Kaplan vs Kornblith
In “Epistemology Denatured” Mark Kaplan critiques the notion of ‘non-regulative justification’ invoked by (some) externalists. He takes it that epistemological justification conceived without constitutive regulative or methodological implications is empty, it is epistemological justification robbed of its true nature. What is the epistemological naturalists’ brief in favor of non-regulative justification? What is Kaplan’s argument against it? Explain and assess.
The main body of the paper should be preceded by an abstract, single spaced, in italics, and just a few sentences long, which clearly and succinctly says what the paper is about, identifying its main conclusions or lines of argument. Try to restrict your exegesis and setting-up moves to approximately one-half, saving the rest for critical discussion or one or two points of your own that you wish to develop, or questions that you wish to raise
“INVESTIGATING KNOWLEDGE ITSELF”(2002)
p.1 Philosophy does not consist in an investigation and understanding of our concepts. The subject matter of epistemology is knowledge itself, not our concept of knowledge.
p.2 Our intuitions about various imaginary cases should not carry a great deal of weight in philosophical matters.
George Bealer’s View: naturalism makes no room for appeals to intuition. So much the worse for naturalism. The very practice of philosophy is incompatible with naturalistic epistemology.
p. 3 Kornblith will defend naturalistic epistemology against Bealer, arguing that philosophical theorizing may flourish while assigning a significantly smaller role to appeals to intuition.
1.1 Appeals to intuition: The phenomenon.
Examples: the Gettier Problem and would-be solutions.
p. 4 intuitions vs. empirical investigation. “No empirical investigation is called for, it seems. Each of us can tell, immediately and without investigation of any kind, whether the case described involves knowledge or justified belief or neither.”
1.2 Bealer’s account of intuition and the standard justificatory procedure.
p. 6 intuitions as prima facie evidence. “When we speak of intuitions we mean a priori intuition”…not just ‘commonsense’ beliefs. But epistemological naturalists of course reject the a priori (Quine et. al.) All justification is empirical.
[But what does Kornblith mean by “a priori”? Does he just mean what Quine meant? He doesn’t say, although there is one clue – see below.]
Bealer’s characterization of naturalism, as self-defeating in practice. (see p. 7).
Kornblith proposes to make room for the practice of appeals to inituition in epistemology, but rejects that there is anything a priori about this.
1.3 A naturalistic account of appeals to intuition.
p. 8 The intuitions to which philosophers appeal are widely shared…
p. 9 The intuitions of the majority, while not definitive, carry substantive evidential weight, at least in comparison those of a single individual, such as oneself.
It is a big mistake to treat these shared intuitions as delineating the boundaries of our shared concepts. That changes the subject, from the nature of knowledge top the nature of our concept of knowledge.
Kornblith’s critique of “deflationary”, “social constructivist accounts” of knowledge: see p. 10, 1st para. [Why is it about power, rather than about prediction and control to the benefit of human happiness and well-being, a la Haslanger? And why would the explanation of that social function have to be in terms of some underlying natural kind, knowledge?]
p. 12 top A causal theory of reference for natural kind terms being assumed. (Is it also being assumed that a causal theory of reference wouldn’t be appropriate for terms picking out social constructs?)
p. 12 middle: our intuitions about cases aren’t a priori, even if widely shared and obvious.
p. 13 Indeed, they are “…corrigible and theory-mediated” (emphasis mine). Aha!. (Discuss)
p. 14-15 The relation between intuition and theory to be understood dynamically. The appeal to intuitions at an early, pre-theoretic stage of investigation should give way over time to more empirical investigations of external phenomenon; such as the psychology of inference, concept formation, cognitive development, the social distribution of cognitive effort, the social structure of science, etc. This is what naturalistic philosophers do. [Well, that certainly seems to rule out any constructive approach….]
p. 17 The passing away of Cartesian intuitions…(Really?? Dream on!)
p. 18-19: Goldman’s ‘middle ground’: preserving a role for conceptual analysis of our folk-epistemological concepts. (Read.) Rejected on the grounds that Goldman is simply assuming that knowledge is not a natural kind. “Why should our folk epistemological notions be of any more interest to epistemologists than our folk chemical notions are to chemists?” Just as modern chemistry skips straight over folk chemistry to the project of understanding real chemical kinds, so we should consider “…the possibility that a similar strategy might be equally fruitful in epistemology.”
p. 20-21. “The use of intuitive judgement does not disappear at any stage of [naturalistic] theorizing. Instead, old intuitions give way to well-integrated theoretical judgements, and in addition to new intuitions about matters not yet fully captured in explicit theory.” So Bealer’s view is short-sighted. He gets off-track when he assumes that intuitions are a priori.
[Of course, Kornblith doesn’t comment upon the would-be nature and role of appeals to intuition in mathematics, logic, and various branches of linguistics. Those are hard cases for the epistemic naturalist.]
1.4 Naturalism and rules of inference
p. 22 Takes a potshot at Kaplan’s appeal to a priori standards of cogent argument: meeting a priori standards is simply irrelevant. Reliable rules of inference may not be a priori, while a priori rules of inference may be unworkable in practice or hopelessly mired in problems of computational complexity.
But what is view about modus ponens, or conjunction introduction? What is the argument here? Is it that if they had proven not to be reliable then we would have rejected them? And how would such unreliability have manifested itself? Can we make coherent sense of that?
1.5 Naturalism and epistemic terminology
p. 24 “…the legitimacy of epistemic terminology depends on its properly latching on to genuine theoretically unified kinds.”
The normative dimension of knowledge, its value to us, will be revealed by our empirical investigation of the natural kind. We will see why it is worth having.
[So, the question is: Why is knowledge (a natural kind) worth having?; not “What should we cognitively aim for (call it “knowledge”)?”.]
p. 25: Naturalism about knowledge and other epistemic properties does not imply theoretical reduction
1.6 The autonomy of philosophy
p. 26 The autonomy of a discipline consists in dealing with a distinctive set of questions, or in approaching certain phenomena with a distinctive set of concerns.
So, epistemology might be more autonomous than, say, chemistry. Like chemistry,it is continuous with the rest of empirical science. But like chemistry, it has its own distinctive questions and concerns. But unlike chemistry, its subject matter needn’t be theoretically reducible to physical theory. Wow.
MIDWEST STUDIES IN PHILOSOPHY, XIX (1994)
If it was not already obvious from the reception of Alvin Goldman’s work
on justification in the last fifteen years or so,’ if it was not already patent
in the success of Naturalizing Episremology, Hilary Kornblith’s collection of
papers (now in its second edition),z then the appearance of Philip Kitcher’s
“The Naturalists Return,” in the centenary issue of the Philosophical Review3
has made it abundantly clear naturalist accounts of epistemic justification are
very much in vogue.
That this should be so is doubtless due in part to Quine’s influencebut
by no means entirely. Most of the work on justification that has made its
way under the naturalist banner has been quite un-Quinean in its commitments,
full of unreconstructed talk of belief and yoked to the enterprise of conceptual
analysis in a manner much more reminiscent of Chisholm than of Quine.4 What
seems to have done the most to animate the re-emergence of naturalism in the
theory of justification is the influence of one particular criticism that has been
leveled at the theories of justification naturalism has sought to displace.
Those theories hold that a person’s belief in p is justified just if there is
a good argument for p that can be constructed from propositions she already
believes.5 This much granted, the task of the theory of epistemic justification
is clear: it is to specify what is required for an argument to count as good. The
debates that task has spawned are familiar enough. For example, foundationalists
maintain that there is a special class of self-justifying propositions such
that an argument can count as good only if it has ultimate premises that belong
to that class. Coherentists counter that no self-justifying premises are available
or necessary: an argument counts as good so long as it establishes that p bears
the appropriate relation of coherence with the person’s system of beliefs.
EPISTEMOLOGY DENATURED 351
But while naturalists tend to have sympathy for certain of the positions
that have been defended in the course of the debate (they tend, for example,
to endorse the coherentists’ denial that justification admits of the sort of
foundations upon which foundationalists insist), it is what is not under dispute
between foundationalists and coherentists that has most bothered naturalists.
What bothers the naturalists is the idea that the mere fact that there is a certain
sort of argument for p constructible frGm propositions she believes should
suffice to render a person’s belief that p justified. It is an idea Kornblith has
called the arguments-on-paper thesis.6
After all, suppose the person in question repudiates that argument. Worse
still, suppose she believes p only because she likes the sound of one of the
sentences that expresses p. Surely, say the naturalists, we are not going to say
in such circumstances that she is justified in believing p ! And why? Because,
they suggest, it is to the etiology of her belief that p that we are going to look
to assess whether that belief is justified. When we look at its etiology, we see
that the constructibility of the argument for p from her beliefs is in no way
causally responsible for her harboring her belief that p. And we reckon that the
process that is causally responsible is not the sort that confers justification on
the beliefs it produces and sustains.’
The moral the naturalists draw is that what determines whether a person’s
belief that p is justified is not the constructibility of a good argument for p from
her beliefs. What determines (at least in large part) whether her belief is justified
is, rather, the process by which her belief was caused and is sustained. The
theory of epistemic justification thus has no particular call to concern itself
with the anatomy of good argument. Its task is, rather, to specify the features
a process must have in order for the beliefs it causes andor sustains to count
as justified and to identify the processes that have these features.
The idea (quite plausible so long as the arguments-on-paper thesis was
in place) that one might determine from the comfort of one’s armchair the
epistemic propriety of a principle or rule of inference, is (the naturalists conclude)
fundamentally mistaken. Determining how justified beliefs are arrived
at is a thoroughly empirical matter. It is only through the scientific study of
the processes that cause and sustain our beliefs that we can determine by what
means we can and do acquire justified beliefs.* In particular, we must be open
to the possibility that this scientific study will reveal that principles of inference
that seem compelling in our armchairs set the wrong standards for appraising
the inference of actual human being^.^
Now, I hold no brief for the arguments-on-paper thesis and I am no
friend of foundationalism or coherentism. But it seems to me that the foregoing
critique of the arguments-on-paper thesis, so widely disseminated and
apparently so influential in naturalist circles, is itself fundamentally mistaken,
that the moral drawn from it is fundamentally misguided. My purpose here is
to explain why.
First a confession. I believe that Abraham Lincoln was assassinated. I believe
that the sun does not orbit the earth. I believe that Walter Mondale would not
have defeated Ronald Reagan in 1984 even if he (Mondale) hadn’t promised
at the nominating convention that he would raise federal taxes. I harbor all
these beliefs but I have not the slightest idea what the etiology of any of
them is. I cannot even remember when, or in what circumstances, I first came
by these beliefs. Nor can I be said to have done anything resembling serious
scientific research into the matter of what causal processes sustain the beliefs
now. Whatever the features are that a causal process must have in order to
produce justified beliefs, I find myself in no position to say whether any of
the three beliefs to which I have just confessed was caused (or is causally
sustained) by a process that has these features.10
Now, I expect that few naturalists will find my confession remarkable or
disturbing. It is no part of the naturalist credo that I should be able casually to
tell which of my beliefs are justified. Nor is it any part of the naturalist credo
that only such beliefs whose fortunate causal provenance I am aware of can
count as justified. But, in the light of the naturalist credo, it seems to me that
my confession should count as very disturbing indeed.
For, insofar as I find myself in no position to say, of any of the three
beliefs to which I have confessed, whether it enjoys the sort of fortunate
etiology required for it to count as justified by naturalist lights, then I cannot
but conclude that, should I embrace the naturalist credo, I will find myself
in no position to say, of any of these beliefs, whether that belief is justified.
That is. I will find myself in no position to say, of any of these beliefs, that
its propositional object is any more worthy of my belief than the propositions
reported as news on the front page of the National Enquirer-in no position
to say why I should not do what I can to divest myself of these beliefs.
But this is methodological madness. It is madness to suppose that I should
be trying to divest myself of these beliefs. And that is because it is madness
to suppose that I need to be in a position to say that a belief of mine enjoys a
happy etiology before I can feel that it is intellectually respectable to harbor the
belief. If the naturalist conception of justified belief is suggesting otherwise,
then it is surely mistaken.
I suspect that few naturalists will be womed. Their response will be that
their conception of justification suggests no such thing. The outline of this line
of thought is fully present in Goldman’s writings. “Evaluative, or normative,
discourse appears in different styles,” he writes. And one
division of styles distinguishes regulative and nonregulative normative
schemes. A regulative system of norms formulates rules to be consciously
adopted and followed, for example, precepts or recipes by which an
agent should guide his conduct. A nonregulative system of evaluation,
by contrast, formulates principles for appraising a performance or trait,
or assigning a normative status, but without providing instructions for the
EPISTEMOLOGY DENATURED 353
agent to follow, or apply. They are only principles for an appraiser to
utilize in judging.
He goes on to say that,
[flor the most part, my evaluative approach will be non-regulative.. , .
This non-regulative style will be adopted even in talking about justificatory
My argument against the naturalist account of justification, the naturalist can
thus claim, is a product of a mistake. I have mistaken the naturalist’s nonregulative
account of when a belief is justified for a regulative one. Having made
that mistake, I have gone on to complain that this account has outrageous
methodological consequences. But, once the mistake is exposed, my complaint
about the naturalist account of justification collapses. For that account means
only to capture what we are saying when, taking a nonregulative approach to
epistemic appraisal, we say a person’s belief is justified. And such an account
cannot have outrageous methological consequences. Being an account of the
nonregulative use of “justified,” it has no methodological consequences at all.
But this response to my complaint is problematic-and in two ways. The
first problem lies in the way Goldman distinguishes between regulative and
nonregulative systems of evaluation. The importance of a regulative system of
evaluation is clear, and made clear, by Goldman: it provides rules for agents
to follow in the conduct of inquiry. And what does a nonregulative system of
evaluation provide? The only answer offered is: “principles for an appraiser
to utilize in judging” a performance or trait “without providing instructions for
the agent to follow or apply.” But to what end, in what interest, is an appraiser
supposed to be making these judgments? Why is it supposed to be of any
importance that anyone make these judgments? Goldman tells us nothing.
The problem is that, if it is characteristic of the nonregulative use of
“justified” that it has no methodological import-that there is nothing in the
set of rules to be consciously adopted by an inquirer that calls for her to
determine whether any of her beliefs is justified in the nonregulative use of
that term-then it is hard to see what point there could be to a system of
evaluation dedicated to saying when it is that a person’s belief is justified in
the nonregulative sense. After all, in the sense of “justified” in question, it
would seem that we might engage in inquiry as scrupulously and carefully as
anyone can, yet never have occasion (and never suffer for our failure) even to
inquire into whether our beliefs are justified!
Moreover (and here lies the second problem with the naturalist response),
if it is characteristic of the nonregulative use of “justified” that it has no
methodological import, it is hard to see on what basis one could judge (as
the naturalists do in their critique of the arguments-on-paper thesis) that the
expression, “justified,” is being incorrectly applied.
To see why this is so, suppose that someone were to advance the following
proposal: your belief that p is justified (in the nonregulative sense) iff you either
find it psychologically impossible to doubt p’s truth (p is indubitable for you)
or you can produce on demand a derivation of p by truth-preserving rules from
a set of propositions each member of which is indubitable for you. The proposal
is a draconian one. Assuming that you are fairly normal psychologically, it is
a proposal that would deem most of your beliefs–certainly those about the
future, about everyday matters, about science-unjustified.
But is the proposal mistaken? Our immediate reaction may well be that
it is. Surely any proposal that deems all these beliefs unjustified is, eo ipso,
mistaken. But mistaken about whar? The proposal is not, after all, suggesting
that it is incumbent upon you to revise your doxastic attitudes toward the future,
toward matters everyday and scientific. It is not suggesting that you should be
uncomfortable about the fact that you harbor those doxastic attitudes. If it were
suggesting that, then we would do well to regard it as mistaken. It would be
issuing manifestly unacceptable methodological advice.
But, of course, the proposal cannot possibly be issuing unacceptable
methodological advice. It is, after all, a proposal whose only pretension is
to instruct us in nonregularive epistemic appraisal. Its only pretension is to tell
us when, in the nonregularive sense of “justified,” a person’s belief counts
as justified. That being the case, the draconian proposal cannot have any
But, if this is so, then we are perfectly free to endorse the draconian
proposal and continue (without fear of bad conscience) to persist in all the
practices of belief-formation in which we are wont to engage. All our endorsement
of the proposal would commit us to is a style of epistemic appraisal-a
style of appraisal whose judgments are of absolutely no import to our conduct
of inquiry-n which the practices in which we ordinarily engage in the course
of inquiry do not yield justified belief.
And if that is all our endorsement of the draconian proposal would commit
us to, then it is hard to see how we can possibly go wrong by making such an
endorsement. Indeed it is hard to see how we could go wrong endorsing any
other of the proposals with which the draconian proposal pretends to compete
(such as the foundationalists’, the coherentists’. and the naturalists’). With no
facts about what we are prepared (and not prepared) to believe capable of
undermining any of these proposals, we can endorse any one we care to (or any
pair or trio-why not commit ourselves to a variety of nonregulative styles?)
without fear of having made a mistake. A mistake is only possible when there
is some constraint on what counts as getting matters right. And here there is
no such constraint. Cut loose from the obligation to answer to our practice of
inquiry and criticism, proposals about how to make nonregulative judgments
about what beliefs are justified answer to nothing.
But don’t such proposals answer, at the very least, to facts about what we
know? Goldman writes that “a prime motivation for a theory of justifiedness
is the desire to complete the theory of knowledge”; that “justified belief is
necessary for knowing”; that he is “interested in the conception of justification
most closely associated with knowledge.”’* Can’t we look to the facts about
EPISTEMOLOGY DENATURED 355
what we know for evidence as to whether a proposal that would proscribe the
nonregulative use of “justified” is correct?
The answer is: “No.” As Gettier has taught us, no matter how scrupulously
a person weighs evidence and how carefully she adheres to the best
methodological principles, it remains an open question whether even the true
beliefs she forms as a result count as instances of propositional knowledge.
And that means that no adequate principle for evaluating whether a person
knows that p-no principle that would distinguish what she knows from what
is merely one of her well-reasoned true beliefs-an be regulative. After all,
since no person is in a position to make that distinction in the case of beliefs
she currently holds, the principle cannot serve as part of a recipe she can use
to guide her conduct. And, whatever might conceivably be at stake in how we
distinguish between those propositions she knows .and those in which she has
but well-reasoned true belief, it can have nothing to do with how we would
have her conduct inquiry. We have already granted that her conduct has been
exemplary: she has scrupulously weighed the evidence, she has adhered to the
best methodological principles. Thus, even when used by us to evaluate her,
no principle that would distinguish knowledge from well-reasoned true belief
can count as a regulative 0ne.13
But that is just to say that there is nothing more to constrain our attributions
of propositional knowledge than there is to constrain our nonregulative
attributions of justified belief. We are free to answer the question of whether our
agent’s well-reasoned true belief counts as knowledge however we like-we
are free to adopt any position concerning the nature and extent of propositional
knowledge-without fear that there will be something in our (or anyone else’s)
conduct of inquiry that will have to be altered to conform to our verdict. We
are free to take any position we like on the nature and extent of propositional
knowledge without consequence.14
And that means that we cannot hope to support proposals concerning the
nature and extent of justification (in the nonregulative sense of the term) by
appeal to facts about the nature and extent of propositional knowledge. For
it means that, bearing not at all on our conduct of inquiry and criticism, any
substantive thesis as to the extent and nature of propositional knowledge will
itself be every bit as immune to evidential constraint as are the proposals about
the extent and nature of justified belief it is called upon to support.
But surely, it will be replied, such proposals are subject to a very substantial
evidential constraint: they must answer to the rather sizable corpus
of ordinary, commonsense intuitions we have about which beliefs count as
justified. And what better proof can there be of this than that the arguments-onpaper
thesis founders precisely because it runs afoul of one of those intuitions?
But does it really founder? Recall that the naturalist account of justification
also seemed to founder on the reef of ordinary intuition-for example,
the intuition that I am in a position to say that my belief that the earth orbits
the sun is justified. Yet, with the aid of the distinction between regulative and
nonregulative styles of epistemic appraisal, the naturalist account was saved.
That distinction in hand, we were able to explain both why the intuition appears
to have such force and how the naturalist account can survive it unscathed. The
apparent force of the intuition derives from the fact that, taking “justified” in
the regulative sense, the intuition is quite correct. I am indeed in a position to
say that my belief is justified, in the sense that my belief meets the standards of
methodological hygiene. But (and this is why the intuition fails to do damage),
the naturalist account of justification does not claim otherwise. The naturalist
account is nonregulative in nature. Thus, it is only in the nunregulative sense
of “justified” that (should I embrace the naturalist account) I will find myself
in no position to say whether my belief is justified. The thing that I will find
myself in no position to say has nothing to do with the methodological propriety
of my belief.
But notice that we can float the arguments-on-paper thesis off the shoals
of apparently recalcitrant intuition by exactly the same maneuver. The charge,
recall, is that the arguments-on-paper thesis founders on the intuition that,
regardless of whether a good argument is constructible for p from the propositions
she believes, a person who believes p because she likes the sound of
a sentence that expresses p is unjustified in that belief. But now, equipped
with the regulativehonregulative distinction, it is easy to say why the intuition
appears to have such force and why the account it would seem to undermine
nonetheless manages to emerge unharmed. The apparent force of the intuition,
we can say, derives from the fact that, taking “justified” in its regulative
sense, the intuition is quite correct: no sound methodology worthy of the name
would sanction forming a belief in the manner described. But, we can say, the
arguments-on-paper thesis does not claim otherwise. The arguments-on-paper
thesis is nonregulative in nature: it is committed only to the claim that, in the
nunregulative sense of “justified,” her belief is justified. And this claim implies
nothing about what sound methodology sanctions.
Of course, what we have done to rescue the naturalist account and the
arguments-on-paper thesis from apparent disconfirmation by intuition, we could
do just as easily for the draconian proposal and any number of other proposals
that would tell us under what conditions a belief is justified in the nonregulative
sense of that term. Theses about justification in the regulative sense are fated to
survive only at the pleasure of our methodological intuitions. But theses about
justification in the nonregulative sense are answerable to none.
Thus, it seems to me, matters stand as follows. Maintaining that the
naturalist account of justification is a nonregulative one does indeed provide
refuge from the charge that the account has methodologically mad consequences.
But it purchases that refuge only at the price of undermining the
naturalist’s claim to have shown that the arguments-on-paper thesis is mistaken.
For that claim presupposes that it is possible to be mistaken in making a
nonregulative judgment as to whether a person’s belief is justified, that there is
something to which such a judgment is called upon to answer. And neither of
these presuppositions appears to be true. The conclusion seems unavoidable:
maintaining that the naturalist account of justification is nonregulative saves it
EPISTEMOLOGY DENATURED 357
from the charge that is methodologically mad only at the cost of rendering the
Not that I think that the foundationalists and coherentists-the advocates of
the arguments-on-paper thesis-are any better off. Consider why it is that they
are advocates of the arguments-on-paper thesis. The reason is that they are
committed to the view that no ordinary perceptual belief, such as my belief
that I am seeing a table before me, is self-justifying. Any such belief stands in
need of justification, a justification that requires a substantial bit of argument.
Foundationalists and coherentists recognize, however, that they cannot expect a
person to be able to produce on demand that substantial bit of argument. Apart
from serious students of epistemology, few people can be expected to have
much idea of what sort of argument is called for.15 And even those who do
have some idea of what sort of argument is called for are guaranteed to find the
task of filling in the details of such an argument difficult, if not impossible.16 So,
rather than have to claim that we have few, if any, justified perceptual beliefs
(a claim that they find unpalatable), foundationalists and coherentists opt for
the arguments-on-paper thesis-they hold that it is sufficient for a belief to be
justified that the right sort of argument for the belief merely exist.”
Now it seems to me quite clear that foundationalists and coherentists are
right to shrink from requiring, as a condition of our being justified in harboring
perceptual beliefs, that we be able to produce on demand the sort of arguments
these philosophers think are required to justify the beliefs. The consequence of
such a requirement would indeed be intolerable. Patently unable to do what is
required to render our perceptual beliefs justified, we would be forced (insofar
as we embraced foundationalism or coherentism) to draw the methodologically
mad conclusion that we have no business harboring these beliefs-that we
would be well advised to do what we could to effect their abandonment.
But how does the arguments-on-paper thesis avoid this mad methodological
consequence? Only by telling us that we may with good conscience
persist in a belief we know stands in need of substantive justification even as
we recognize that we cannot now, and may not ever be able to, produce the
justification it requires. But, while this is certainly a convenient and innocuous
doctrine when applied to ordinary cases of perceptual belief–cases in which
no reasonable person would ever challenge us to produce some justification
for our beliefs-the doctrine is pernicious if universally applied. For it licenses
a methodology on which one is free to persist in believing anything at all,
about any matter at all, with no regard for whether one can produce even
the most meager argument in the belief’s favor. In opting for the argumentson-
paper thesis, foundationalists and coherentists simply exchange one mad
methodological commitment for another. 18
To be sure, the foundationalist and coherentist can seek to block this
unhappy verdict by claiming their doctrines are being misread-that, properly
understood, they do not have any methodological import. But in this the foundationalist
and coherentist will fare no better than the naturalist did. Like their
naturalist relative, the foundationalist and coherentist theories of justification
can be stripped of methodological consequence only at the cost of rendering
them immune to any and all evidential constraint. The conclusion we drew
in the case of the naturalist account of justification applies just as well to the
foundationalist and coherentist ones: if they are not methodologically mad, then
they are em~ty.1~
What has gone wrong with the naturalist critique of the arguments-onpaper
thesis is, thus, not that the critique places the naturalist on the wrong side
of the issue. On the contrary, as I have just finished arguing, the argumentson-
paper thesis is, as the naturalists maintain, mistaken. Where the naturalists
go wrong is in their account of why the arguments-on-paper thesis is mistaken.
Far from criticizing the thesis for its mad methodological consequences, the
naturalists take for granted that the theory of justification to which the thesis
means to contribute is not supposed to be of methodological consequence. The
naturalists simply think that the foundationalist contribution is in error.
What we have seen is that this cannot be right. What we have seen is that,
insofar as a theory of justification is not of methodological consequence, then it
is simply empty. And insofar as it is of methodological consequence then it is
adequate only to the extent that its methodological consequences are acceptable.
What that would seem to mean is that the only way a theory of justification
can make an honest living is by providing methodological insight andor good
methodological advice. Properly conceived, the job of a theory of justification
is just to provide rules for (and insights into) the proper conduct of inquiry.
The mistake the naturalists make (a mistake they share with the advocates of
the arguments-on-paper thesis) is to suppose that a theory of justification can
do anything else.
But if the naturalist approach to the theory of justification is inspired by the
mistake of thinking that whether a belief is justified is determined (at least in
part) by the processes by which it is caused and/or sustained, it is not defined
by that mistake. Naturalists are, after all, also advancing an independent, and
substantive, conception of just how to go about improving the conduct of
inquiry. It is a conception that means to make quite a departure from the
traditional view of how methodological innovation is to be arrived at.
The traditional view is that at the heart of any good methodology must lie
a conception of good argument, of evidential support. What we are looking for,
after all, is beliefs for which we can muster good arguments, for which we have
adequate evidential support. Coming up with an adequate conception of good
argument and evidential support is largely a reflective exercise: a rehearsal of
possible cases, possible scenarios, possible objections all by the light of our
epistemic intuitions. Thus it is that, on the traditional view, a significant portion
EPISTEMOLOGY DENATURED 359
of the theory of justification as I have claimed it is properly conceived (as a
theory that means to provide us with methodological insight and advice) can
profitably be pursued in the comfort of an armchair.
But not on the naturalists’ view. According to the naturalists, the methodological
ambition of the theory of justification is to be pursued (in the words
of PhiIip Kitcher) “by describing processes that are reliable, in the sense that
they would have a high frequency of generating epistemically virtuous states
in human beings in our world,’’20 where prominent among those epistemically
virtuous states is true belief. It is an empirical, scientific task, the naturalists
maintain-a task for which the armchair approach to methodology is entirely
useless, hamstrung by its apriorism, i.e., by its nonempirical approach.2’
The apriorism involved in the traditional sort of armchair methodological
research, maintain the exponents of naturalism, founders on three fundamental
difficulties. First, armchair apriorism must necessarily proceed in ignorance of
the actual cognitive capacities of the human agents whose cognitive activities it
seeks to govern. This renders it singularly ill suited to determine what cognitive
processes are apt to be reliable in actual human agents. Second, having no
ingress to what is apt to work in the actual world, armchair apriorism has no
choice but to restrict itself to prescribing cognitive strategies that are optimal no
matter how the world may be. This is a quixotic endeavor that leaves aside and
(because it sets so high a standard of success) may even frustrate our attempts
to find what we really want: strategies that will serve us well in the world
we happen actually to inhabit. Third, armchair apriorism assumes that the fact
that a methodology is in accord with our epistemic intuitions shows that it
is worthy of our allegiance. But there is no reason why a methodology that
satisfies some other set of epistemic intuitions, a set of intuitions harbored (say)
by members of some other distant culture, might not serve our cognitive aims
better. The fact that a methodology accords with our intuitions concerning
rationality and evidential support is entirely compatible with its being quite
unreliable at delivering us cognitively virtuous states in our world-much less
reliable than some feasible alternative.22
Now, I confess that I am in considerable sympathy with the foregoing
line of thought. The project which the naturalists favor-to devise strategies
that are apt to lead beings such as we to achieve our goals in inquiry in the
world we happen to inhabit-is surely a worthy one. And it is just as surely
one that cannot plausibly be executed from an armchair. There are things we
need to know about our environment, about our cognitive capacities, and the
relationship between the two. (Perhaps a strategy that is apt to lead to disaster
in most circumstances is pretty effective in the environments in which we are
apt to use it.) This knowledge is not available apriori.
But I am at the same time struck by the fact, to show us all this, that
what the proponents of epistemological naturalism have offered us is a series
of arguments. And I am struck by the fact that they have not offered us any
scientific evidence that, were we to believe what they would have us believe as
a causal consequence of their having exposed us to these arguments, our belief
would be the result of our having instantiated a process that is apt to have a
high frequency of generating epistemically virtuous states in human beings in
our world. (No surprise here: there is not-and probably never will be-any
evidence of this sort available,) I am struck by the fact that they apparently
think that their arguments are sufficiently cogent to show us that they are right.
But are their arguments cogent? So long as the naturalists mean to be
showing their audience in spoken word and in print that their doctrines are
correct23 this question will be an urgent one. But how are we supposed to go
about trying to answer it? What are we to do-what can we do-to decide
whether the naturalists’ arguments are cogent?
It is hard to see what we can do except evaluate these arguments by
the light of the very sorts of epistemic intuitions which the naturalists-are so
eager to disparage. Granted, we might well have had other intuitions.24 Granted,
had we those intuitions, we might well have very different views about what
counts as a cogent argument. Granted, we do not know that heeding those other
intuitions would not yield a higher frequency of epistemically virtuous states
in human beings in our world than we achieve by heeding the intuitions we
actually have .25
But none of this poses any worry that either we or the naturalist can
afford to indulge. Were we unwilling to take the guidance of our epistemic
intuitions until it had been shown scientifically that there is no other set of
intuitions we might be capable of harboring that, if heeded, is apt to generate a
higher frequency of epistemically virtuous states in human beings in our world,
we would be forced to deny ourselves-now and for the foreseeable futurethe
only resources we have available with which to judge the cogency of any
argument, the naturalists’ arguments included.26 Like it or not, the naturalists’
attempt to show us the errors of aprioristic methodology depends for its success
on our consulting, and finding naturalist arguments in accord with, the very sorts
of armchair intuitions whose advice the naturalists would have us ignore.
Some naturalists will doubtless fail to be impressed by this line of argument.
The day is coming, they will respond, when argument will not have the
importance it now has, when our current preoccupation with the propriety of our
propositional attitudes will seem quaint. It will be other attitudes towards other
(perhaps hitherto unimagined) sorts of information-bearers that will occupy our
attention.*’ But suppose they are right. Suppose there will come a day when
we will give up all our dialectical chat for some yet-to-be discovered way of
tweaking our cognitive mechanisms. It is still true that, until that day comes
(if it ever does), the marshaling of arguments for and against the propriety of
harboring various propositional attitudes towards hypotheses will continue to
be central to the conduct of inquiry-for all of us. naturalists included. And,
if so, there is still a place for a theory of justification that seeks to clarify
and improve the standards by which we judge whether these arguments are
cogent. It is precisely this place that the traditional apriori armchair approach
to methodology occupies.
EPISTEMOLOGY DENATURED 36 1
This last point, that the traditional approach to methodology is to be understood
as offering a contribution to the understanding and improvement of the standards
of cogent argument, deserves emphasis. It is a point that many naturalists
apparently have failed to appreciate. as some of their criticisms of the traditional
For example, naturalists have often complained that the advice issued
by the traditional approach-for example, that a person’s beliefs should be
consistent, that they should reflect her total evidence, that her states of confidence
should be probabilistically constrained-is entirely unfollowable by any
human being. We simply do not have the cognitive capacity. That the traditional
approach issues such patently unfollowable advice only serves, they argue, to
illustrate the fruitlessness of armchair apriorism.28
Now this much can surely be granted. The three pieces of advice just
described are humanly unfollowable. And thus, if we think of these three
pieces of advice as describing conditions for the violation of which a person is
to be called on the carpet as a bad inquirer, then it is advice that is obviously
mistaken. We cannot reasonably call someone on the carpet for failing to follow
advice it was not in her capacity to follow. But if we think of the advice as
contributing to a theory of cogent argument-if we think of the three pieces of
advice as describing conditions the violation of which opens a person’s state of
opinion to legitimate criticism-matters look rather different. Then, what the
advice comes to is this: we should consider an inconsistency in a person’s set
of beliefs, a failure of her beliefs to reflect her total evidence, a failure on her
part to invest confidence in propositions in accord with axioms of probability
calculus, grounds for criticizing the cogency of her beliefs (or, in the last case,
of the way in which she has invested confidence in propositions). And, to judge
from naturalist writings (thick as these writings are with arguments to the effect
that, on pain of inconsistency, we must conclude so-and-so; with arguments to
the effect that, given some unduly neglected bit of evidence, we must conclude
such-and-such), this is advice to whose general propriety naturalists are (rightly)
very much committed.
Is the advice correct in its every detail? Some may want to dispute
that it is. They may, for example, think that the inconsistency involved in
believing that at least some of your beliefs are false does nor open your beliefs
to legitimate criticism. Or, they may think that, the last piece of advice to
the contrary, there is nothing wrong with investing more confidence in an
explanatory theory than we do in a proposition that merely describes one of its
observational consequences.29 In my view, it is a mistake to think either of these
things. But this is not the place in which to adjudicate this dispute.30 It suffices
to observe that the dispute is an intramural one, that how we adjudicate seems
to depend entirely on how we deploy the resources of the traditional approach to
methodology. Once we understand the traditional approach to methodology to
be offering a contribution to a theory of cogent argument, we can see that
362 MARK KAPLAN
considerations having to do with the cognitive capacities of human inquirers
have absolutely no bearing on questions concerning the propriety of the advice
the traditional approach issues.
The same thing goes for considerations having to do with the extent to
which our cognitive tendencies are well adapted to our world. Consider one
of the more conspicuous and powerful products of the traditional armchair
approach to methodology: the canons of statistical inference. A large number
of experiments conducted in the last twenty-five years or so have convincingly
shown that inquirers (even trained inquirers) tend systematically to violate
universally accepted canons of good statistical inference.31 Many of those
responsible for the experiments have found their results disheartening. But
some naturalists beg to differ. Writing about two prominent purveyors of gloom,
Hilary Kornblith writes that they
compare the logical form of our inferences with the logic of statistical
inferences, and on that basis, declare us sinners. Given the standards of
proper statistical inference, our inferences receive a failing grade. But is
the logic of statistical inference a reasonable standard against which to
measure our own inferences?3*
He thinks it is not. He argues, in the case of one sort of systematic violation,
that we are cognitively adapted toward our environment in just such a way
that the offending inferences we actually make (given the circumstances in
which we tend to make them) are quite reliable. Kornblith concludes that the
principles of statistical inference do not set appropriate standards by which to
appraise human inference.33
Now, if he means by this that it is inappropriate to condemn people for
their violations of the principles of statistical inference,34 then Kornblith is
surely right. Understood as a set of principles for the violation of which a
person should be called on the carpet, the principles of statistical inference do
seem quite unwarranted. But this, of course, is not how these principles are
to be understood. They are, rather, to be understood as a set of principles the
violation of which opens a person’s inferences to cogent criticism. And, unless
we are prepared to treat it as a matter of indifference whether the arguments
we produce in word and print on behalf of our views are statistically fallacious
(and I would wager that there is not a naturalist writing today who is prepared
to do this), then we have little choice but to conclude that, so understood,
the principles of statistical inference provide a perfectly reasonable standard
against which to measure our own inferences.
By way of closing, let me emphasize that I do not mean to be suggesting that
a scientific study of human cognitive mechanisms offers no hope of methodological
advance. On the contrary. The better we know the biases, tendencies,
strengths, and weaknesses of a tool, the more effectively we can deploy it to our
EPISTEMOLOGY DENATURED 363
ends. What I am arguing is that (many naturalists to the contrary) this scientific
study of cognition is no substitute for the traditional approach to methodology.
The methods and results of the traditional approach are indispensable to the
scientific study of cognition, as they are to all inquiry. An epistemology that
would pretend otherwise would be an epistemology denatured.35
1. See, for example, “What is Justified Belief’?’’ Justification and Knowledge, edited by
George S. Pappas (Dordrecht, 1979), 1-13; “The Internalist Conception of Justification,”
Midwest Studies in Philosophy V: Studies in Epistemology (Minneapolis, 1980), 27-5 1;
Epistemology and Cognition (Cambridge, Mass., 1986).
2. Hilary Kornblith, Naturalizing Epistemology (Cambridge, Mass., 1985 and 1992).
3. ‘The Naturalists Return,” Philosophical Review 101 (1992): 53-1 14.
4. See, for example, Goldman, Epistemology and Cognition, 38-39. Goldman expresses
some misgivings about the enterprise in “Psychology and Philosophical Analysis,” Proceedings
of the Aristotelian Society 89 (1988): 195-209. So does Hilary Kornblith, in his
Inductive Inference and its Natural Ground (Cambridge, Mass., 1993). 4-5.
5. More properly, they hold that a person’s belief that p is justified just in case, if p is
the sort of proposition lhot requires justification by argument (and for foundationalists, not
all propositions are of this sort), then there is a good argument for p that can be constructed
from propositions she already believes. But in what follows, I will continue for the sake of
convenience to omit the italicized qualification.
6. “Beyond Foundationalism and the Coherence Theory,” Journal of Philosophy 77
(1980): 597-612, 509.
7. For evidence of the prominence this sort of argument has in the naturalist canon,
see Gilbert Harman, Thought (Princeton, N.J., 1973), 30-31; Goldman, “What is Justified
Belief?” 9, and Epistemology and Cognition, 8 1-93; Kornblith, “Beyond Foundationalism
and the Coherence Theory,” 599-602; Kitcher, ‘The Naturalists Return,” 60.
8. These morals are spelled out with particular vividness in Kitcher, “The Naturalists
9. See, for example, Kornblith, Inductive fnference and Its Natural Ground.
10. I do not mean here to be denying that I find myself in a position to say things like
‘This belief is sustained by memory” or “This belief was produced by reasoning from some
sort of evidence.” But I do not for a minute suppose that this constitutes anything like a
scientifically adequate way of identifying cognitive processes. And it seems to me that, in
ignorance of how properly to identify cognitive processes and in the (consequent) absence
of any investigation (serious or othenvise) into which cognitive processes thus identified
have the appropriate justification-confening property, it would be folly of me to suppose,
nonetheless, that the three beliefs to which I have confessed are caused andor sustained by
cognitive processes that have that justification-conferring property. Alvin Plantinga makes
much the same point in his Warrant: The Current Debate (Oxford, 1993), 202.
11. Epistemology and Cognition, 25-26. See too his ‘The Internalist Conception of
12. Epistemology and Cognition, 76.
13. For a fuller elaboration and defense of the points made in this paragraph, see my
“It’s Not What You Know that Counts,” Journal offhilosophy 82: 350-63.
14. This consequence of the nonregulative nature of knowledge attributions has, indeed,
been embraced by some contemporary philosophers of skeptical bent, who rightly see in
it a way to wax skeptical about the extent of our propositional knowledge without thereby
opening themselves up to the claim that they are urging us to act as if we (say) don’t
know that the phlogiston theory is false or as if we don’t know that we have teeth. See, for
example, Barry Stroud, The Significance of Philosophical Scepticism (Oxford, 1984), ch. 2
364 MARK KAPLAN
and Bredo Johnsen, “Relevant Alternatives and Demon Skepticism,” Journal of Philosophy
84 (1987): 643-53, esp. 652-53.
15. Indeed, even the most eminent of epistemologists have had difficulty saying to their
own satisfaction (let alone to the satisfaction of others) what forn of argument is called for.
Roderick Chisholm, to take the most eminent among them, has been revising his account of
how the argument should go for over four decades.
16. Laurence BonJour concedes that this is true of the argument from the coherence of
a person’s belief system he favors, allowing that no person has a sufficient explicit grasp of
her system of beliefs to furnish such an argument (The Structure of Empirical Knowledge
[Cambridge, Mass., 19851 151-52). Given that there is no foundationaiist alternative that
does not itself involve some strong coherence constraint (see, for example, the coherence
constraint that Chisholm imposes in chapter 6 of his Theory ofKnowledge, 3rd ed. [Englewood
Cliffs, N.J., 1989]), it would seem that a comparable concession is warranted from
the foundationalist quarter as well.
17. For an expression of this line of thought, see (for example) William Alston, “Level-
Confusions in Epistemology,” Midwest Studies in Philosophy V, 1 44-45.
18. The commitment is not, however, so mad that it has failed to find philosophical
voice. See, for example, Isaac Levi, The Enterprise ofKnowledge (Cambridge, Mass., 1983),
1-2; Gilbert Harman, Change in View (cambridge, Mass., 1986). ch. 4. But see David
Christensen, “Conservatism in Epistemology,” Nois 28 (1 994): 69-89.
19. The last four paragraphs constitute a compressed version of the argument that
dominates my “Epistemology on Holiday,” Journal of Philosophy 88 (1991): 132-54.
20. Kitcher, “The Naturalists Return,” 75-76.
2 1, I am using the term apriorism here (and apriori below) only because it has been
used as an epithet by such naturalists as Kitcher and because, understood in its colloquial
sense, the use of the tern makes vivid the contrast between a methodology that is itself an
empirical, scientific enterprise and one which is a reflective one that can be pursued in an
armchair. Nothing here will turn, or bear, on any views about what, if anything, counts as
22. For a lucid bill of indictment in this vein, see Kitcher, ‘The Naturalists Return.”
23. As opposed to merely trying to sway their audience.
24. Our personal histories teach us that our epistemic intuitions tend to change over time.
And the history of human inquiry teaches us that what epistemic intuitions are widely held
in the scientific, and other investigative, communities has changed over time.
25. And we do know that some people have had intuitions that are-and that support
methodological rules that are-quite mad. See chapter 4 of Stephen Stich, The Fragmentation
of Reason (Cambridge, Mass., 1990) where Stich takes observations such as these to
undermine the integrity of armchair methodology.
26. Indeed, we would be forced to deny ourselves these resources even if we demanded
only that it be scientifically shown that our intuitions, when heeded, are apt to generate a
reasonably high frequency of epistemically virtuous states in human beings in our world.
27. See, for example, Stephen Stich, From Folk Psychology to Cognitive Science (Cambridge,
Mass., 1983); Patricia Churchland, “Epistemology in the Age of Neuroscience,”
Journal of Philosophy 84 (1987): 544-52; Paul Chruchland, A Neurocomputational Perspective
(Cambridge, Mass., 1989).
28. See, for example, Christopher Cherniak, Minimal Rutiodity (Cambridge, Mass.,
1986); Kornblith, Inductive Inference and Its Natural Ground, 5; Kitcher, “The Naturalists
29. Kitcher, together with a great many others of both naturalist and nonnaturalist
persuasion, thinks the first of these two things (Kitcher, ‘The Naturalists Return,” 85). Clark
Glymour (Theory and Evidence [Princeton, 19801, 83-84) thinks the second. The latter is
by far the more serious misgiving, striking at the heart of the last piece of advice.
EPISTEMOLOGY DENATURED 365
30. I address that task in “Believing the Improbable,” Philosophical Studies, (forthcoming)
and “Confessions of a Modest Bayesian,” in Reconstructing Philosophy? New Essays
in Meraphilosophy, edited by J. Couture and K. Nielsen, Cunadiun Journal of Philosophy
Supplemenrary Volume 19 (1993): 315-37.
3 1. See, for example, Daniel Kahnemann, Paul Slovic, and Amos Tversky, eds., Judgmenr
under [email protected] (Cambridge, 1982).
32. Kornblith, Inductive Inference and Irs Natural Ground, 90-9 I .
33. Ibid., ch. 5.
34. I follow Kornblith in writing as if there is agreement on what those principles are.
There is not. See Vic Bamett, Comparative Staristical Inference, 2nd ed. (New York, 1982).
35. I would like to thank Ken Gemes, Hilary Kornblith, Jamie Tappenden, and Joan
Weiner for helpful and timely comments on this essay and Peter Haddawy for useful
discussion in the early stages of its writing.