable empirical generalizations (such as
“All ravens are black”) which are understood to apply to a potentially inﬁnite number of
cases, which would surely transcend the bounds of possible experience. He then
modiﬁes the criterion to read: A sentence is meaningful if, and only if, the speaker of the
sentence can identify what observation or set of observations would “be relevant to the
determination of its truth or falsehood.”12 No statement, other than an analytic statement,
can be thought to be true come what may, and can be thought to cover a potentially

inﬁnite number of cases. The veriﬁcation principle is, however, supposed to apply to all

22

sentences — a potentially inﬁnite number of sentences —- and to have identiﬁed them as
meaningless in virtue of certain facts about them that transcends the bounds of possible
experience. That is to say, the veriﬁcation principle must be either analytic or a priori. It
cannot be analytic if it is to concern matters of fact, but it also cannot be a priori by dint
of Ayer’s empiricism. The only way it could possibly serve as the criterion of
meaningful utterance is as a synthetic a priori truth about all utterance. '3 The veriﬁcation
principle itself would then seem to commit the positivist to a metaphysical postulate as
unwarranted, by her lights, as any statement involving the Absolute.

Since the veriﬁcation principle was supposed to justify the positivists’
empiricism, this would be admitting that empiricism cannot defend its own conception of
philosophy except by subjecting it to empirical conﬁrmation, and thus by begging the
question. The principle cannot, of course, establish empiricism, since the principle is
itself, at bottom, merely an expression of a commitment to empiricism about meaning
and reference.

This result should, naturally, be embarrassing in the extreme for the positivist, and
it may seem a bit surprising. I believe, however, that what has happened here can be
accounted for as follows. The positivist begins with the assumption that no belief is
warranted unless it can be empirically warranted. This assumption, by itself, begs no real

questions. It is, of course, perfectly consistent with empiricism in general, and positivism

 

‘2 Ibid., p. 38.

'3 It could be objected that the veriﬁcation principle should be understood as a principle of sentence
construction, so that what the principle covers is all constructible sentences, recursively deﬁned. Even if
we were to grant this construal of the principle, however, it would make no difference, for the following
reason: the principle is a criterion of sentence meaningfulness. Even if we regard it as a principle of
sentence construction, what it says is: “any future constructible sentence, S, will be meaningful if and only
if S is constructed so as to be empirically veriﬁable.” The principle cannot say this, even about potentially
constructible sentences, without either begging the question against the metaphysician or simply stipulating

23

is a brand of empiricism. The positivist then introduces the principle of veriﬁcation,
which is intended to adjudicate the meaningﬁrlness of sentences. The principle, however,
just embeds the original empiricistic assumption. Non-veriﬁable statements are then
identiﬁed as meaningless and banished to the land of wind and ghosts. This is just to
draw the net of empiricism over the entirety of discourse, and to forbid by ﬁat any
statement to be meaningful which does not admit of empirical veriﬁcation or tautological
self-evidence. Thus, the positivist has ruled out the possibility that any statement have,
as its object, any aspect of realty as such — any facts about reality as it is in itself — which
is just to say that the positivist claims to have knowledge of the structure of the entirety
of meaningful discourse in advance: that none of it refers to any objects or facts which
are not possible objects of sense, and thus that there can be no facts about reality as such,
since the meaning of a factual statement is simply to be equated with what would be the
case if it were true, and since there can be no meaningful statements about reality as such
(and since, of course, the ideal language would say everything that can be said about the

world).

As we have seen, however, the veriﬁcation principle commits the positivist to the
very thing which the principle was, in part, designed to criticize and extirpate: speculative
metaphysical claims about the mind-independent structure and nature of the world.

Could the positivistic distinction between statements within and without the language of

 

a deﬁnition of “meaningful” that rules out metaphysical assertions. I am grateful to Matt McKeon for
raising this objection.

24

science be understood in such a way as to avoid relying on the principle of veriﬁcation?
That is, could we offer some other reasons for saying that the question of the reality of
the total system of entities, or of the “ﬁt” of our conceptual framework with reality in
itself are impermissible?

Rudolph Carnap, in “Empiricism, Semantics, and Ontology,” proposes to do just
that. He had, by that time, relinquished his hope of a successful rational reconstruction of
our knowledge of the external world - of establishing our knowledge of the world on the
basis of a reduction of statements about physical objects to statements about sense data
and observation sentences — and had accepted the conclusion that the veriﬁcation
principle cannot be empirically established. He adopted a pragmatic approach to the
acceptance of linguistic frameworks in general (including, presumably, the positivistic
framework), and argued that positivism simply made the best sense of our total
experience. 14 In doing so, he denied the need for our entire epistemological enterprise to
be (metaphysically) grounded. Carnap’s distinction between questions that are “internal”
to our “thing language” and questions “external” to this framework as a whole presages
the neo-pragmatist’s denial, as developed later by Quine and Rorty.

This denial has since been formulated as a refusal to “answer the skeptic;” that is,
since it is, presumably, only the skeptic who even thinks to raise the questions of how and

whether our entire conceptual (or linguistic, or whatever) framework “ﬁts” or “hooks on

 

” Carnap’s “Empiricism, Semantics, and Ontology” appeared, in fact, nearly simultaneously with Quine’s
“Two Dogmas of Empiricism,” in which Quine, of course, also advocated a pragmatic approach to revision
of our overall belief structure, though not in terms of the acceptance of “linguistic frameworks,” which
Quine thought to be too reliant on Camap’s retention of the troubled analytic-synthetic distinction. It is in
this way that pragmatism follows (for committed empiricists) from positivism once the principle of
veriﬁcation is discarded: without the distinction between analytic and synthetic statements, no statement
can be thought to be immune from revision, and by what other criteria than simplicity, fecundity, breadth,
and so forth, are we to decide which revisions are appropriate? All that continued to divide Carnap and

25

to” or “reﬂects” reality in itself, the question of the possibility of our epistemology being
grounded can be understood as “the skeptical question,” and answers can be understood
as offering to “answer the skeptic.”lS Thus, it has been contended by neo-pragmatists
that we need not answer the skeptic. In other words, they have maintained not that the
answer to the skeptical question is that our epistemological enterprise cannot be securely
metaphysically grounded, but that we cannot ask or answer this question - that the
question itself is meaningless, or unanswerable, or misguided, or unintelligible.l6
Positivists and neo-pragmatists both held this position for a variety of reasons, some of
which I shall be examining in detail.17 As I shall argue below, however, any possible
reason we could have for refusing to answer the skeptic will require a comprehensive
epistemology, and thus will require metaphysical assumptions or assertions or

justiﬁcation.

 

Quine was Camap’s retention of the analytic-synthetic distinction, and Quine’s acceptance of Duhemian
holism.

'5 This terminological usage has the unfortunate terminological consequence that even those who believe
that our entire epistemological enterprise can be securely metaphysically grounded are skeptics, merely for
raising the question whether our epistemology can be so grounded, not for the reason that they think that
our beliefs about the world ultimately cannot be justiﬁed (which was, of course, the earlier usage of
“skeptic,” so that to say that one was a skeptic implied that one endorsed a negative answer to what is now
known as “the skeptical question”).

'6 Richard Rorty (1979) seems, incredibly, to maintain both, even though it would seem too obvious to need
stating that answering the skeptical question in the negative requires metaphysical commitments of a sort
that would render any refusal to answer the skeptic otiose. I will deal with Rorty on this subject in Chapter
Five.

'7 It may seem that the positivist’s refusal to answer the skeptic is inconsistent with her commitment to
grounding our epistemological enterprise. It must be remembered, however, that positivists believed that
our epistemological enterprise could be grounded without the need to appeal to a secure metaphysics (or
any metaphysics, for that matter), by translating observation statements into reports of sensory stimulations,
where physical objects are taken to be groups of these sensory stimulations. This would answer the
skeptical question by defusing it. Positivists did not take this construal of the term “physical object” to
involve or require any metaphysical assertions about the real (i.e. mind-independent) nature of objects. As
we have seen, however, the apparatus required for such a construal itself requires a synthetic a priori
judgment about the nature of utterance. Carnap’s distinction between “internal” and “external” questions
was an attempt to save the positivist’s linguistic reduction of metaphysics to logic by treating objects
generally as intrasystemic or intralinguisitic posits and treating all talk of objects as part of a non-cognitive
adoption of a linguistic framework.

26

Carnap proposes, among other things, to salve the consciences of those who wish
to make use of abstract entities in their philosophical and scientiﬁc researches, but who
would ordinarily shrink from doing so because of their nominalistic scruples, by assuring
them that the question of whether or not to accept the existence of numbers (for instance)
is not a matter of positing the existence of any entities at all. Rather, the question is,
Carnap maintains, a matter of whether or not to accept a system of rules that will govern
the use of statements containing number-terms, and this question is to be answered in
terms of the utility of this regulative system to our scientiﬁc endeavors. That is to say,
the question of whether or not to “accept the existence of numbers” is really only the
question of whether or not we should adopt (or maintain) “number talk” as part of our
apparatus for advancing our knowledge of the world. As part of this apparatus, however,
we should not assume that numbers have any separate existence, or that there is really
any meaningful sense in which numbers exist outside of the system of linguistic rules
which govern the use of “number talk.” Of course, this makes the issue of whether
numbers exist entirely trivial — numbers exist just in case we utilize a system of rules
which govern the use of statements containing number-terms — but this is precisely
Carnap’s point; the issue becomes trivial once we arrive at a proper understanding of the
nature of questions concerning the existence of some entity or class of entities.

Carnap argues that we cannot meaningfully ask whether the entities or system of
entities identiﬁed by (or named within, or speciﬁed by) our conceptual framework really

exist(s), as this is an “external question” — a question about our conceptual framework

27

and its objects as a whole — and external questions are “without cognitive content.”18

Carnap insists that the external question of “the reality of the thing world itself,”

is raised neither by the man in the street nor by scientists, but only by
philosophers.l9 Realists give an afﬁrmative answer, subjective idealists a
negative one, and the controversy goes on for centuries without ever being solved.
And it cannot be solved because it is framed in a wrong way. To be real in the
scientiﬁc sense20 means to be an element of the system; hence this concept cannot
be meaningfully applied to the system itself.21

Of the “external” question of whether or not there really are numbers — that is, outside of
our language of things —— Carnap says that “unless and until [the traditional
metaphysicians] supply a clear cognitive content” to this question, we “are justiﬁed in our
suspicion that their question is a pseudo-question, that is, one disguised in the form of a

theoretical question while in fact it is non-theoretical.”22 What basis does Carnap have

 

'3 Carnap 1964, p. 241.

'9 Which means, I suppose, that neither the person in the street nor the scientist is a philosopher. We could,
perhaps, take Carnap here to mean that the scientist in her capacity as a scientist is not a philosopher, but
then how do we make the required correction for the person in the street? That in her capacity as a person
in the street she is not a philosopher? Carnap here seems to echo the Wittgensteinian/Austinian conviction
that philosophy is a rareﬁed and highly abnormal activity. Plato also held this view, but only on empirical
grounds (most people do not make philosophy their life), not on the theoretical grounds that philosophy is a
particular sort of aberrant human activity or a disease that needs to be cured, as did Wittgenstein and
Austin. While most people do not make philosophy their life, and do not spend a great deal of their time
asking themselves philosophical questions, it is simply false that the person in the street and the scientist
are not philosophers in the sense that they do not ask themselves the questions that philosophers do, as can
be quite easily empirically conﬁrmed by simply asking the person in the street or the scientist some
philosophical questions and noting their replies. One ﬁnds that while the average person has few or no
opinions about modal logic, she will have quite ﬁrm opinions about freedom of the will, for instance, or
survival after death, or whether the existence of God can be established by demonstrative proof. Plato’s
objection was not to philosophy itself, as was Austin’s and Wittgenstein’s, but to people placing so much
more importance on things other than philosophy in their lives. It is not sufﬁciently frequently noted that
Wittgenstein and Austin were both anti-philosophers, and were anti-philosophical and anti-philosophy in
their dispositions.

2° We must remember that the question of the grounding of our entire epistemological enterprise is often
construed as the grounding of our scientific knowledge of the world — that is, our empirical knowledge, or
our knowledge of objects — however far that extends. As Descartes put it, he could not have “establish[ed]
anything ﬁrm and lasting in the sciences” if he had not ﬁrst secured their foundations.

2' Carnap 1964, p. 235.

2’ Ibid., p. 237.

28

for saying that questions about our conceptual framework and its objects as a whole can
have no cognitive content?

Carnap’s view seems to be that to ask such a question is to make a sort of
category mistake — to ask of the entire system of entities what we sometimes ask of
certain sorts of entities: namely, whether they exist. Carnap maintains that we cannot
meaningfully ask this of the entire system.23 Again, why not? Because to talk of any
entities whatever is simply to introduce (or to have introduced, or to utilize) a conceptual
or linguistic framework governed by logical rules and “empirical methods” (by which, I
think, Carnap means veriﬁcation).24

Firstly, it is not altogether clear what Carnap means in saying that external
questions and metaphysical assertions have no cognitive content. Presumably it would be
too strong a formulation of the claim to suppose that it means that such questions are
simply nonsensical in both their terminology and their syntax, for, if this is what Carnap
intends, he cuts himself off from the possibility of maintaining this position by himself
making clear both what is the intended meaning of such questions (namely, that they
mean to inquire as to the overall ﬁt between our language of things and the world itself),
and why such questions cannot be meaningfully formulated. To a ﬁrst approximation,
one may, perhaps, say that Carnap himself has already speciﬁed the cognitive content of
external questions for us. After all, in order to formulate the distinction between internal

and external questions, we must be clear about which sorts of questions are of which

 

23 Ibid. pp. 235 and 245.

2" Ibid., p. 234. I hope that I have here captured some of the ﬂavor of redundancy and circularity that
suffuses “Empiricism, Semantics, and Ontology.” The paper is not, of course, strictly circular in its
argumentation, but it seems so largely because Carnap takes the schema of positivism so much for granted
that he does not seem to think that he needs to argue for this schema, but merely to expound it. He argues
for it, in essence, only when he eventually makes it clear that he thinks that the positivistic framework is the

29

kind, and we must be able to employ criteria which would enable us to distinguish
questions of one type from questions of the other type. Thus, in order to identify a
particular question as an external question, we must be able to identify it as inquiring as
to the overall ﬁt between our language of things and the world itself. If we can see that
external questions have this content, then it cannot be that they have no cognitive content
at all.

This is perhaps too easy a victory. In saying that external questions and
metaphysical assertions have no cognitive content, Carnap does not, in fact, intend to say
that such questions and assertions are meaningless as such. Rather, his point seems to be
that there can be no way of determining how external questions (and metaphysical
assertions) are to be evaluated or understood, since there are no agreed-upon standards of
evidence for settling them. Thus, such questions and assertions are non-cognitive, which
is to say that nothing that can be said about them can be thought to be rationally
evaluated in terms of any available evidence to which all parties can point. Carnap

remarks that he

cannot think of any possible evidence that would be regarded as relevant to both
[the realist and the nominalist], and therefore, if actually found, would decide the
controversy or at least make one of the opposite theses more probable than the
other. Therefore I feel compelled to regard the external question as a pseudo-
question, until both parties to the controversy offer a common interpretation of the
question as a cognitive question; this would involve an indication of possible
evidence regarded as relevant by both sides.25

 

one we should adopt, “based upon the content of our experiences” (p. 236). He scarcely argues for this
claim, however, but then, how could he?
’5 Ibid., p. 246.

30

This, of course, means that the adoption of any linguistic framework governing the
construction of statements about objects is also a non—cognitive matter — it is not a
straightforward question of simply weighing the relevant evidence, not “a question
simply of yes or no,” but is, as Carnap puts it, “a matter of degree.”26 It is a non-
cognitive evaluation or judgment or decision as to the relative simplicity, fecundity,
breadth, and so on, of adopting one linguistic framework rather than another.

It is worth noting, to begin with, how signiﬁcant a concession to the
metaphysician Carnap has here been driven to make by the demotion of the veriﬁcation
principle. Presumably there would be nothing to prevent our adoption of a metaphysical
realist linguistic framework, if it were judged to be simplest, fecund, sufﬁciently broad,
and so forth, which is to say that such a framework would not be meaningless or
nonsensical. In fact, Carnap seems to be saying that a framework becomes meaningful
when it is adopted as a means of accounting for the available evidence. Carnap simply
wishes to promote the positivist framework as best ﬁtting the pragmatic bill. It is
difﬁcult, however, to know how to evaluate his advocacy of positivism, since this
advocacy is itself presumably non-cognitive.

Still, though it is non-cognitive, it is presumably not irrational. Carnap’s
standard for rationality involves the availability of mutually available, mutually relevant
evidence. Positivism, he thinks, best ﬁts the available, agreed-upon evidence. A realist
about numbers could, however, argue, that she and the nominalist would agree to the
following: if it should happen that any attempt to formulate a system of real numbers
should result in contradictions, that would make the actual (metaphysical) existence of

numbers less probable. The realist and the nominalist could both interpret this as

 

’6 Ibid., p. 247.

31

potential evidence of the failure of apparent a priori awareness (of the existence and
nature of numbers, in this case) to provide knowledge of the nature and structure of
reality as it is in itself. In that case, however, there is mutually agreed-upon and available
evidence to which both the realist and the nominalist can point, that would be relevant to
settling their dispute. Carnap would presumably reject the very idea of a priori
awareness of the nature and structure of mind-independent reality, but then it would seem
that his only possible reason for doing so is that he has made the (non-cognitive) choice
of a positivistic framework within which a priori awareness is impossible. Why make
this choice? Carnap pays too high a price for the preservation of the positivists’ rejection
of metaphysics. By viewing the adoption of any conceptual scheme as non-cognitive, he
withdraws whatever support there might have been for the claim that positivism should
be preferred to other schemes. If positivism is just one scheme among many, what basis
do we have for saying that there is no evidence that could settle metaphysical questions?
Carnap’s claim that we cannot say how and whether our language of things ‘ﬁts ” the
world as such is itself a claim about the nature of the relationship of language to the
world, in which case it is an external claim. Thus, the basis for drawing the distinction
between internal and external questions in the ﬁrst place depends upon a particular (and
foregone) answer to a particular external question — namely: What can be said about the
relationship between our language of things and the world as such?

The point is this: as soon as one embarks upon the task of distinguishing between
meaningful and meaningless statements, or between what is cognitive and non-cognitive,
rational and irrational, knowable and unknowable, one must develop criteria which will

provide a means of doing the distinguishing, which will require saying something

32

positive about the content of meaningless expressions. Not only will any attempt to
establish a boundary between what can be known and what cannot be known require
metaphysical assumptions or assertions, but any attempt to refuse to ground our
epistemological enterprise metaphysically will require us to establish such a boundary,
which will, in turn, require metaphysical assumptions or assertions. Thus, both the
rejection of metaphysics and the rejection of a metaphysical grounding for our
epistemological enterprise themselves require epistemological commitments whose
support can only be metaphysical.

Anytime we would hope to establish a boundary between what can and cannot be
said, or what can and cannot be meaningful, or what can and cannot be known, we are in
the business of locating an epistemological frontier beyond which human reason (or
reason simpliciter) cannot pass. In order for the placement of this frontier not to seem to
be simply arbitrary, we must offer reasons for its being just where it is,27 but the only
sorts of reasons that could possibly suffice would be metaphysical in nature, since they
will ineluctably involve claims that go beyond direct experience (and even all possible
experience). After all, they must be able to explain why reason cannot pass beyond this
limit, which will involve providing an account not only of why we can say or mean or
know what we can, but of why we cannot say or mean or know what we cannot.

Moreover, if the limit is to be an epistemological limit, and not merely a
pragmatic or psychological one, then providing the sort of explanation just called for will
require us to make claims about all possible experience, or all possible knowledge, or all

possible meaning, or all possible statements, the justiﬁcation of which claims (if they are

 

27 Quine would, of course, deny this claim, or at least refuse to be tempted by it (see his “Epistemology
Naturalized” (1985)). I will deal with Quine on this matter in Chapter Three.

33

justiﬁed at all) could only be a priori. As I shall argue in the chapters that follow,
we have no choice but to establish such boundaries, and, thus, no choice but to develop a
comprehensive epistemology: an epistemological account that attempts to establish a
boundary between what can and what cannot be known.

The foregoing was intended as offering a preliminary argument for this
conclusion, but we can see that the case can be generalized as follows: to deny either that
the skeptical question can be asked or that it can be answered, or even just to claim that
this question is nonsensical or unintelligible or beyond our ken, we must be able to offer
a reason why this is so, and the reason must be capable of showing that the skeptical
question is on the far side of a line separating what is askable (or answerable) and what is
not. Such a line would constitute an epistemological boundary whose location could only
be metaphysically justiﬁed.

Of course, to establish the truth of this claim would require, if I am right, a secure
metaphysics, and perhaps a transcendental argument — at any rate, something more than
what I am prepared to offer here. What I can and will offer, however, is what I take to be
a quite strong inductive case to the effect that this claim’s being true would go a long way
toward explaining the failure of both positivism and neo—pragmatism to successfully
escape from or eliminate metaphysics. I have already shown that Ayer and Carnap were
unable to do so, and I shall, in the following chapters, show that Quine, Putnam and

Rorty fare no better, and for similar reasons.

34

CHAPTER 3

QUINE’S REJECTION OF THE A PRIOR]

In Language, Truth and Logic, A. J. Ayer emphasizes the extent to which logical
positivism is indebted to Hume, and follows Hume, even though the central distinction of
logical positivism — between analytic and synthetic statements -— had in fact been codiﬁed
by Kant. Hume refers, of course, to relations of ideas and matters of fact, but this
distinction itself did not originate with Hume either, so why then do the positivists father
their project so much upon Hume? Hume was the most radical of the modern empiricists,
and in advertising their allegiance to Hume, positivists were, no doubt, signaling two
things: their determination to follow Russell and Moore in rejecting post-Kantian German
Idealism, and their rejection of metaphysics more generally, including Kantian
metaphysics. It is true that they owed their formulation of the distinction between
analytic and synthetic statements to Kant, but they believed that they could put the
distinction to a Humean use, and complete the Cartesian epistemological project on
empirical grounds alone, without recourse to speculative or critical metaphysics. They
needed only (1) to treat the phenomenal world as consisting of sense-data, (2) to treat
legitimate statements as either contingent observational reports (reducible to, or
translatable as, statements about sense-data) or as necessarily true axioms governing the
logic of statement-making, and (3) a means of distinguishing legitimate from illegitimate
statements.

Positivists had already begun working to secure themselves against the charge
that their means of distinguishing legitimate from illegitimate statements — the criterion

of veriﬁability — was incompatible with (or at least not apparently expressible within)

35

positivistic empiricism, even before Willard Van Orman Quine asserted that the
distinction between analytic and synthetic statements is untenable. As Quine himself
points out in “Two Dogmas of Empiricism,” however (where he makes his now famous
argument to this effect), the unveriﬁability of the veriﬁcation principle and the
untenability of the analytic-synthetic distinction come to much the same thing. The idea
that statements could be veriﬁed in isolation from other statements depends upon the
possibility of reducing observation statements to statements about sense-data (so that the
meaning of a statement consists in its veriﬁcation conditions), and on axiomatic
statements to be veriﬁed come what may.1 Since the reduction project is hopeless, no
statement can have its meaning in isolation from other statements, and thus no statement
can be veriﬁed in isolation, or veriﬁed come what may.2 Quine argues that the analytic-
synthetic distinction additionally suffers from the fact that the notion of analyticity is
empirically unspeciﬁable.

Thus, the Cartesian epistemological project, and modern rationalism along with it,
would seem to have failed. In fact, this was Quine’s view of things, as he makes plain in

his “Epistemology Naturalized”:

Carnap’s constructions, if carried successfully to completion, would have enabled us
to translate all sentences about the world into terms of sense data, or observation, plus
logic and set theory. But the mere fact that a sentence is couched in terms of
observation, logic, and set theory does not mean that it can be proved from
observation sentences by logic and set theory. The hopelessness of grounding
natural science upon immediate experience in a ﬁrmly logical way was
acknowledged. The Cartesian quest for certainty had been the remote motivation of
epistemology, both on its conceptual and its doctrinal side; but that quest was seen as
a lost cause.

 

'Quine 1980, p. 3742.
2 Ibid.
3 Quine 1985, p. 18.

36

Quine thus sets about, beginning with “Epistemology Naturalized,” sketching the
consequences for the Western epistemological project of the impossibility of translational
reduction and of rational reconstruction, and of the consequent collapse of positivism.
Quine notes that following the failure of translational reduction, former positivists and
others were tempted to despair, and began to consider excising epistemology from
philosophy as they (thought they) had excised metaphysics. Quine says that “Carnap and
the other logical positivists of the Vienna Circle had already pressed the term
‘metaphysics’ into pejorative use, and the term ‘epistemology’ was next.”4 Quine
maintains that this attitude is, however, unnecessarily pessimistic. We need not eliminate
epistemology as such, but merely eliminate comprehensive epistemology (and thus
foundational epistemology). Epistemology can no longer aspire to ground both itself and
our knowledge of the external world} but this does not mean that it is left with nothing to
do. Though rational reconstruction is (mostly) out, epistemologists can turn to the
evidences of psychology for data on the basis of which to formulate theories of belief-
formation.6 As we shall see, however, this result depends on the positivistic reduction
project being the last best hope for foundational epistemology, which in turn depends

upon both the positivists’ rejection of metaphysics and the Kantian psychologistic

 

4 Ibid, p. 23. In fact, Quine sees the collapse of positivism as the impetus behind the Philosophy-as-
Therapy movement: “Wittgenstein and his followers, mainly at Oxford, found a residual philosophical
vocation in therapy: in curing philosophers of the delusion that there were epistemological problems”
(Quine 1985, p. 23). I think he is right about this, as I suggested in the ﬁrst chapter, but Quine does not, at
least in “Epistemology Naturalized,” puzzle much over why Wittgenstein and Austin and so on did not
follow him in his naturalistic pragmatism. He seems to think that they just lapsed into unwarranted
despondency following the death of positivism, and never recovered. Quine calls this tendency
“epistemological nihilism” (Quine 1985, p. 27). He does, however, spend part of the rest of his career
defending his version of empiricism against other neo—pragmatists who offer alternative theories of science
and scientiﬁc progress, such as Kuhn and Feyerabend.

5 At least, epistemology can no longer hope to ground itself and our knowledge of the external world on
immediate experience and its contents.

37

reconstruction of the a priori.7 The support for rationalism collapses with the demise of
analyticity only if analyticity is the last possible home of necessary truth. Is it?

It is important to notice that the force of “Two Dogmas of Empiricism” against
rationalism (and thus the force of “Epistemology Naturalized,” inasmuch as the
arguments of the latter depend upon those of the former) depends entirely on the
maintenance within the epistemological tradition of representationalism, and on the
Kantian attempt to rescue rationalism and metaphysics by treating a priori knowledge as
knowledge of the structure of cognition. As I suggested in the ﬁrst chapter, this way of
understanding the a priori runs counter to the philosophical tradition, and we should, at
the very least, be aware of the fact that it is part of the legacy of representationalism.
Other rationalist epistemologies may be possible and defensible. That is to say, Quine’s
case in “Two Dogmas” does not tell against rationalism generally, but only against the
legacy of Cartesian rationalism as embodied in positivism. Nor does “Two Dogmas”
support empiricism against rationalism, except insofar as it contributes to making Quine’s
empiricism (and naturalism, and pragmatism) more perspicuous.

These last two claims should not seem so surprising; after all, Quine makes it
clear in the course of the article that his aim is to cleanse modern empiricism of its
dogmas. Rationalism is his target only insofar as he believes that positivistic empiricism
is still infected with traces of rationalism - the “unempirical dogma[s] of empiricists,
[their] metaphysical article[s] of faith.”8 Moreover, “Two Dogmas” does not offer

arguments in favor of empiricism (over against rationalism) because it assumes

 

‘5 Quine 1985, p. 24.
7 Discussed in Chapter One.
‘ Quine 1980, p. 37.

38

empiricism and labors to preserve it in the face of the collapse of positivism.9 Consider:
Quine does not argue that there can be no a priori awareness of any kind, but only that
the notion of analyticity cannot be speciﬁed. We could, as was noted above, reject
analyticity as does Quine, and still go looking for an alternative rationalist epistemology
that would escape the fate of representationalism. Does Quine offer any arguments
against rationalism generally, or against the possibility of an alternative foundational
epistemology?

Laurence BonJour, in his In Defense of Pure Reason, suggests that the answer to
this question is no. He canvasses Quine’s major works — those that are generally taken to
offer Quine’s most developed arguments against rationalism (minimally, against the
possibility of a priori awareness or knowledge or justiﬁcation), and he concludes that
these works either do not really offer any arguments against the possibility of a priori
awareness generally, or beg the question against the rationalist.lo Of “Two Dogmas,”
BonJ our says, “it reaches the conclusion that there is no a priori justiﬁcation only by
adopting a conception of epistemic rationality [namely, holism] that already tacitly
assumes this is so.”11 BonJ our makes similar remarks concerning Quine’s argument from
radical translation and Quine’s arguments in “Epistemology Naturalized,” ultimately
turning to more general considerations of his own concerning what he calls “radical

empiricism.” This is, I believe, unsatisfactory and unacceptable. Surely Quine, the

 

9 “Two Dogmas” argues for empiricism over against rationalism only to the extent that it rejects the
lingering rationalism of the positivists as untenable, and its problems as intractable. Of course, if Quine
can successfully defend his naturalism, or if this position turns out to be more tenable than every version of
rationalism, then obviously rationalism should be rejected. In order to establish this, however, we would
need at least an elaboration of Quine’s position, and either (1) some arguments against the tenability of
rationalism in general, or (2) some arguments to the effect that Quine’s position has fewer problems
(intractable or otherwise) than every version of rationalism. “Two Dogmas” only provides arguments
against one version of rationalism, and part of an elaboration of his position, in outline.

'0 BonJour 1999, pp. 62-88.

39

originator of neo-pragmatism and the arch-empiricist of the twentieth century, must have
had reasons for accepting empiricism and rejecting rationalism. Surely he must have
seen that the collapse of positivism need not have been the end of foundationalism once
and for all, and surely he must somewhere offer a more thoroughgoing rejection of
rationalism that explains his adherence to empiricism. Surely not all of his best
arguments to this effect could simply beg the question or fail truly to address the issue.
In fact I believe that we can extract the outlines of such a rejection of rationalism from
Quine’s work, and, though there are reasons to believe that a wholesale rejection of
rationalism is out of the question for Quine, this will provide us with two ways of

understanding Quine’s legacy in analytic philosophy.

Quine describes his naturalism as “boiling down to the claim that in our pursuit of
truth about the world we cannot do better than our traditional scientiﬁc procedure, the
hypothetico-deductive method.”l2 Another way of putting this would be to say that
philosophy, and certainly science, cannot hope for (and should not hope for) an
overarching epistemology that would comprehend, in its scope, the general outlines of
what grounds any and all knowledge available to any knowing beings of any description.
This, Quine suggests, seems like a nice dream, but it is no more than a dream. To think
that we could develop such a ﬁrst philosophy on the basis of which to explain the

possibility of science is to think that we could “ﬁrst fashion a self-sufﬁcient and infallible

 

” Ibid., p. 77.
'2 Quine 2004a, p. 281.

40

lore of sense data, innocent of reference to physical things, and then build our theory of
the external world somehow on that ﬁnished foundation.”l3 Presumably one of the
questions epistemology wants answered — what is, in fact, probably the central question
of all epistemology, and the question that provides it with its primary impetus — is how
we human beings can and do know anything. Rather than trying ﬁrst for an overarching
theory that would contain human knowledge (natural science), why not try simply for an
account of that knowledge itself, on its own terms? Or, rather, since we have tried, and
failed, at the overarching account, why not be content with what we have in the case of
ourselves as knowers and believers? That is, why not settle for psychology (as Quine
puts the matter in “Epistemology Naturalized”)?

Have we failed, and must we always fail, at the overarching account? Quine
clearly thinks so, and with respect to the Cartesian tradition of representationalism that I
have been tracing, he seems to be right. I have suggested, and will argue further in later
chapters that representationalism is untenable as a foundational epistemology. Descartes
set the standard for the modern epistemological project; he simply failed himself to meet
it. Of course, so has everyone in the tradition since Descartes ﬁrst published his
Meditations. Should we, then, conclude that the standard is unmeetable, revise our
standards, and rest content with more modest aspirations? I have been arguing that we
should not, and that we should not be tempted to do so on the basis of the failure of the
epistemological tradition to deliver a successfully secured foundational epistemology,
because that tradition has been systematically engaged in making the same mistake for

the past three hundred and ﬁfty years: namely, the mistake of promoting

 

'3 Ibid., p. 276.

41

representationalism. We need, I suggest, a better foundational epistemology. Why does
Quine think that no foundational epistemology is called for, or even possible?

Firstly, Quine clearly seems to think that positivism was the last best hope for not
merely Cartesian rationalism, but for rationalism (and foundationalism) simpliciter. He
seems, however, to have and to offer no reasons for thinking so. Even though
representationalism collapsed (again) with the defeat of positivism, why not think that we
just need a stronger rationalism?l4 Quine does not say. Of course, Quine is an
empiricist, but this does not answer the question. Why opt for empiricism over
rationalism? Quine is committed to holism, but, again, holism only follows from the
failure of positivism if our goal is the preservation of a satisfactory empiricist
epistemology. Holism explains the failure of sentences to be verifiable in isolation from
other sentences, according to the positivistic criterion of veriﬁcation, but does so only by
way of proposing an alternative empiricistic criterion of veriﬁcation. Holism does not,
strictly speaking, follow from the fact that sentences are not empirically veriﬁable one-
by-one; it can play a role in an account of this fact — which is, in fact, the use to which
Quine puts it — but it is not a logical consequence.ls We are, then, still left with the
question why Quine seems to think that foundationalism breathed its last with the demise
of Carnap’s reduction project.

Quine offers a number of tantalizing hints. Quine had, of course, profound

respect for Carnap, though it was Quine more than anyone else who was responsible for

 

‘4 As Eliot Sober points out, “it is worth noticing that Quine’s advice ‘Since Carnap’s foundationalism
failed, why not settle for psychology’ carries weight only to the degree that Camapian epistemology
exhausts the possibilities of epistemology” (Sober 1978, p. 166).

'5 See, especially, “Epistemology Naturalized.” Quine observes that “in giving up hope of [reductive]
translation, then, the empiricist is conceding that the empirical meanings of typical statements are
inaccessible and ineffable. How is this inaccessibility to be explained? Simply on the grounds that the

42

identifying the fatal ﬂaws in Carnap’s reduction program. Indeed, Quine seems to think
that it is not merely an understandable, but even a sensible, desire, near to his own heart
as an epistemologist, that natural knowledge could be grounded on sensory experience;
his “only reservation,” he says, “is that I am convinced, regretfully, that it cannot be
done?” He points out, in “Two Dogmas of Empiricism,” and again in “Epistemology

6‘

Naturalized,” that Carnap’s treatment of physical objects fell short of reduction not
merely through sketchiness, but in principle,” and that we “must despair of any such
reduction.”l7 Still, why think that such a reduction is the only hope for foundational
epistemology? “Transcendental argument, or what purports to be ﬁrst philosophy, tends
generally to take on rather [the] status of immanent epistemology insofar as I succeed in
making sense of it,” Quine remarks.18 “To account for the external world as a logical
construct of sense data—such, in Russell’s terms, was the program. It was Carnap, in his
Der logische Aufbau der Welt of 1928, who came nearest to executing it,” but, when his
efforts proved unsuccessful, “the Cartesian quest for certainty was seen as a lost
cause.” In other words, Quine does not appear even to consider the possibility of an
alternative rationalistic foundational epistemology that could meet Descartes’ standard.
Before we chalk this up to a lack of imagination, however, Quine offers two
independent lines of argument, both of which are intended to support the conclusion that
foundational epistemology is a lost cause. The ﬁrst line of argument is, I think, not

compelling, and is in fact based on a misunderstanding of the Cartesian epistemological

project as “immanent epistemology.” The second line of argument is stronger, and

 

experiential implications of a typical statement about bodies are too complex for ﬁnite axiomatization,
however lengthy? No; I have a different explanation” (Quine 1985, p. 21).

'6 Quine 2004b, p. 247.

‘7 Quine 1980, p. 48; Quine 1985, p. 20.

43

provides us with a clue to understanding both Quine’s legacy and the neo-pragmatist’s
position in rejecting metaphysics and foundational epistemology.

Firstly, Quine suggests that, though “we should like to be able to translate science
into logic and observation terms and set theory,” there is also something a bit absurd
about the whole idea in the ﬁrst place, since it seems just daffy to think that we could
derive science from a foundation of sense data.‘9 His objection seems to be that science
is not now, and never was (or even could have been), in need of a ﬁrst philosophy to set
out its tasks, deﬁne its scope and projects, and dictate its content in advance of actual
empirical investigation. In “Things and Their Place in Theories,” Quine remarks that “it
is within science itself, and not in some prior philosophy, that reality is to be identiﬁed
and described;” and in “Naturalism; or Living Within One’s Means,” Quine asserts that
“the naturalistic epistemologist dismisses [the] dream of prior sense-datum language,
arguing that the positing of physical things is itself our indispensable tool for organizing
and remembering what is otherwise, in James’ words, a ‘blooming, buzzing
confusion.’”20 Quine seems to spell this out most fully in “The Scope and Language of

Science,” where he remarks that

It would be unwarranted rationalism to suppose that we can stake out the business of
science in advance of pursuing science and arriving at a certain body of scientiﬁc
theory. Thus consider, for the sake of analogy, the smaller task of staking out the
business of chemistry. Having got on with chemistry, we can describe it ex post facto
as the study of the combining of atoms in molecules. But no such clean-cut
delineation of the business of chemistry was possible until that business was already
in large measure done.21

 

'3 Quine 2004b, p. 246.

'9 Quine 1985, p. 19.

2° Quine 2004b, p. 246; Quine 2004a p. 276.
2' Quine 1976d, pp. 234-5.

44

Surely, though, the ﬁrst philosophy (metaphysics) that supports a foundational
epistemology does not posit physical entities at all, or any entities of the sort with which
science is concerned. If it did, it would not be first philosophy. The goal and the work of
foundational epistemology is to discover “a foundation for natural science, ﬁrmer than
science itself,” as Quine himself observes in “Naturalism,” not to stake out the business
of science in advance or to posit its objects prior to its going about its job.22 There must
be some sort of confusion here.

The confusion seems to me to be this. Quine says, for instance, that “if the
epistemologist’s goal is validation of the grounds of empirical science, he defeats his
purpose by using psychology or other empirical science in the validation. However, such
scruples against circularity have little point once we have stopped dreaming of deducing
science from observations.”23 He likewise speaks of deduction or derivation in other
places in “Epistemology Naturalized,”24 as in the following: “It was sad for
epistemologists, Hume and others, to have to acquiesce in the impossibility of strictly
deriving the science of the external world from sensory evidence.”25 This is putting
things backwards. As we have just noted, and as Quine in some places appears to be
aware, foundational epistemology is “concerned with the foundations of science,” not
with the speciﬁc content of science as such. It may seem that foundational epistemology
is concerned with the speciﬁc content of science, however, if we think of foundational
epistemology as consisting in the reduction (or translation) of statements about physical

objects to constructions out of sense data. If such a reduction could be carried out,

 

’2 Quine 2004a, p. 281.

2’ Quine 1985, pp. 19.

2’ Ibid., p. 21: “The empiricist made one major concession when he despaired of deducing the truths of
nature from sensory evidence.”

45

certain observation statements (and thus scientiﬁc theories and laws) would, in fact, be
deducible from the constructions of sense-data, and thus certain of the contents and
objects of science would be likewise deducible. It might even seem that the totality of
science could then be deduced from “a self-sufﬁcient and infallible lore of sense data.”26
This, however, is a mistake. Sense data are still data; what would be secure and infallible
would be the reduction apparatus and the rules governing its use. These would ground
our scientiﬁc enterprise because they would guarantee that our statements about the
world are translatable as descriptions of the relations of its elements, but we would be in
exactly the same position as we ever were with respect to its elements: we become aware
of them only through experience. Quine’s objection to foundational epistemology on the
grounds that science needs no prior philosophy to tell it what to do seems to arise out of
his slipping into thinking that the provability of the statements of natural science from its
foundations implies its deducibility in advance from those foundations. This was never

either anticipated or hoped-for by foundational epistemologists.27

 

2’ Ibid., p. 19.

2" Quine 2004a, p. 276.

27 This might seem an odd thing to say, or it might even seem simply false. Wasn’t it in fact precisely
Descartes’ hope for foundationalism that our empirical knowledge be deducible from the epistemic
foundations, such that the entire ediﬁce would have the security of demonstrative proof and logical
deduction? Yes, indeed, but we must be clear about the sense in which our empirical knowledge is
deducible from the foundations. Not even Descartes could seriously have believed that all of our scientiﬁc
knowledge of the external world could simply be derived by entailment in advance of actual experience
from its epistemic foundations. Our scientiﬁc knowledge could, once it had been gathered, be seen
retroductively to be in principle deducible from the foundations, but this would not (indeed, could not)
make such knowledge available in advance, since it involves an irreducibly contingent element: namely,
our experience of the world from our perspective at a speciﬁc place and time. The foundations would of
course contain, in some sense, all empirical knowledge from the outset, but this does not mean that we
would be aware at the outset, simply in virtue of being familiar with the foundations, of all of their
implications and entailment relations. We are still temporal beings, and must work out these entailments in
the fullness of time. Moreover, our epistemic situation is that, short of some leap forward in our cognitive
abilities, we must appeal to experience in the process of working out the implications of any metaphysical
account of ultimate reality and our relation to it.

46

Quine’s second line of argument against foundational epistemology, however, is
more powerful, and more elusive. He says, in “Epistemology Naturalized,” that “two
cardinal tenets of empiricism [remain] unassailable One is that whatever evidence
there is for science is sensory evidence. The other, to which I shall recur, is that all
inculcation of meanings of words must rely ultimately on sensory evidence.” The ﬁrst of
these tenets seems to me indeed unassailable if it is to be understood as claiming that the
only evidence there is for speciﬁc scientiﬁc claims is sensory evidence, but false if it is to
be understood as claiming that the only evidence there is for the truth of scientiﬁc claims
qua scientiﬁc claims is sensory evidence. I take it, however, that Quine intends the ﬁrst
reading (although he would probably endorse both), so I will let this be. The second
tenet, however, I ﬁnd neither unassailable nor even very well supported anywhere in
Quine’s work. It is true that Quine goes to great lengths to work out a consistent and
comprehensive behavioristic theory of language acquisition and use, but he nowhere
appears to argue directly and explicitly against the possibility of a priori awareness,
knowledge, or justiﬁcation, which, if possible, could serve as a non-sensory source of
word meaning and conceptual content. The ﬁnal two sections of “Two Dogmas of
Empiricism” would appear to be the likely candidates for the locus of Quine’s arguments
to this effect, but as I noted earlier, Quine’s target in those two sections is the possibility
of a statement’s being true, come what may, in virtue of its being analytic, where this
notion is explicitly an artifact of positivism, by way of Kant. Moreover, while Quine
offers holism as an account of the meaning of the statements of science, he clearly

assumes an empirical attitude toward both meanings and language use, and does not

47

argue for it.28 What, then, does Quine have to say in support of his claim that “all
inculcation of meanings of words must rely ultimately on sensory evidence?”

In “Posits and Reality,” Quine asserts that “society teaches us our physicalistic
language by training us to associate various physicalistic sentences directly, in
multifarious ways, with irritations of our sensory surfaces, and by training us also to
associate various such sentences with one another.”29 Again, in “On Mental Entities,” he
says that “we get the system [of science], in its main lines, from our forebears. As
children learning the language, we get on to various simple terms and key phrases by
direct association with appropriate experiences.”30 In “Things and Their Place in
Theories,” he says that “our talk of external things, our very notion of things, is just a
conceptual apparatus that helps us to foresee and control the triggering of our sensory
receptors in the light of previous triggering of our sensory receptors. The triggering is,
ﬁrst and last, all we have to go on;” also, “there is no extrasensory perception.”3 1
Moreover, in “The Scope and Language of Science,” Quine remarks that “we cannot
signiﬁcantly question the reality of the external world, or deny that there is evidence of
external objects in the testimony of our senses; for to do so is simply to dissociate the
terms ‘reality’ and ‘evidence’ from the very applications which originally did most to

invest those terms with whatever intelligibility they may have for us.”32

 

28 Quine says, simply, that “once the theory of meaning is sharply separated hour the theory of reference, it
is a short step to recognizing as the primary business of the theory of meaning simply the synonymy of
linguistic forms and the analyticity of statements; meanings themselves, as obscure intermediary entities,
may well be abandoned” (Quine 1980, p. 22). Well, yes, if one already expects to be able to get a complete
and satisfactory theory of meaning on minimalist empiricist grounds. The question is not whether such
grounds are possible, available, or good, the question is whether Quine argues for an empiricist approach.
He does not.

2" Quine I976c, p. 253.

3" Quine I976b, p. 222.

3‘ Quine 2004b, p. 229.

’2 Quine I976d, p. 229.

48

These, however, are not arguments. Even a persuasive defense of behaviorism
about language acquisition and language use would offer at best a negative case against
the possibility of a priori awareness — by suggesting that language acquisition and use
can be accounted for entirely without the need to appeal to a priori awareness. Lacking a
complete account of language acquisition and use, however, does Quine offer any
positive arguments against the possibility of a priori awareness? He does, but we will
need to elaborate them a bit.

In “Things and Their Place in Theories,” Quine suggests that it is a “conﬁrsion” to
think that “we can stand aloof and recognize all the alternative ontologies as true in their
several ways We must speak from within a theory, albeit any of various.”33 One is
reminded of Carnap, and of his distinction between legitimate “internal” questions and
illegitimate “external” questions. Quine, however, rejects the basis upon which Carnap
grounds this distinction.34 What basis, then, does Quine have for making this claim?
What prevents us from having any awareness that transcends our theory of the world (or
our conceptual scheme, or whatever) as expressed in statements whose content is “ﬁrst
and last” determined by our sensory experience alone?

In “The Limits of Knowledge,” Quine considers the question whether there are
things that humankind can never know, whether there are any limits to human
knowledge. He begins by rephrasing the question. Surely, he notes, there are things we
will never know, such as how many automobiles will have entered our city between now
and midnight tonight, but this is not the issue. For one thing, this is something we could

know if we took the trouble to investigate (we could arrange a stake-out), and for another

 

3’ Quine 2004b, p. 246.
3’ As he makes plain in “Ontological Relativity,” (Quine 1999) pp. 54-55.

49

thing, surely the question is not whether there are things we could know but won’t, but
whether there are things we could not know at all. Secondly, Quine suggests that the
question cannot rightly be whether there are things we could not know at all, since things
are neither the sorts of things we know (propositionally, as it were), nor again quite to the
point of the question. Of course there are things — stars on the far side of our galaxy —
that we may never encounter, but again, we could in principle encounter these things
under suitable conditions. Thus, the question, according to Quine, is: are there questions
that we could never, in principle, answer?

His reply to this question, incredibly, is no. This is not, however, because he
believes that our intellect is so powerful as assuredly to be capable of encompassing
everything there is (or could possibly be) in its grasp, nor because he believes that we
know already everything there is to know, or know more or less what knowing
everything will be like. Quine does not believe that it makes any sense to talk about
reality beyond the bounds of our language for navigating it, so he thinks that there are no
questions that we in principle could not answer because any question that we in principle

could not answer would be a question that we in principle could not ask.

[T]here is a reluctance to assign meaning to strictly unanswerable questions.
Questions, let us remember, are in language. If a question could in principle
never be answered, then, one feels that language has gone wrong; language has parted
its moorings, and the question has no meaning.

Again, Quine says,

We have to work within some conceptual scheme or other; we can switch schemes,
but we cannot stand apart from all of them. It is meaningless, while working within a

 

3’ Quine 1976a, p. 67.

50

theory, to question the reality of its objects or the truth of its laws, unless in so doing
we are thinking of abandoning the theory and adopting another.36

The bounds of meaning, then, according to Quine, as well as the bounds of
knowledge and awareness, are the bounds of language. Have we ﬁnally got an argument
against the possibility of a priori awareness or knowledge — awareness or knowledge that
would transcend our conceptual scheme and certify its accuracy with respect to reality
itself? Not quite. Since Quine’s view of language acquisition and use is wholly
empiricist, then to say that the bounds of awareness, knowledge, meaning and
intelligibility are the bounds of language is just as much as to reiterate his commitment to
empiricism about meaning, which would simply beg the question against the foundational
epistemologist and the rationalist. To what else might Quine appeal in order to claim that
the bounds of meaning, intelligibility and awareness are the bounds of language (where
the meanings of words are given empirically)? The only way in which Quine can draw
the limit of our knowledge at the edge of our (empirical) use of language would be for
him to claim that there can be no possible conceptual scheme in which language
acquisition and use proceeds along rationalist, as well as empiricist, lines. In short, he
would have to claim that a priori awareness is not, and cannot be, a feature of any
conceptual scheme whatever. Since, however, such a claim would itself cover and
transcend what are, no doubt, an inﬁnite (or at least indeﬁnitely large) number of
conceptual schemes, this claim would itself be an example of the very sort of a priori

awareness the argument was meant to disallow.

 

3" Ibid., p. 65.

51

Have I, then, convicted Quine of the same mistake I accuse the positivists of
having made — that in attempting to eliminate metaphysics they will inevitably need to
appeal to a metaphysical postulate which supports limiting our knowledge so as to place
metaphysics forever off limits? Quine does not believe that it makes sense to speak of
reality as such, apart from theories, so for him to offer a transcendental argument against
the tenability of rationalism or the possibility of a priori knowledge would be worse than
self-vitiating, it would be inconsistent with his commitment to naturalism, and would in
fact render his naturalism incoherent. Quine asks of his own position, in “Naturalism; or,
Living Within One’s Means,” what his naturalism really rules out. He makes it clear that
his naturalism involves the “abandonment of the goal of a ﬁrst philosophy, prior to

science,” but does he rule out ﬁrst philosophy?37

Descartes’ dualism between mind and body is called metaphysics, but it could as well
be reckoned as science, however false. If I saw indirect explanatory beneﬁt in
positing sensibilia, possibilia, spirits, a Creator, I would joyfully accord them
scientiﬁc status too, on a par with such avowedly scientiﬁc posits as quarks and black
holes. What, then, have I banned under the name of prior philosophy?38

Quine does not answer the question directly, but in refusing to do so, he makes it clear
what he thinks his naturalism is about. It is, in the end, about the blurring of boundaries,
among other things, so he is hardly one to say in advance what is permissible and what is

impermissible.

 

37 Quine 2004a, p. 275.
38 Ibid., italics added.

52

Demarcation is not my purpose. My point in the characterizations of naturalism that I
quoted is just that the most we can reasonably seek in support of an inventory and
description of reality is testability of its observable consequences in the time-honored
hypothetico-deductive way—whereof more anon. Naturalism need not cast
aspersions on irresponsible metaphysics, however deserved, much less on soft
sciences or on the speculative reaches of the hard ones, except insofar as a firmer
basis is claimed for them than the experimental method itself.39

Now what justiﬁes this concluding caveat? We have seen that it would be inconsistent
for Quine to reject rationalism outright, but this is as much as to say that it would be
inconsistent for him to reject foundational epistemology — and the ﬁrst philosophy that
goes with it — as well. This is a signiﬁcant concession, because it means that Quine (and
neo-pragmatists generally) can have no basis for rejecting metaphysical claims out of
hand, and cannot simply insist that every theory must submit itself to trial under the
experimental method, since the question whether sensory awareness is our only
awareness is precisely what is at issue, and since the experimental method tacitly assumes
that sensory awareness is our only awareness. That is, to insist upon the experimental
method as the means of testing every theory is simply to beg the question against the
rationalist, since the experimental method is itself part of the apparatus of empiricist
epistemology. This also means that Quine and other neo-pragmatists cannot consistently
treat metaphysics or metaphysical questions or claims as meaningless, since they cannot
consistently offer a reason why the sort of awareness that would support such claims is
off limits. As I suggested in the last chapter, if we are going to claim that certain sorts of
statements are unintelligible or meaningless, we must at least be able to offer reasons
why this is so. My claim is that these reasons can only be metaphysical, but it is enough

for the moment to note that Quine can neither rule out metaphysics in the way he hopes,

 

’9 Ibid., p. 276, italics added.

53

nor offer any reasons why it must be ruled out. It should, I would hope, seem very
disappointing to be told that we cannot transcend the bounds of our language, whose
inputs are wholly sensory, but that we cannot say why this is so. In the end, Quine leaves
us with the insistence that natural science is the best we can do, while, in taking this
position, he cuts himself off from any possibility of providing a rationale for this
limitation. Quine offers, of course, reasons for thinking that we need not appeal to a
priori awareness in order to explain any of our knowledge, the success of science, or the
possibility of science, in the form of his attempts to explain all of this through his
naturalism and empiricism alone. I shall consider these attempts in later chapters, but
note that Quine must meet the rationalist on level ground; neo-pragmatists cannot scruple
to dismiss rationalism because of its commitment to metaphysical realism. We shall see,
in fact, that theory choice in epistemology must be driven by both a posteriori sensory

evidence and a priori principles — as ever.

54

CHAPTER 4

PUTNAM ON DIRECT REFERENCE AND CONCEPTUAL CONTENT

Scientiﬁc realism is a version of metaphysical or external realism that adheres to
representationalism as its epistemology of choice, but despairs of establishing the
accuracy of its epistemology on the basis of a secure metaphysics. Descartes believed, of
course, that he could deductively demonstrate the existence of God, and that certain facts
about God’s nature certiﬁed representationalism, both because God would not have
allowed us to be deceived about the true nature of the world, and because God is in a
position to bring it about that ideas represent objects —- that our minds reﬂect reality.
Scientiﬁc realists accept that the existence of God cannot be deductively demonstrated,
and are aware that representationalism cannot be grounded on the reduction of the
physical world to constructions of sense-data, but they maintain (contra Hume and contra
Quine) that representationalism is the best available explanation of the success of our
epistemological endeavor. After all, the scientiﬁc realist would say, something must
account for the fact that science not only enables us to explain the operation of physical
phenomena, but provides us with ever more powerful and comprehensive explanations of
these phenomena. This cannot simply be an accident.

Hilary Putnam used to be a scientiﬁc realist; though, as he points out, he was
never really committed to the traditional representationalism of Descartes. As he says in
“Why is a Philosopher?”, his “realism was not simply a revival of past ideas,” since “it
consisted in large part in an attack on conceptions which had been central to realism from

”I

the seventeenth century on. He was, after all, aware that traditional representationalism

 

' Putnam 1990a, p. 108.

55

required that the mental representations of each individual be produced by their
corresponding objects, and that our knowledge of these objects is thus a function of these
representations and their contents. Putnam has famously argued that the contents of an
individual’s mental representations cannot possibly sufﬁce to account for his or her
knowledge of certain objects, however, since the contents and even the existence of our
representations are often not the product of our contact with the relevant objects at all, but
are the product of our contact with others (and their reports of their experiences) who
have had certain kinds of contact with these objects.2 For most of us, our knowledge of
gold, and our ability to speak meaningfully about gold and its chemical properties are, for
instance, not a result of our own interaction with gold and its production in us of the
relevant (chemically expressible) concept. Our concept of gold, for the most part, goes as
far as “that bit of metal there, and other bits like it,” and our knowledge of its chemical
properties (and thus our ability to distinguish gold from sliver, or what have you — in
other words, to “pick gold out”) is ﬁlled in by those who possess the relevant expertise.3
The upshot of this would seem to be that our concept of gold (as opposed to the chemist’s
concept) neither originates in the way speciﬁed by Descartes, nor could possibly be
robust enough to represent gold as such (or anything speciﬁcally else, for that matter).
That we can refer to gold at all, Putnam suggests, is a result of the fact that there is some
speciﬁc thing (gold) to be identiﬁed in terms of its essential properties, and that our

intention to refer consists of an intention to refer to “bits of metal that are the same (in

 

2 Putnam 199015.
3 Ibid.

56

whatever sense of sameness is essentially relevant) as this bit here,” where we defer to
the appropriate experts to say what senses of sameness are essentially relevant.4

The fact that traditional representationalism required this modiﬁcation was not,
however, by itself enough to persuade Putnam to abandon representationalism altogether.
It convinced him that the traditional representationalist view could not be the whole story,
but not that representationalism was unsalvageable. Indeed, it would be difﬁcult to see
how this modiﬁcation could require us to reject representationalism altogether. For one
thing, so long as certain of our concepts represent their objects (in the traditional way),
and serve both to secure our scientiﬁc practice and to support the balance of our belief
structure, then the fact that some of our concepts contain promissory notes for their
content to be spelled out by the relevant experts (and by future science) does not defeat
representationalism, but merely calls for an adjustment. This was not Putnam’s position,
however. Again, as a scientiﬁc realist, he had abandoned the view that there could be
any privileged representations or a priori certiﬁcation of representationalism. Instead, he
maintained that short of accepting that there is a ﬁxed realm of mind-independent entities
with determinate properties, whose natures we can discover by means of scientiﬁc
investigation, there is no way to account for the success of science and our scientiﬁc
practices — that is, of our epistemic enterprise. In “The Meaning of ‘Meaning’? Putnam

observes that

It is beyond question that scientists use terms as if the associated criteria were not
necessary and sufﬁcient conditions, but rather approximately correct characterizations
of some world of theory-independent entities, and that they talk as if later theories in
a mature science were, in general, better descriptions of the same entities that earlier
theories referred to. In my opinion the hypothesis that this is right is the only

 

‘ Ibid.

57

hypothesis that can account for the communicability of scientiﬁc results, the closure
of acceptable scientiﬁc theories under ﬁrst order logic, and many other features of
scientiﬁc method.5

Scientiﬁc realism is often criticized on the grounds that the success of science
does not stand in need of any explanation, or that any explanation we could give would
either be circular (by taking for granted the criteria of success already implicit in science
itself) or otiose (since the explanation would neither improve our scientiﬁc practice nor
our understanding of this practice).6 This last criticism was, in fact, apparently what
convinced Putnam to abandon scientiﬁc realism and representationalism.
Representationalism began to seem to him not to constitute a serious explanation of

anything:

How does the familiar explanation of what happens when I ‘see something red’ go?
The light strikes the object (say, a sweater), and is reﬂected to my eye. There is an
image on the retina (Berkeley knew about images on the retina, and so did Descartes,
even if the wave aspect of light was not well understood until much later). There are
resultant nerve impulses (Descartes knew there was some kind of transmission along
the nerves, even if he was wrong about its nature—and it is not clear we know its
nature either, since there is again debate about the signiﬁcance of chemical, as
opposed to electrical, transmissions from neuron to neuron). There are events in the
brain, some of which we understand thanks to the work of Hubel and Wiesel, David
Marr, and others. And then—this is the mysterious part—there is somehow a ‘sense
datum’ or a ‘raw feel’. This is an explanation?

An ‘explanation’ that involves connections of a kind we do not understand at all
and concerning which we have not even the sketch of a theory is an explanation
through something more obscure than the phenomenon to be explained.7

In fact, Putnam offers at least three related lines of argument against representationalism

as an epistemology, and, like Quine, apparently concludes that the failure of

 

5 Putnam 1975, p. 237.

6 See, for instance, Rorty 1979, pp. 282-284 and 339-341, Williams 1980, “Coherence, Justiﬁcation, and
Truth,” pp. 271-272, and van Fraassen 1980, pp. 39-40.
7 Putnam 1987, pp. 7-8.

58

representationalism is the failure of rationalism once and for all. I shall argue that
Putnam’s criticisms of representationalism are unsuccessful (though, again, I believe that
representationalism fails for other reasons, to be discussed in the next chapter), and that,
like Quine, Putnam does not address himself to rationalism generally in a way that would
motivate his empiricism and his adoption of pragmatism over every version of
rationalism. His arguments against rationalism in general only tell, in fact, against
representationalism speciﬁcally, and we shall see that Putnam’s criticisms of
representationalism fail for the same reason that they fail truly to engage rationalism —
namely, they all assume empiricism. We shall also see that, as was the case with Quine,
Putnam’s pragmatism prevents him from being able to rule out either rationalism or

metaphysics.

The First Argument

What is, perhaps, Putnam’s most famous argument against representationalism is
his “Brains in a vat” argument. What seems often to go unnoticed, or at least
unappreciated (or underappreciated) about this argument is that it is aimed only at
representationalism as a rationalist epistemology, and is effective only against
representationalism (if it is effective at all, which it is not, as presented by Putnam).
Putnam as much as acknowledges this in Reason, Truth and History when he pauses to
“consider why it seems so strange that such an argument can be given (at least to

philosophers who subscribe to a ‘copy’ conception of truth)?8

This is signiﬁcant because
even if Putnam’s argument were successful in defeating representationalism, it would not

succeed in defeating rationalism simpliciter, so long as there are other possible versions

59

of rationalism. As we have seen, the possibility of other versions of rationalism is an
open question, inasmuch as none of the arguments we have considered so far that have
been adduced to rule it out can offer a non-metaphysical rationale for doing so, and
inasmuch as the reason for this would seem to be that only a metaphysical justiﬁcation
could be offered for ruling out rationalism (which would vitiate the justiﬁcation).

Putnam’s argument, however, is intended to show that even if we are brains in a
vat, we can neither know nor even so much as truthfully say that we are brains in a vat,
since there is no way for our brain-in-a-vat word “vat” to “hook onto” the vat that
contains us. If this argument were successful, Putnam would have succeeded in showing
that there is no way to meaningfully formulate the skeptical hypothesis that serves as the
basis of the development of representationalism as an epistemological position (which
hypothesis representationalism is designed to defeat). Putnam’s argument is relatively
straightforward, and it goes like this: suppose that we and all sentient creatures are in fact
no more than brains ﬂoating in vat of nutrient solution and tended by a mad scientist who
supplies us with stimulations that give us the (ex hypothesi) mistaken impression that we
have bodies, inhabit cities, and so on. Now this situation seems physically possible, but
if it is not logically possible then it cannot, as a matter of fact, be physically possible after
all, and this is the question that concerns us at present: is this situation logically possible
(meaning, I presume, that such a situation can be meaningfully and truthfully conceived
of and stated by us to be actual)? Putnam’s answer is no, it is not.

Putnam claims not only that if we are brains in a vat we cannot possibly know or
(meaningfully, truthfully) say that we are, but that since this is so, it cannot possibly be

true that we are brains in a vat, since we cannot even meaningfully formulate the

 

8 Putnam 1981, p. 8.

60

hypothesis that we are. This is because, if we are brains in a vat, then when we use the
term ‘brain’ or the term ‘vat’ or the term ‘in’, we are referring not to the brains that we
really are and the vat we are really in, but to the brain-image or vat-image or containment
concept that we have in our vat-world generated by the mad scientist. This, in turn, is
because “there is no connection” between the word ‘brain’ or the word ‘vat’ “as used by
these brains” and actual brains and actual vats.9 Thus, “if we are brains in a vat, then the
sentence ‘We are brains in a vat’ says something false (if it says anything)? since
obviously it is false in the vat-world — the only world, according to Putnam, to which the
brains can meaningfully refer.lo “In short, if we are brains in a vat, then ‘We are brains
in a vat’ is false. So it is (necessarily) false.”ll
To begin with, this last claim certainly does not follow. Putnam’s point is that if,
in fact, we are brains in a vat, then there is no way for us to know, represent, or express
the fact that we are, and any such expression will be false (in the vat world) and
otherwise meaningless. But what justiﬁes the claim that it is otherwise meaningless?12
What gives Putnam license to claim that “We are brains in a vat” is necessarily false?
The answer is: nothing; if there are facts we cannot, even in principle, express,'3 or
thoughts that we cannot even in principle know to be true, our inability to express them

or know that they are true does not mean that they are not facts (or true) all the same, so

Putnam is not entitled to the claim that my thought that I am a brain in a vat is necessarily

 

9 Ibid., p. 12.

'0 Ibid., p. 15.

” Ibid.

'2 If it is, in fact, meaningless, then Putnam should not even be able to meaningfully formulate its refutation
in the manner he has, since the claim expressed in the antecedent of his conditional — “If we are brains in a
vat” — would be equally meaningless. (See David 1991, p. 892.)

'3 Whether there are such facts, is, of course, at stake in this debate.

61

false. '4 To say that there can be no facts we cannot (even in principle) express (or that
the very idea of such facts makes no sense), because they are not expressible by us,
seems to suggest that the limits of intelligibility for us are the limits of reality, as Quine
claims. What gives Putnam license to say that?ls

In fact, it turns out that Putnam’s “proof” that we are not brains in a vat at best
proves that we cannot think or conceive or imagine or express that we are brains in a vat.
In order to get from the latter conclusion to the former, we would need an additional
argument, but Putnam does not offer us one. Crispin Wright makes a similar same point
in his “On Putnam’s Proof that We Are Not Brains in a Vat.” Wright observes that
Putnam’s proof does not succeed in ridding us of “the real specter to be exorcised” from
our nightmares of radical error, because the real specter “is that of a thought whose truth
would make a mockery of humankind and its place in nature” — a thought that Putnam’s
proof cannot assure us is false.16 Putnam, in a reply to Wright, does, however, suggest
that “if we are supposed to be able to think that we are in some state of affairs of a very
general kind,” such that we cannot conceive of its exact nature, “we have to be able to
say what type of situation that is.”17 Putnam insists that the very notion that there even
are unexpressible facts must be “cashed out” within some theory of reference. This

seems to miss the point. The idea is that there may be facts about the nature and structure

 

'4 To be sure, one might want to know how we could know that there are such facts at all, but we need not
be troubled by this question here. The point is simply that if such a possibility is conceivable, and I cannot
see that it isn’t, then Putnam is wrong to say we cannot possibly be brains in a vat. (See also, Forbes 1995,
p . 219-222, and Wright 1994, pp. 237-240.)

As Bryan Magee observes, “The only plausible possibility of a reality completely corresponding to our
conceptions of it rests on the possibility that reality itself could be mind-like, or could be created by a mind,
or by minds. Paradoxically, it is those philosophers who most conﬁdently do not believe this, that is to say
those in the empiricist tradition, who are most tightly wedded to an epistemological criterion of reality.
That is one of the incoherences of their position” (Magee 1999, p. 357).

'6 Wright 1994, pp. 239-240. See Forbes 1995, pp. 219-222, for another argument to the same effect.
‘7 Putnam 1994, p. 286.

62

of reality that we cannot (now, within our conceptual scheme as it stands, or perhaps
ever) express and maybe cannot even think. Putnam remarks that “the notion of how
things are makes no sense apart from the way in which we interact with those things.”18
This seems to imply the highly implausible thesis that we cannot conceive of the
possibility of there being facts about the world we cannot comprehend. The question is:
how could such a thesis be supported? It could perhaps be supported by a general theory
about the relationship between language (or concepts) and its (their) referents, to the
effect that, as Putnam puts it, “reference is transactional.”19 It is very hard to see,
however, how such a theory about the relationship between language and reality could be
defended, short of offering a transcendental argument. Indeed, Putnam, in defending
internal realism as a sort of Kantian transcendental idealism without the transcendence,20
seems to need just the sort of argument he has barred himself from having.

What about the claim that the brains in a vat cannot (meaningfully, truthfully)
think of themselves as brains in a vat, because their words do not refer to the right
objects? Putnam says that the brain-in-a-vat’s word ‘brain’ does not, and cannot, refer to
actual (non-vat-world) brains because “there is no connection” between the brain-in-a-
vat’s word ‘brain’ and actual brains. Says who? This would be the case if, and only if,
the only kind of permissible connection is an empirical one.” But Putnam admits that he

has built this empiricist assumption into the argument: his procedure, he notes, involves

“assumptions which might be described as ‘empirical’ (e. g. the assumption that the mind

 

'8 Ibid., p. 288.

'9 Ibid. See footnote 45 below for further discussion of this point.

2° See Putnam 1983c, p. 85, where he says that his internal realism is Kant’s “empirical realism” without
Kant’s corresponding metaphysical realism.

2‘ Again, one may ask what other sort of connection is possible or actual, but we need not be troubled by
this question here. That is a question for a rationalist epistemology to answer.

63

has no access to external things or properties apart from that provided by the senses)?’22

He claims that in developing this argument he has been reasoning a priori, but by this he
can only mean that he has been reasoning by means of logic alone.

Who is this argument intended for? Putnam suggests that it is intended for
philosophers who adhere to representationalism, but it will convince few rationalists,
given its “empirical assumptions.”23 It can be intended only for scientiﬁc realists, who
Putnam presumably hopes will follow him in his rejection of representationalism.
Putnam’s point seems to be that representationalism cannot be established empirically,
and that the conditions of the possibility of representationalism are empirically self-
refuting. Again, this should little impress rationalists. It is not clear, however, why it
should convince scientiﬁc realists either.24 Scientiﬁc realism never had any aspirations
of establishing representationalism within the bounds of empiricism — this was

positivism’s goal; rather, scientiﬁc realism appeals to representationalism as an inference

 

22 Putnam 1981, p. 16. One is reminded of Quine’s insistence that “there is no extrasensory perception.”

2’ Putnam (1994), in his reply to Crispin Wright’s (1994) “On Putnam’s Proof that We Are Not Brains in a
Vat,” suggests that his proof (in the ﬁrst chapter of Reason Truth and History) is aimed at the “internal
skeptic” who aims to “convince us, on the basis of assumptions we ourselves hold, that all or a large part of
our claims about the empirical world cannot amount to knowledge,” not at the “inﬁnitely regressive
skeptic” who “may raise a skeptical doubt only so that, no matter what premises one may rely on in
answering the doubt, he or she can respond, ‘and how do you know that?”’ (p. 284). This is as much as to
acknowledge that Putnam’s proof is not directed at rationalism simpliciter.

Putnam observes that the inﬁnitely regressive skeptic could argue that “transcendent reference is
possible,” but he claims that inﬁnitely regressive (i.e. Cartesian) skepticism cannot be answered or refuted
(p. 285). It could be answered or refuted, however - the inﬁnite regress of justiﬁcation could come to an
end — if transcendent reference (viz., a priori awareness) could be shown to be not only possible but self-
certifying. I shall discuss this possibility further in Chapters Six, Seven and Eight.

2" There is a prodigious literature on the question whether Putnam’s argument even succeeds in refuting the
sort of skeptical possibility that would be necessary for the formulation of scientiﬁc realism, but inasmuch
as I have no particular stake in the success of scientiﬁc realism and since I believe that Putnam’s anti-
skeptical argument does not succeed in refuting representationalism anyway, I shall not attempt to
disentangle the many and various threads of this debate here. For some representative samples of the
arguments for and against Putnam’s anti-skeptical proof, see Tymoczko 1989, Warﬁeld 1998, and Haukioja
2002 (for), and Wright 1994, Forbes 1995, and David 1991 (against).

64

to the best explanation (of the success of science).25 Moreover, the claim that the
conditions of the possibility of representationalism are empirically self-refuting involves
the possibility of our knowing or saying (that we can empirically verify) that
representationalism is true, which, again, scientiﬁc realists do not assert.26 Let us move
on to an argument that is supposed to defeat representationalism at its central point: the

claim that our ideas represent the objects of the world.

The Second Argument

Putnam’s second argument against representationalism is, in fact, a generalized
version of his “Brains in a Vat” argument. Putnam invites us to consider the following
hypothetical situation: “An ant is crawling on a patch of sand. As it crawls, it traces a
line in the sand. By pure chance the line that it traces curves and recrosses itself in such a
way that it ends up looking like a recognizable caricature of Winston Churchill.”27 He
then asks, “Has the ant traced a picture of Winston Churchill, a picture that depicts
Churchill?”28 Putnam thinks that the answer is clearly no, and his point is that pictures
(images, signs, symbols, representations) of things do not naturally, inherently, or
necessarily represent the things they picture. How then do they picture them? By having

a particular role and use in a particular context, which itself arose in the particular history

 

25 It is worth pointing out, however, that Putnam seems to be right to say that the scientiﬁc realist’s
semantic extemalism can never be “worked out,” because this theory of the causal relationship between
concepts and their referents is not a part of our overall scientiﬁc explanatory framework, but is a
metaphysical account of the success of that framework that the scientiﬁc realist presumably holds on
inductive grounds. Any scientiﬁc realist who genuinely believes that the causal theory could indeed be
worked out is thus mistaken, since there is no way to wotk out, from within the scientiﬁc framework, the
metaphysical account of the success of that very framework. On this point Putnam is entirely correct. I
would add that scientiﬁc realists who are genuinely committed to realist metaphysics should adopt
foundationalist metaphysical realism at once.

26 This is because, according to Putnam’s argument, the truth conditions of “We are brains in a vat” are its
conditions of empirical veriﬁcation. See Clark Glymour (1982), “Conceptual Scheming.”

65

of the development of a particular linguistic or symbolic scheme, as spoken or utilized by
a particular group of speakers. Signs, symbols, representations, images and so on have
the meaning they do only in that context. They could be put to meaningful use by other
speakers of other languages in other historical situations, but this would not be the same
meaningful use (even if these other speakers just happened to use the sign or symbol or
whatever in question to refer to the ‘same’ entity or entities as the original speakers did).
The idea that any sign or symbol could inherently or necessarily refer to a speciﬁc object
is, Putnam says, a holdover of “magical thinking? “some primitive people believe that
some representations (in particular, names) have a necessary connection with their
bearers; that to know the ‘true name’ of someone or something gives one power over

it.”29 Moreover, he calls any theory of reference30 that adheres to the notion of inherent

or necessary reference a “magical theory of reference.”3 1

Now I should like ﬁrst of all to observe that this bit of Putnam’s argument is (as I
should hope would be obvious) nothing more than insulting rhetorical grandstanding, the
force of which would seem to be as follows. It would be “primitive” to believe that signs
or symbols can have a necessary (“magical”) connection with their referents. You’re not
primitive, are you? You don’t believe in magic, do you? Well, then, I’m sure you’ll
agree that representationalism is wrong!

The move here is clear: simply associate the view one is espousing with

sophisticated reasoning and the view one is criticizing with superstition. Nobody (no

 

2’ Putnam 1981,p. 1.

2‘ Ibid. ,

29 Ibid., p. 3.

3° For Putnam, the theory of reference is a successor theory to foundational epistemology. An adequate
theory of reference would explain (1) the nature of those of our mental contents that are responsible for our
contact with the objects of the world, and (2) how, through the mechanism of language, our concepts
enable us to refer to (or “pick out”) things in the world.

66

philosopher, anyway) wants to be unsophisticated. Of course, this does not provide one
with a reason to reject representationalism, but with a motive to reject it: if you defend
representationalism, you will seem to your fellow philosophers to be primitive. Perhaps
no one can even speak with you (or perhaps no one is really addressing you), since you
do not even occupy the same realm of legitimate theory as the rest of us.

Fortunately, two can play at this game. Some primitive people believe that some
beliefs are true, or are made true, just because they and everyone else in their tribe adhere
to them, focus their energy and attention on them, or meditate on them. A theory of truth
or of reference that maintains that what is true is what is warranted assertable within a
particular community of speakers, or that what there is just depends on what we say there
is in our best available theories, is just a holdover of this sort of “because we say so”
thinking. Such “collective energy” theories of truth and reference suffer from the
problem that they don’t really explain what they set out to explain. Hilary Putnam, for
instance, says that “it is trivial to say what any word refers to within the language the
word belongs to, by using the word itself. What does ‘rabbit’ refer to? Why, to rabbits,

of course!”32

(As though a speech act, series or set of speech acts, or disposition to
certain speech acts could make it true that ‘rabbit’ refers to rabbits!) This is an
explanation?

Of course, pragmatists and meaning holists have produced (to all appearances)
sophisticated theories of reference and truth that are meant to clarify just how meaning

can be a function (or product, at any rate) of context and linguistic behavior.

Wittgenstein, for instance, offers us the related concepts of language-game and form of

 

3' Putnam 1981, p. 3.
’2 1616., p. 52.

67

life. I could complain, perhaps, that such theorists have a lot of work left to do to explain
just how the “rules” that govern warranted assertability or stipulation or grammar are
introduced, how these rules ﬁx themselves within the linguistic or conceptual framework
of a given community, and how they function between speakers, but this would really be
beside the point. It would be pure historiographical speculation on my part to claim that
such theories of reference and truth are holdovers of primitive “collective energy”
theories (as it would be pure speculation on Putnam’s part to claim that
representationalism is a holdover of “magical theories of reference”), and in any event it
wouldn’t matter even if this were true. What matters is what current pragmatists and
holists can and do make of their theory, whatever its origin, just as what matters about
representationalism is not where it came from but where it is going. To think otherwise
is to commit the genetic fallacy.

Putnam has, or appears to have, a point, however, about the referential use of
images, signs, symbols, and representations. Unless (I) all of our representations refer
necessarily to their objects, or (2) some of our representations refer necessarily to their
objects, and ﬁx the usage of the other representations in our scheme, then there is no way
to say what refers to what. What seems to us to be a picture of someone or something
does not just “automatically” refer to that someone or something. For that matter, a
‘picture’ may not really even be a picture unless it functions as a symbol in a
representational scheme, as the thought experiment involving the ant seems to show.
Putnam considers further counterfactual examples that would suggest that an individual
could have all of the requisite representations and still not refer to the same things we

refer to by way of these same representations, or not succeed in referring to anything at

68

all. Thus, it is not just having the representations (even the “requisite ones”) that does the
trick, but having a context of use and criteria of use — the right interactions with the right
objects under the right circumstances.

Has Putnam simply taken the notion of a representation too literally, as though
representations are really tiny pictures inside our heads? Of course no picture of a tree is
a picture of a tree unless there is a set of interpretative practices or procedures that makes
the picture one of a tree and not a one of a groundhog, but while this holds for pictures, it
need not hold for every kind of representation. Some representations could be
necessarily linked to their objects through a similarity of formal features that is non-
pictorial. Moreover, some representations could be linked to their objects simply in
virtue of the fact that those representations are causally linked to those objects: the
objects in question gave rise to the representations in question. In fact, there are
philosophers who have proposed that our mental representational schemes function
holistically, and that representations are assigned a conceptual role in our scheme
according to how they were caused and with what other objects or experiences they are
causally (or emotionally, or what have you) associated.33 Putnam acknowledges this, and
observes that: “if there are mental representations that necessarily refer (to external
things) they must be of the nature of concepts and not of the nature of images.”34

His reply, however, is that the notion of a concept fares no better than the notion
of a representation. We could just as easily imagine, he says, someone having the

requisite concept and being unable to refer as we can imagine someone having the

requisite representation and being unable to refer. Having a concept, Putnam insists,

 

33 See, for instance, Ned Block 1995.
3‘ Putnam 1981, p. 17.

69

involves having the “ability to use sentences in situationally appropriate ways,” and he
then proceeds to offer examples of the exercise of this ability.35 The problem is that all of
his examples — pointing to a tree, for instance — involve the use of concepts whose source
can only be empirical. In fact, his discussion of the nature and use of concepts is meant
to support his claim that “one cannot refer to certain kinds of things, e. g. trees, if one has
no causal interaction at all with them,” but this is clearly empiricist.“ Putnam does not
even consider the possibility that some concepts necessarily refer a priori, and instead
takes empirical reference as the paradigm case of “situationally appropriate use” of a
concept. A justiﬁcation that assumes empiricism can provide no support for the
empiricist assertion it was meant to defend. Ruling out a priori concepts that necessarily
refer would be consistent with scientiﬁc realism, but Putnam is not here defending
scientiﬁc realism, he is abandoning it. Moreover, in failing even to consider the
possibility of privileged representations, or a non-representationalist rationalism as an
alternative, Putnam’s argument neither defeats representationalism nor defeats

rationalism generally, and for the same reason: it assumes empiricism.

The Third Argument

In “Why There Isn’t A Ready-Made World,” Putnam in fact offers no general
reasons for thinking that there isn’t a ready-made world (a mind-independent world of
objects possessing determinate properties), but rather argues that the idea that there is a
ready-made world is not, contrary to the hopes of scientiﬁc realists, consistent with a

commitment to materialism. As was the case with the ﬁrst two arguments, then,

 

3’ Ibid., p. 19.
36 Ibid., p. 16.

70

Putnam’s third argument is aimed primarily, if not exclusively, at scientiﬁc realism. Here
we at last get a glimpse of Putnam’s reasons for apparently ignoring rationalism
altogether. “The reason that I am going to focus my attack on materialism,” he says,
is that materialism is the only metaphysical picture that has contemporary ‘clout’.
Metaphysics, or the enterprise of describing the “furniture of the world”, the
“things in themselves” apart from our conceptual imposition, has been rejected by
many analytic philosophers and by all the leading brands of continental
philosophy. Today, apart from relics, it is virtually only materialists (or

“physicalists”, as they like to call themselves) who continue the traditional
enterprise.

I should again like to begin by pointing out that Putnam’s choice of words here
amounts to nothing more than insulting rhetorical grandstanding. Perhaps in this case it
would seem to depend rather on one’s point of view: is Putnam being derisive by calling
traditional metaphysicians “relics,” or is he is merely being descriptive? After all,
rationalists nowadays are few and far between, and now perhaps appear as remnants of
the last vestiges of early analytic philosophy in the tradition of Frege and Russell.

This is simply no longer the case (and of course I dispute the assumption that the
only possible remaining versions of rationalism are versions of representationalism).

J arrold J. Katz, whose version of rationalism I consider in Chapter Six, went over to
rationalism aﬁer having spent most of his career and indeed most of his life working on
problems in the philosophy of language in the (empiricist) analytic tradition. Laurence
BonJour, whose version of rationalism I consider in Chapter Seven, went over to
rationalism after working to defend a coherentist account of empirical knowledge.38

Moreover, to call someone a relic is not merely to describe them, but to insult

them. The force of the rhetoric here would seem to be as follows. Traditional

 

3’ Putnam 1982, p. 144.

71

metaphysicians are behind the times, out of date or out of fashion, near curiosities in this
day and age, and out-of-step with the movement of philosophical thought and history.
Nobody wants to be behind the times. Of course, this does not provide one with a reason
to reject metaphysics, but with a motive to reject it: if you defend traditional metaphysics,
you will seem to your fellow philosophers to be a relic. Perhaps no one can even speak
with you (or perhaps no one is really even addressing you), since you do not even occupy
the same realm of legitimate theory as the rest of us.

Besides the fact that Putnam’s language is loaded, however, the attitude toward
traditional metaphysics and rationalism nakedly revealed here is simply unwarranted, if
the argument I have been developing over the course of the last four chapters is correct.
Like Quine, Putnam seems to take for granted that rationalism breathed its last with the
death of positivism - that once representationalism is defeated, rationalism is ﬁnished.
As I have shown, however, the arguments advanced by both positivists and (neo-
)pragmatists in the twentieth century that were designed to demonstrate the untenability
of rationalism and the impossibility of metaphysics at most demonstrate the untenability
of representationalism, and provide (and can provide) no basis for the rejection of
metaphysics or rationalism generally. Indeed, Putnam seems to assume that there can be

no a priori basis for representationalism (other than a demonstration of the existence of
God by reason (logic) alone, which is of course impossible), and does not so much as
consider the possibility of a priori awareness of the structure and nature of reality in non-
representationalist terms. Putnam says, “the situation was summed up by Kant: Kant

held that the whole enterprise of trying to demonstrate the existence and nature of a

 

3" BonJour 1985.

72

39 Today, as I remarked,

supersensible world by speculation leads only to antinomies.
only a few relics would challenge this conclusion, which put an end to Rationalism as
well as to the medieval synthesis of Greek philosophy with revealed religion.” There’s
that word again! The point here, however, is that Putnam seems to take for granted (or is
at least quite conﬁdent) that to defeat representationalism in its apparently ﬁnal refuge
(scientiﬁc realism — “materialism,” or “physicalism”) is to defeat rationalism and
metaphysics pretty much once and for all. Putnam believes representationalism to be
doomed to failure, but this is because he thinks that the only way in which the traditional
metaphysical hope of a privileged standpoint can be preserved is by way of the
establishment of a correspondence relation (or something like it) between our words or
concepts and the objects of the world, or some of them.

For Putnam, then, the ﬁnal fate of rationalism and metaphysics rests squarely
upon the hopes of a physicalistically speciﬁed causal theory of direct reference to things-
in-themselves, which, incidentally, Putnam thinks is not to be had. Putnam argues at
length that the nature of cause cannot be physicalistically speciﬁed or deﬁned, and that if
we have to take a mentalistic notion like causation to be an irreducible, non-deﬁnable
fundamental element of our current or future science, this is not merely an adjustment of
materialism but an abandonment of it.40

What is worse, however, is this. Causation, Putnam observes, is no better off than any of

our other terms or concepts from the standpoint of the possibility of getting our concepts

 

39 Putnam 1982, p. 154. It’s not clear why our only recourse is speculation, but presumably this is because
speculation is all that we have unless the mind is simply capable of having contact with or awareness of
reality as such — as Putnam puts it, of having an “intellectual intuition” or intellektuelle Anschauung.
Whether these are our only two options, and what we should make of this possibility is a subject I take up
in Chapters Six, Seven and Eight.

‘° Ibid., p. 152.

73

or terms to correspond with reality. To see why this is so, consider the following. How
do we get our concepts to hook on to the objects of the world? Putnam observes that “an
act of will (or intention) won’t work. I can’t simply pick one particular correspondence
relation, because in order to do that I would need already to be able to think about the
correspondence C — and C, being a relation to things which are eternal and mind-
independent, is itself something outside the mind, something ‘external’ !”41 This,
however, is putting things backward. No representationalist assumes that we must get
our concepts to hook on to reality. The view of the representationalist is that the objects
of the world cause our representations of them, and so of course the correspondence
relation is causally grounded. Intentionality isn’t an explanation of the possibility of
contact with things-in-themselves, it is (if it is anything) evidence for this contact. Now
it is true that in order for us to identify and specify a single relation as the correspondence
or representation relation, we must already be capable of contact with things-in-
themselves. As Putnam says in Reason, Truth and History, “you can’t single out a
correspondence between two things just by squeezing one of them hard (or by doing
anything else to just one of them); you cannot single out a correspondence between our
concepts and the supposed noumenal objects without access to the noumenal objects.”42
This correspondence or representation relation, however, must obtain ex hypothesi for
representationalism even to get off the ground. The representationalist has not begged

any questions at this point. She must then, however, be able to certify that the

representation relation is as her epistemology says it is.

 

4' Ibid., p. 143.
‘2 Putnam 1981, p. 73.

74

This is what Putnam says the materialist cannot do, and on this point he is correct.
Here’s why. Any concept we have is simply part of our overall empirical explanatory
framework, and it would be hopeless to try to justify any concept or cluster of concepts
by appealing to another concept, since they are all in the same boat, epistemically
speaking. So long as we adhere to an empiricist account of language learning and
language use, none of our concepts can refer to anything (any object, any entity, or any
fact) other than those that are available to us through our senses. But then, in that case, it
is clear that none of our concepts is in a special or privileged position to certify the
operation of our senses, and thus none of them may serve as a ground for any of the
others. Therefore, no concept in our explanatory framework can be provided with a
framework-independent or framework-transcendent meaning.

Again, however, scientiﬁc realists are not in the business of certifying
representationalism, and though Putnam contends that representationalism is inconsistent
with materialism, he achieves this result only by way of a demonstration of the fact that
there is no way to establish representationalism within materialism (or natural science)
alone. Who, other than positivists, says there is? Moreover, since he again neglects to
consider the possibility of a priori certiﬁcation of representationalism, he has not
defeated it, but merely demonstrated that it is empirically uncertiﬁable, which was
already established with the death of positivism, and which is the reason why scientiﬁc
realists despair of supporting representationalism deductively.

Our consideration of Putnam’s position need not end there, however. In The
Many Faces of Realism,43 Putnam considers the question whether there is (or whether it

makes sense to say there is) one and only one way the world truly is (but many ways of

75

superﬁcially “cutting it up”), or there could be two (or more), non-compatible
interpretations of the structure and contents of reality, both (or all) of which could be
right or true in some sense. He observes that “metaphysical realists to this day continue
to argue about whether points (space-time points, nowadays, rather than points in the
plane or in three-dimensional space) are individuals or properties, particulars or mere
limits, etc.,” and he offers the following: “my view is that God himself, if he consented to
answer the question, ‘Do points really exist or are they mere limits?’, would say, ‘I don’t
know’; not because his omniscience is limited, but because there is a limit to how far
questions make sense.”44 Very well, but what is the limit? If, as Putnam suggests in
Reason, Truth and History, the limit is that of human intelligibility, and if the accounts of
the contents of our concepts and of our use of language are empiricist, then we arrive at
the same result as we did in our consideration of Quine’s arguments against rationalism
and the possibility of metaphysics.45 Either Putnam’s view assumes an empiricistic
account of concept acquisition and so begs the question against the rationalist, or his
pragmatism (internal realism) prohibits him from really ruling our rationalism and
metaphysics, since the only justiﬁcation he could offer for locating the limit of
meaningfulness at the limit of our language (or concept-scheme) would have to be

metaphysical in nature.

 

’3 And in Realism With a Human Face.

4“ Putnam 1987, p. 19.

’5 Putnam (1994) insists that “as John Dewey urged in Experience and Nature and elsewhere, all reference
is transactional,” and he claims that this leads us to “agree with Rorty that the idea of the mirror of nature —
the language that is nature’s own language, intrinsically, the language that is ‘correct’ for reasons which
have nothing to do with how we interact with nature, apart from how we are embedded in nature — makes
no sense” (p. 288). This, again, assumes representationalism, or at least assumes some sort of non-monadic
pluralist metaphysics, but we need not go along with these assumptions. For further discussion of this
point, see Chapter Eight.

76

But enough. To go on like this would be tiresome, and the point has, I think, been
made. The survival of representationalism as apparently the only possible version of
rationalism has made it seem that rationalism itself is untenable. This is not the case, but
since the possibility of rationalism now looks like the last hope for traditional
metaphysics (given the ascendancy of empiricism and naturalism in the wake of the
demise of rationalism’s “only” version), empiricists have proclaimed the death of
metaphysics since the dawn of the twentieth century. Representationalism is not the only
possible rationalist epistemology, however, and the arguments that have been advanced
by both positivists and pragmatists to demonstrate that metaphysics is impossible and
rationalism is dead can only be supported by metaphysical assertions of the kind these
arguments were supposed to reveal to be impossible or meaningless.

I myself have insisted over and again, however, that representationalism is
untenable, and that we need a better rationalist epistemology, but I have not yet offered
an argument against representationalism, nor offered an alternative rationalist
epistemology. It is high time I did so, and moved on to a consideration of non-
representationalist rationalisms. There is, I believe, an effective demonstration of the
ultimate untenability of representationalism available, and I think that this demonstration
is pretty well clinched by Richard Rorty in his Philosophy and The Mirror of Nature.
(The germ of this demonstration is implicit in Putnam and in Quine, in their observation
that, short of a priori certiﬁcation of the correspondence of certain of our concepts or

representations with reality as such, there is no way to “get outside our own heads,” or

77

outside of our own language or conceptual scheme, in order to “check its ﬁt with

reality”) It is to Rorty and to this demonstration that I now turn.

78

CHAPTER 5
RORTY: THE END OF REPRESENTATIONALISM AND THE ELIMINATION
OF EPISTEMOLOGY

There is a feeling of apparent rightness about representationalism; it just seems so
simple and straightforward. So long as we imagine that there are objects in the world, or
that there is a way the world is, it seems natural to suppose that the relationship between
our minds and what’s not in our minds consists of our having accurate mental images of
these things.1 This account enjoys such pictorial obviousness that it has attracted many,
if not most, Western philosophers, and has remained at the center of epistemological
inquiry for 2000 years. Hilary Putnam notes that representationalism was the
epistemological position that “ancient and medieval philosophers attributed to Aristotle,”
and, as I noted in the ﬁrst chapter, Descartes’ version of representationalism borrowed
elements of its central conceptual apparatus from Scholastic philosophy, which itself has
its origins in the work of Aristotle.2

While the representationalist schema seems intuitively natural and structurally
modest, it rather quickly leads us to a host of counter-intuitive consequences, and

requires accommodations and adjustments of great complexity. Indeed, its problems are

 

' It is worth noting here that this schema appears to take for granted that we are distinct from the world and
its objects — in the sense of being distinct entities or substances - so that we must be gotten into a certain
sort of relationship with the world (or its objects) in order to know anything about it. In that case, however,
it would seem somewhat mysterious that conscious beings alone have a certain special power to “picture”
other distinct entities and their relations to one another. To dispel the mystery requires one of two things:
either that (l) the world is itself a mind, or was created by a mind, or is mind-like, or is otherwise
sufﬁciently similar to the mind to make a mind-world connection possible, or that (2) our minds possess a
special sort of intuition or faculty that puts us in contact with the world. The traditional apparatus of
representationalism, then, takes metaphysical pluralism for granted. This is not necessarily to say that
metaphysical pluralism requires representationalism as its epistemology. Perhaps metaphysical pluralism
can be supplied with a non-representationalist rationalist epistemology; the point here is that the question of
the tenability of rationalism is clearly bound up with the ancient metaphysical problem of the one and the
many. I shall explore this issue further in Chapter Eight.

2 Putnam 1981, p. 57.

79

many and varied, and serious difﬁculties arise at every point of its theoretical apparatus:
at the point of the objects that constitute the source of our representations, at the point of
the nature of the representations themselves, and the point of our use of these
representations. Many of the problems of representationalism are by now quite familiar.
One of the most obvious of these is that if we are aware of objects and their properties in
virtue of mental representations that register these properties, how can we ever be in error
about the properties or nature of a given object? Plato was aware of this problem with
representationalism (among others), and rejected representationalism as an account of
knowledge on the grounds that it could not accommodate the possibility of error (so long
as we understand representations fairly literally, as little pictures in the head).3 The
problem of error has greatly exercised contemporary mental representation theorists as
well, who have been working to explain how a system of representations, understood as a
formally (that is, symbolically or syntactically) expressed computational scheme, could
go wrong.4

There are many more problems. If representations are literally little pictures in
our heads, where are they? Are they in principle observable by others? That is, could
someone else view, say, my representation of elephants, and see that it looked like a
picture of an elephant? Since any one picture of an elephant, or of anything, is only a
picture of one instance of that kind of thing (of one elephant, for instance), how does the

representation function as a representation of all of the things of that kind?

 

3 See Plato 1997a, l9lc-l96d. Plato seems, in fact, to have rejected representationalism generally, though
the idea that the mind grasps independently-existing forms through which it apprehends particulars and
their properties bears more than a superﬁcial similarity to representationalism. Plato nowhere offers an
account of mental entities that would reveal form-grasping to be representational in nature, however.

4 See, for instance, Dretske 1995, and Fodor 1995.

80

Even if we reject this sort of simple-minded understanding of representationalism,
however, there are problems. If representations are formally expressed, it is still not clear
how representations reﬂect or capture or represent the properties of their objects. If my
representation of The Grapes of Wrath captures the fact that it is hundreds of pages long,
how does it do that? I may be able to conjure an image of the book in my head, but this
image cannot be my representation of the book, if we reject the simple-minded approach.
(Besides, how will I know that this image represents the book as shorter than Moby Dick,
if they look about the same thickness in my mind’s eye? How could I know which book
is really longer unless my representations contain information about the number of pages
in standard type-face of each book, and do not merely picture the thickness of my copies
of them?) Does my mental representation generator assign numeric values to property
variables contained in my representations, such as “number of pages?”5 If so, how does
it do that? And how does it assign numeric values to certain variables and the
appropriate qualitative values to certain others? How, for instance, does my mental
representation reﬂect both the fact that The Grapes of Wrath is hundreds of pages long
and the fact that it dramatizes the plight of displaced Oklahoma farm families during the
Dust Bowl years? How is this information “encoded?”

This last question hints at another cluster of problems concerning the mind’s
employment of representations. The idea that representations are pictures would seem to
suggest that they do not require any interpretation (though of course this is false, as

Wittgenstein and Putnam so strenuously insist), since we can just see what it is they

 

5 Hilary Putnam and Saul Kripke have both argued that many of our concepts of things do not contain
enough information to refer uniquely to their objects anyway. Moreover, Kripke (1980) notes that even if
our concepts involve quite a lot of information about a given thing, their objects might have lacked many or

81

picture, and how the things so pictured are being depicted. If our representations are not
little pictures, however, and if the information they contain is then somehow “encoded”
in them, how do we make use of this information? That is, how is the mind able to

“rea ” this information? If the mind requires a language or set of rules for decoding the
information contained in one’s representations, how are the language or rules expressed?
If they are themselves expressed as representations, then we have an inﬁnite regress; if
they are “built-in,” how can we know that this built-in encoding and decoding system
“hooks on” to the real properties of objects and represents them as they really are?

This last problem was familiar to both Descartes and Locke, and while Descartes
thought that appealing to God’s existence guaranteed the potentially accurate operation of
our representation generating system, both philosophers noted that certain properties of
objects appear differently at different times and places to the same observer, as well as at
the same time to different observers. For instance, a bit of stone that appears red in
sunlight may appear orange under ﬂorescent light, or may look red from one angle and
orange from another, even in the same light. Thus, color (among other things) was
thought by Descartes and Locke not to be a property of objects themselves, but a feature
of our perception of objects caused by other real properties of those objects, such that
they have the propensity to cause us to experience them as colored. This, however, leads
at once to the counter-intuitive result that many of the properties we sense as belonging to
material objects — including solidity, warmth, and roughness — do not actually belong to
any objects (under any description) at all. Moreover, once this distinction between real or

“primary” properties and “secondary” properties is granted, representationalism is open

 

even most of the properties we ascribe to them and still have been the same objects for all that, which
suggests that even a richly detailed representation can fail to represent its object uniquely.

82

to the objection, advanced by Berkeley, that primary properties are every bit as observer-
relative as are secondary properties, and that there is no way to reduce primary properties
to invariantly speciﬁable qualities attributable to objects.6 This would appear to lead to
the even more counter-intuitive result that there are no material objects whatever.

None of these problems of representationalism would sufﬁce, however, to
demolish it. Representationalism could survive all of the objections implicit in the
foregoing puzzles. All that would be required is a sufﬁciently sensitive adjustment of
one or more of the elements of the theory to accommodate the claims of reason or the
claims of observation. Indeed, the intuitive appeal of representationalism has prompted
many philosophers to take precisely this approach. Some philosophers,7 would rather
preserve representationalism at virtually any cost than to take the risk of abandoning it,
presumably because abandoning it would mean that the entire project of Western
epistemology as handed down from Descartes was forfeit. I shall argue, however, over
the course of this and the next three chapters, that representationalism must be abandoned
because it is internally incoherent, but that we need not therefore relinquish all hope of
providing our epistemological enterprise with a secure metaphysical foundation.
Advances in our scientiﬁc understanding of nature (which such a metaphysics is intended

to underwrite) might seem to rule out the possibility of our having the sort of contact with

 

6 Berkeley also famously observed that ideas could not possibly represent material objects anyway, because
ideas are immaterial. For something to represent something else, Berkeley points out, the representation
must be similar to what it represents, and immaterial things are in no way like material things. Putnam
(1981) suggests that Berkeley is mistaken about this, since “everything is similar to everything else in
inﬁnitely many respects” (p. 64). This, however, would presumably not be true of two distinct substances.
Moreover, Putnam (1982), following Hume, insists that properties like causation and disposition are not to
be found in nature but are supplied by the mind, and are in fact mental constructions, from which he
concludes that to locate these properties in nature is to believe that reality is mind-like. This, however, is as
much as to say that the world must be similar to the mind in point of just those structural forms that mirror
the structure of cognition for it to be conceived as rationally ordered.

7 Carnap, for instance, and positivists generally.

83

reality that would make metaphysics (understood as an independent ﬁeld of inquiry into
the most general facts about the nature and structure of reality as such) possible. I shall
argue, however, that we cannot consistently renounce the Cartesian epistemological
project. This, in turn, means that we must accept the Cartesian epistemological standard,
and must therefore either have recourse to a secure metaphysics if possible, or to a

speculative metaphysics otherwise.

Hume, like Descartes, Berkeley, and Leibniz before him, saw that
representationalism requires independent incontrovertible certiﬁcation, but he argued
(contra Descartes, Berkeley, and Leibniz) that the existence of God cannot be
demonstrated deductively, and that we can have no other means of verifying the accuracy
of the operation of our so-called faculty of rational intuition. Thus, Hume declared, we
cannot possibly establish that our ideas stand in any given relation to the objects of the
world. Hume was prepared to accept Descartes’ standard, but also to reject
representationalism, and to develop an empiricism that makes do with the consequences.

More recently, Quine and Putnam have argued that one cannot stand apart from
one’s own (or from every) conceptual scheme, or raise oneself out of one’s own
language, in order to check its ﬁt with reality as such, which is another way of putting the
argument ﬁrst advanced by Hume. This argument, if sound, would mean the end of
representationalism, and I contend that it is, ultimately, sound, though not in the forms

advanced by Quine, Putnam, or Hume. There are three ways of understanding and

84

developing this argument. The ﬁrst two are the most common, though the ﬁrst way is
entirely trivial, and the second cannot successfully defeat representationalism. The third
way is advanced by Richard Rorty, and demonstrates the untenability of
representationalism.

The argument is sometimes put in terms of the claim that in order to speak of
anything or refer to anything we must use our own language, and that there is no way to
step outside of the bounds of one’s own language to see that it ﬁts with, or ﬁts more or
less well with, reality. This is, of course, to express matters ﬁguratively. We must have
recourse to language, and generally our own language, in speaking about our direct
contact with the world, but this does not rule out the possibility of such contact, which is
had (if at all) independently of the language we use to talk about it.8 What difference
would it make if we used a different language? Only if we already suppose that the
language of a community of speakers constitutes their ontology — their way of carving up
and experiencing objects and the world — would it make any difference to the possibility
of our being aware of mind-independent features of reality that (our) language is the only
means we have for expressing this awareness. In other words, our dependence on
language only matters if we beg the central questions concerning meaning and reference.
If the question is not begged, then the fact that we have to use our own language, or some
language or other, in expressing metaphysical realism is as trivial as the fact that we have
to use language to order french fries. Telepathy aside, how else would we do it?

Besides, no metaphysical realist ever suggested that we could step outside of our

own language in order to check its ﬁt (or lack of ﬁt) with reality. Establishing the

85

correctness of representationalism would indeed require either that (1) we can appeal to
someone or something that could certify, from its “perspective,” that mental
representations reﬂect reality, or that (2) we can identify certain representations that
necessarily refer to their objects, and that ﬁx the reference of our other representations.
In neither case, however, must we be able to step outside our own language or system of
representations or concepts. How could we (if it even makes sense to talk about such a
thing)? Why would anyone suppose that the metaphysical realist would commit herself
to a position that requires her to take up a perspective no one could possibly occupy?9

On the other hand, if neither of the two options outlined above is available (as
Putnam and Quine believe), then taking up this impossible position would be our only
other choice, which is clearly out of the question. At most, then, this claim is shorthand
for the claim that there can be no scheme-independent certiﬁcation of
representationalism, since the only remaining means of doing so is patently off limits.
The fact that the argument is couched in terms of our use of our language to describe the
world is irrelevant, as the metaphysical realist hastens to point out.

What, then, of the claim that there can be no scheme-independent certiﬁcation of
representationalism? Rather than repeat the arguments advanced in the last two chapters,

I will simply note that this claim, as expressed by Hume, Putnam and Quine, fails to

 

8 As William Alston (1999) remarks, “of course, when we use the word ‘true’ or any other term, we are
using our language, if we know what we are talking about. Who else’s language might we be using?” (p.
625)

9 The idea that the rationalist is committed to such a view has nevertheless been, I fear, the source of many
attacks on rationalist epistemology. (See, for instance, Maria Lugones’ “Purity, Impurity and Separation”)
It is sometimes claimed that the rationalist, in order to justify her rationalism, must be able to take up a
perspective or viewpoint, or position outside of all perspectives (or viewpoints, or positions) in order to
survey the whole (including her own position) and reﬂect upon it, which is clearly something no one could
possibly do, since the very idea of a conscious being occupying a perspective-of-no-perspective is
incoherent. This just begs the question against the rationalist, by assuming without argument that taking up

86

defeat representationalism, unless it can be established empirically, as it cannot, that there
can be no a priori awareness. In demonstrating the impossibility of such awareness, the
neo-pragmatist or empiricist either begs the question against the rationalist by assuming
an empiricist account of concept acquisition and use, or else must offer a metaphysical
justiﬁcation of the claim that no word in any language could refer necessarily to its
referent. This, of course, they cannot do without refuting the claim they are trying to
establish. The third way of developing this argument, however, succeeds in defeating
representationalism, and it runs as follows.

Richard Rorty argues that there can be no “privileged access” to the world that
can justify the belief that our ways of thinking and talking about it “hook on” to it at the
appropriate points - that is, represent it correctly. However, Rorty’s point is that our
supposed means of direct contact with the world itself will be, must be, conceived in
terms of other aspects of our total representational framework, which means that our
speciﬁcation of the means of contact is not merely in the same boat as the rest of the
framework, but must exhibit the same form as the rest of the ﬁamework. Rorty, claiming
to be paraphrasing Putnam, observes that “the attempt to get a set of nonintentional
relationships (such as those offered by a causal theory of reference or by a Sellarsian
notion of ‘more adequate picturing’) is always vitiated by the fact that those relationships
are simply further parts of the theory of the world of the present day.”lo This
observation, however, represents a signiﬁcant advance over any of the formulations
offered by either Putnam or Quine. Rorty’s argument here implies not that the only

means we can have for certifying any of our concepts would just be more (empirically

 

such a perspective is the rationalist’s only recourse, and that perspectives consist in nothing more than
discrete, equally “valid” ways of carving things up.

87

speciﬁed) concepts, but that, if getting the world right amounts to a certain sort of
picturing or mirroring relation between representations and the world, then our picture of
the picturing relation must itself accurately reﬂect reality (even if this picture of the
picturing relation is supposedly certiﬁable a priori), which means that it must itself be a
picture, and leads to the conclusion that the representationalist is trapped in an inﬁnite
regress. The real force of the claim lies in the observation that if we suppose that
representations mediate our awareness of the objects of the world, then the relationship
between minds and the world cannot be that representations picture the world, for then it
would be pictures all the way down. In other words, representationalism is internally
incoherent.

A number of representationalist responses seem possible here, but none seems
plausible. The representationalist could argue that the second-order certiﬁcation of the
ﬁrst order picture need not itself be pictured accurately, since, if the relationship between
language and the world consists of the former picturing the latter, then once we could see
that we had got the picture of picturing right, we would need to look no further.

However, this is to miss the force of Rorty’s argument. Our “seeing that we had got it
right” would itself just consist of another picture. The representationalist could claim that
the picture of the picturing relation enjoys a special status because of the place it occupies
within the representationalist framework. However, this has no more plausibility than the
positivist’s claim that the veriﬁcation principle enjoys a special status because of its
function as the arbiter of meaningfulness. The only apparent recourse for the
representationalist at this point is to block the regress by arguing that the picturing

relation itself does not require a picture, but this would be to reduce the picture theory

 

'0 Rorty 1979, p. 298, italics added.
88

itself to a piece of representationalist dogma, and would force the representationalist to
begin anew the search for foundations, this time seeking to tie both our total framework
and the picture theory to reality.

It gets worse. One last reply the representationalist might make is that if there
were something between a representation and its object, the presence and character of
which could not be doubted, something that was not itself a representation but could
underwrite or otherwise certify the content of a representation (such as an impression or
sense datum or raw feel or “what is given”), then representationalism could still be saved.
It is bad enough that, as Quine demonstrated, the positivistic hope of reducing physical
objects to sense data is vain. What is still worse, however, is that Wilfrid Sellars
demonstrated that if something (like sense data) did not have the form of a representation,
then it could not possibly certify the content of a representation, and if it is a
representation (or is representation-like), then it would stand in need of the same

certiﬁcation as the rest of our scheme. ”

Rorty, like Quine and Putnam before him, draws the conclusion that no direct
contact with the world is possible (though, of course, this conclusion is not supported by
the argument, since it is aimed only at representationalism). Unlike Quine and Putnam,
however, who still endeavor to identify the boundary between sense and nonsense,
meaningful and meaningless discourse, Rorty advocates a meta-philosophical stance. He

maintains that the history of modern philosophy since Descartes can be viewed as a

89

sustained, but failed, attempt to rationalize all discourse by providing an account of the
aim of discourse as the certiﬁcation of its own foundations, and of our knowledge of the
world, through the identiﬁcation of our means of direct or privileged contact with
reality.12 Here again he goes further than either Quine or Putnam, and not only equates
representationalism with foundational epistemology as such, but goes on to argue that
epistemology is not the central project of philosophy, and that the project of rationalizing
discourse should be abandoned altogether.

Since the rise of empiricism, Rorty observes, science has been taken to be our
means of contact par excellence, and we portray the past of humanity within our
epistemological tradition as leading naturally to the present moment, by way of a series
of ever more accurate, if sometimes crude and awkward, attempts to mirror the structure
of the world through scientiﬁc and philosophical activity.l3 Rorty maintains that this
attempt to rationalize discourse through the idea of human beings (and their activities) as
mirrors of nature generates a metaphysics that provides human beings with one way of
conceiving of themselves, but that other ways are possible, and that there is no way of
grounding any such total metaphysical framework in the world through some form of
privileged access to it, and thus no way of saying that any total framework is more

rational than any other, or deserves special status.14 Rorty therefore proposes that the

 

” Sellars 1963, Chapter 5, esp. pp. 127-134.

'2 Rorty also sees this attempt at rationalization in the ancients, and, like Heidegger, he traces the roots of
this conception of human beings as mirrors of nature back to Plato and Aristotle, but Rorty focuses
primarily on the epistemological project of philosophy as it has been carried out since Descartes, and most
especially, since the rise of positivism (or a little before; since the Linguistic Turn, anyway), partly
because, according to Rorty, “Aristotle did not have—did not feel the need of—a theory of knowledge”
(1979, p. 263). It should go without saying that I believe Rorty is wrong about this.

'3 Rorty 1979, p. 282.

"’ It should be noted that this meta-philosophical approach takes complete metaphysical conceptions of the
world as its unit of currency, and notes the historical contingencies which generate trade in them and
among them.

90

attempt at rationalization through the grounding of our epistemological project should be
renounced, and should not be replaced with any new philosophical method (where to
have a method would involve another attempt at the rationalization of discourse, the
speciﬁcation of the bounds of meaning or sense). Instead, Rorty insists that we strive for
ediﬁcation through hermeneutics, which would not constitute a new philosophic method,
but would study the relationship between normal (rationalized) discourse and abnormal
(initially “irrational,” because alien) discourse, and would concentrate upon, and
generate, abnormal discourse, so as to keep the human conversation moving.

It is not clear, however, how Rorty’s hermeneutics is supposed to offer an
alternative to philosophic method. Presumably, hermeneutics is not itself a method
which seeks, or needs (or wants), grounding in reality. The temptation, though, is to ask:
“Yes, but is this meta-philosophical perspective true?” That is, is the account it offers of
totalizing metaphysical frameworks getting things right? Perhaps this is a bit philistine.
Rorty would simply respond by saying that this question indicates the extent to which we
are still trapped within the representationalist’s picture theory, which offers only one
metaphysical account of our relation to the world among many possible accounts. Unless
we are to beg the question against Rorty, our sense of “true” or “right” cannot just
assume representationalism. Rorty would argue that “true” here can mean nothing more
than “more fruitful,” or, in Rorty’s idiom, “edifying.”15 Very well.

What, then, makes Rorty’s hermeneutic meta-philosophical perspective a true
history (by his lights) of the attempt to rationalize human discourse? If Rorty is to be
consistent, the answer is that his hermeneutic meta-philosophical approach is true in the

sense that it is edifying (if it is edifying). But of course this approach itself is simply one

91

approach among many possible approaches from the standpoint of a meta-meta-
philosophical approach to accounts of the history of Western thought, and this approach
can be no more privileged than any other. Why adopt one of these approaches over any
other? Why, for that matter, adopt the approach of hermeneutic meta-philosophy over
the way of (non-representationalist) rationalization and foundational epistemology?

Rorty notes that

Gadamer develops the notion of wirkungsgeschichtliches Bewusstein (the sort of
consciousness of the past which changes us) to characterize an attitude interested not
so much in what is out there in the world, or in what happened in history, as in what
we can get out of nature and history for our own uses. In this attitude, getting the
facts right (about atoms and the void, or about the history of Europe) is merely
propaedeutic to ﬁnding a new and more interesting way of expressing ourselves and
thus of coping with the world. From the educational as opposed to the
epistemological, or the technological point of view, the way things are said is more
important than the possession of truth.”

Firstly, this seems to imply an unnecessary opposition; why does one of these have to be
more important than the other? Can’t we be — aren’t we in fact — concerned with both?
Why must we equate either with philosophy itself? Can’t epistemology be one of
philosophy’s jobs, and self-realization be another? Plato thought so. Secondly, if self-
realization is more important than the possession of truths, does that go also for the
(presumably true if edifying) claim that self-realization is more important than the
possession of truths? What, then of Rorty’s recommendation? Why should we accept
this claim?

There does not seem to be any way to resist this conclusion, or in fact to resist the

conclusion that any method or theory or description or activity or whatever is not, from

 

'5 Rorty 1979, p. 360.

92

the perspective of a still higher meta-philosophical schema, simply one such method
among many possible. What, then, is Rorty on about? He certainly ought not be too
worked up over the attempt, as he sees it, of Western philosophers to ground their
epistemology in the world (even if doing so looks impossible), for, as Rorty himself
admits, this attempt is one way which human beings have had of conceiving of
themselves. What difference could theoretical considerations concerning the
impossibility of direct contact with reality possibly make to the success of Western
epistemology as a mode of self-representation, so long as it is edifying? (Rorty could
argue that it is not edifying, but then what reason could we have for adopting one set of
criteria of ediﬁcation over another, excepting that one set is more edifying?) Unless
Rorty intends to say that his meta-philosophical approach to Western epistemology is
right, what recommends his claim that the impossibility of grounding it should serve as a
reason for abandoning it?17

This is just the old accusation that relativism is self-defeating, aimed now at
Rorty’s hermeneutics, and his attempt to renounce the Cartesian epistemological project
of distinguishing what can be known from what must remain unknown. If no one total
perspective is better than (is to be preferred over, is privileged with respect to) any other,
then that must of course also go for the perspective that insists that no one perspective is
better than any other, in which case, any reasons we might have had for taking up this
perspective go right out the window. The only hope the relativist, or the neo-pragmatist,
or hermeneuticist (as the case may be), has at this point is to claim, as implausibly as the

representationalist and the positivist before her, that her perspective is not a perspective at

 

'6 Ibid., p. 359.

93

all. We can insist that hermeneutics is not a perspective or a method all we like, but if it
involves human beings engaging in self-realization, then it is in the same boat as is
Western epistemology (assuming Rorty thinks that Western epistemology was even part
of an attempt, if a failed one, at self-understanding). If it is not a perspective, then we
will want to know why hermeneutics can claim for itself this status as the only activity
that is free of the methodological assumptions of all other self-conceptualizing human
activities, while metaphysics cannot.

Moreover, Rorty commits himself to the following substantive metaphysical
assumption: that there are a potentially inﬁnite (or at least indeﬁnite) number of
allowable totalizing metaphysical frameworks, none of which hooks on to reality. If this
assumption does not simply beg the question against the metaphysical realist, then, at the
very least, it involves Rorty in metaphysical speculation on a par with the realist. Where
the metaphysical realist wants to say that there is, in the end, one and only one way the
world is, and that we can get it (or at least some of it) right, the neo-pragmatist wants to
say not merely that we cannot know this to be true, but that there is no way for any
possible metaphysics to get the world right. It would seem that the neo-pragmatist is
entitled to say that there is (and can be) no mind-independent world of the sort spoken of
by metaphysicians to get right in the ﬁrst place, since this notion of a mind-independent
world is itself a concept within metaphysical frameworks. This, however, either
presupposes that a priori awareness of features of mind-independent reality is impossible,
and thus begs the question, or requires a synthetic a priori claim about an invariant aspect

of the structure of every possible metaphysical framework: namely, that none of them can

 

'7 As Putnam (1990b, p. 25) remarks, “It seems more likely to me that, most of the time anyway, Rorty
really thinks that metaphysical realism is wrong.”

94

have direct access to reality as such. Rorty claims, as do Quine and Putnam, to have
information about the necessary character of the relationship between language or
concepts and the world that should, strictly speaking, be prohibited by his pragmatism.
Up to now I have been suggesting that it is a bit of a disappointment to be told
that there is a limit to what questions can be meaningfully asked, and that questions such
as “how can we be aware of the nature and structure of reality as it is in itself” fall on the
nether side of that limit, while at the same time being told that we can offer no reasons
why this is so. Such questions look perfectly meaningful, after all, and not just in an
“oceanic” way. Moreover, while it may often be difﬁcult to get undergraduate
philosophy students to see the bite of such questions, similarly supposedly antinomous
questions often arise for the ﬁrst time in childhood. Even Hilary Putnam notes
(interestingly, in “Why There Isn’t a Ready-Made World”) that “when I was 7 years old
the question ‘If God made the world, then who made God?’ struck me one evening with
vivid force. I remember pacing in circles around a little well for hours while the awful
regress played itself out in mind.”18 These are the sorts of questions that can drive one to
become a philosopher. Are we to say that they occur to us ﬁrst when we are children
because they are in fact the product of childish confusion (as so many positivists and
linguistic analysts did actually say of classic philosophical questions of the past, as
though Plato and Aristotle were mere kids by comparison to us)? Or should we say that
such questions must clearly have a sense, since even children can understand them? As
we age and learn, we ﬁnd that often our inability to solve (or resolve) such questions was
a result of our youthful lack of ingenuity at identifying slippage in the formulation of the

problem, where we can drive in a wedge. Are we to say instead that such questions are in

95

fact insoluble because meaningless? This is frankly worse than the replies of the adults I
pestered with such questions in my youth. They thought, like Callicles in Plato’s
Gorgias, that such questions were meaningless in the sense of being impractical, and so
not worth bothering about. We are surrounded by philosophers today who consider such
questions literally unintelligible. They are not, and cannot be, unintelligible on their face.
How could they be, if philosophers — intelligent men and women, the ﬁnest minds of
their times — for two thousand years took such questions seriously? To say that such
questions are unintelligible without offering anything more than an empiricist account of
language acquisition strikes me as a renunciation of inquiry itself. We cannot renounce
comprehensive epistemology - the attempt to distinguish meaningful from meaningless
utterance, to distinguish sense from nonsense, to distinguish what can be known from
what cannot — as Rorty suggests, because we can offer no reasons but only motives for
doing so, and because even these motives have nothing to recommend them, if we
abandon all hope of scheme-independent truth. If, however, we cannot abandon this
epistemological project, then we must provide a (non-question-begging) account of what
it is that we know about language, concepts, and meaning that earlier philosophers in our
tradition, who took metaphysical realist questions to be meaningful, did not know.
Arguments or observations will need to be offered that appeal to facts about the logical or
hidden linguistic structure of such questions; but these, in turn, will require metaphysical

assertions about meaning and reference.

 

'3 Putnam 1982, p. 144.

96

Given its intuitive appeal, representationalism has seemed to many philosophers
to be the only possible hope for a foundational epistemology, or even the only hope for
any epistemology at all, so that many who have been driven to reject representationalism
reject also the very possibility that we could know anything about the nature and structure
of reality as such, and even deny that reality as such has a nature or structure.19 This is as
much as to say that it has seemed to many philosophers that representationalism,
metaphysical realism, and foundational epistemology come to the same thing. Rorty, for
instance, maintains that “to think of knowledge which presents a ‘problem,’ and about
which we ought to have a ‘theory,’ is a product of viewing knowledge as an assemblage
of representations—a view of knowledge which, I have been arguing, was a product of
the seventeenth century.”20 It is my contention that this is false. The point here,
however, is that the intuitive pull of representationalism is in fact so strong as to make it
seem to some philosophers altogether impossible that we could conceive of another
metaphysical realist epistemology.

I shall not speculate as to the psychological forces that contributed to this attitude

among contemporary philosophers. It is, however, my contention that this attitude is

 

'9 Putnam makes this last claim in “Why There Isn’t a Ready-Made World.” Strictly speaking, rejecting
representationalism does not entitle us to say that the world has no inherent structure. We may have no
way of getting our language or thoughts to hook up with the world as it really is, and yet there may still be
a way that the world really is, outside of our ways of thinking and talking about it, for all that. This would,
of course, be a terribly weak and rather idle sort of realism, but it does not just logically follow from the
fact (if it is a fact) that we cannot meaningfully discuss the structure of reality, that it has no structure. The
neo-pragmatist’s claim, however, is that what we mean by “structure” (along with what we mean by
anything, by any term or concept) is itself just a part of our linguistic framework or conceptual structure,
and can thus have no purchase at all where “reality as such” is concerned. Consequently, there is nothing
that it could even mean to say that reality itself has a structure; there is no way to talk about our concept
(“structure”) applying to the world. I have been insisting on the heavily empiricist assumptions that seem
to be at work in this argument: namely, that the terms of our language (or our concepts) can only have such
content as is consistent with what we can sense. The argument, however, is also deeply Kantian; we have
no right to claim that our (phenomenal) notions apply to (noumenal) reality as it is in itself. All this shows
is that we must abandon the entire apparatus and legacy of representationalism (including the Kantian
psychologistic reduction of the a prion) if we are to have any hope of defending a robust realism.

97

more a product of psychology than of philosophy, since the current disregard for
rationalism and metaphysical realism does not rest on a successful demonstration of their
impossibility, contrary to what many philosophers today appear to believe. We must
reject representationalism, but this need not imply the death of rationalism and
metaphysical realism generally. This leaves open the possibility of developing a non-
representationalist rationalism that could serve as a foundational epistemology, and
provide us with a defense of metaphysical realism. Such a non-representationalist
rationalism would, of course, have to defend itself against alternative epistemologies, but
it need not defend itself against the claim that epistemology (as an attempt to secure the
metaphysical foundations of, and thusly to explain the possibility of, our knowledge of
the world) is impossible, since the arguments that have been offered on behalf of this
claim equate representationalism with foundational epistemology. We need do no such
thing. Let me be quite clear about this: metaphysical realism and rationalism do not get a
free pass that relieves them of the need to look like respectable philosophical theories in
light of advances in science, logic, mathematics, evolutionary theory, and so forth. They
do, however, get to compete without bias, because the conviction that such positions have
been deﬁnitively ruled out depends on metaphysical realism’s retention of
representationalism, and on the mistaken equation above. This conviction is the product

of an illusion.

 

2° Rorty 1979, p. 136.

98

CHAPTER 6

JERROLD J. KATZ’S REALISTIC RATIONALISM

Rationalism would be established if there were anything that we could know
about any aspect or feature of reality as such, through reason alone, without need of (and
in the absence of any possibility of) sensory awareness of this aspect of reality. That is to
say, rationalism would be established if we can have any knowledge a priori. Of course,
thoroughgoing empiricists wish to deny that any such knowledge is possible. As we have
seen, however, no argument can be offered to this effect, for two (related) reasons: no
argument that rests ultimately on sensory evidence alone can warrant the conclusion that
a priori awareness or knowledge is impossible; and the only sort of argument that could
establish this conclusion would require a means of distinguishing what can from what
cannot be known, which would in turn require us to appeal to premises whose
justiﬁcation can only be a priori (thus vitiating the argument). Thus, empiricists simply
cannot rule out (as it were, a priori) the possibility of a priori awareness or knowledge.
Empiricists have, however, produced a number of other arguments that would, if
successful, very nearly rule out a priori awareness, and rationalism along with it. Any
rationalist epistemology, even to get off the ground, must be able to meet these
arguments, in addition to being able to defend its account of the fumiture of the universe,
and our means of awareness of it.

The standard empiricist arguments against the existence of any actual a priori
awareness or knowledge involve a combination of inductive and dispensability
arguments, and they run as follows. Any number of statements have, in the history of

Western thought, been identiﬁed as being a priori knowable, and as constituting a priori

99

knowledge, and many of these have not stood the test of time. Statements that have been
held to be a priori knowable necessary truths have turned out to be a posteriori necessary
falsehoods. To take a particularly famous example: Kant, who was trying to save
metaphysics from Hume, claimed that Euclidean geometry is the a priori science of
space, but in fact it is neither an a priori science of actual space, nor even an accurate
science of actual space. What we think is a priori (even what we think is demonstrably a
priori) is not always really a priori, and this leaves the rationalist in an uncomfortable
position: to say that we know anything a priori, we would appear to need an independent
means of establishing that a statement or belief is a priori, but this means must itself be
either a priori or a posteriori. If it is a posteriori, the force of the claim that the belief is
a priori is lost, but if it is a priori, we are launched upon an inﬁnite regress. Since it
seems likely that we cannot really know when a belief (or when our knowledge or
awareness) is a priori, and that beliefs that are purported to be a priori have often turned
out, and will often turn out, to be otherwise, the notion of a priori knowledge is really
doing no epistemic work for us,1 and the claim that there is any a priori knowledge at all
is inductively suspect.

Secondly, empiricists argue that we need not appeal to a priori awareness to
account for the knowledge we have.2 Even mathematical knowledge, empiricists claim,

can be fully explained without recourse to (synthetic) a priori knowledge. There are a

 

‘ Komblith 2000, pp. 71 and 86.

2 Again, empiricists cannot argue that we need not appeal to a priori awareness to account for any
knowledge we could have; thus, empiricists lack what I have been calling a comprehensive epistemology.
Presumably, this would not trouble someone like Quine, for instance, who maintains that a comprehensive
epistemology, while a nice dream, is out of the question anyway. The point, however, is that Quine is not
entitled to this claim, and that we cannot rule out comprehensive epistemology, for the same reason we
cannot rule out rationalism: the only rationale for doing so would require an appeal to metaphysical
statements that would vitiate the attempt. As I argued in the last chapter, we cannot give up on
comprehensive epistemology, because we can offer no rational reasons for doing so.

100

number of competing empiricist projects aimed at providing such an account, ranging
from treating mathematical truth as a type of truth-in-ﬁction and mathematical
knowledge as knowledge of a certain story, to treating mathematical statements as
ranging over sets, understood as theoretical posits, and mathematical knowledge as
consisting of a web of empirically well-supported statements that enable us to cope with
reality. The point on which all such projects and their defenders agree, however, is that
some such empiricist project will be able to account for all (actual) human knowledge
without once appealing to a priori awareness of anything. Following on the heels of this
claim, empiricists argue that a successful empiricist account of human knowledge will
also be able to explain (or will explain away) the appearance of the necessity of some
truths.3 Thus, a priori awareness, a priori knowledge, and rationalism generally, are
dispensable.

Lastly, empiricists argue that rationalists can offer no comprehensible account of
the possibility of rational insight, or a priori awareness of the structure and nature of
reality as such, and thus that a rationalist metaphysics could be nothing other than

speculative.4 They maintain that even if there seemed to be good reasons for thinking

 

3 Quine (1980), for instance, suggests that certain statements, such as elementary arithmetical statements,
only seem necessarily true because the chances that we would be compelled to revise them seem so
vanishingly small, given their centrality to our epistemological framework. Such statements are, however,
in principle revisable according to Quine, since (1) we could be confronted with reasons for revising them
by recalcitrant experience, and (2) in such a case, the rational course would be to revise them, rather than
preserve them at the cost of compromising the phenomena. 1 would argue that the fact that Quine’s
epistemology requires him to admit that it is very unlikely that we would have to revise such elementary
statements is primafacie evidence in favor of the rationalist’s contention that the structure of reality is
rationally discemable. The development of this claim will, however, have to wait for my next dissertation.
" Empiricists sometimes develop this objection along the lines of the claim that rationalists cannot produce
a respectable causal account of the possibility of a priori awareness, but since the demand for a respectable
causal account often involves a demand for a causal account that empiricists can respect on their own
terms, the demand begs the question, as Katz (2000) argues, and as we shall see later in this chapter. The
demand for a philosophically comprehensible account is, of course, perfectly reasonable; the demand for an
account that falls neatly within the scope of current science and physicalist scruples is not, for reasons that
should be obvious.

101

that we could have a priori awareness of some aspect of reality, we have no way of
making sense of how this could be possible, other than by positing some sort of “mental
grasping” or “mental seeing” of that aspect of reality, that seems more akin to mind rays
and occult powers (or “revealed knowledge”) than to any of the other forms and modes of
information reception of which we are aware.5 If there were any such mode of
awareness, surely we would know something more about it by now. Since there does not
seem to be any way to make sense of the possibility of a priori awareness, then, the
prospects for rationalism seem dim, even if we had what would otherwise look like good
reasons for thinking that some of our knowledge is a priori.

Jerrold J. Katz offers what he calls a realistic rationalism (pun intended) in which
he defends a version of metaphysical realism and offers a version of rationalism to
account for our knowledge of this ontology, and in which he answers each of the
empiricist arguments outlined above. We shall examine Katz’s realistic rationalism to
determine whether he has developed a version of rationalism that satisﬁes the demands
we have speciﬁed that any epistemology must meet to be acceptable.‘5 We shall see that
while Katz’s replies to the empiricist arguments are often quite strong, or at least suggest
a direction in which a stronger reply could be developed, not all of his replies are fully

satisfactory, and his realistic rationalism will not meet our demands.

 

5 I shall argue, later in the chapter, and again in Chapter Eight, that this sort of criticism on the part of
empiricists is not only unwarranted but actually unmotivated, and amounts to little more than scientiﬁc
provincialism (also oﬁen involving a straw man caricature of a priori awareness). I shall also argue that
the best way to understand Katz’s “mystery versus mysticism” argument in Realistic Rationalism (2000,
pp. 32-34), is as a rebuke of this criticism along the lines 1 shall develop. . I will also return to this issue in
Chapter Eight, where I shall attempt to make the very idea of a priori awareness seem more metaphysically
lausible.
gAs a reminder, these demands are as follows: to be acceptable, any epistemology must be capable of
answering the Cartesian skeptical question (viz., what, if anything, enables us to have any knowledge
whatever?) and of thereby providing a ground for our knowledge (or some foundational part of it), as well

102

Katz’s reply to the inductive argument is actually fairly strong, though at ﬁrst it
appears rather weak. He is aware, of course, that the challenge to rationalism, and to the
possibility of a priori knowledge in general, rests in part on historically famous cases of
purportedly a priori knowledge turning out to be neither a priori nor even true.
“Everyone is familiar,” Katz observes, “with the standard problems with transcendental
idealism, such as the fact that Euclidean geometry, alleged to be an a priori necessary
truth, turned out to be an a posteriori contingent falsehood.”7 Moreover, Katz
acknowledges that “not only did the fate of Euclidean geometry deal a heavy blow to
conﬁdence in the Kantian explanation of synthetic a priori knowledge generally, but the
role of geometry in relativity theory provided a paradigm for treating other alleged a
priori necessary truths as a posteriori contingent truths—highly theoretical, but

nonetheless empirical in nature.”8

His answer, however, is that the fact that Euclidean

geometry was taken to be known a priori to be the science of actual space does not mean
that it really was a priori. Katz observes that “to be sure, people once believed that it is a
priori that Euclidean geometry is a true theory of physical space, but Euclidean geometry

was never determined a priori to be a true theory of physical space.”9

He goes on to say
that “from the fact that it is believed that p is a priori, it does not follow that p is a priori.

In fact, the grounds for accepting Euclidean geometry as a true theory of physical space

 

as a metaphysical justiﬁcation for the location of the boundary between what can be known and what
cannot.

7 Katz 2000, p. 8.

8 Ibid.

103

were straightforwardly a posteriori. The abandonment of Euclidean geometry in physics
was a revision of an empirical theory.”10

This reply seems simply to miss the force of the objection. The objection is that
while some statements have appeared to be a priori knowable, they have turned out not to
be a priori or true, and that there does not seem to be any way for us to know when we
are really in possession of an a priori truth, even when we seem demonstrably to be in
possession of such truth. To the extent that the Kantian example is especially salient, it
does matter whether Kant (and others) thought he had determined a priori that we do
know a priori that Euclidean geometry is the science of actual space. Katz’s claim is that
this was never in fact determined. It is not clear, however, whether he means that nobody
(including Kant?) took Kant to have determined it,11 or that even if it had been believed
to be demonstrated, it was not. In any event, the former seems unlikely to be true, and
the second still misses the point. If Kant (and others) had merely thought that he had
demonstrated that we know a priori that Euclidean geometry is the science of actual
space, but were wrong, the empiricist will ask how we can ever know that we are truly in
possession of a priori knowledge. If Kant could be wrong about this, could we ever
know that we got it right?

Katz’s claim is that it should have been obvious, even in Kant’s time, that Kant’s

epistemology did not really enable him to show that anything is a priori known to be

necessarily true. He makes the point in the context of a discussion of mathematical

 

” It is beyond doubt that Kant, at least, thought that he had determined a priori (or that a determination
could be made) that Euclidean geometry is the science of actual space, because Kant thought that he had
shown a priori that “space is not a concept that has been drawn from outer experiences,” and that “space is
a necessary representation, a priori.” From this it would follow that, as Kant puts it “the apodictic certainty

104

necessity, but the point may be extended to any of Kant’s claims concerning a priori
knowledge, where that knowledge is supposed to ﬂow from the facts about the structure
of pure cognition. Katz observes (citing Frege as the source of the observation) that
“locating the grounds of necessity within us does not explain the necessity of
mathematical truth. It at best explains why we naturally take mathematical truths to be
necessary?”2 If this criticism is right, then Kant never had any grounds for claiming that
we know anything a priori to be necessarily true. This would appear to answer the
empiricist’s objection, and if this is a fair characterization of Kant’s epistemology, it
should soften the impression that Kant’s epistemology constitutes an acute
embarrassment for rationalists.

These considerations can, however, only serve to soften the impression; they
cannot eliminate it, and Katz’s answer does not provide us with a satisfactory reply to the
empiricist. We shall see that Katz’s realistic rationalism does only a little better than
Kant’s transcendental rationalism in its account of a priori knowledge of necessary truth,
but, worse than this, the empiricist can still maintain that the objection has not been
answered. Even if we could (or should) have noticed this consequence of Kant’s
epistemology, the point is that we didn’t (until now, anyway). Once this consequence is
conceded, it explains why we should not think that an a priori necessary truth was

falsiﬁed (since it was, apparently, never really a priori to begin with), but it does not

change the facts that there were intelligent philosophers who thought that a demonstration

 

of all geometrical principles and the possibility of their a priori construction are grounded in this a priori
necessity [of space)” (Kant 1998, pp. 157-158).

'2 Katz 2000, p. 9. Here Katz makes something close to the criticism of Kant I offer in Chapter One: that
Kant’s attempt to salvage metaphysics requires abandoning the traditional conception of the a priori as
consisting of awareness of aspects of reality as it is in itself (outside the mind) in favor of a conception that
understands a priori knowledge as consisting of awareness of those features of cognition that determine the

105

of a priori knowledge had been successfully carried out, and that they were wrong. This
produces a general worry about claims to a priori knowledge or knowability: such claims
seem vulnerable to future revision and even falsiﬁcation; how, then, can we ever say
conﬁdently that we are in possession of an a priori necessary truth (or simply that we
know anything a priori)?13

Katz does not answer this question directly, since he seems to think that the
empiricist’s concern is that it looks as though some a priori necessary truths can be, and
have been, falsiﬁed. Katz dispatches this worry easily, but then, this worry is easily
dispatched. The problem is not how to handle this worry, but is rather that the most
obvious way of handling it leads immediately to more serious problems for the
rationalist. It is easy to say, simply, that no necessary truth has been falsiﬁed because, as
it turns out (lucky for us!), the truth (or falsehood) in question was not actually necessary
after all. Very well, but if this is our answer, how do we deal with the result that we are
committed to saying that a priori awareness of necessary truth can be mistaken? Katz
has an answer, but it must be extracted from his argument that apodictic principles of
formal sciences can be known a priori, not because they themselves can be established a
priori, but because they can be shown to be “essential to the best systematization of the

”14

body of [formal] knowledge in question. The reason that their being essential to such

 

structure of experience as such. Kant was, in this respect, one of the very ﬁrst philosophers to propose
naturalism as an alternative to speculative metaphysics.

'3 Of course, this general worry does not warrant thoroughgoing skepticism with respect to the possibility
of a priori awareness. If it could be established that such awareness is possible, then we could have reason
to think that we, at least sometimes, succeed in having it. Still, so long as our putatively a priori
knowledge or awareness plays only a systematizing and organizing role within our overall system of
beliefs, the status of such putative a priori knowledge remains in jeopardy, so long as some other element
of the system could imply the need for revision of an (ostensibly) a priori belief. As 1 shall argue further in
the next two chapters, at least some a priori awareness must be self-certifying for us to have any assurance
that any given (putatively) a priori truth is really unrevisable.

" Ibid., p. 48.

106

a best systematization renders them a priori is that “knowledge of basic formal facts is a
priori, the step from that knowledge to transcendent knowledge of formal laws and
theories is also a priori,” and “ﬁlling gaps and correcting errors is a priori too,” so
“formal knowledge always has the a priori warrant of pure reason.”'5

At last we have our answer. Katz adds that “the revisability of a priori
propositions is revisability in the light of further pure ratiocination, not revisability in
the light of further empirical discoveries.”16 Even though we cannot be sure that we are
in possession of bona ﬁde a priori knowable necessary truth at any given moment, and
even though what is thought to be known a priori is open to revision, we can only be
persuaded to change our minds about the status of such beliefs by additional a priori
considerations. Thus, while we can be mistaken about what is a priori known or
knowable, our only means of correcting these mistakes is itself a priori, so the worry that
we never know anything a priori is unfounded, and we need no independent means of
establishing that our belief is genuine a priori knowledge, because even if it is not
genuine knowledge, a priori it remains.

This conclusion depends, however, on all a priori knowledge consisting of
knowledge of the facts in the domains of the formal sciences. This seems plausible
enough, on its face, inasmuch as it would imply that a priori knowledge is concerned
with non-empirical facts. I shall, for the moment, put off asking whether all a priori
knowledge does, in fact, consist of knowledge of the facts in the domains of formal

sciences (as Katz understands the formal sciences). More signiﬁcant is the fact that even

if Katz is right about this, he would seem simply to have evaded, or answered by ﬁat, the

 

'5 Ibid.
'6 Ibid.

107

empiricist claim that it looks as though claims to a priori knowledge can be defeated by
empirical considerations. Katz claims that we could be wrong about what we know, but
to the extent that what we think we know a priori is justiﬁed (even if it is in fact false),
its justiﬁcation is a priori, and is only defeasible through further a priori considerations.

Though Katz does not develop this claim further, it appears to be undermined by
the case of the status of Euclidean geometry. It would appear to have been empirically
discovered that Euclidean geometry cannot be known a priori to be the necessarily true
science of actual space, since it was demonstrated not to be the science of actual space by
the replacement of Newtonian mechanics with Einstein’s general theory of relativity. If
Katz is right, then we could only have known a priori that Kant’s epistemology did not
warrant his conclusions concerning the necessity of truths concerning the structure of
pure cognition. Katz could reply that while Einstein’s general theory of relativity is
clearly an empirical theory, it is not this theory that requires us to abandon Euclidean
geometry as the science of actual space, but the principle of non—contradiction (along
with other logical principles): observation gives us reason to believe that space is not
Euclidean only because we infer from these observations that Euclidean geometry as a
theory of actual space requires predictions about the structure of actual space that are not
borne out in the event. It is not Einstein, then, but the principle of non-contradiction that
tells us that space cannot be both Euclidean and non-Euclidean. Such inferences require
logical principles that are themselves known a priori; therefore, a priori beliefs are
revisable only in light of further pure ratiocination, as Katz claims.

The difﬁculty with this line of reply is, of course, that it either threatens to

collapse the distinction between a priori and a posteriori knowledge altogether, or fails to

108

address the problem. To say that our beliefs about the structure of actual space are a
priori because we must appeal to a priori principles when revising them would make it
difﬁcult to imagine which of our beliefs are not a priori. On the other hand, if Katz
wishes to claim that, in the end, our (supposedly) a priori belief about the structure of
actual space was revised only in light of other (genuinely) a priori principles, then this
simply would be to ignore the apparent contribution of empirical observation to the need
for belief revision in the ﬁrst place. This would leave our other (putatively) genuine a
priori principles as vulnerable as the supposedly a priori Kantian claim.

Katz does not, however, develop this reply any further, and it seems difﬁcult to
imagine how he could, so long as he hopes to maintain both that all a priori knowledge
consists of facts within the domains of purely formal sciences and that a priori
knowledge is only revisable in light of further pure ratiocination. I shall return to the
question whether genuine a priori knowledge can be refuted (or even challenged) by
empirical observation (and to the question whether we can ever be secure in the belief
that apparent a priori awareness is genuine), in the next two chapters. Katz’s answer to
the empiricist’s claim that if a priori belief is revisable then it is doing little or no
epistemic work, is that appealing to a priori beliefs about the nature of abstract objects
that constitute the objects of the formal (foundational) sciences plays an indispensable
role in the best epistemological account of our knowledge of mathematics, linguistics,
and logic. Obviously, much depends upon this aspect of Katz’s position, but I shall
return to this matter later in the chapter. For now, let us consider Katz’s replies to the

remaining empiricist arguments against the existence of a priori awareness.

109

Katz’s answer to the empiricist’s claim that empiricism can provide a satisfactory
account of the apparent necessary truth of certain statements is straightforward and
devastating. Quine’s holistic account of the meaning and adequacy of the statements of
our natural scientiﬁc framework is almost universally regarded as the best contender for a
successful account of the apparent necessity of the truth of certain statements, and Katz
argues that Quine’s account is incoherent.

Quine holds that no statement is in fact immune from revision. This is certainly
consistent with thoroughgoing empiricism, since we could presumably have no means of
knowing in advance of all possible future experience, what will stand up to repeated
attempts at disconﬁrrnation and what won’t. This conclusion, however, would seem to
conﬂict with our intuition that some elementary statements — such as the principle of non-
contradiction, or that 1=1 — are certain to such a high degree that we cannot imagine them
being false, and that they are in fact necessarily true; nothing — no experience of any kind
— could conﬂict with such statements, or cause us to give them up or alter them. Quine
concedes that such statements are not generally susceptible to direct observational
disconﬁrrnation, and that it is very unlikely that we will give up these sorts of elementary
statements, but he maintains that it is not impossible. Quine’s epistemology is
coherentist, and he pictures the whole of our (natural scientiﬁc) belief structure as
consisting of a web or network of statements, some of which are relatively closer to direct
empirical experience (and thus more susceptible to direct disconﬁrrnation), and some of
which are relatively further from this experience, including those that constitute our
apparatus for making sense of and organizing that experience — mathematics and logic.

The statements, of mathematics and logic for instance, that are closer to the center of our

110

framework are not only less susceptible to direct disconﬁrrnation but are also generally
less likely to be revised; since their organizing function is so central to the framework as
a whole, the cost of revoking or revising these statements is quite high, and any
disruption among these statements will produce framework-wide disruption, so we
rationally choose to avoid paying such a price, or inducing such disruptions, by making
adjustments elsewhere in the framework.17 Thus, elementary beliefs are not in fact
unrevisable, though they may appear to be because of their role in our framework of
beliefs about the world. To say that they are necessarily true is wrong, but it is
understandable that our intuitions would deliver this judgment. To say that they are
necessarily true is, to paraphrase Wittgenstein, only a slightly hysterical way of saying
that it’s very hard to imagine giving them up, and very unlikely that we will have to.
Katz argues as follows. To regard every statement as revisable, we must,
presumably, include even the statements that govern revisability and the process of
revision. In that case, the principle of non-contradiction is revisable, however unlikely it

is that it may require revision in light of recalcitrant experience. To argue that any

 

'7 One objection to Quine’s picture that Katz does not exploit here, is that while its accuracy as a
description of our actual theory-selecting practices is part of what makes it so attractive, there can be no
warrant for regarding the picture as a normative account of our theory-selecting practices, as Quine seems
to assume. This is because Quine’s thoroughgoing empiricism implies that we can say nothing informative
about the extent to which our actual practice of striking a balance between permitting system-wide
disruptions and making more minor adjustments is rational, since there is no way to know or even to talk
about this balance being apprOpriate to the way the world is. At best, Quine could claim that it is rational
according to current standards of rationality, or that our actual practice seems best given our other beliefs
about the world, as Quine does in fact seem to say, for instance, in “Two Dogmas of Empiricism.” In that
case, however, nothing recommends our current practices, other than the practices themselves; and, what is
worse, nothing recommends Quine’s epistemology as anything other than a description of our current views
about theory choice. It certainly cannot be said to be rational in any normative sense. (As Jaegwon Kim
puts the matter, in “What is Naturalized Epistemology,” “for epistemology to go out of the business of
justiﬁcation is for it to go out of business.” (Kim 2000, p. 306).) This is entirely consistent with Quine’s
naturalized epistemology, to the extent that it is (obviously) non-nonnative, but the point seems to have
been missed by those (including Quine) who recommend Quine’s epistemology on the grounds that
adopting it will lead to knowledge. The only reasons we have for thinking that this is true are the same
reasons we have for thinking that our account of knowledge is a good one: namely, that it’s the one we
endorse. This is the epistemic equivalent of patting ourselves on the back and calling it acclaim.

111

statement or set of statements stands in need of revision, however, we must be able to
appeal to the fact that this statement (or these statements) conﬂict with — that is,
contradict — other statements that we would like to make in order to accommodate some
new experience(s). If one of the statements that we are considering revising or
jettisoning is the principle of non-contradiction itself, then either the argument that
supports its abandonment is invalid, in which case we have no argument for its
abandonment, or it cannot be abandoned (so that the argument may go through), in which
case it is not revisable. Either way, at least some statements are not revisable, on pain of
the consequence that we no longer have a decidable process of revision. As Katz puts the
matter: “It might make sense to contemplate replacing every plank in Neurath’s boat, but
it makes no sense to contemplate replacing the basic principle of ship construction that
says that there have to be planks between us and the water.”‘8 In fact, as Katz points out,
this argument supports the following general epistemological principle: “There is no
actual epistemology that says that everything including itself is revisable.”19
Quine could reply that this sort of argument is ungenerous to his position and
unimaginative about the possibilities for belief-revision. If we needed to revise the
principle of non-contradiction, it’s not as though we would replace it with the counter-
principle: “Two contradictory statements can be true at the same time.” We would
replace it with something we maybe at this point can’t even imagine — but this is
precisely the point; the fact that we can’t imagine its being false is a result of our inability

to imagine what would replace it. We could, however, replace it with some loosened

constraint on statement conﬂict, or some coherence-preserving constraint, or who knows

 

'8 Katz 2000, p. 74.
'9 Ibid., italics added.

112

what. To say that we could not replace it, however, is unsupported, historically and
logically.

Firstly, it should be noted that this statement of Quine’s position requires that we
could not abandon the principle of non-contradiction, but only modify it.20 Secondly, this
reply would seem to suggest that Quine’s epistemology still needs some sort of principle
of coherence, which seems fairly obvious and uninteresting until we notice that this result
supports the principle stated above. Quine’s epistemology cannot be universally
revisable, but it must be to maintain its thoroughgoing empiricism. Thirdly, the claim
that history shows that every belief or statement can in principle be revised has already
been answered in part above, and will be answered in more detail below. F ourthly, the
fact that we can’t now imagine the principle of non-contradiction being false, considering
that human beings have never been able to imagine this,21 as much supports rationalism
as it supports Quine’s thoroughgoing empiricism.

Katz’s conclusion, and mine, is that Quine’s account of the appearance of
necessary truth enjoyed by some statements fails, and that its failure is a result of the fact

that some statements are unrevisable, and are therefore necessarily true.

 

2° Even Hilary Putnam argues that there must be at least one a priori truth: namely, a somewhat weakened
version of the principle of non-contradiction that states that not every statement and its negation can be true
(Putnam 1983d). To deny this principle, Putnam points out, is to give up on rationality itself. BonJour
makes a similar point in In Defense of Pure Reason, when he observes that abandoning all appeals to any a
priori principles would amount “in effect to intellectual suicide” (BonJour 1999, p. 5). Of course Putnam
draws from this observation different conclusions concerning the nature and ground of rationality than I,
and BonJour, do.

2' 1 must qualify what I mean by this somewhat sweeping generalization. Of course dialetheists argue that
some contradictions might in fact be true, or that, within a certain scope (or within certain paraconsistent
logics), some contradictions might be permissible or acceptable. (See, for instance, Beall 2004.) But,
again, this claim is clearly not equivalent to an outright rejection or repudiation of the principle of non-
contradiction, and I maintain that such an outright rejection or repudiation is in fact inconceivable, as Katz
argues, even if the principle of non-contradiction could be (or should be, or must be) modiﬁed in some
way(s).

113

Turning to the other of the two dispensability arguments, then, Katz answers the
claim that we can account for all of our knowledge without appealing to a priori
awareness or knowledge by saying that a rationalist epistemology offers the best account
of knowledge in the formal sciences: logic, mathematics, and linguistics. This claim has
at least some initial plausibility in light of the fact that thoroughgoing empiricism seems
incapable of explaining away necessary truth. As was noted above, however, the
rationalist still owes the empiricist an account of our knowledge of mathematical or
logical truths (for instance) as a priori knowledge. Katz presents such an account.

He says that “a particular realism is an ontological position in the foundations of a
particular formal science,” which means that, for Katz, rationalism is an account of our
knowledge of the objects in that domain (or those domains — however far formal science
extends). He maintains that the best account of our knowledge in logic, mathematics, and
linguistics is a rationalist account that clariﬁes the nature of our cognitive contact with
the abstract objects of whose structure and structural relations knowledge in the formal
foundations of these sciences consists. Knowledge in the natural sciences (including
“mathematics proper, logic proper, and linguistics proper”), then, while partly empirical,
also rests on (or relies upon) these formal foundations. The formal sciences are
“branches of the philosophy of science, concerned with a philosophical understanding of
the results in the particular sciences proper.”

Katz argues that such a realistic rationalism is the best account of our knowledge
in mathematics, logic, and linguistics because, roughly, there are necessary truths (as we
saw above) — namely, certain logical, mathematical and linguistic necessary truths —- and

our only serious prospect for being able to make sense of the possibility of our knowing

114

that a statement expresses a necessary truth is rationalism. His argument receives further
support, as we have seen, from his critique of empiricist accounts of the appearance of
necessity enjoyed by certain statements, and of empiricist accounts of knowledge
generally, as well as from a critique of empiricist objections to rationalism on the grounds
that it is incomprehensible, mystical, and lacking examples of a priori knowledge (this
last critique to be considered shortly). Katz also seems to think that his realistic
rationalism is the best account of philosophical knowledge in general (despite the almost
stipulative nature of his pronouncement that metaphysical realism is “an ontological
position in the foundations of a particular formal science”), and he considers the
prospects of extending his project to cover “philosophical knowledge wholesale.” He
concedes that “the initial problem in extending this account to philosophical knowledge is
that philosophy is not for the most part a study of abstract objects.” I would say that the
bigger problem is that some of Katz’s characterizations of the general nature of
philosophical inquiry equate it with formal science as he understands that term, while
some do not. Thus, it would seem both trivial and non-trivial to extend his account to the
rest of philosophy. Moreover, if philosophy is “not for the most part a study of abstract
objects,” then the “initial problem” of extending his realistic rationalism also seems to be
an insurmountable problem. Worse still, even his most generous characterizations of
philosophy and of rationalism are too narrow, as we shall see. Katz is at pains to
differentiate his realistic rationalism from Quine’s thoroughgoing empiricism (that also
acknowledges abstract objects), but his only real disagreement with Quine is, at bottom,
over whether our knowledge of abstract objects is a priori or a posteriori. Katz’s view is

that rationalism and empiricism are simply opposing approaches to providing a

115

satisfactory account of scientiﬁc knowledge and practice, and that rationalism is
established so long as it is preferable to every available empiricist account of knowledge.
This position is of course too weak to meet our demands. Let us begin, however, with
Katz’s account of our awareness of abstract objects.

Katz begins by noting, quite correctly, that the empiricist’s demand for a
comprehensible account of a priori awareness begs the question against the rationalist to
the extent that it consists of a demand for an account that explains a priori awareness in
terms of sensory contact with reality itself. This begs the question because it is essential
to the rationalist’s position that a priori awareness is not sensory awareness.22 Since this
is so, no account of a priori awareness can be formulated in terms of sensory awareness,
but the demand for such an account rules out the possibility of a priori awareness without
a hearing, on the assumption that any comprehensible account of awareness of any kind
will have to be spelled out in sensory terms. Katz argues that the empiricist demand for a
causal account of a priori awareness is question-begging in precisely this way, because to
demand a causal account is to demand an account rendered in terms of sensory contact or
awareness. Of course, this assumes that causation is a purely empirical affair, and that
reality as it is in itself (or abstract objects, according to Katz’s preferred ontology) is
causally inert.” Katz indeed insists that abstract objects are causally inert, but his
insistence is grounded on his assumption that causation is an empirical affair, so this
insistence cannot amount to an argument for the assumption. Such an assumption is
entirely consistent with a Kantian or a Humean epistemology, but it is not just clearly

correct, and Katz nowhere argues for it, despite the fact that he explicitly rejects both

 

22 I hesitate to say that it is supersensory awareness for reasons that shall become clear presently.
2’ Or that the notion of cause does not apply to reality as it is in itself.

116

Humean and Kantian epistemology. He takes other rationalists to task for offering an
essentially empiricist account of a priori awareness by offering a causal account, but this
accusation rests on his assumption, and while it sometimes seems clear that he has
successfully indicted rationalists for sliding over into empiricism, it is not clear that a
causal account of a priori awareness must be empiricist by deﬁnition. For instance, Katz
accuses Plato of proposing a causal pre-natal account of a priori awareness in the Meno,
but later notes that Plato “characterized the Forms as atemporal, incorporeal, nonsensible,

and transcendent.”24

If they are non-sensible, then why is a causal account of contact
with the forms empiricist (or even coherent, given Katz’s insistence that a causal account
is a sensory account)? The point is that it is not clear why or whether we should regard a
causal account of a priori awareness as necessarily an account rendered in terms that
surrender to the empiricist.25

Katz does, however, also rightly criticize empiricists who object to the very idea
of a priori awareness on the grounds that such awareness would require a sort of mental
power that seems almost magical, if not mystical, in its operation. Katz quotes Dale
Gottlieb as saying that “abstract entities are mysterious and must be avoided at all costs,”
Hartry Field as saying that the rationalist must “postulate some aphysical connection,
some mysterious grasping” and Charles Chihara as saying that appealing to rational

intuition is “like appealing to experiences vaguely described as ‘mystical experiences’ to

justify belief in the existence of God.”26 Katz’s reply is that something’s being

 

2‘ Katz 2000, p. 124; italics added.

25 I shall return to this issue in Chapters Seven and Eight.

26 Katz 2000, p. 32. I have often heard it said that realists are really just theists in disguise, that they harbor
a “desire for God,” or wish to preserve the idea of a “God’s eye view” or a “view from nowhere” that
would underwrite the comforting thought that nature has a plan, or a design, or a rationally discemable
order, that can be surveyed from the appropriate height. (Maria Lugones 1994 adds one more: the realist,
or at least the monist, is a “lover of purity.”) I will not deny that there are similarities between some

117

mysterious is not the same thing as its being mystical. A priori awareness is mysterious,
to be sure, but philosophy is, as Katz points out, “full of such mysteries.” The charge of
mysticism, he says, is unwarranted, because mysticism “involves the claim to have a
means of attaining knowledge beyond our natural cognitive faculties,” and “those who
cry ‘mysticism,’ ‘superstition’ and the like perhaps need to be reminded of the fact that
our sensory faculties do not exhaust those faculties.”

This is altogether too easy. Presumably, empiricists would admit that we have
mental ratiocinative faculties, but this does not settle the matter. They do not need to be
reminded that we have faculties other than our ﬁve senses and our ability to calculate,
they need to be convinced that we do. This is precisely why they cry “mysticisml”: they
maintain that to claim any sort of awareness beyond our ﬁve senses and our ability to
calculate is to claim that we have “a means of attaining knowledge beyond our natural
cognitive faculties.” Moreover, the fact that philosophy is full of such mysteries is cold
comfort by itself; some philosophical “mysteries” really do just consist of poorly framed
questions about the natural world that science could handle if properly understood. Why
should we have conﬁdence that the idea of a priori awareness is not like this? Katz’s

argument is far too thinly developed.

 

versions of realism and some versions of Jewish and Christian theology. Nevertheless, this is not an
argument against realism but a diagnosis of the realist that is meant to explain how he or she has been led
so far from good sense. The idea is that the realist wants something that sophisticated and advanced
contemporary thought shows cannot be had, and that the realist is just hanging on to this outmoded world-
view because it seems comforting, despite the fact that he or she would be better off facing up to the
indifference of the universe and growing out of this attachment to the sort of superstition that was the
hallmark of humanity’s childhood. This is, of course, the old trick of defeating your opponents by
declaring them psychologically deviant and in need of therapy, not further arguments. The assumption
seems to be that the arguments against the possibility of realism are so telling that only irrational persons
could fail to be persuaded by them. As I have shown, however, the arguments are neither telling nor
persuasive.

118

It can, however, be improved. Firstly, empiricists who say such things as that
rationalists are mystics or must appeal to some sort of magical mental grasping are
making rationalism out to be a straw man. Not one of the most well-known rationalists
holds such a ridiculous position. Plato seems to have assumed that there was an already
established correspondence between mind and reality (a la Parmenides). Aristotle may
have been the ﬁrst proponent of representationalism, and most rationalists since then
have cloven to something like representationalism, which is hardly mysticism, whatever
its other faults may be. It is, in fact, only in the wake of the Humean arguments against
the possibility of representationalism being underwritten by a secure metaphysics that
empiricists have begun to regard a priori awareness as “mystical.”

Secondly, to say that the sort of connection to reality required by a priori
awareness is magical or smacks of superstition or seems reminiscent of revealed
knowledge of God’s existence is merely rhetorical grandstanding on the part of the
empiricist. The idea seems to be this: if you think that a priori awareness is possible, you
must think that we possess cognitive faculties that are capable of delivering information
about reality other than our ﬁve senses and our ability to calculate; but this is radically
out of joint with the current scientiﬁc consensus, and seems reminiscent of pre-scientiﬁc
superstitions involving talk of occult “mental powers.” Thus, rationalists are not really
participants in contemporary philosophical discourse about mathematical knowledge, and
rationalism is clearly not a serious candidate for an account of mathematical knowledge.
Perhaps we can’t even talk to them. Of course, it doesn’t matter whether rationalism
appears similar to talk of occult mental powers, or even if it derives from such talk. The

question is what rationalists can do with it now.

119

On the other hand, the empiricist is entirely right to say that the rationalist must
produce a comprehensible account of a priori awareness. The empiricist’s guess is that
the rationalist cannot produce a comprehensible account, since the only modes of
awareness we understand are our ﬁve senses and our ability to calculate, and it’s hard to
see what else there could be (mind rays?). This is not, as in the case of the principle of
non-contradiction, however, hard to see because it is in fact impossible. There is no
metaphysical or logical or grammatical incoherence or impossibility in the idea of
additional means of awareness. I concede that the idea of non-sensory awareness of
aspects of reality seems unlikely (to say the least) and inexplicable if we assume a
pluralist metaphysics. But then, if rationalism seems to be the best way to account for the
fact of necessary truth, perhaps we should reconsider our metaphysics. I have already
offered an argument against representationalism, but further development of this
suggestion will have to wait until chapter eight. I will there try to make the possibility of
a priori awareness seem much more plausible than it now does. For now, we should turn

to Katz’s account of a priori awareness.

Katz’s account of a priori awareness is profoundly unenlightening. He appeals to
Godel, who says that “mathematical intuition cannot be associated with actions of certain
things upon our sense organs, or with something purely subjective, as Kant asserted,” and
that rational intuition “may be due to another kind of relationship between us and

reality.” This is, frankly, lame. I use the word in the clinical sense. To begin with, the

120

“may be due” portion of Godel’s statement is a bit distressing. As rationalists, we like to
just come right out and say “is due.” What is worse, however, is the characterization of
our non-sensory relationship to reality: there is none. This is tantamount to Einstein
saying that since Newtonian mechanics does not account for the invariance of the speed
of light, some other theory of space and time must account for it. Katz’s realistic
rationalism is long on realism but short on rationalism. A number of problems ﬂow from
this gap in Katz’s position, and although it seems like a colossal oversight, we can see
why Katz thinks that it is not.

To begin with, though Katz insists that a Kantian rationalism will not do, because
Kant’s psychologistic rationalism only explains why we take some truths to be
metaphysically necessary, not why they are metaphysically necessary, his failure to
provide a fuller account of a priori awareness and a secure metaphysics leave him
scarcely more able to explain the metaphysical necessity of truths than Kant. After all,
without any assurance that his speculative metaphysics is accurate, it is open to the
empiricist to claim that our knowledge of elementary mathematical and logical
statements consists of our treatment of the statements as axiomatic, and their ranging over
constructed abstract entities. Without an account of a priori awareness, we do not in fact
know that elementary mathematical and logical statements are metaphysically necessary,
and Katz’s realistic rationalism at best explains why taking them to be metaphysically
necessary would be preferable to taking them to be axiomatic.

This, however, is because, as was noted earlier, it is Katz’s view that the job of
epistemology is to explain, not the possibility of knowledge generally, or even the

possibility of knowledge in the sciences, but the fact of knowledge in the sciences. This

121

is why Katz does not trouble himself overmuch about providing an account of a priori
awareness, and why he does not even address the question of Cartesian skepticism until
nearly the end of Realistic Rationalism: he thinks that we cannot help but beg the
skeptical question in the philosophy of science, and that the dispute between rationalism
and empiricism is simply a dispute over which account of scientiﬁc knowledge best ﬁts
the facts. This severely limits both the scope and the value of Katz’s rationalism.

If, according to Katz, a particular realism is “an ontological position in the
foundations of a particular formal science” — namely, the ontological position that the
domain of that science is populated by abstract objects — what is general realism? Katz
says that “general realism is realism in general. It makes the indeﬁnite claim that there
are abstract objects.”27 If this is right, then there is no hope whatever of extending Katz’s
realistic rationalism to cover the whole of philosophy. As Katz himself points out,
“philosophy can be about anything.”28 Katz’s claim about general realism is false, as is
his view of epistemology generally. Philosophy can be about anything, and is in fact
about everything. What we have seen in previous chapters is that epistemology is
concerned, at bottom, with the question of how we can know anything at all, and
metaphysics is concerned with the nature and structure of reality as it is in itself — that is,
with ultimate ontology. Their relationship is that epistemology explains how it is
possible for us to have knowledge of reality — scientiﬁc knowledge, but also metaphysical
knowledge generally — not merely knowledge of abstract objects in a formal science
(unless it can be shown that all metaphysical knowledge reduces to such knowledge, but

Katz does not undertake to show this). As we saw in the last chapter, traditional

 

2’ Katz 2000, p. 2.
2‘ Ibid., p. 190.

122

metaphysical questions cannot simply be dismissed as meaningless or unintelligible, and
questions such as “Why something rather than nothing?”, “What is Being?”, “Is reality
plural or singular?” seem to be clear candidates for metaphysical inquiry.

Katz’s only reply, it would seem, would be to say that we might be able to get
closer to answering these questions by examining their linguistic structure. Katz wisely
does not subscribe to representationalism as his epistemological account of a priori
awareness, and he rejects Kantianism, but he shifts the fundamental metaphysical and
epistemological issues by missing the fact that representationalism has been the problem,
and by picking up where Frege left off, with an ontology of non-Fregean intentional
objects. Katz’s intentional objects are non-Fregean because they play no role in a theory
of reference. Katz notes that this move enables him to evade the indeterminacy
arguments of Quine and Wittgenstein, since these arguments are aimed primarily at
Fregean intentional objects and the role they play in the theory of reference. What Katz
doesn’t seem to have noticed, however, is that Frege tied his theory of senses to the
theory of reference in order to reduce epistemology to the philosophy of language — in
order, in other words, that logicism could be a successor theory to Kantian epistemology.
By snipping the connection between sense and referent, Katz likewise snips the
connection between his formal scientiﬁc ontology and our knowledge of the world. It is
not clear, for instance, how formal knowledge in the “foundations” of mathematics serves
to underpin natural knowledge in mathematics proper, since Katz nowhere thinks to say

how tokens of numbers are tokens of their formal ontological types.29

 

‘9 The fact that abstract and concrete objects co-exist in the same world, as Katz hastens to insist, does
nothing to solve this problem. Indeed, it makes the problem worse, since, if abstract objects co-exist with
concrete objects, one will want to know: where, exactly, are these things? It is precisely to avoid this sort

123

Even though Katz’s view of linguistics as a formal science is that its domain
consists of abstract (linguistic) objects, this proposal strikes me as scarcely an advance
over linguistic analysis, or the idea that philosophy does not explain the possibility of our
knowledge or produce new knowledge, but merely helps us to clarify what we already
know. Sufﬁce it to say that I ﬁnd it highly unlikely that, if there is a fact of the matter
concerning questions such as “Is reality singular or plural,” the facts consist of facts
about the language in which I ask it. Thus, as it stands, Katz’s realistic rationalism is not
a particularly good candidate for a speculative metaphysics, and he does not offer to
secure it.

Katz acknowledges that nothing he has said in Realistic Rationalism secures his
epistemology from attack on the grounds that it does not secure access to the ontology he
propounds, but he offers some considerations that are meant to meliorate this situation.
For one thing, he says, the empiricist, appearances aside, is not in any more secure a
position with respect to skepticism than is the rationalist. It may seem that by turning
their backs on the skeptic, empiricists have defused the skeptical argument, but this is not
so. We have already arrived at this same conclusion, inasmuch as it is incumbent upon us
all to provide a comprehensive metaphysics. Katz, however, also points out that “the
Humean [skeptical] problem resurfaces as a problem of what right we have to think that
[our] inductive practices are reliable in the long run.”30 This is an important point. He
adds that while we cannot answer the skeptic, we are at least on the right track, since
rationalism is the only epistemology that can have any hope of answering the Cartesian

skeptic. He points out that “Cartesian and Humean skepticism show that, if there is an

 

of problem that Plato is sometimes taken to have postulated a “separate realm” for the forms. The idea is
that if forms are atemporal, non-spatial, and eternal, then they do not exist in our phenomenal world.

124

argument that can meet Cartesian or Human skepticism, it cannot be one based on
experience.”3 1 This is also an important point.

In the end, however, Katz contents himself with the following. He says, without
offering any argument, that to answer the Cartesian skeptic, “we would have to prove the
existence of God.”32 Since this very probably cannot be done, we’ll need another
approach, and Katz offers no reasons for thinking that there isn’t one (other than that
answering the skeptic will be very, very hard, if it even can be done). The fact that if we
go down, we all go down together, empiricists and rationalists alike, is, however, only
satisfying if we are already convinced, as Katz is, that the important questions are the
“internal” questions of whether our standards support the claim that we have knowledge,
and what the best account is of these standards. This is not the basic epistemological
issue. We feel conﬁdent that we have knowledge, of course, but to the extent that this
conﬁdence rests on our own standards, then our conﬁdence amounts to little more than
our feeling pretty darn proud of ourselves. The epistemological question is: how is it
possible for us to have any knowledge at all? Even Descartes thought it unlikely that it

would turn out to be impossible to rule out the existence of an evil deceiver who

 

3° Katz 2000, p. 189. See footnote seventeen, above, for a discussion of a similar point.

3' Ibid., p. 203.

32 Ibid., p. 204, n. 4. Katz observes, however, that Robert Nozick has shown that even establishing the
existence of a benevolent God would not do the trick. Nozick (1981) argues that even if we were created
by a benevolent God, so long as we hold God’s motives to be inscrutable (as Descartes does), we cannot
know that such a good God does not “have his own reasons for deceiving us.” God might, Nozick claims,
“deceive us temporarily” for “a period which includes all of our life thus far (but not the afterlife),” or
might even “be motivated to allow another powerful ‘demon’ to deceive and dominate us” (p. 202). These
do indeed seem to be legitimate possibilities. I submit, however, that if we could demonstrate the existence
of a benevolent God, then we would have reason to believe that any divine deception must serve a long-run
(benevolent) purpose, and that this would, in itself, be enough to defeat Cartesian skepticism —— though the
intended certiﬁcation of our knowledge of the external world might in this case be ultimately disappointed.
Should we lose heart? I don’t see why we should. Nozick has here simply demonstrated the extent of the
difficulty of defeating Cartesian skepticism with respect to our knowledge of the external world.

125

systematically distorts our sensory awareness. Are we right to feel so conﬁdent? Why?

I hope to draw nearer to answers to these questions in the next two chapters.

126

CHAPTER 7

LAURENCE BONJOUR’S DEFENSE OF PURE REASON

Though J. J. Katz’s realistic rationalism did not provide us with a secure
metaphysics (or even an adequate speculative metaphysics) and an acceptable
rationalism, Katz’s arguments did provide us with (at least partial) answers to the
empiricist arguments against the actual existence of any a priori awareness or
knowledge. Some of these answers require further development, but the beginnings of an
answer to each of the three empiricist arguments1 are now available, and I take the ﬁrst
dispensability argument to have been answered pretty decisively in favor of rationalism.
Laurence BonJ our defends what he calls a “moderate rationalism,” in the course of which
defense he offers arguments that, as we shall see, will assist us in reﬁning and reinforcing
our answers to the inductive and the second dispensability arguments. BonJour’s
moderate rationalism will, as we shall also see, prove to be only somewhat more
satisfactory than Katz’s realistic rationalism -— which is to say not very — and it will not
meet our demands for an acceptable epistemology. BonJour’s rationalism will, however,
point the way to invaluable insights regarding the form that any acceptable epistemology

must take.

 

' As a reminder, the three empiricist arguments against the actual existence of any a priori awareness or
knowledge are: the inductive argument that many claims to a priori awareness or knowledge have
historically turned out to be mistaken; the ﬁrst dispensability argument that empiricism can perfectly well
account for the apparent necessity of the truth of some statements; and the second dispensability argument
that an acceptable empiricist account of human knowledge can be offered that need not ever appeal to a

127

BonJ our, in his In Defense of Pure Reason, offers what might seem to be a
surprising reply to the empiricist’s inductive argument against the actual existence of any
a priori awareness or knowledge. He claims that a priori awareness or knowledge is
fallible and corrigible, and thus not immune to rational revision. This may seem
somewhat surprising for two reasons. Firstly, as BonJour himself observes: “it is a
familiar fact that an overwhelming majority of the historical proponents of rationalist
conceptions of a priori justiﬁcation and knowledge regarded such knowledge as certain —
where the primary content of the notion of certainty was that a proposition that is justiﬁed
a priori, via rational insight, cannot fail to be true, that a priori justiﬁcation is
infallible.”2 Secondly, if a priori awareness or knowledge is fallible and corrigible, then,
in the absence of any means of knowing when we are genuinely in possession of a priori
knowledge, no claim to such knowledge could ultimately serve the function of securing
our epistemological enterprise. This is indeed a worry, and BonJour does not, in the end,
do enough to address it. We shall see, however, that the fallibility and corrigibility of
(some) a priori awareness is neither so serious a problem as it would appear to be, nor so
surprising as it might at ﬁrst seem. In fact, the fallibility and corrigibility of a priori
awareness go a long way toward explaining the apparent failure of rationalism
historically, and also serve to illuminate the differences between a priori belief that is
open to correction and belief that is not.

BonJ our acknowledges that there are a number of apparent counterexamples to
the supposed infallibility of a priori knowledge or awareness. In fact, he says that there

are “at least three classes” of such counterexamples: the historical examples of apparently

 

priori awareness or knowledge of any kind whatever. See Chapter Six for a development of these
arguments in somewhat greater detail.

128

necessarily true statements that were taken or claimed to be known a priori that turned
out to be false; the claims to indubitable a priori knowledge on the part of rationalists
down through the ages that have conﬂicted with one another; and the ordinary errors in
judgment and calculation to which any of our beliefs is liable, including those that are
supposed to be the result of direct rational insight or awareness (or the result of
inferences or calculations involving the use of such supposed insights).3 BonJour further
admits that we must, of course, account for these apparent counterexamples to preserve
rationalism against the empiricist’s inductive argument. What BonJ our proposes,
however, is that we break with the traditional view that a priori awareness is infallible,
and adopt a more moderate rationalism, according to which a priori awareness or insight
is both fallible and corrigible.

There are two problems with this proposal. Firstly, it does not initially seem to be
any more promising than Katz’s reply to the inductive argument. As we have seen, the
problem is not that it is not open to us simply to say that statements that appeared to be
knowable a priori but turned out to be false were not really cases of a priori knowledge
in the ﬁrst place; the problem is what we should do about the consequences for
rationalism once we have said this. We glimpsed some of these consequences in the last
chapter, and have already the beginnings of a solution. BonJour does not seem, however,
to appreciate the potential gravity of these consequences, and seems to think that the
adoption of a more moderate rationalism by itself constitutes the bulk of the answer to the
empiricist’s inductive argument. BonJour does address some of the difﬁculties generated

by the concession that not all apparent instances of a priori awareness will turn out to be

 

2 BonJour 1999, p. 110.
3 Ibid., p. 111.

129

actual instances of a priori awareness, though his discussion, like Katz’s, needs
elaboration. We will consider BonJour’s attempt to address these difﬁculties shortly. Let
us, however, ﬁrst turn to the other problem with BonJour’s proposal.

BonJ our asserts that we must “concede, contrary to the main historical tradition,
that what appears subjectively to be [an a priori] seeing or grasping or apprehending may
fail to be one, most strikingly in the case where the proposition that seemed to be
necessary turns out to be false.”4 This of course implies that, according to the main
rationalist tradition, what appears subjectively to be an a priori seeing or grasping or
apprehending or whatever is (always?) in fact objectively such an a priori seeing or
grasping or apprehending or whatever. The problem with this claim is that it is simply
false. Few if any among even the most strident rationalists held such a view of a priori
awareness. Descartes did not. He held, famously, that beliefs arrived at through clear
and distinct perception are infallible, because God would not allow us to be deceived
about anything we could perceive clearly and distinctly. Even so, Descartes allowed that
we might be mistaken as to whether a particular perception is clear and distinct, and that
careful direction and training of one’s intellect are required to eliminate these sorts of
errors of judgment. He observes that “the giving of our assent to something unclear
happens because we imagine that we clearly perceived it on some occasion,” and adds
that “we must concentrate our mind’s eye totally upon the most insigniﬁcant and easiest
of matters, and dwell on them long enough to acquire the habit of intuiting the truth
distinctly and clearly.”5 Indeed, the mere fact that Descartes singled out one particular

sort of cognition — clear and distinct perception — as constituting genuine rational

 

" Ibid., pp. 15-16, n. 18.
5 Descartes 1988, pp. 174 and 171.

130

intuition (as opposed, presumably, to apparent rational intuition) suggests that he was not
only aware of such a distinction but endorsed it. There are a number of famous
difﬁculties associated with how we are to distinguish clear and distinct perception from
other sorts of cognition, but the point is that Descartes proposed a criterion of
genuineness for a priori awareness, which suggests that he knew that he needed one, and
that he held that not all apparent a priori awareness is actual.

Plato, too, endorsed the distinction. He contents that there is genuine knowledge
to be obtained about the nature of reality as such, and that this sort of knowledge can be
achieved only through dialectic, if at all (and not through empirical observation as such),
which is concerned with the intelligible as opposed to the sensible. He notes, however, in
many places in the dialogues how difﬁcult such knowledge is to obtain. A number of
dialogues end in aporia — that is, in confusion —- without a positive result. Moreover, in
the Republic, Plato remarks that the highest knowledge to which one can aspire, after
metaphorically leaving the cave of the shadows of the merely sensible — namely,
knowledge of the purely intelligible —— is at ﬁrst no more certain and indubitable than is
the perception we enjoy on ﬁrst emerging into the sunlight after so many years spent in
darkness. Plato says of the person who has just left the cave, that “he’d need time to get
adjusted before he could see things in the world above,” and he makes it plain that this
also applies to our journey beyond the bounds of sense.6 “In the knowable realm, the
form of the good is the last thing to be seen, and it is reached only with difﬁculty.”7 Plato
does not think that everything we believe we know about reality as such is genuine

knowledge, and he plainly thinks that genuine knowledge in this area is hard to come by,

 

‘5 Plato 1997b, 516a.
7 Ibid., 51715.

131

and requires years of careful thought, reﬂection and attention (not to mention dialectic!)
to obtain.

Indeed, I suspect that the rationalist tradition is rather one of moderate rationalism
than of radical rationalism, as BonJ our would have it. The only obviously plausible
reasons anyone could even have for thinking that all apparent cases of a priori knowledge
are bona ﬁde cases of such knowledge, and that we could know that this is so, are that (1)
our mental states and their contents are entirely transparent to us, including their
inferential relations to one another, and that (2) the faculty we possess for generating
such states never errs when reality as such is the only matter under consideration.8 These
are pretty ambitious expectations for any epistemologist to have regarding human
cognitive faculties, and the only apparently likely candidate for a rationalist whose
expectations are this loﬁy would be Descartes, inasmuch as he is committed both to
cognitive transparency and to the infallibility of clear and distinct perception. But, as we
have seen, even Descartes does not accept (1) and (2). Thus, BonJour’s moderate
rationalism should seem neither surprising nor novel.

It may fairly be argued that rationalists have often failed to devote much of their
attention to developing the distinction between merely apparent a priori awareness and
genuine a priori awareness, but this may be, as BonJour notes, partly due to the fact that

the existence of a priori awareness was largely taken for granted by most philosophers

 

8 BonJour curiously does not even consider the holding of these reasons by other rationalists to be a
possible explanation for what he takes to be the historical consensus that a priori awareness is infallible.
Rather, he suggests that perhaps it was the fact that a priori beliefs are often beliefs about necessary truths
that led rationalists to think that we could not be wrong about them. This is at least plausible, if we add, as
BonJ our does not, that perhaps it was thought that beliefs about necessary truths could not be mistaken
because such truths, and hence the truth of the belief that they are true, is guaranteed. Otherwise, this
confusion would indeed involve a gross fallacy, as BonJour puts it, but this alone should seem to be enough
of a reason to doubt the explanation. BonJour simply notes that while “such an explanation of a deeply

132

until Hume,9 so the absence of such a distinction was not regarded as a matter of serious
concern. Moreover, rationalists’ accounts of the proper means of attaining certain
knowledge (and of correcting cognitive errors) could be taken to supply all that is needed
in the way of a distinction between apparent and genuine a priori awareness. Indeed, this
shall be our contention here: that once we have provided for the possibility of the purely
a priori correction of a priori belief, and for the possibility of the identiﬁcation of some a
priori beliefs that are not revisable, we will have answered the empiricist’s inductive
argument. Let us return, then, to BonJour’s discussion of both the fallibility and the
corrigibility of a priori beliefs.

As we saw in the last chapter, Katz insists that genuine a priori awareness or
belief can be neither challenged nor refuted by empirical considerations. Of course, he
concedes that what was thought to be known a priori, but is in fact a posteriori, can be
challenged and refuted by empirical considerations, but this does no damage to
rationalism as such.10 What does threaten rationalism, however, as neither Katz nor
BonJ our seem sufﬁciently to appreciate, is the fact that, in such cases, the statement in
question that was thought to be known a priori seemed to us to be a priori, and was in
other respects indistinguishable from other supposedly genuine instances of a priori
knowledge or belief. Concerning the (apparently empirical) defeat of Euclidean
geometry, Hilary Komblith observes that “the Newton-Einstein example is a clear case of

empirical theorizing forcing a rational change in a belief which earlier had all the

 

entrenched historical claim is unsatisfying,” he has “no better account to offer” (1999, p. 111). Of course,
since there is no such historical consensus anyway, there is nothing that really needs explaining.

9 BonJour 1999, p. 17.

'0 Excepting the erosion of conﬁdence in rationalism, and in a priori awareness generally, provoked by
salient instances of the abandonment of beliefs once thought to be a priori. I shall deal with this difﬁculty
shortly.

133

hallmarks of a priority.”ll If Katz is right in his claim that it should have been obvious,
even in Kant’s time, that Kant’s epistemology did not really enable him to show that
anything is a priori known to be necessarily true, then Komblith is wrong about this.
This is encouraging, given the salience of the example. Still, the question is whether
claims such as Katz’s are generally or always true of cases of failed claims to a priori
knowledge, since this would mean not only that no necessary truths had turned out to be
false but also that no a priori awareness was actually involved (or overturned by
experience). What reasons do we have then, if any, for thinking that genuine a priori
belief cannot be challenged or refuted by experience?

Drawing upon Philip Kitcher’s discussion of a priori belief in his The Nature of
Mathematical Knowledge, BonJ our considers three possibilities: an a priori belief could
be directly challenged by experience, theoretically challenged as a result of observation,
or socially challenged by the relevant professional specialists (mathematicians,
physicists, etc.) who reject it for reasons internal to their practice. BonJ our begins by
noting that in none of the standard cases is the purported a priori knowledge directly
challenged or refuted by experience or observation. It’s not as though we just look and
see that Euclidean geometry is not the science of actual space. What we see are a series
of experimental results or empirical observations that, when coupled with our predictive
expectations or certain other statements that form a part of our physical theory, suggest
that our theory of actual space (or some part of it) is wrong. This, however, suggests that
any theoretical or social challenge to our a priori belief posed by experience is indirect at
best, and is in fact only prompted by the apparent conﬂict between observation and

certain other theoretical commitments we already have. BonJ our points out that the

 

“ Komblith 2000, p. 76.

134

conﬂict is only generated by an inference from our observation, and that “any such
inference would have to rely, tacitly if not explicitly, on some underlying premises or
principles of inference connecting the experience in question with this further result,”
these premises or principles of inference themselves being, in the end, a priori.'2
Observation gives us reason to think that (what we take to be) an a priori belief is false
only in cases where the belief in question was not really a priori to begin with; in other
words, genuine a priori awareness cannot be refuted or even challenged by empirical
observations, because if it were, it would not be genuine a priori awareness after all.
Ultimately, logical and epistemic principles that are themselves known a priori are
revisable (if at all) only in light of further a priori considerations. So long as at least
some of our a priori beliefs can be demonstrated to be unrevisable, we need not worry
that some of our other (putative) a priori beliefs may require revision in light of their
coherence with certain of our empirical beliefs. It should still be troubling that certain
beliefs that we had good reason to think were a priori turned out not to be a priori at all
(and that still others may yet turn out not to be a priori as well), but at least in such cases
we can be assured that no genuine a priori belief was overturned. Of course, the
rationalist still owes the empiricist an argument for the claim that we do in fact know at
least some logical and epistemic principles a priori, but we shall return to this issue
below in the context of our discussion of the second dispensability argument. We must
ﬁrst address the question whether any of our a priori beliefs is in fact immune from

revision. Even if our a priori beliefs are revisable, ultimately, only in light of other a

priori beliefs, however, at least some of them must be entirely immune from revision if

 

'2 BonJour 1999, p. 122.

135

our metaphysics is to be anything other than speculative, and if our epistemology is to be
comprehensive.

Kornblith observes, quite correctly, that

once the real possibility of error is recognized even for a priori claims, these too
stand in need of justiﬁcation by integration with one’s other beliefs.” He goes on
to say that “the claims of reason no longer stand somehow outside the web [of
belief], adjudicating once and for all what inferential moves are legitimate.
Instead, they too are brought down to earth, forced to earn their epistemic worth
just like any other claims, through integration with our total body of beliefs.13

In order for the appeal to a priori awareness or knowledge to accomplish its task of
accounting for the very possibility of knowledge, and thus for our means of awareness of
the furniture of the universe, at least some of our a priori awareness — namely, the portion
that consists of the account of a priori awareness itself — must be impervious to error.
What is needed, in short, is some sort of foundationalism. Katz rejects this claim because
he maintains that we need not answer the skeptic to claim that we have any knowledge at
all, or any knowledge in the sciences or the foundations of the sciences. As we have
seen, however, the basic epistemological question is not whether we have any knowledge
at all, but is rather how knowledge is possible. If we fail to answer this question, or are
incapable of answering it, then our evolving cognitive project can have no claim to the
title “knowledge,” our license to call any of our beliefs “knowledge” must be revoked,
and the skeptic carries the day. The possibility of answering the basic epistemological
question, however, is still open as things stand, and it is not answered by noting that we

have knowledge in the sciences and the foundations of the sciences and by providing a

 

'3 Komblith 2000, p. 82.

136

plausible account (even the maximally plausible account) of what we can take scientists,
mathematicians, linguists, and so on to be talking about when they practice their crafts.
BonJ our seems to acknowledge, in some places, the need for a foundational
account of a priori awareness, but he nowhere offers one in In Defense of Pure Reason.
Quite apart from everything I have just said, there is another entirely conclusive reason
why an account of a priori awareness or knowledge (or belief or justiﬁcation) must be
incorrigible and indefeasible: a system of a priori belief cannot be justiﬁed in virtue of its
internal coherence, because “any conception of coherence, however restricted, will
presuppose certain fundamental premises or principles that deﬁne the conception in
question and cannot be assessed by appeal to it.”14 BonJour recognizes that any account
of a priori awareness must be foundationalist, but he seems to think that his own account
is foundationalist. He says that “a thoroughgoing coherence theory of a priori
justiﬁcation and knowledge is impossible, so that a theory of a priori justiﬁcation must
be, like the present one, essentially foundationalist in character.”ls
It is not at all clear, however, why he takes his account in In Defense of Pure
Reason to be foundationalist. Although he couches his account of a priori awareness in
terms of a system of a priori beliefs, he nowhere identiﬁes any speciﬁc a priori beliefs
that we should take to be incorrigible. It is true that he often argues that logical principles

should be taken to constitute a priori knowledge, and he argues that no argument could

possibly proceed without them, but he does not say precisely which logical principles are

 

'4 BonJour 1999, p. 118. This is similar to Katz’s dictum: “There is no actual epistemology that says that
everything including itself is revisable” (2000, p. 74). Closer attention to his own dictum should have
prompted Katz to modify and further develop his own account of the rational revisability of a priori beliefs.
5 Ibid. For another argument to the effect that any coherence theory of knowledge or justiﬁcation must, in
the end, rest or rely upon a priori principles that constrain or require its long-run stability, see BonJour
1985, pp. 169-188.

137

incorrigible, if any. As Komblith and BonJour himself both point out, even if they are
only revisable according to their coherence with other a priori beliefs, the fact that their
warrant remains strictly a priori16 does not by itself entitle them to act as the adjudicators
of all belief. Their coherence with our empirical beliefs will, in that case, also be a
factor, and we could never have any reason in principle for preferring a revision of our
total system that preserves a priori beliefs over one that preserves empirical beliefs.
Moreover, BonJour does not, in the course of his discussion of the fallibility and
corrigibility of a priori beliefs, have anything to say about which a priori beliefs are
incorrigible, or how we could identify them.

BonJ our does note the difﬁculty in identifying them, and in distinguishing
between merely apparent and genuine a priori belief generally. Presumably genuine a
priori belief would be incorrigible; the question is how we know when one of our
apparent a priori beliefs (or our a priori awareness) is genuine. “If rational insight is
indeed fallible,” BonJ our observes, “then it is natural to think that some ﬁrrther,
epistemically prior criterion or standard is needed in order to distinguish genuine rational
insights from merely apparent ones.”17 The problem with this, as he immediately makes
clear, is that “the need to appeal to such a criterion would deprive a priori insight of most
or all of its cognitive value.” This is because “any such criterion or standard would itself
have to be somehow justiﬁed; and only a little reﬂection will show that there is no
possible way in which it could be justiﬁed without either impugning the a priori status of
the claims that are justiﬁed by appeal to it (if it is justiﬁed empirically) or else being

guilty of obvious circularity (if it is justiﬁed a priori)?’ There can be no useful epistemic

 

'6 Or au fond a priori, at least.
'7 BonJour, p. 115.

138

criterion that enables us to distinguish genuine from merely apparent a priori awareness.
What is needed is a secure metaphysics that would underwrite both the possibility (and
actuality) of incorrigible a priori awareness, and our comprehensive rationalist
epistemology along with it. BonJour does not offer a secure metaphysics in In Defense of
Pure Reason, though he does offer a speculative one. I shall consider this speculative
metaphysics below. We will not have provided a thorough or a satisfactory answer to the
empiricist’s inductive argument until we have produced a secure metaphysics. This
discussion must, however, wait until we have considered BonJour’s reply to the second
dispensability argument.

BonJ our offers no reply to the ﬁrst dispensability argument, although it may be
that he thinks he doesn’t need one. He does not argue, as does Katz, that radical
empiricism a la Quine is incoherent and therefore incapable of accounting for the
apparent necessity of certain truths. Rather, he argues that Quine’s empiricism is wholly
inadequate as an epistemology, which would presumably imply that it is not a serious
contender as an account of apparent necessary truth, or of knowledge generally.’8 Thus,
he proposes, in effect, to answer the ﬁrst dispensability argument by answering the
second. He begins by noting that Quine’s web metaphor, as it is presented in “Two
Dogmas of Empiricism” begs the question against the rationalist and against the

possibility of a priori knowledge.

 

'8 BonJour argues that Quine’s epistemology is wholly inadequate, however, because it leads immediately
to a nearly total skepticism, though he considers the possibility that Quine would not have shrunk from this
result, and may even have intended it. In that case, however, it could be argued that skepticism is the best
we can do, and though Quine’s epistemology would be disappointing, it would not be unacceptable.
Moreover, it might then offer the only available and plausible account of apparent necessary truth, so that
showing that Quine’s position leads to skepticism answers neither dispensability argument, strictly
speaking. It seems to me beyond question that Quine did not intend his epistemology to be skeptical,
though of course it might be. BonJour’s argument, though, even if it is correct, only does damage to

139

What follows from the Duhemian view is only that the revisions prompted by
recalcitrant experience need not be conﬁned to the observation periphery [of our
web of beliefs], that is, that the demands of experience can equally well be
satisﬁed by revisions in the non-observational interior, so that there can be no
experiential test of a single sentence in isolation. But to conclude from this that
any sentence can be rationally given up it must be assumed that epistemic
rationality is concerned solely with adjusting one ’s beliefs to experience Thus
Quine’s main argument in “Two Dogmas of Empiricism” against the possibility
of a priori justiﬁcation and knowledge turns out to be totally lacking in force: it
reaches the conclusion that there is no a priori justiﬁcation only by adopting a
conception of epistemic rationality that already tacitly assumes that this is so.19

Hilary Komblith, in his “The Impurity of Reason,” insists that this is not “the right way to
understand the Quinean move.” Komblith observes that “the web metaphor is designed
to show that there is available a way of looking at things which makes sense of a
thoroughgoing rejection of the a priori.”20 Komblith is entirely right to say that the web
metaphor is intended to constitute an empiricist epistemology that accounts for human
knowledge and requires no appeal whatever to a priori awareness or knowledge. On the
other hand, BonJ our is also entirely right to say that the metaphor is question-begging as
an argument for empiricism over rationalism or as an argument against the very
possibility of a priori awareness. The metaphor can only serve as an argument for
empiricism and against rationalism insofar as it constitutes an elaboration of a potentially

plausible empiricist epistemology, but to that extent it assumes empiricism (as, of course,

 

Quine’s position if it can also be shown that skepticism is avoidable, and that, of course, can only be shown
by means of the production of a secure metaphysics.

'9 BonJour 1999, p. 76. Of course, Quine’s argument in favor of universal revisability in “Two Dogmas of
Empiricism” is also presumably buttressed by his arguments against the possibility of drawing a distinction
between analytic and synthetic statements, in addition to the support it receives from Duhemian holism.
The two arguments are intended jointly to entail universal revisability. Even taken together, they don’t
entail universal revisability, however, because Quine’s arguments against the analytic-synthetic distinction
do not tell against the possibility of a priori knowledge, as we saw in Chapter Three.

2° Komblith 2000, p. 74.

140

it would).21 Likewise, it cannot possibly operate as an argument against the very
possibility of a priori awareness without either begging the question or undermining
itself in the attempt, as we saw in chapter three. Komblith is right to say that Quine’s
epistemology is, in part, an attempt to make sense of “a thoroughgoing rejection of the a
priori,” but it cannot be construed as making sense of a rejection of the a priori in
principle and in advance.

Construed in the ﬁrst way, however, the argument is that the empiricist can
account for human knowledge without needing to appeal to a priori awareness or
knowledge. Of course, this argument depends upon the empiricist’s account of human
knowledge being acceptable as an epistemology, and BonJour maintains that it is not.

For one thing, Quine’s view can offer nothing to recommend it as an account of epistemic
justiﬁcation: we can have no reasons whatever for thinking that if our beliefs are
generated in ways that are consistent with Quine’s picture of epistemic rationality, they
will be true.22 Moreover, the only reasons we could have for accepting our current
standards as conducive to knowledge would ultimately rest on those standards
themselves. This may seem to be a bit unfair, since Quine does not think that it is
possible to say whether our beliefs are true if what we mean by saying a belief is “true” is
that it “gets the world right.” As we have seen, however, Quine is not entitled to this

wholesale rejection of metaphysical realism, and he can offer no reasons for abandoning

 

2‘ Indeed, Quine and his followers observe that we are right to prefer our conceptual framework to that of,
say, the ancient Greeks, because ours is more comprehensively explanatory — it makes sense of more of our
experience than does theirs. But why think that adherence to our framework (rather than theirs) is rational?
If explanatory power were the only goal of our epistemological enterprise, then indeed greater explanatory
power would be the most that we could hope for; but if this is just what we mean by rationality, then the
claim that our adherence to our framework is rational is tautological. To paraphrase Wesley Salmon, we
can call ourselves rational, and isn ’t that nice!

22 I offered essentially the same argument in Chapter Six, footnote seventeen.

141

comprehensive epistemology. BonJour’s point, then, is not question-begging so long as
Quine cannot rule out metaphysical realism, which he cannot.

Komblith replies that BonJour’s argument makes it sound as though the self-
certiﬁcation of our epistemic standards is a simple and straightforward matter — merely
patting ourselves on the back. He suggests that, according to BonJour’s picture of the
situation, we “use our methods to investigate whether our methods are truth-conducive,
conclude that they are, and then stop.” Nothing, Komblith insists, could be further form
the truth: “the Quinean project of investigating the methods which we actually use, and
then attempting to explain how they might actually work, turns out to be a constructive
project with tremendous potential for revision of our standards.” For good measure, he
adds that “epistemic justiﬁcation, on this view, is a highly non-trivial achievement.”

The rationalist’s claim that using our standards to investigate our standards is
question-begging does not, however, imply that such investigation is trivial, or that it
cannot or will not lead to changes in our standards. After all, such investigation, to the
extent that it is empirical, will require contact with occasionally stubborn experience,
which may force (or seem to force) a shift in our standards and our practice. The point,
however, is that we can have no basis for saying that such shifts are rational in that they
are more likely to lead to the truth. Even if our standards shift as a result of an
investigation of our standards, this is, in the end, nothing to the point, because we can
only be conﬁdent that the shift was justiﬁed if we can be conﬁdent that our original
standards were justiﬁed. This either leads to an inﬁnite regress or to the claim that our
standards simply can never be justiﬁed in the way hoped for by the rationalist. This

claim, however, cannot be supported by the empiricist.

142

The empiricist can, of course, point to the success of our cognitive endeavors as
evidence that our practices are truth-conducive, or at least rational. Nevertheless, this
does not, as Komblith seems to suppose, warrant the conclusion that “given the success
of the scientiﬁc enterprise, whatever methods have been used in the development of
scientiﬁc theorizing are ones which we should ourselves adopt if we wish to have more

accurate beliefs.”23

We can have no reason whatever, on Quine’s view, for believing this,
since we can have no way of talking about the reasonableness of conservatism as an
epistemic principle in light of the ﬁt of our actual beliefs with reality. Komblith says that
“the Quinean practice of self-examination” is not aimed at anyone who is a “radical
skeptic about the entire practice of science,” but as we have seen, the empiricist cannot
simply refuse to answer the skeptical question. Again, the Cartesian skeptical question,
and the basic epistemological question, is not whether we have any knowledge at all, but
how our knowledge is possible. Komblith, like Katz, takes the basic or only task of
epistemology to be the development of the best possible account of scientiﬁc knowledge,
but this is simply a mistake. This is not to say that providing an acceptable account of
scientiﬁc knowledge is not among the tasks of epistemology, but it is not the only, and
certainly is not the most basic, task. Komblith maintains that
the empiricist’s doubts about the importance of a priority, in the end, come down
to this: the important philosophical questions we wish to ask call for substantive
and illuminating intellectual answers. The appeal to a priority seems to avoid,
rather than address, the questions about which we care so deeply. If the appeal to

a priority is to solve these questions, then a great deal more will need to be said
about the kind of illumination which such appeals are meant to provide.24

Since the appeal to a priori awareness is meant to account for the possibility of

knowledge of the world, and thus to answer the basic epistemological question, it is not

 

23 Komblith 2000, p 77.

143

altogether surprising that someone who does not regard this as a serious philosophical
question would ﬁnd appeals to a priori awareness unilluminating. What, however, are
“the questions we care about so deeply?” Questions such as “To what extent are human
beings prone to fundamental attribution errors?” or questions such as “Can we be aware
of mind-independent features of reality as such, and if so, how?” I do not wish to suggest
that no one is, or can be, driven by questions like the ﬁrst one, but it should, I think, go
without saying that questions such as the second are more fundamental, and can only
partially be answered by way of getting answers to the ﬁrst sort of question. Both are
legitimate (and important) epistemological questions. My point is not that questions of
the ﬁrst sort are unimportant, or that questions of the second sort should be our only, or
even our primary, focus; my point is that these are two essentially different sorts of
epistemological question, that the second sort of question cannot be ruled out, as
Komblith seems to suppose without argument, and that the second sort of question is not
exhausted by answering all the questions there are of the ﬁrst sort.

To return, then, to BonJour’s critique of Quinean epistemology, he further argues
that Quine’s view is ultimately unacceptable because it leaves us with no way of making
sense of the need to revise our beliefs in light of anything. The reason for this is that
since any belief can be revised or abandoned, and since no belief has its meaning (or
empirical import, or epistemic import) in isolation, there is nothing that actually generates
the tension or conﬂict required to demand a revision. Even if I hold two contradictory
beliefs, along with the belief that two contradictory statements cannot both be true, I am
only in a position to assign meaning to the three statements in light of experience, and in

light of their relation to one another, but then I need a fourth statement that tells me to

 

2‘ Ibid., p. 87.

144

interpret the apparent conﬂict as actual, unacceptable conﬂict, but then I need a ﬁfth
statement that tells me to interpret the fourth statement as true in light of my other beliefs,
but then Since the demand for background sentences is inﬁnite, the conﬂict can never
even get started, and Quine’s view is a non-starter.

Korblith’s ﬁrst reply to BonJour on this point is tu quoque: “this is an odd claim
for BonJ our to be making, given his commitment to the view that even a priori
justiﬁed beliefs may be rationally revised in light of further evidence. if Quine’s view
leads to total skepticism for this reason, BonJour’s moderate rationalism does as well.”
This would not be the case, however, if at least some of our a priori beliefs or awareness
were not open to revision. Komblith’s second reply, however, is no better. He argues
that BonJour’s criticism of Quine “does not square with the actual conduct of science.”
The point, I take it, is that in actual practice the need for revision and the process of
revision does indeed gain traction somewhere, and our beliefs are revised in light of our
experience and observation. Again, however, if Quine’s account is merely a description
of actual practice, then it cannot serve as a prescription for how to approach belief
revision in the future, since any belief can be given up, and since the argument that the
success of science warrants continuing our actual practice assumes the scheme-
independent rationality of induction. BonJour’s point once again revolves around the
issue of justiﬁcation: he does not say that revision could not in practice get started, he
says that revision could not in principle be shown to be rational.

Komblith remarks that “ﬁt with empirical beliefs is always rationally at issue,

even if there should be cases of beliefs which somehow could not be given up in the face

145

of empirical evidence.”25 He concludes from this that “the appeal to a priority can do no
epistemological work.”26 This is misleading, and his conclusion does not follow (or, if it
does, then it should follow equally well for empirical beliefs, which would be incoherent,
since then nothing would be doing the epistemic work). If some of our genuine a priori
knowledge is unrevisable, then it would of course be correct to say that these beliefs will
cohere with our empirical beliefs, but this is putting things the wrong way round: the
coherence goes the other way. Thus it is simply wrong to say that the appeal to a priori
beliefs can do no epistemological work; this appeal is doing the work of permitting
inferences beyond immediate experience. The rationalist wishes to claim that we can
have knowledge that goes beyond, or transcends, all possible future experience, in the
sense that such knowledge can be known in advance necessarily to apply to all possible
future experience, and in the sense that such knowledge cannot and will not be refuted by
any future experience. The empiricist challenges the rationalist to demonstrate that we do
in fact have such knowledge, but the challenge for the empiricist would seem to be to
explain how we can have any knowledge that goes beyond immediate experience.

To the extent that the empiricist admits only our ﬁve senses and our ability to
calculate as potential modes of awareness and conduits of knowledge and information, it
is only our ability to calculate (and reﬂect)27 that would enable us to escape the bounds of
the present moment. If, however, our reasons for saying that our ability to calculate can

provide us with knowledge are themselves empirical, then the problem becomes even

 

2’ Komblith 2000, p. 71.

26 Ibid.

27 I have so far excluded memory from my discussion of empiricism and rationalism as one of our modes of
awareness, but this is only because if the empiricist is correct in thinking that our ﬁve senses and our ability
to calculate are the only modes of awareness that we really have, then memory cannot count as a distinct
mode, since the contents of our memories can only consist of recollected sensations and calculations.

146

more acute. In that case, even inferences from present experience are justiﬁed, in the
end, by present experience (or past experience, which is simply the present experience of
yesterday), so we are left with the question how we can know anything that goes beyond
present experience. BonJour considers whether “an argument of any sort [could] be
entirely justiﬁed on empirical grounds,” and he observes that “it seems clear on reﬂection
that the answer to this question is ‘no.’” After all, “any purely empirical ingredient [in an
argument can be] formulated as an additional empirical premise,” and “when all such
premises have been explicitly formulated, either the intended conclusion [of the
argument] will be explicitly included among [the premises] or it will not.” If it is
included, no argument is required; if it is not, no argument is possible, since “the needed
inference clearly goes beyond what can be derived entirely from experience.” Thus,
BonJ our concludes, “the repudiation of all a priori justiﬁcation is apparently tantamount
to the repudiation of argument or reasoning generally, thus amounting in effect to
intellectual suicide.”28
Komblith wonders who is the intended audience of this argument, since “the
argument seems to be addressed only to those who already accept its conclusion.” He
notes that the argument presupposes foundationalism, so that no one other than
foundationalists could possibly take it seriously. This assessment of the argument,
however, ignores all of the trouble that BonJour undertakes to demonstrate that some sort

of foundationalism is the only tenable epistemology. Thus, BonJour takes the ﬁrst and

the second dispensability arguments to be answered, and so shall I.

 

28 BonJour 1999, p. 5.

147

We must now return to our reply to the inductive argument, and to BonJour’s
speculative metaphysics. Since, as we have seen, only a secure metaphysics can
ultimately answer the inductive argument, BonJour’s speculative metaphysics will not
provide the reply we need, or the basis for a comprehensive epistemology, and will thus
not meet our demands. If BonJ our offered an argument to the effect that the best we can
do is a speculative metaphysics, then we could perhaps accept that his metaphysics is the
best that we can do (if it were acceptable on other grounds), but in fact BonJour admits
that a secure metaphysics is needed and provides none. He does suggest that it is
“relatively uncontroversial” that “the Cartesian goal cannot be attained for ‘natural
knowledge,”’ but he does not elaborate.

Still, his speculative metaphysics is instructive. BonJour rejects
representationalism in its current incarnation as the computational theory of cognition,
primarily because this version of representationalism would seem to entail that we can
never be aware of the contents of our own thoughts. BonJour ﬁnds this consequence of
the computational theory so repellant and counter-intuitive that he takes it as a reductio of
the theory itself. This is an intriguing line of argument (though perhaps it comes to little
more than a version of John Searle’s Chinese Room argument), but inasmuch as we have
already undertaken an effective demolition of representationalism in Chapter Five, we
shall not pause to evaluate it.

BonJ our realizes that the problem with representationalism involves the
representation relation itself. “No surrogate or stand-in of any sort will do,” he says,

“since any account of the relation between such a surrogate and the property itself would

148

raise anew all the same difﬁculties that afflict the symbolic theory (of which any such
view would in effect be an instance)?’ I am not so sure that every version of
representationalism is a version of the computational theory of cognition, but every
version of representationalism is arguably a version of a symbolic theory of mental
content, however that content is processed, and it is ultimately this theory to which

BonJ our objects. His conclusion, however, is that for some mental states to have intrinsic
content, the forms or nature of the objects of thought must themselves “literally occur,” in
some sense, in those states. Thus, “the view is that when I think of triangular things, the
form triangularity informs my mind in a special way that is different from the way in
which it informs triangular things: it has esse intentionale as opposed to esse naturale.”
BonJour hastens to add that he thinks it is more sensible to construe the account as
involving the existence of a separate intentional form of triangularity, rather than
construing it as involving two instantiation relations — one for thoughts and one for
things. It is hard to see how this improves the situation, however. BonJour’s realism is at
least more comprehensive than Katz’s, and it at least (perhaps) implies a theory of
reference, but in the end BonJour’s realism is only slightly more plausible than Katz’s. I
shall offer an explanation of where and why Katz and BonJour both go wrong, and

consider how we may begin to put things right, in the next chapter.

149

CHAPTER 8
MOVING BEYOND REPRESENTATIONALISM: SKETCH OF A
PROGRESSIVE RATIONALISM

What, exactly, is the nature of a priori awareness? How does it operate, and how
does one have it? A comprehensive and satisfactory account of the nature of a priori
awareness would, of course, require a secure metaphysics, because any successful
account of the possibility of a priori awareness will require metaphysical assertions about
the link between the mind and reality, and because these assertions themselves can only
be known (if at all), and known to be true, a priori. Thus, as we have noted as far back as
the ﬁrst chapter, a priori awareness must, in the end, be self-certifying, or otherwise
underwrite the possibility of its own certiﬁcation. That is to say that at least some of
what we know a priori must be itself concerned with the facts about reality that make it
possible for us to have any knowledge at all of the sort that a priori knowledge professes
to be. I am afraid, however, that I cannot, at present, offer a secure metaphysics that
would underwrite a priori knowledge.

Unlike BonJ our and Katz, 1 do not intend simply to offer up a speculative
ontology and ignore the question of the possibility of a secure metaphysics. As I have
argued, if a secure metaphysics is not to be had, then a speculative one must sufﬁce,
because we cannot do without it. Nevertheless, the possibility of a secure metaphysics
cannot be ruled out, for the same reason that a priori awareness cannot be ruled out. To
rule out the possibility of a secure metaphysics, we would need to appeal to metaphysical
assertions that would entail that no metaphysical assertions can be known to be

necessarily true. Of course, if these assertions are themselves not necessarily true, then

150

they cannot establish the impossibility of a secure metaphysics, and if they are
necessarily true, then they would constitute a secure metaphysics, and would thus be self-
vitiating.

We cannot do without metaphysical realism, however, because we cannot
abandon the project of comprehensive epistemology, and because comprehensive
epistemology cannot be undertaken without metaphysical realism. The arguments
intended to rule out the very possibility of a priori awareness fail because, to be
successful, they would themselves require a priori knowable metaphysical assertions.
Moreover, the arguments against comprehensive epistemology fail because they cannot
rationally recommend themselves. No empiricist epistemology can, however, hope to be
comprehensive; thus, the only possibly successful epistemology is rationalist, and
foundationalist. If, at the end of the day, we cannot answer the skeptic, then we can offer
no answer to the question “How is knowledge possible?”, but this would also mean that
we could not say why we cannot answer the skeptic, because any answer would require
us to be able to know why we can have no means of seeming our access to reality as it is
in itself, and this knowledge would itself transcend our cognitive limits. This would not
mean, however, that the Skeptic’s position is incoherent, as is frequently assumed; it
would mean that we could never say, to all eternity, whether any of our beliefs are in fact
rational.

1 do not believe, however, that such pessimism is warranted. We shall see, in the
course of the following discussion of the nature and possibility of a priori awareness, that
a potentially securable metaphysics is available, because there are certain unrevisable a

priori truths that we cannot do without.

151

Before we begin to discuss the possibility and actuality of a priori awareness,
however, it will be well to try to get a bit clearer about the nature of a priori awareness.
As was noted in Chapter Six, empiricists frequently contend that the very idea of a priori
awareness is incomprehensible — or at least that rationalists have offered no
comprehensible account of it — and that the very idea seems more like a mystical or
occult power than a cognitive faculty.I In response to this challenge, rationalists have
often done more harm than good. For one thing, many rationalists persist in using terms
such as “rational insight” or “rational intuition,” that sound mystical. What is worse,
however, is that the terms “insight” and “intuition” are themselves notoriously unclear
and imprecise, so that to preﬁx such terms with “rational” offers little in the way of
clariﬁcation, and in fact tends to make things muddier. I cannot offer anything like a
comprehensive and exhaustive analysis of any of these terms,2 but we can get quite a lot
clearer about all of them.

What, ﬁrst of all, is intuition?3 Much has been made recently of the fact that

intuitions are neither universal nor reliable — that they are culturally shaped, historically

 

' In fact, some philosophers, like Austin, have gone so far as to claim that they do not even understand what
is meant by the question whether there are such things as “a priori concepts” (Austin 1970, pp. 32 and 47).
Austin’s insistence that he does not understand the question seems, however, to be little more than a glib
way of saying that once the question is properly analyzed we shall see that any serious attempt to clarify the
origins of our concepts will amount to “either nothing or nonsense” (p. 46).

2 And I would not do so even if 1 could.

3 I shall not speciﬁcally address mathematical intuition (or mathematical intuitionism) in this discussion,
since mathematical intuition (or intuitionism) is generally conceived either (1) as a form of rational insight
akin to what I have been calling “a priori awareness,” in which case mathematical intuition is simply a
speciﬁc instance of the general form of this sort of awareness, and does not require separate treatment here,
or (2) as some sort of Kantian cognitive awareness of the form of inner intuition of temporal succession, in

152

conditioned, context-sensitive, and context-relative. Used as a court of ﬁnal appeal, they
can operate as a philosophical cudgel, a means of convenient dismissal of opposing
views, a weapon of discrimination, or a mechanism of group identiﬁcation and
exclusion.4 To the extent that intuition is something like our pre-reﬂective, non-
discursive and spontaneous disposition toward the meaning or truth-value of statements
and their contents, all this would seem to be true. Yet, philosophers — particularly
analytic philosophers — rely, or claim to rely, on their intuitions all the time, and take
them actually to constitute evidence in favor of certain claims or theses. If intuition is so
unreliable, why do they do this?

In the context of a discussion of the notions of necessary and contingent
properties, Saul Kripke claims, in Naming and Necessity, that “if someone thinks that
[one of these notions] is a philosopher’s notion with no intuitive content, he is wrong.”5
He goes on to observe that “of course, some philosophers think that something’s having
intuitive content is very inconclusive evidence in favor of it,” but that he thinks “it is very
heavy evidence in favor of anything.”6 In fact, Kripke adds, “I really don’t know, in a
way, what more conclusive evidence one can have about anything, ultimately speaking.”7
That’s a pretty strong statement. Why does Kripke assign such importance to intuition,
and what does he mean by it?

Since he is insisting that the notions of necessary and contingent properties are

not simply “philosophers’ notions,” it would seem that in asserting that these notions

 

which case it is something more like a speciﬁc instance of the sort of thing contemporary philosophers are
talking about when they speak generally of “intuition,” and again does not require separate treatment.

4 See the discussions of Putnam’s rhetorical dismissal of rationalists in Chapter Four.

5 Kripke 1980, pp. 41-42.

6 Ibid., p. 42.

7 Ibid.

153

have “intuitive content” he means to claim, as against those who wish to deny that
ordinary speakers or laypersons distinguish between necessary and contingent properties,
that this distinction is a perfectly commonplace one — a common sense distinction. Of
course, it is straightforwardly empirical matter whether ordinary speakers make this
distinction, or one like it, and it is not at all clear how this bears on the question of the
meaningfulness of the notions in question, or the meaningfulness of the distinction itself.
Isn’t it possible for ordinary language and common sense to employ illegitimate or
erroneous distinctions? Why would Kripke say that he doesn’t know “what more
conclusive evidence one can have about anything, ultimately speaking?”

David Lewis observes that “one comes to philosophy already endowed with a

stock of opinions,” and he suggests that

it is not the business of philosophy either to undermine or to justify these
preexisting opinions, to any great extent, but only to try to discover ways of
expanding them into an orderly system. A metaphysician’s analysis of mind is an
attempt at systematizing our opinions about mind. It succeeds to the extent that
(1) it is systematic, and (2) it respects those of our pre-philosophical opinions to
which we are ﬁrmly attached. There is some give-and-take, but not too much:
some of us sometimes change our minds on some points of common opinion, if
they conﬂict irremediably with a doctrine that commands our belief by its
systematic beauty and its agreement with more important common opinions.8

Now I would argue that this is simply false. As we have seen in the last two chapters, it
is precisely the job of comprehensive epistemology to inquire into the justiﬁcation of our
beliefs about the world, and into the very possibility of knowledge, not to take our
knowledge, or common sense, for granted. Kripke and Lewis, like positivists and

pragmatists, appear to want to turn their backs on the skeptic.

 

8 Lewis 1979, p. 186.

154

In much of the discussion of skepticism that has taken place in analytic
philosophy over the course of the last hundred or so years, “the skeptic” has tended to be
rather literally personiﬁed, as though he or she were some real, ﬂesh-and-blood heretic,
who simply is not aware that his or her own communicative commitments undermine the
very skepticism being expressed, by requiring the skeptic to rely upon the shared
assumptions of discourse (or common sense) that he or she is calling into question. No
doubt this is because philosophers who wished to turn their backs on the skeptic actually
had other speciﬁc philosophers in mind as representatives of the sort of skeptical attitude
that they wanted to reject. It may also be a consequence of the inﬂuence of thinkers like
Austin and Wittgenstein, who emphasized the need for a context for meaningful
utterance, and who pointed out that in most contexts nobody would think to question the
fact that they are standing on the ground, and that they have two hands. This, however,
has the effect of treating philosophers as though they were aliens. What the
personiﬁcation of the skeptic ignores is that we are the skeptic. Descartes assumes the
skeptical position in order to see whether and how he can refute it in its strongest
imaginable form. This is not to say that there are not, in fact, actual skeptics, or that there
never have been any. Rather, the point is that we assume skepticism as a means of
testing the strength of our own reasons for believing what we do. In any argument, we
take up the stance of the opposition to see whether our arguments can meet those that are,
and could be, offered on the other side. Epistemologically speaking, the opposition is the
skeptic.

It is not clear why it is at all relevant that in most contexts nobody would think to

question whether there are external objects, or whether they have two hands. In most

155

contexts (if one really thinks about what this comes to), nobody thinks to wonder whether
quarks are the smallest unit of matter, but this should not prevent us, and it did not
prevent Wittgenstein, from saying that particle physicists are not talking nonsense. There
is clearly a context within which some people question whether they have two hands, and
this is not (usually) because they really doubt that they do, but because they want to know
how it is that we can be secure in this knowledge, and because they are convinced that we
ought to be able to have this security but cannot see how to come by it. As we have seen,
no argument (that is not self-undermining) can be offered to the effect that we hope for
such security in vain. To begin by afﬁrming that we have the knowledge we would hope
to secure, and that therefore the skeptic has no point to make, is to turn our backs on
philosophical inquiry itself. If there are no meaningful contexts in which someone would
or could question that she has two hands, then philosophical inquiry is meaningless. It is
not sufﬁciently frequently reiterated that this was precisely Wittgenstein’s view: he was
an anti-philosophy philosopher.

Philosophy questions and examines our common sense beliefs, it does not
enshrine them. In fact, this is true of science as well. This is why Quine is both right and
wrong to say that science is continuous with common sense. Science and philosophy
both grow out of common sense, for they arise in the context of the quite ordinary human
activity of wonder, and common sense embodies (to some extent) our pre-reﬂective
attempt to organize and understand the world. But both philosophy and science go
beyond common sense, and are not to that extent continuous with it, because neither takes

common sense for granted, because they both involve an examination and critique of

156

common sense (as a pre-reﬂective “theory” of the world), and because they both require
reﬂection of a sort that sometimes forces common sense convictions to be given up.
Lewis acknowledges this, but insists that while “there is some give-and-take,”
there is “not too much.” Common sense, like the intuition that reﬂects it, is provincial in
both time and place; it is context-bound. Certain common sense notions, such as that we
each have two hands, have probably been held by all persons in all places and times.
Many common sense notions, however, such as that the earth stands still, depend upon
our pre-reﬂective view of things from where we stand. This notion is no longer part of
our common sense view of the world, which suggests that common sense is corrigible.
To treat common sense as though it were largely right, with the appropriate adjustments,
however, is to treat science and philosophy as the corrective arms of common sense, and
this is to ignore the difference between our culturally, historically, and perspectivally
conditioned pre-reﬂective intuitions and the self-correcting project of reﬂection.
Common sense, as Russell put it, “leaves us completely in the dark as to the true intrinsic
nature of physical objects, and if there were good reason to regard them as mental, we
could not legitimately reject this opinion merely because it strikes us as strange. The

”9

truth about physical objects must be strange. As Bryan Magee observes:

modern science has shown that behind our moment-to-moment experience of the
everyday world teem truths and realities that common sense is totally unaware of,
that are frequently astounding and often counter-intuitive, and sometimes deeply

difficult to grasp even when we know them to be true. In these circumstances
attempts to defend the commonsense view of the world are doomed before they
begin. '0

 

9 Russell 1959, p. 38.
'0 Magee 1999, p. 42.

157

Lewis and Kripke deserve credit for being among the ﬁrst philosophers in recent
analytic philosophy to insist on the fact that ordinary people understand, have and utilize
metaphysical notions - that such notions are not the sole province of philosophers-qua-
specialists, or qua-misusers-of-language. If ordinary people did not have such notions,
then philosophers would not be ordinary people, even contingently, or initially, before
their “conversion” to philosophy. They also deserve credit for noting that our possession
and acquisition of metaphysical notions are largely non-discursive (though not pre-
reﬂective, or necessarily innate). In their (understandable) zeal to reinstate metaphysical
notions, however, Lewis and Kripke concede too much to empiricism by construing
intuition in terms of common sense, so as to avoid skepticism and preserve a conception
of philosophy that takes its chief business to be the clariﬁcation of what we already know
or believe. Unfortunately, this licensed the rather regrettable practice among analytic
philosophers of appealing to their intuitions as a last resort in the justiﬁcation of their
views, and in the refusal to treat with those who do not share their intuitions, prompting
the counter-attack mentioned earlier. Intuitions understood as the concretization of
common sense are clearly neither trans-cultural nor trans-historical, and we have no more
reason for thinking that they are a reliable guide to justiﬁed belief than we have to think
that conservatism in theory choice is rational in light of our current epistemic standards.

If, on the other hand, we construe intuition as a non-discursive awareness of
necessity or of necessary truth, then we are getting closer to an acceptable clariﬁcation of
the nature of a priori awareness. To call such intuition “rational” is perhaps somewhat
redundant, but the usage is no doubt intended both to distinguish this sort of intuition

from other sorts (like the concretization of common sense), and to call attention to the

158

fact that it operates independently of sensory awareness or experience (though not
without any initial sensory input whatever). Given the apparent redundancy of “rational
intuition,” however, where “intuition” is construed as suggested above, and given the
ambiguity of the term “intuition” used by itself, we shall prefer to retain the term “a
priori awareness.” This sort of awareness is universal and trans-historical.ll Understood
in this way, however, how can we make sense of the possibility of a priori awareness?
Isn’t it still mystical?

Inasmuch as we cannot abandon the project of comprehensive epistemology, we
must, in developing this project, appeal to the possibility and actuality of a priori
awareness to account for the possibility and actuality of our having knowledge of reality
as such. Thus, we must have an account of the possibility of a priori awareness that is
either secure now or potentially securable — that is to say, one that secures, or can be
developed in the direction of securing, the possibility (and actuality) of a priori
awareness through a priori knowledge that is self-certifying, or that underwrites the
possibility of its own certiﬁcation. We do not have such an account at present, but we
can see how we may move in the direction of developing one. As we have seen, at least

some of the principles of epistemological theory choice must be in principle unrevisable

 

” There are serious and compelling arguments on offer in feminist and social and political philosophy, as
well as in the work of “continental” thinkers such as Haberrnas, Merleau-Ponty, Adomo and Derrida
(among others), that this is not so, and that norms of rationality are constituted communicatively among
agents participating in social forms of life. Another argument that is sometimes pressed in this context is
that it is meaningless to say that humans are, by nature, capable of a priori awareness, because there is no
such thing as universal human nature. I cannot begin to do justice to these arguments here. Sufﬁce it for
now to say that I think the problem is misstated. Even if humans are not by nature capable of a priori
awareness (trivially, because there is no universal human nature), this does not mean that not everyone is
capable of a priori awareness, given the appropriate preparation. Norms of rationality are perfectly
objective, even if they are only realized discursively. The battle is not truly joined here, but at the much
deeper level of the very possibility of knowledge of reality as it is in itself, which many of the thinkers and
traditions mentioned above reject. To that extent, I would (eventually) need to deal with their arguments in
the same way we dealt with the arguments of the positivists and the pragmatists, in Chapters Two, Three,
Four, and Five.

159

in order for our epistemology to avoid skepticism, so some of these principles must be
necessary truths about reality as such. Such principles as the principle of non-
contradiction, for instance, are not, and cannot be, self-certifying, but they constitute the
unrevisable principles that guide and constrain theory choice as we select, not
speculative, but hypothetical, metaphysical principles and postulates that form part of our
self-correcting project of accounting for the unrevisability of the principles themselves,
for the possibility and actuality of a priori awareness, and for the possibility of
knowledge in general. This is a metaphysics that does progress, pace Kant, positivism
and pragmatism. Some of the metaphysical postulates that we select (and perhaps some
of the principles of theory choice) will be revisable, and will require revision, but they
will be revisable only in light of further pure ratiocination12 and in light of at least some
unrevisable principles without which reason could not operate. Moreover, this process of
revision will be self-correcting because there are minimally agreed-upon principles of
theory choice to which any acceptable metaphysical theory must conform. Of course,
within the constraints of these principles, much is possible and permissible, which is
perhaps why metaphysical debates have seemed to some philosophers to be interminable
and pointless. Fortunately, they are neither. Two related concrete contributions to the
advancement of metaphysics were, in fact, made in the modern period by empiricists.
One was the observation that the existence of God, or of anything, cannot be established
through rational argumentation alone. The other was the observation that existence is not

a predicate. These are not, and cannot be, simply facts about the way we use the words

 

'2 See Chapters Six and Seven for an elaboration and defense of the claim that a priori beliefs are
(ultimately) revisable only in light of further a priori beliefs.

160

“God” and “exists”; they are constraints on the very possibility of what we can know
through reason alone.

Even Hilary Putnam notes that while he takes it “as a fact of life” that “there is a
sense in which the task of philosophy is to overcome metaphysics,” there is also “a sense
in which it its task is to continue metaphysical discussion.” He says that “in every
philosopher there is a part that cries, ‘This enterprise is vain, frivolous, crazy—we must
say ‘Stop’!’ and a part that cries, ‘This enterprise is simply reﬂection at the most general
and abstract level; to put a stop to it would be a crime against reason?”'3 It would be
worse than a crime against reason, it would preclude the very possibility of reasoning
altogether. Putnam, in the course of developing his famous “brain-in-a-vat” argument,
admits that his reasoning has proceeded along lines that are best characterized as a
priori.14 This is Laurence BonJour’s insight in his In Defense of Pure Reason. As
BonJ our puts it: “the argumentative transition, in thought or discourse, from the premises
to the conclusion is an inference,” and the inferential steps of any argument are non-
discursively apprehended mechanisms of reasoning that can be themselves be expressed
as necessary rational principles; thus, “the repudiation of all a priori justiﬁcation is
apparently tantamount to the repudiation of argument or reasoning generally.”ls

BonJ our concludes that this argument “surely constitutes a strong primafacie

reason for regarding the idea of a priori justiﬁcation as philosophically and intellectually

 

'3 Putnam 1990b, p. 19.

'4 Putnam 1981, p. 16. Of course, Putnam insists that he is reasoning “not in the old ‘absolute’ sense” of a
priori, because he doesn’t “claim that magical theories of reference are a priori wrong,” and he further
notes that he is reasoning a priori “in the sense of inquiring into what is reasonably possible assuming
certain general premisses, or making certain very broad general assumptions.” The point here, however, is
that Putnam recognizes that the inferential steps in his argument, and much of its evidence, depend upon or
consist of rational principles, and we have seen that our knowledge of such principles cannot be accounted
for without appealing to a priori metaphysical realist assumptions or assertions.

'5 BonJour 1999, p. 5.

161

indispensable,” and he observes that “the argument for this conclusion is extremely
straightforward and obvious, so much so that it is very hard to understand the widespread

failure to appreciate it.”16

The argument has not been appreciated because its primafacie
appeal has been taken to have been overridden by the incomprehensibility of a priori
awareness and the apparent availability of an empiricist epistemology that treated rational
principles as revisable as a last resort (which explains the appearance of their being
necessary). This is where BonJ our and Katz go wrong: what is needed is an ultimafacie
reason for thinking that a priori awareness is philosophically and intellectually
indispensable, and we have seen that the impossibility of abandoning comprehensive
epistemology provides us with such a reason. BonJ our and Katz maintain that we need
only develop a rationalist epistemology that can be defended as preferable to every
available empiricist epistemology. Thus, their chief concern is to answer the empiricist,
not the skeptic, and to show that an appeal to a priori awareness makes better sense of the
fact of our scientiﬁc and philosophical knowledge than do the available versions of
empiricism. As a result, their accounts of a priori awareness are developed only enough
to support the claim that rationalism is tenable, not to support the claim that rationalism is
philosophically and intellectually indispensable. BonJour and Katz emphasize the prima
facie importance of a priori awareness, but it remains open to the empiricist to argue that
their accounts of a priori awareness are unenlightening, and that empiricism therefore
remains preferable. What is needed is an argument to the effect that a priori awareness,
and thus rationalism and metaphysical realism, are inescapable, and this is what I have

provided. Thus, while a priori awareness may still (at this point) seem somewhat

mystical, we cannot do without it. In the remainder of this discussion, I shall do what I

 

'6 Ibid., p. 6 and pp. 3-4.
162

can to make a priori awareness seem comprehensible, or at least to rid it of the

appearance of mysticism.

I have described metaphysics as inquiry into ultimate ontology — the discovery of
“the fumiture of the universe.” This description of metaphysics provides Quine and other
naturalistic and scientistic philosophers with another argument in favor of the elimination
of metaphysics, and in favor of viewing philosophy as concerned, not with reality directly
or as such, but with scientiﬁc discourse about reality. The argument is that it is science,
not philosophy, that postulates objects, and that the sciences cover knowable reality
exhaustively. Philosophy draws distinctions, but it takes as it subject matter the concepts
(objects) introduced and utilized by, the sciences. This is partly true: metaphysics does
not, in general, postulate objects, so the description above is somewhat misleading. Only
somewhat, however, because we need not construe inquiry into ultimate ontology as
concerned with the discovery of speciﬁc objects, but as concerned with the general nature
of objecthood and worldhood — with the principles that govern the structure and the limits
of reality. While science is concerned with the construction and limits of what is actual
and sensible, metaphysics is concerned with what is necessary and intelligible. Thus
metaphysics, and philosophy, take reality itself as their subject matter. Moreover, the
scientist and the metaphysician do not study reality in the same way, or study the same
aspects of it, except occasionally. Philosophy is of course a science in concepts, but it is

not, as Quine would have it, the science of concepts.

163

A priori awareness can only be a feature or function of cognition that is capable of
being in receipt of information about the necessary features or organization of reality
itself. Does that make it mystical? I shall offer three reasons why it does not. Firstly, as
Putnam points out, we do not have anything like a comprehensive, or really even a very
satisfactory account of ordinary, empirical sensory awareness, when it comes to the
questions of how sensory input becomes mental content, and how that content is
processed, accessed and utilized. Philosophy of mind still abounds with terminology
such as “raw feel,” “after-image,” “sense data,” and “qualia.” Concerning the process(es)
through which sensory input informs thought, we remain almost totally in the dark.

Secondly, the reason for this is that we lack an account of conscious awareness.
Presumably, if the mind is capable of being aware of at least some of the necessary
features of reality, this capacity will be exercised through certain conscious cognitive
processes, and will not be associated with the operation of any empirically observable
sensory organs. To insist that if there is such a thing as a priori awareness we should
know something more about it by now is to insist that we should already have a more
complete and satisfactory account of the nature of consciousness. In fact, there are some
philosophers who insist on just such a thing, and who moreover argue that the persistent
impenetrability and mysteriousness of consciousness suggest that there simply is no such
thing. I can offer nothing like a fully developed challenge to this argument here, but I
will say that this sort of argument strikes me as unnecessarily hasty and as in fact
unmotivated. The only reason we could have for thinking that if consciousness is a bona
ﬁde phenomenon our science should already have been able to explain it is that

consciousness is not sufﬁciently complicated to elude explanation by current methods.

164

We can have no possible warrant for this view, which involves a conﬁdence in the
explanatory powers of current science that is supported neither historically nor by the
results of current research on the brain. If we are to have an account of a priori
awareness as a cognitive capacity, it will surely arrive no sooner than a rather more
developed account of the nature of conscious awareness.

Thirdly, an explication of the operation of a priori awareness must ultimately
involve an account of the link between mind and reality. Any account of this link will
require metaphysical assumptions or assertions. This is not a short argument for dualism,
or for the claim that this aspect of conscious awareness (or of conscious awareness in
general) will forever remain inexplicable in scientiﬁc terms because it is a metaphysical
concept. Rather, it is an observation that there will remain an irreducibly philosophical
component to any satisfactory account of a priori awareness, which account must be a
joint product of scientiﬁc and metaphysical inquiry. Therefore, not only do we have no
reason to think that we should already be in possession of an account of a priori
awareness, but we can see that such an account must be developed in the process of
providing an account of the nature of consciousness. If a priori awareness is mystical,
consciousness is also mystical, and so is philosophy of consciousness.

What is needed, then, to provide an account of a priori awareness, is a
metaphysical account of the relationship between the mind and the rest of reality. In
Chapter Six I conceded that it seems unlikely that such an account can be offered by a
pluralist metaphysics. I say this because a pluralist metaphysics must, in the end, offer an
account of how ideas reveal to the mind facts about the order, nature and structure of

distinct, mind-independent entities. That is to say, a pluralist metaphysics must explain

165

the possibility (and actuality) of the mind having contact (in some sense of contact) with
things-in-themselves, and it is at best far from clear how this may be done. It is not
obvious that, as Katz suggests, we would have to prove that God exists. Of course, this
was Descartes’ strategy, and if we could establish the existence of God, and establish that
He would guarantee the actuality of contact, then we would indeed have (the beginnings
of) an account of a priori awareness. God would, at least, guarantee the possibly proper
operation of rational insight, and that would be enough to get started.” In any event,
God’s existence cannot be established by reason alone, but this does not mean that the
pluralist has exhausted her options. On the other hand, a pluralist metaphysics implies, at
minimum, that we and the other objects of the world are distinct entities, so the pluralist
must somehow overcome the metaphysical problem of interaction. Hilary Putnam has
argued that the only way to make sense of the idea that our minds are capable of being
aware of the nature of reality as it is in itself is to assume that reality itself is a mind, or is
like a mind, or is rational, or rationally ordered.‘8 This takes for granted Berkeley’s
dictum that ideas really can only be like other ideas. It is not clear that this is true, but if
it is not, only a successful solution of the metaphysical problem of interaction could show
that and why it is not. Moreover, pluralists cannot revive representationalism as an
account of interaction, because representationalism is untenable. There have been, in the
history of philosophy, pluralist metaphysicians of genius,19 who have produced

thoroughgoing and compelling metaphysics, so I do not intend to rule anything out.

 

’7 Of course, we would then need a Cartesian science of psychology as a science of immaterial substance.
This may seem to suggest that any acceptable account of a priori awareness does imply dualism after all,
but this would only be the case if proving the existence of God really were our only hope for providing a
secure metaphysics, and I submit that this is not obviously the case.

'3 Putnam 1982, p. 154.

’9 Leibniz comes to mind.

166

A monistic metaphysics would seem, on its face at least, to stand a better chance
of providing a metaphysically comprehensible account of a priori awareness. Though
monism has been the minority metaphysical position in Western thought since Plato,
there have also been monist metaphysicians of genius — notably Spinoza, but also
arguably Plato (and of course Parrnenides, though little of his work survives). If, as
Spinoza suggests, thought is a mode of reality,20 then it is not ontologically reducible to
physical entities, or to a sequence of interactions between physical entities. Thus we can
respect the arguments of those like Searle, earlier put forward by Leibniz in the

Monadology, as follows:

We must confess that the perception, and what depends on it, is inexplicable in
terms of mechanical reasons, that is, through shapes and motions. If we imagine
that there is a machine whose structure makes it think, sense, and have
perceptions, we could conceive it to be enlarged, keeping the same proportions, so
that we could enter into it, as one enters into a mill. Assuming that, when
inspecting its interior, we will only ﬁnd parts that push one another, and we will
never ﬁnd anything to explain a perception. And so, we should seek perception in
the simple substance and not in the composite or the machine.21

On the other hand, mind is not, for Spinoza, a distinct substance from body. Spinoza
argues that mind and body are co-extensive, that they are in fact identical — the same
aspect of reality conceived in different ways.22 If thought is simply a feature of one total
reality, then there is no problem of interaction. We need only explain how thought is
constituted as a feature of reality that also apprehends the necessary principles of the

construction of reality.

 

2° Spinoza 1992, p. 64.
2‘ Leibniz 1991, p. 70.

167

Wittgenstein claims, at the outset of the T ractatus, that in that work, the
remaining problems of philosophy “have in their essentials been ﬁnally solved.”23 This
being, naturally, the main value of the work, its secondary value is that “it shows how
little has been done when these problems have been solved.”24 Wittgenstein says this not
because he means to demonstrate his sensitivity to the fact that solving all of the
outstanding problems of philosophy puts food in no one’s mouth, clothes on no one’s
back, and shelter over no one’s head, and makes no one safe from violence or harm or
injustice. As philosophers, we are, and must always be, acutely aware that philosophy
rarely if ever can do these things. Wittgenstein says this, however, because his solution
to the problems of philosophy, in the T ractatus, and again in Philosophical
Investigations, involves taking the position that there never really were any actual
philosophical problems to begin with. It would in fact have been better, or at least more
accurate, if Wittgenstein had said that the Tractatus dissolves all of the problems of
philosophy, by declaring them pseudo-problems or merely apparent problems. A favorite
metaphor of Wittgensteinians is that philosophy exists, and continues to exist, only as a
ladder we may use to climb out of philosophical problems. Once we have climbed out,
we kick the ladder down behind us. This of course suggests that we are climbing to the
end of philosophy — a period in human development wherein, after two-and-a-half
thousand years of self-vexation, we can ﬁnally rid ourselves of the gadﬂy once and for

all, and show him the way out of the bottle and out of our heads.

 

2’ Spinoza 1992, pp. 66-67, 70-76, and 80-81.
23 Wittgenstein 1999, p. 29.
2“ Ibid.

168

I could not possibly claim for the present work that it solves all of the problems of
philosophy. If it shows anything, however, what it shows is that there are genuine
philosophical problems which require thoughtful metaphysical solutions. As Bryan
Magee observes, “anyone who believes that the real task of philosophy is to clarify
utterance must believe that non-linguistic reality presents us with no philosophical
problems. And this is precisely what Austin and the older Wittgenstein, and their

followers, did believe.”25

If we are to treat metaphysical questions as senseless or
meaningless, we must have readyeto-hand a criterion or theory or account of sense and
meaning — and ultimately a theory of knowing — that not only shows that these questions
are meaningless, but explains why they are meaningless. Any criterion or theory or
account of sense and meaning (and ultimately any theory of knowledge) must, however,
appeal to the very sort of metaphysical assumptions or assertions to which it is meant to
deny us access. There can be no hope of ruling out rationalism or metaphysics.
Metaphysical realism is philosophically indispensable.

It may be said that my arguments here will convince no one, and that it is not
clear whether they are even needed. Empiricists will not be persuaded by my
demonstrations and protestations, and rationalists are already convinced. Besides, as I
myself have pointed out, metaphysics and rationalism have returned to philosophy, even
to analytic philosophy, at the end of the twentieth, and the beginning of the twenty-ﬁrst,
century. They are outcasts no longer. While positivism is dead, however, it lingers like a
specter in philos0phy today, a shade of its hegemony in early twentieth century Anglo-

American philosophy, and the empiricist revolution of which positivism was a part has

continued, and has retained the conception of philosophy as analysis (or as philosophy of

 

’5 Magee 1999, p. 89.

169

science) for which positivism was the justiﬁcation, even long after positivism breathed its
last. This conception of philosophy must go. My goal in this work was not to defeat
empiricism, but to take some of the air out of the wholly unjustiﬁed empiricist conﬁdence
that has been generated and sustained by the supposed defeat of rationalism. Moreover,
philosophical arguments have always had two functions (among many others): to
challenge opposing positions, and to contribute to the development and the continued
emergence, production, and reproduction of the position being defended. I intend this
work to challenge my opponents, but I also mean it as an invocation, an effort to bring
rationalism into being and to nourish the being that it has, an incantation that reveals a
place for metaphysical questions, deeply and fundamentally human questions — my

questions — in a world of science and sense.

170

WORKS CITED

Alston, William. 1999. “Yes, Virginia, There is a Real World,” in Metaphysics: An

Anthology. Jaegwon Kim and Ernest Sosa, eds. New York: Blackwell Publishers,
Inc. Pp. 620—633.

Austin, J. L. 1970. “Are There A Priori Concepts?”, in Philosophical Papers, Second

Edition. J. O. Urrnson and G. J. Warnock, eds. New York: Oxford University Press.
Pp. 32-54.

Ayer, Alfred Jules. 1952. Language, Truth and Logic. New York: Dover Publications,
Inc.

Beall, JC. 2004. “Introduction: At the Intersection of Truth and Falsity,” in The Law of
Non-Contradiction: New Philosophical Essays. Graham Priest, JC Beall and Bradley
Arrnour-Garb, eds. Oxford: Oxford University Press. Pp. l-l9.

Block, Ned. 1995. “Advertisement for a Semantics for Psychology,” in Mental
Representation: A Reader. Stephen P. Stitch and Ted A. Warﬁeld, eds. Cambridge:
Blackwell Publishers, Inc. Pp. 81-141.

BonJour, Laurence. 1985. The Structure of Empirical Knowledge. Cambridge: Harvard
University Press.

BonJour, Laurence. 1999. In Defense of Pure Reason. Cambridge: Cambridge
University Press.

BonJour, Laurence. Forthcoming. “Analytic Philosophy and the Nature of Thought.”

Bordo, Susan. 1987. The Flight to Objectivity: Essays on Cartesianism and Culture.
Albany: State University of New York Press.

Carnap, Rudolph. 1964. “Empiricism, Semantics and Ontology,” in Philosophy of
Mathematics: Selected Readings. Paul Benacerraf and Hilary Putnam, eds.
Englewood Cliffs, NJ: Prentice-Hall, Inc. Pp. 233-248.

David, Marian. 1991. “Neither Mentioning ‘Brains in a Vat’ nor Mentioning Brains in a
Vat Will Prove that We Are Not Brains in a Vat,” in Philosophy and
Phenomenological Research 51(4): 891 -896.

Descartes, René. 1979. Meditations on First Philosophy. Donald A. Cress, trans.
Indianapolis: Hackett Publishing Company.

171

Descartes, René. 1988. Principles of Philosophy. In Descartes: Selected Philosophical
Writings. John Cottingham, Robert Stoothoff and Dugald Murdoch, trans.
Cambridge: Cambridge University Press. Pp. 160-212.

Dretske, Fred. 1995. “Misrepresentation,” in Mental Representation: A Reader.
Stephen P. Stitch and Ted A. Warﬁeld, eds. Cambridge: Blackwell Publishers, Inc.
Pp. 157-173.

Forbes, Graeme. 1995. “Realism and Skepticism: Brains in a Vat Revisited,” in The
Journal of Philosophy 92(4): 205-222.

Glymour, Clark. 1982. “Conceptual Scheming, or, Confessions of a Metaphysical
Realist,” in Synthese 51: 169-180.

Goodman, Nelson. 1978. Ways of Worldmaking. Indianapolis: Hackett Publishing
Company.

Haukioja, Jussi. 2002. “Water, Phlogiston, Brains, and Vats,” in Sorites. 14:16-20.

Heidegger, Martin. 1962. Being and Time. John Macquarrie and Edward Robinson,
trans. New York: Harper & Row, Publishers.

Hume, David. 1993. An Enquiry Concerning Human Understanding. Eric Steinberg,
ed. Indianapolis: Hackett Publishing Company.

Kant, Immanuel. 1998. Critique of Pure Reason. Paul Guyer and Allen W. Wood,
trans. and eds. Cambridge: Cambridge University Press.

Kant, Immanuel. 2001. Prolegomena to any Future Metaphysics. Paul Carus, trans.
James W. Ellington, rev. Indianapolis: Hackett Publishing Company.

Katz, Jerrold J. 2000. Realistic Rationalism. Cambridge: The MIT Press.
Kim, Jaegwon. 2000. “What is ‘Naturalized Epistemology’?”, in Epistemology: An

Anthology. Jaegwon Kim and Ernest Sosa, eds. New York: Blackwell Publishers,
Inc. Pp. 301-313.

Kenny, Anthony. 1968. Descartes: A Study of His Philosophy. New York: Random
House.

Kolenda, Konstantin. 1992. “Rethinking the Teaching of Philosophy,” in Teaching
Philosophy 15(2): 121-132.

172

Komblith, Hilary. 2000. “The Impurity of Reason,” in Pacific Philosophical Quarterly
84: 67-89.

Kripke, Saul. 1980. Naming and Necessity. Cambridge: Harvard University Press.

Latour, Bruno and Steve Woolgar. 1979. Laboratory Life: the Social Construction of
Scientific Facts. Beverley Hills: Sage Publications.

Leibniz, G. W. 1991. Monadology. In Discourse on Metaphysics and Other Essays.
Daniel Garber and Roger Ariew, trans. Indianapolis: Hackett Publishing Company.
Pp. 68-81.

Lewis, David. 1979. “Possible Worlds,” in The Possible and the Actual: Readings in the
Metaphysics of Modality. Michael J. Loux, ed. Pp. 182-189.

Lugones, Maria C. 1994. “Purity, Impurity and Separation,” in Signs: Journal of Women
in Culture and Society 19(2): 458-479.

Magee, Bryan. 1999. Confessions of a Philosopher. New York: Random House.

Nielsen, Kai. 1994. “Philosophy Within the Limits of Wide Reﬂective Equilibrium
Alone.” Iyyun 43: 3-41.

Nozick, Robert. 1981. Philosophical Explanations. Cambridge: The Belknap Press.

Oliver, Alex. 2000. “A Realistic Rationalism?”, in Inquiry 43(1): 111-135.

Plato. 1997a. Theaetetus. In Plato: Complete Works. M. J. Levett, trans. Myles
Bumyeat, rev. John M. Cooper, ed. Indianapolis: Hackett Publishing Company. Pp.
157-224.

Plato. 1997b. Republic. In Plato: Complete Works. G. M. A. Grube, trans. C. D. C.
Reeve, rev. John M. Cooper, ed. Indianapolis: Hackett Publishing Company. Pp.
971-1223.

Putnam, Hilary. 1975. “The Meaning of ‘Meaning’,” in Mind, Language and Reality:
Philosophical Papers, Volume 2. Cambridge: Cambridge University Press. Pp. 215-
271.

Putnam, Hilary. 1981. Reason, Truth and History. Cambridge: Cambridge University
Press.

173

Putnam, Hilary. 1982. “Why There Isn’t a Ready-Made World,” in Synthese 51: 141-
167.

Putnam, Hilary. 1983a. “Analyticity and Apriority: Beyond Wittgenstein and Quine,” in
Realism and Reason: Philosophical Papers, Volume 3. Cambridge: Cambridge
University Press. Pp. 115-138.

Putnam, Hilary. 1983b. “Models and Reality” in Realism and Reason: Philosophical
Papers, Volume 3. Cambridge: Cambridge University Press. Pp. 1-25.

Putnam, Hilary. 1983c. “Reference and Truth,” in Realism and Reason: Philosophical
Papers, Volume 3. Cambridge: Cambridge University Press. Pp. 69-86.

Putnam, Hilary. 1983d. “There is at Least One A Priori Truth,” in Realism and Reason:

Philosophical Papers, Volume 3. Cambridge: Cambridge University Press. Pp. 98-
114.

Putnam, Hilary. 19836. “Why Reason Can’t Be Naturalized,” in Realism and Reason:

Philosophical Papers, Volume 3. Cambridge: Cambridge University Press. Pp. 229-
247.

Putnam, Hilary. 1987. The Many Faces of Realism. LaSalle, IL: Open Court Publishing
Company.

Putnam, Hilary. 1990a. “Meaning and Reference,” in The Philosophy of Language, 2nd
ed. J. P. Martinich, ed. New York: Oxford University Press. Pp. 308-315.

Putnam, Hilary. 1990b. “Realism With a Human Face,” in Realism With a Human Face.
Cambridge: Harvard University Press. Pp. 3-29.

Putnam, Hilary. 1990c. “Why is a Philosopher?”, in Realism With a Human Face.
Cambridge: Harvard University Press. Pp. 105-119.

Putnam, Hilary. 1994. “Comments and Replies,” in Reading Putnam. Peter Clark and
Bob Hale, eds. Cambridge: Blackwell Publishers, Inc. Pp. 242-295.

Quine, W. V. O. 1976a. “The Limits of Knowledge,” in The Ways of Paradox and
Other Essays. Cambridge: Harvard University Press. Pp. 59-67.

Quine, W. V. O. 1976b. “On Mental Entities,” in The Ways of Paradox and Other
Essays. Cambridge: Harvard University Press. Pp. 221-227.

Quine, W. V. O. 1976c. “Posits and Reality,” in The Ways of Paradox and Other
Essays. Cambridge: Harvard University Press. Pp. 246-254.

174

Quine, W. V. O. 1976d. “The Scope and Language of Science,” in The Ways of Paradox
and Other Essays. Cambridge: Harvard University Press. Pp. 228-245.

Quine, W. V. O. 1980. “Two Dogmas of Empiricism,” in From a Logical Point of View.
Cambridge: Harvard University Press. Pp. 20-46.

Quine, W. V. O. 1985. “Epistemology Naturalized,” in Naturalizing Epistemology,
Hilary Komblith, ed. Cambridge: The MIT Press. Pp. 15-29.

Quine, W. V. O. 1999. “Ontological Relativity,” in Metaphysics: An Anthology.

Jaegwon Kim and Ernest Sosa, eds. New York: Blackwell Publishers, Inc. Pp. 45-
62.

Quine, W. V. O. 2004a. “Naturalism; or, Living Within One’s Means,” in Quintessence:
Basic Readings ﬁom the Philosophy of W. V. Quine. Roger F. Gibson, Jr., ed.
Cambridge: The Belknap Press. Pp. 275-286.

Quine, W. V. O. 2004b. “Things and Their Place in Theories,” in Quintessence: Basic
Readings from the Philosophy of W. V. Quine. Roger F. Gibson, Jr., ed. Cambridge:
The Belknap Press. Pp. 229-248.

Rawls, John. 1985. “Justice as Fairness: Political not Metaphysical,” in Philosophy and
Public A ﬂairs 14(3): 223-251.

Rorty, Richard. 1979. Philosophy and the Mirror of Nature. Princeton: Princeton
University Press.

Russell, Bertrand. 1959. The Problems of Philosophy. London: Oxford University
Press.

Schlick, Moritz. 1959. “Positivism and Realism,” in Logical Positivism. A. J. Ayer, ed.
Glencoe, IL: The Free Press. Pp. 82-107.

Sellars, Wilfrid. 1963. Science, Perception and Reality. New York : Humanities Press.

Sober, Elliott. 1978. “Psychologism,” in Journal for the Theory of Social Behavior 8(2):
165-191.

Spinoza, Baruch. 1992. The Ethics. Samuel Shirley, trans. Seymour Feldman, ed.
Indianapolis: Hackett Publishing Company.

175

Stroll, Avrum. 2000. Twentieth-Century Analytic Philosophy. New York: Columbia
University Press.

Tymoczko, Thomas. 1989. “In Defense of Putnam’s Brains,” in Philosophical Studies
57(3): 281-297.

van Fraassen, Bas C. 1980. The Scientific Image. Oxford: The Clarendon Press.

Warﬁeld, Ted A. 1998. “A Priori Knowledge of the World: Knowing the World by
Knowing Our Minds,” in Philosophical Studies 92(1-2): 127-147.

Williams, Michael. 1980. “Coherence, Justiﬁcation, and Truth,” in The Review of
Metaphysics 34: 243-272.

Williams, Michael. 1986. “The Elimination of Metaphysics,” in Fact, Science and
Morality: Essays on A. J. A yer ’s Language, Truth and Logic. Graham MacDonald
and Crispin Wright, eds. New York: Blackwell Publishing. Pp. 9-25.

Wittgenstein, Ludwig. I999. Tractatus Logico-Philosophicus. C. K. Ogden, trans. New
York: Routledge.

Wright, Crispin. 1994. “On Putnam’s Proof that We Are Not Brains in a Vat,” in
Reading Putnam. Peter Clark and Bob Hale, eds. Cambridge: Blackwell Publishers,
Inc. Pp. 216-241.

Young, Iris Marion. 1990. Justice and the Politics of Difference. Princeton: Princeton
University Press.

176

       

MIC GAN TATE U VERSTY

111111111 11111111111111111111111.1171

3 1293 02956 8056