Plantinga's Proper Function Account
of Warrant
Alvin Plantinga has finally told us
what he considers to be the sober truth about one central epistemological
concept, the concept of warrant.[i] The concept he explores, which he terms
"warrant", is that warrant is that quantity enough of which, together
with true belief, yields knowledge, and he argues that warrant is to be
understood primarily in terms of proper function. To have a warranted belief, on this account,
is to have a belief that results from cognitive equipment that is properly
functioning, that is functioning as it was designed to function. Plantinga thus maintains that there is a
design plan for various parts of our cognitive apparatus, and the central
element of the account Plantinga defends is that warrant depends on cognitive
equipment functioning as it was designed to function.
Plantinga thus offers an approach
that begins by assessing the faculties or abilities of a cognitive system or
agent. Once such an assessment is
complete, the epistemologist is in a position to infer the epistemic status of
the doxastic products of those faculties or abilities. If the faculties are suitably virtuous or
excellent, or deemed to be functioning in an ideal way or at least in an
adequate way, the beliefs that result from the use of those faculties pass
epistemic muster, they receive some type of positive epistemic evaluation. The crucial element of such a view is the
primacy of evaluating the capacities of a system, and the derivative way in
which epistemic concepts are applied. We
might term such an approach "Aristotelian", for it is reminiscent of
Aristotelian approaches in ethics, where the fundamental evaluation applies to
persons and their characteristics, and derivatively to their actions. In the epistemological case, the derivative
evaluation applies to the beliefs that are products of systems determined to
have epistemically virtuous or adequate capacities.[ii]
Plantinga formulates his theory as
one concerning the epistemic notion of warrant, but I do not wish to pursue
whether one should prefer this notion or other notions in an account of
knowledge, notions such as justification, rationality, having the right to be
sure, etc.. Rather, I want to focus on a
more general issue, an issue concerning where to begin constructing a theory of
any of these epistemological concepts.
As we shall see, Aristotelianism in epistemology offers a distinctive
answer to the question of where to begin, and the point of this paper is to
argue that this Aristotelian answer is incorrect.
To determine where to begin, we might
first ask what kinds of things can have epistemic properties. Regarding the range of epistemic concepts
that have been proposed, at least two answers are available: beliefs (or believings) sometimes receive
epistemic support (warrant, justification, rationality, etc.), but propositions
(sentences, statements) also sometimes receive such support (for a particular
person in a particular situation.
Descartes' belief that he exists is epistemically supported for him, as
is the claim (proposition) that he, Descartes thinks, whether he believes it or
not. The first kind I will call
"doxastic support," and the second kind "propositional support."[iii]
This distinction is a common one in
recent epistemology, and the argument for making the distinction is
simple. Consider Holmes and Watson, both
of whom are in the same total epistemic situation regarding a particular case
they are working one (i.e., each has shared all they know with the other about
the case in question). As the case is
drawing to a close, they may even the the same belief about who is guilty. Both believe, let us suppose, that the butler
did it. There is a crucial difference,
however. Watson holds the belief on the
basis of a hunch he has that the butler did it; Holmes sees how the information
he and Watson share shows that the butler did it. He has, as he is wont to say,
"deduced" it.
Given the distinction between
propositional and doxastic support, we can express the difference between
Holmes and Watson as follows. Both
possess propositional support for the claim that the butler did it, but only
Holmes' believing is doxastically supported.
For Holmes believes that the butler did it because of, or on the basis
of , the information they both possess.
Watson believes only because he has a hunch, and so does not see how the
information shows that the butler did it.
The argument for the distinction and
its epistemological importance is thus straightforward, and once one notices
the distinction, a natural question to ask is how the different applications of
epistemic concepts are related. The
Aristotelian viewpoint on the matter is that the fundamental application of
epistemic concepts is to the products of epistemically suitable cognitive
systems, i.e., to beliefs. This
viewpoint falls directly out of the distinctive Aristotelian commitment that we
begin epistemological inquiry by evaluating cognitive machinery. Once such evaluation is complete, the
epistemic status of the products of that machinery, i.e., beliefs, is
determined by how adequate the equipment is.
So, for the Aristotelian epistemologist, the primary application of
epistemological concepts is to beliefs, not propositions. Plantinga affirms this viewpoint by saying,
According to the central and
paradigmatic core of our notion of warrant (so I say) a belief B has warrant
for you if and only if (1) the cognitive faculties involved in the production
of B are functioning properly . . . (2) your cognitive environment is sufficiently
similar to the one for which your cognitive faculties are designed; (3) . . .
the design plan governing the production of the belief in question involves, as
purpose or function, the production of true beliefs . . .; and (4) the design
plan is a good one: that is, there is a
high statistical or objective probability that a belief produced in accordance
with the relevant segment of the design plan in that sort of environment is
true.[iv]
The feature here that is important for my purposes is that
according to Plantinga, "the central and paradigmatic core of our notion
of warrant" is that "a belief B has warrant for you." In this regard, Plantinga is a paradigmatic
Aristotelian in epistemology--he holds that the primary application of the
concept of warrant is to beliefs, to the products of an adequate cognitive system.
Several other proposals in recent
years in the theory of knowledge have maintained the same viewpoint Plantinga
adopts. One example is particularly
instructive for my purposes. Alvin
Goldman, in "What is Justified Belief?", claims that the primary type
of justification attaches to beliefs in virtue of having been produced by a
reliable belief-producing mechanism.[v] Goldman's proposal is interesting because
after clarifying what he considers to be the fundamental application of the
concept of justification, he asks how we might understand talk of justification
when it applies to propositions. Goldman
argues for the completeness of his understanding of justification by claiming propositional
justification is derivative on doxastic justification. Goldman thus defends a version of what I will
call "doxasticism" according to which the basic kind of justification
is the kind that attaches to beliefs, and any other kind of justification,
including that which attaches to propositions, can be defined in terms of the
doxastic kind.
Plantinga’s work on the concept of
warrant includes no such effort, for Plantinga never attempts to characterize a
notion of propositional warrant to supplement the notion of doxastic warrant on
which he focuses. In fact, there is some
reason to think that Plantinga would deny that there is any such thing as
propositional warrant at all, for Plantinga begins his study of warrant by
claiming, “My topic is warrant: that, whatever precisely it is, which together
with truth makes the difference between knowledge and mere true belief.”[vi] And in the Preface to the second volume, he
characterizes warrant as “this elusive quality or quantity enough of which,
together with truth and belief, [that] is sufficient for knowledge.”[vii] If warrant is that which plugs the gap
between true belief and knowledge, or if warrant is that which plugs such a gap
when there is enough of it, then there will be no kind of warrant which is propositional
in nature.[viii] For no amount of propositional support alone
will be sufficient for plugging the gap between true belief and knowledge.
This latter point may be missed if
one thinks only of cases in which one has propositional support for claims one
does not believe. In such a case, it may
be tempting to think that if one believed the claim in question and there was
enough propositional support for it, one would have knowledge. But one can not only have propositional
support for claims one does not believe, one can also believe a claim on the
basis of something other than that which propositionally supports it. In such cases, knowledge is lacking precisely
because of the failure to base one’s belief on that which supports it
epistemically. Because of this latter
possibility, no amount of mere propositional support is sufficient to close the
gap between true belief and knowledge.
Still, Plantinga does not simply deny
the possibility of propositional support even though his remarks about warrant
imply that there is no kind of warrant which is propositional in
character. In circumscribing his topic,
Plantinga not only claims that beliefs or believings are subject to epistemic
appraisal, he also notes quite rightly that we appraise not only a person’s
beliefs “but also her skepticisms or (to use another Chisholmian term) her
withholdings, her refrainings from belief.
An unduly credulous person may believe what she ought not; an unduly
skeptical (or cynical) person may fail to believe what she ought.”[ix] Such remarks indicate that the topic of
Plantinga’s work must take into account not only some property, complex or
otherwise, that attaches to beliefs, but also some property, complex or
otherwise, that attaches to things other than beliefs.
Unfortunately, we do not find in
Plantinga's two‑volume work any discussion of this further property. What we do find is an account of warrant on
which it is a property of beliefs, leaving unaddressed the question of the
nature of this property that attaches to something other than beliefs. At first glance, however, the supplementation
looks simple. Warrant, we are told, is
roughly the property a belief has when it is the product of properly
functioning cognitive equipment. This
same property can attach to other cognitive attitudes as well, attitudes of the
sort mentioned above. For example, a
person’s skepticism about a claim might be the result of properly functioning
cognitive equipment, and thus the withholding of belief in such a case would be
warranted. The resulting view would then
hold that warrant is not just a property that attaches to beliefs but also to
other cognitive attitudes such as disbelief and withholding.
Such an approach is a broadened form
of the Aristotelianism described earlier. Instead of claiming that fundamental epistemic
evaluation applies to beliefs, it claims that it applies to intentional
attitudes; thus I will distinguish between doxasticism, which claims that
fundamental evaluation applies to beliefs and that other kinds of epistemic
evaluation must be understood in terms of this fundamental kind, and
Aristotelianism, which claims that fundamental epistemic evaluation applies to
intentional attitudes besides beliefs, and that any other kind of epistemic
evaluation is to be understood in terms of the kind that applies to intentional
attitudes. The passages cited show that
Plantinga is committed to the view that epistemic evaluation applies to
intentional states other than belief, so he would be a doxasticist only if he
has some way of explaining such evaluation in terms of the kind of evaluation
that applies to beliefs, e.g., if he could explain the warrantedness of withholding
in terms of the warrantedness of believing. Plantinga never attempts such an explanation,
nor does he ever address the distinction between terms of epistemic appraisal
applying to propositions rather than beliefs.
So we are left in the dark about precisely what view he takes on the
matter. Of this, however, we can be
sure: he his at least an Aristotelian, for he thinks epistemology is to be done
by beginning with an evaluation of our cognitive machinery and applying the
results of this evaluation to beliefs and other mental states. Whether he is merely an Aristotelian or a
doxasticist as well, we cannot know from his writings.
Non-doxasticist Aristotelians would
seem to have some advantage over doxasticists, if only because they have less
to explain. The non-doxasticist
Aristotelian needs only to explain how any kind of propositional support that exists
can be understood in terms of the epistemic evaluation that applies to
intentional attitudes. Furthermore, this
non-doxasticist theory does seem able to handle much of what motivates the
distinction between propositional and doxastic support. When there is propositional support for a
claim a person does not believe, the non-doxasticist Aristotelian can claim
that such support is to be understood in terms of the unwarrantedness of that
person’s withholding regarding that claim.
Unfortunately, things are a bit more
complicated. Since withholding,
disbelieving, and believing are all types of intentional attitudes, the above
account of propositional warrant will fail.
For it is simply false that we take one of these intentional attitudes
toward every proposition whatsoever, for a necessary condition for taking any
of these intentional attitudes toward a proposition is that one also conceive
of that proposition. Since there are
many propositions which we never conceive of, there are many propositions we
never believe, disbelieve, or withhold concerning. Nonetheless, such propositions can be the
recipients of epistemic support.
So the success of Aristotelianism in
epistemology is not so easily secured.
There are of course other options for the Aristotelian to try here, an
obvious one being the counterfactual approach on which propositional support is
to be understood in terms of some attitude that would be justified or
warranted if it were held. We will
consider such proposals later, but the point to note here is that some
complications will have to be introduced in order for Aristotelianism to
succeed. I believe, however, that the
situation is worse, for I believe that Aristotelianism in epistemology cannot succeed. I believe, that is, that no version of
Aristotelianism, the approach that begins by evaluating cognitive faculties and
applies terms of epistemic appraisal to beliefs or other intentional attitudes
on the basis of this evaluation of faculties, can succeed in explaining how
terms of epistemic appraisal apply to propositions. Furthermore, I believe that an approach that
claims that terms of epistemic appraisal apply fundamentally to propositions
can explain the possibility of epistemic evaluation of beliefs and other
intentional attitudes. I will give the
name “propositionalism” to this view.
As stated, both the Aristotelian and
the propositionalist agree that there is a fundamental kind of epistemic
appraisal, and seek to understand the other kind in terms of this fundamental
kind. Of course, assuming
interdefinability has its risks. The
alternative view, that terms of epistemic appraisal are ambiguous between
propositional and doxastic uses, is not initially attractive, however. For it seems quite easy to understand
doxastic uses in terms of propositional:
doxastic justification, for example, is just propositional justification
plus proper basing, i.e., propositional justification where one holds a belief
on the basis of, or because of, that which justifies it or is epistemically relevant
to it. It will not do, however, for
Aristotelians to appeal to this definition, for it is a version of
propositionalism--it defines doxastic support in terms of propositional rather
than the other way around. Nonetheless,
the point remains that the interdefinability assumption is supported by this
propositionalist proposal, for in order to defend the ambiguity thesis, one
would have to argue against the plausibility of this account.
So, the interdefinability position is
quite attractive. Regardless of whether
it is the most attractive or the least attractive, however, any complete
epistemological theory should say what the relationship is between
propositional and doxastic terms of epistemic appraisal. Plantinga does not address the question, and
the quick proposal above to supplement his account by allowing warrant to apply
fundamentally not only to beliefs but to other propositional attitudes is not
successful. Still, the propositionalist
account above may offer hope to the Plantinga and other Aristotelians, for if
one can define X in terms of Y, one can often define Y in terms of X.[x] In this case, however, it cannot be
done. I will argue that the basic kind
of epistemic support cannot attach to beliefs or other intentional attitudes,
as Plantinga implies, and that if it is not the basic kind, the right
conclusion to draw is that proper function is at best of ancillary interest to
the epistemological project of understanding the terms of epistemic
appraisal.
I will begin by considering the narrower
view, doxasticism, and then extend the discussion to Aristotelianism more
broadly. Moreover, in the discussion
that follows I will use Plantinga’s favored term ‘warrant’, writing of both
propositional and doxastic warrant, even though, as we have seen, Plantinga’s
explicit remarks about warrant imply that there is no kind of warrant that is
propositional in character. I could have
spoken more circumlocutorily to avoid the appearance of violating Plantinga’s usage,
but there is little to be gained by it.
Those more wedded to Plantinga’s usage may feel free to interpret my the
relationship between Plantingian warrant--that which plugs the gap between true
belief and knowledge--and propositional warrant akin to how we treat talk of
Senators and former Senators. I do not
care if one wants to insist that propositional warrant is not really a kind of
warrant at all; what matters here is that terms of epistemic appraisal apply to
propositions, and I will speak of propositional warrant in discussing that phenomenon.
The Problem for Doxasticism
Consider how one might attempt to understand
propositional warrant in terms of doxastic warrant. An initial difficulty is that sometimes one
has warrant for things one does not believe; that is, propositional warrant can
obtain without doxastic warrant. The way
to proceed, however, is fairly obvious:
in such cases, we imagine what things would be like were one to believe
the claim in question. If things go well
for the doxasticist, we will be able to say of the imagined situation that it
is one where doxastic warrant obtains.
So, the place to begin is with a counterfactual: a particular propositional content is
warranted for you, say, when you do not believe it, if you would have doxastic
warrant for the belief had you believed it.
Such an approach faces first the
Problem of Cognitive Admirability.
Consider the case of Sally the Scrupulous Scientist. Sally is scrupulous about scientific beliefs,
so scrupulous that she would not hold such a belief unless it were warranted
for her. Among the things she presently
holds no belief about is whether superstring theory is correct. Since she is so cognitively admirable, she
would not believe this claim unless believing it would be warranted for
her. According to the simple
counterfactual approach stated above, she thereby has propositional warrant for
the claim that superstring theory is correct.
For it is true of Sally that if she were to believe that superstring
theory is correct, her belief would be warranted. Since she has no such warrant, the simple
counterfactual approach does not specify a sufficient condition for
propositional warrant.
This problem is easy to explain from
a propositionalist's point of view, that point of view according to which the
basic kind of warrant is propositional warrant.
In the imagined counterfactual situation in which Sally believes that
superstring theory is correct, Sally's total epistemic situation is radically
different. Such a difference is implied
by the description of Sally as unusually meticulous about conforming belief to
evidence. Because of this difference, it
is not surprising that her doxastic warrant in the counterfactual situation
does not line up with what she actually has by way of propositional
warrant.
If the doxasticist could make use of
these claims, there would be a simple way to restrict the simple counterfactual
approach. Instead of saying that
propositional warrant obtains when and only when a person's belief in that
proposition would be justified if held, the doxasticist could say that
propositional warrant is to be understood in terms of having a doxastically
warranted belief in counterfactual situations where one holds the belief in the
same total epistemic situation, in that situation in which one's total evidence
remains the same.
Unfortunately, this account is
circular. Concepts such as the concept
of evidence or total epistemic situation must be understood in terms of
propositional warrant. One thing is
evidence for another provided the former warrants the latter, and this warrant obtains in the presence of
the former whether or not one believes the latter. Hence, the kind of warrant imparted is
clearly propositional warrant.
Similarly, two individuals are in the same total epistemic situation if
and only if they have precisely the same evidence. So the problem raised for the doxasticist is
to try to restrict the simple counterfactual approach to avoid counterexamples
like the case of Sally without relying on notions like evidence and total
epistemic situation that make the resulting account circular.
Here is how Goldman addresses this
problem. Instead of restricting the
counterfactual situation to those in which the person has the same evidence, he
restricts it to those in which the person's cognitive state is pretty
much the same as it presently is. He
says a person S is propositionally justified in believing p
"if and only if there is a reliable belief-forming operation available to S
which is such that if S applies that operation to his total cognitive
state at t, S would believe p at t-plus-delta (for
a suitably small delta) and that belief would be [doxastically]
justified."[xi] Since the concept of a cognitive state can be
specified without appeal to any epistemic concepts at all, it follows trivially
that it can be specified without appeal to the epistemic concept of
propositional warrant.
Of course, doxasticists cannot
require identity of cognitive state in the proposed counterfactual situation,
for such situations involve an added belief.
Goldman is somewhat cavalier about the problems involved in specifying
adequate overall similarity of total cognitive state. From the passage just quoted, all we are told
about which counterfactual situations are relevant is that they are ones in
which an "available" reliable belief-forming operation is applied to
"one's present cognitive state".
Such an approach hardly faces the problem. Some people hold beliefs in the face of
direct perceptual information to the contrary.
The "available" reliable belief-forming mechanism of
perception has already been applied, and they've ignored it. Presumably, that implies that if it were
applied, they would not have a doxastically warranted belief concerning what perception
shows (because they would not have a belief).
Furthermore, if perception is always
an "available" reliable mechanism in those who have the capacity to
see, then when my eyes are closed, I have propositional warrant for the
existence of everything around me within my range of vision. If I simply opened my eyes, the application
of the belief-forming operation of perception would result, more or less
immediately, in a doxastically warranted belief. But surely I do not always have propositional
warrant regarding the existence of every visible thing around me, e.g., when I
am sleeping or daydreaming.
My point, though, is not to try to
argue that refinement which addresses these concerns is impossible, but only to
show the difficulty of the task. In what
follows, I will assume that the problems can be solved, and thus will assume
that some doxasticist surrogate for total epistemic situation in terms of
(sufficiently similar) total cognitive state can be found. For there are other problems that are deeper
and more general.
One problem arises out of recent
literature which shows that belief can undermine evidence or reasons for
belief. One such case, due to Richard
Foley, goes as follows. You are close
enough to graduating that you know you will graduate if and only if you pass a
final exam. You have no beliefs about
the matter, but you do have evidence that you will pass. This initial description, then, is just
another case in which you have fully adequate evidence, i.e., propositional
warrant, for a claim you do not believe.
Suppose, however, that your examiners have it in for you, and will
change the exam to be so difficult you will not pass it should you come to
believe that you will pass it (and they will change the exam at the precise
instant of belief formation).
Furthermore, you know that this is so.
Then you are in a position where you have adequate evidence for the
claim that you will pass the exam, but you also know that if you were to
believe that claim, you would not rationally believe it.[xii]
There are also propositional contents
which themselves conspire to make doxastically warranted belief
impossible. One can have propositional
warrant, for example, for the claim that one has never considered the
proposition that the square root of 625 is 25 (perhaps you have just begun
learning about exponents and the like and you know you've not progressed past
single-digit numbers). Believing the
claim, however, would undermine the evidence you have for it, with the result
that no one can be doxastically warranted in believing that they have never
considered the proposition that the square root of 625 is 25.
What has not been noticed is the
implication of such cases for doxasticism.
In all these cases, the counterfactual analysis central to doxasticism
fails. If you were to believe that you
have never considered the proposition that the square root of 625 is 25, your
belief would not be warranted. Yet, you
have propositional warrant to that claim.
And if you were to believe you will pass the test, your belief would not
be warranted, in spite of the fact that you now have warrant for the claim that
you will pass.
The only way out for doxasticists is
to deny that the relevant counterfactuals are false. How might they responsibly do so? They cannot do so by denying that one can
have warrant for something one does not believe, nor can they responsibly deny
that there are cases in which such warrant would be undermined by belief. What is left, then, is to focus on the
antecedent of the counterfactual, which requires believing the proposition in
roughly the same total cognitive state.
The only possibility left open for doxasticists is to say that the cases
in which believing undermines reasons for belief do not involve belief in
roughly the same total cognitive state.
As already noted, the doxasticist
cannot insist on identity of total cognitive state, for some changes have to
occur for the belief to be added--minimally, adding the belief itself. The doxasticist idea is to minimize the changes
in such a way as to be able to mimic the propositionalist's notion of total
epistemic situation. The doxasticist
idea then is to define what it is for total mental states to be sufficiently
similar to each other, and hope to find such a definition which implies
that the mental states cannot be sufficiently similar when adding a belief that
undermines reasons for believing. If
this result can be obtained, doxasticists can deny the falsity of the
counterfactual by which they define propositional warrant in terms of doxastic
warrant by appeal to the standard semantics for counterfactuals. According to the standard semantics, all
counterfactuals with necessarily false antecedents are trivially true. Some object to this implication of that
semantics, but the semantic theory is a very useful and powerful theory, and
jettisoning a few intuitions because of these virtues may be worth the
price. The result is that doxasticists
may be able to rescue their view from counterexamples concerning belief which
undermines warrant by taking refuge among standard semanticists regarding
counterfactuals. For example,
doxasticists might claim that it is simply impossible for there to be a
sufficiently similar cognitive state to the actual one in which you believe
that you will pass the test, or in which you believe that you have never
considered the proposition that the square root of 625 is 25. Because such a state is impossible, the
counterfactuals employed by the doxasticists all come out true, thus yielding
the appropriate consequence that you have propositional warrant for the claim
that you will pass the test and that you have never considered that the square
root of 625 is 25.
This response opens doxasticism to
further problems, however, problems that are irremediable. One problem concerns claims of the form I
have never considered the proposition that p propositionally warranted,
even when the claim is known to be false by the person in question. For the same reasons as those given above for
thinking that it is impossible that one be in a cognitive state sufficiently
similar to the actual on in which one believes that one has never considered
the proposition that the square root of 625 is 25 will be reasons for think
that there are no possible states sufficiently similar to the actual one in
which one believes any proposition of the form I have never considered p. For to consider such a proposition is,
transparently to all, to consider p; it is intrinsically the kind of
proposition that would require radical alterations of our constitution to
believe. So the reasons for claiming
that the antecedent of the required counterfactual is necessarily false in the
case of I have never considered the proposition that the square root of 625
is 25 are also reasons for thinking that the required counterfactual is
necessarily false for any proposition of the form I have never considered p. Obviously, however, some claims of this sort
are propositionally warranted and some are not, in spite of the implication of
this defense of doxasticism that all of them are propositionally
warranted. For example, the proposition I
have never considered the claim that doxasticism is false is not
propositionally warranted for me.
Another problem for this defense of
doxasticism arises from noting that cases in which believing undermines
evidence are just the reverse of cases in which belief itself makes the content
of the belief true or likely to be true.
Believing that you will succeed makes success more likely, and knowing
that this is so increases the warrant we have for the claim that we will
succeed when we believe that we will. In
such cases, we will be unable to explain the degree of warrant a proposition
has when not believed in terms of what degree of warrant it would have if it were
believed.
Such cases raise problems for
doxasticists similar to the problem raised by cases in which belief undermines
warrant. Regarding the latter problem,
doxasticists must say that it is impossible to come to hold a belief that
undermines warrant in a cognitive state that is suitably similar to the actual
state. Any reasons for saying so,
however, will be reasons for saying the same thing about about cases in which
believing creates warrant. They will
have to say that there is no possibility of holding the belief that success is
imminent without making changes to one's total cognitive state that rule out
the counterfactual total cognitive state as being sufficiently similar to one's
actual total cognitive state. The
implication of such a claim, however, is that the counterfactual in question
will always be trivially true, yielding the result that we always have
propositional warrant for thinking we will succeed at anything whatsoever. Robert Schuler or Norman Vincent Peale might
think so, but they’ve apparently not thought about trying to put a Cadillac up
your nose.
There is a deeper problem as
well. I'll begin with some specific
cases, and then discuss what they have in common. Consider the proposition that I do not exist. I doubt that I can believe that proposition,
and if I cannot, then any doxasticist proposal resting on the standard
semantics is bound to fail. For if I
cannot believe that claim, any counterfactual having in its antecedent the
requirement that I believe the claim will be trivially true according to the
standard semantics.
Suppose however that I can believe
that I don't exist. Even if I can
believe that I don't exist, Plantinga (and anyone else committed to a
traditional western conception of God) is committed to the view that God cannot
believe that he does not exist.
Plantinga believes that God exists and necessarily so, and is
essentially omniscient, and that eliminates the possibility of God being an
atheist. If so, however, any doxasticist
proposal resting on the standard semantics for counterfactuals will count the
proposition God does not exist as being propositionally warranted for
God. Perhaps there are theology
departments that would welcome such as theological novelty rather than
philosophical debacle.
Finally, suppose certain
psychological states are self-presenting, implying that it is impossible to
consider whether you are in such a state and be wrong about it. Suppose further that you are in some such
state P, and consider the proposition I am not in psychological state P. Since there is no world in which I believe
that proposition in the same total cognitive state I am presently in, any
counterfactual whose antecedent stipulates that I believe that claim is a
counterfactual with an impossible antecedent.
Each of these counterexamples
illustrates the Problem of Essential Cognitive Admirability. A person of such a nature is essentially such
that he or she would not believe any of some range of propositions without
warrant, i.e., it is not possible for that person to believe any such proposition
without holding a doxastically warranted belief. No counterfactual account of propositional
warrant can be adequate to such possibilities if it accepts the standard
semantics for counterfactuals. For no
matter how one restricts the antecedent of the relevant counterfactual, the
only situations in which such a person holds the belief in question are worlds
in which the belief is warranted. This
latter fact holds even for propositions that the person presently knows are
false, as long as those propositions are within the range of propositions about
which the person is essentially cognitively admirable.
Examples concerning essential
cognitive admirability are especially telling against Plantinga, for Plantinga
believes that there is an individual who is essentially cognitively admirable
over the entire range of propositions.
Since God is essentially omniscient, every proposition is such that God
would believe it if and only if it were doxastically warranted for him. Given the doxasticist proposal that rests on
the standard semantics for counterfactuals, this piece of theology implies that
every proposition whatsoever is propositionally warranted for God, even those
known by God to be false.
The Problem of Essential Cognitive
Admirability does not depend on appeals to omniscience, however. For we need not imagine cases in which a
person is essentially cognitively admirable over all propositions in order to
see the problem. More limited kinds of
cognitively admirability are fully sufficient to show that such accounts must
fail. The history of epistemology is
rife with examples of theorists who claimed that human beings are essentially
cognitively admirable over some range of propositions. For many epistemologists have thought that
human beings are incorrigible or infallible about their own sensory states, or
appearance states, or sense‑data.
Others have implied that we are essentially cognitively admirable about
our own existence or the fact that we are a thinking thing. According to such epistemologists, humans are
essentially such that they know the truth of certain claims if those claims are
true and are believed to be true. Such
epistemologists might be wrong, but it is surely possible that beings exist who
are essentially cognitively admirable in these or other ways, and the mere
possibility of such essentially cognitively admirable beings is sufficient to
undermine doxasticism.
So what is a self‑respecting
doxasticist to do? In a word (actually
two), give up. Clearly, no account of
propositional warrant in terms of doxastic warrant is going to work. First, one has to get a belief into the
picture when as things actually stand there is no such belief. To do so is to imagine a non‑actual
situation, thereby landing us in the domain of counterfactuals, and doxasticism
is impaled on the horns of a dilemma concerning how to interpret certain
counterfactuals. Doxasticists can either
accept or reject the feature of the standard semantics that counterfactuals
with necessarily false antecedents are trivially true. If they reject the semantics, their view
founders on the problem of belief undermining reasons for believing; if they
accept the semantics, their view founders on the problem of essential cognitive
admirability and on cases in which belief itself creates evidence.
The Problem for Aristotelianism
The problems for doxasticism are
easily generalized to Aristotelianism more broadly conceived. Recall that doxasticists claim that terms of
epistemic appraisal apply fundamentally to beliefs, whereas Aristotelians are
committed only the the application of these terms to cognitive states, some of
which are beliefs but others of which are not.
In particular, broader Aristotelians allow terms of epistemic appraisal
to apply to withholdings or skepticisms in addition to believings. Such a broader view is of little help,
however, in addressing the problems faced by doxasticists.
Broader Aristotelians may initially
seem to have some resources that doxasticists lack. For example, suppose there is propositional
warrant for a claim that one does not believe.
If one takes the intentional attitude of withholding concerning that
proposition, one might be tempted to think that the propositional warrant in question
could be explained in terms of the unwarrantedness of withholding. If such an account were successful, broader
Aristotelians would have an advantage over doxasticists, for doxasticists are
forced to apply some counterfactual analysis to such cases, and we have seen
the difficulties that such analyses encounter already. The above suggestion is that broader
Aristotelians could escape the need for a counterfactual analysis in such
cases.
The appearance is misleading for two
reasons. First, it is not clear that the
above account is successful even for the kind of case in question. For withholding concerning a proposition
would be unwarranted if either that proposition or its negation were warranted,
so the account of propositional warrant suggested above for cases in which one
withholds concerning that proposition fails to provide a sufficient condition
for propositional warrant. Furthermore,
broader Aristotelians will not be able to avoid the need for a counterfactual
analysis to handle this difficulty as well as difficulties arising from cases
in which there is propositional warrant for a claim toward which one takes no
intentional attitude whatsoever.
Moreover, the problems for doxasticism concerning belief that undermines
warrant, belief that creates warrant, and cases of essential cognitive
admirability will constitute the same plague of death for broader
Aristotelianism as they do for doxasticism.
Broader Aristotelianism provides no haven in which to take refuge in the
face of the problems for doxasticism.
What is left, then, is either to
become a propositionalist, one who holds that the basic kind of warrant is
propositional warrant, or to defend an ambiguity thesis about the relationship
between propositional and doxastic warrant.
As we have already seen, however, the prospects for the ambiguity thesis
are not good. For there is an obvious
way to understand doxastic warrant in terms of propositional warrant:
DW:
S's belief that p is doxastically warranted if and only if (i) p is
propositionally warranted for S and (ii) S's belief is based on that which
propositionally warrants it.[xiii]
DW provides a solid argument on behalf of the claim that
there is only one sense of `warrant', a sense which covers the application of
the term both to propositional contents and to beliefs. Moreover, this propositionalist proposal is
attractive in part because it handles with such ease the cases that are
problematic for the Aristotelian. In
cases in which belief undermines warrant, propositionalists can simply note
that propositional warrant was present prior to the formation of the belief and
absent after, yielding the appropriate result that the belief in question would
not be doxastically warranted if formed.
In cases in which belief itself is warrant-imparting, propositional
warrant appears with the formation of the belief, allowing the result that no
propositional warrant was present prior to the belief, but that the belief was
doxastically warranted when formed.
Finally, cases of essential cognitive admirability create no problem for
the propositionalist, either. So DW
presents an attractive alternative to Aristotelianism, one that provides a
forceful argument against the ambiguity thesis needed to preserve
Aristotelianism and one that is able to explain successfully the anomalies that
plague its competitors.
The Lesson for Plantingian Epistemology
One is tempted at this point to begin
applying the above discussion to Plantinga’s epistemology by pointing out that
he begins, as do all Aristotelians, with the wrong kind of warrant. The matter is more complicated than that,
however, for Plantinga can defend himself here by claiming that he has defined
warrant to be that quantity which fills the gap between true belief and
knowledge. As we have seen, propositional
warrant simply does not do that. Instead
of focusing on the kind of warrant in question, I believe there is a more
useful approach. As I see it, the heart
of Plantingian epistemology is found in the claim that the central feature of
an adequate account of knowledge, apart from the requirement of true belief, is
the concept of proper function: we know only when our faculties are functioning
properly, in accordance with their design plan.
The lesson to be learned from the above difficulties faced by
Aristotelianism is where in a full account of knowledge an appeal to proper
function might be appropriate.
To approach this issue carefully, we
must first distinguish between the third and fourth conditions for
knowledge--the warrant condition, as I will call it (though I reject the Plantingian
stipulation that warrant is what fills the gap between true belief and
knowledge), from the Gettier condition.
One place that an appeal to proper function might be appropriate is in
giving an account of the Gettier condition for knowledge. Some of Plantinga’s examples suggest just
such a role. For example, in arguing
against coherentism, Plantinga constructs the following counterexample:
Perhaps I have been captured by Alpha
Centaurian cognitive scientists, who make me the subject of a cognitive experiment;
their aim is to give my a system of beliefs in which falsehood and coherence
are maximized. They succeed in giving me
a thoroughly coherent set of beliefs, but in a few cases they slip, giving me a
true belief rather than a false one. . .
. In such cases my beliefs may have a great deal of coherence, but they will
have very little warrant. Those that are
true, are true just by accident, and surely do not constitute knowledge for me.[xiv]
As an attempt to refute coherentism, such a counterexample
invites the following reply.
Coherentists can claim that they never intended their theory to fill the
gap between true belief and knowledge, and that cases such as the above will be
ruled out as cases of knowledge by an adequate response to the Gettier problem,
leaving untouched the coherentist understanding of the third condition for
knowledge.[xv] The coherentist might agree with Plantinga,
however, about the importance of an appeal to proper function in explaining why
the above case is not a case of knowledge.
So if proper function is central to knowledge, one way that might be
true is for proper function to play a role in the fourth condition for
knowledge, that condition aimed at solving the problem of accidentally true,
but justified, belief made famous by Gettier.[xvi]
The concept of proper function might
also play an important role in the third condition itself. Note first that propositional warrant is
surely not all there is to such a third condition, even if we are convinced that
it is the fundamental notion of warrant.
For improperly based beliefs that have propositionally warranted
contents are not cases of knowledge. So
once we see the failure of Aristotelianism and the attractiveness of
propositionalism, two quite different roles should be distinguished for the
concept of proper function to play in the third condition. One might claim that the concept of proper
function is central to an adequate understanding of warrant itself, or to the
concept of proper basing. The first
suggestion is just a version of Aristotelianism, a view sufficiently undermined
by previous argument to warrant ignoring here.
The other option, that the concept of proper function is important in
understanding the concept of properly basing a belief on that which warrants its
content, is not affected by those arguments and has much to recommend it. On this latter alternative, for a belief to
be properly based, it would need to result from properly functioning cognitive
equipment.
If we find Plantinga's appeal to
proper function persuasive and pursue this line of thought, we might think of a
proper function condition as a solution to the problem of deviant causal chains
that plague causal theories of the basing relation. According to such theories, to base your
belief on your evidence is for your awareness of the evidence to be causally
responsible for your belief. The problem
of deviant causal chains arises when one notices that awareness of evidence can
cause belief in a variety of ways. For
example, Watson may fall down the stairs as a result of the excitement upon
being told the evidence for the claim that the butler committed the
murder. The fall might induce an
unshakeable conviction that the butler did, thereby yielding the result that
awareness of the evidence was a cause of the belief in question. Sherlock's belief, however, is caused in a
more straightforward way--namely, by Sherlock's coming to see that the evidence
implies that the butler did it. Watson's
belief, though caused by the evidence just as much as Sherlock's belief, is not
caused in the right way. So, the causal
theory needs to be able to say what particular kinds of causal chains yield
doxastically warranted beliefs, for not just any causal path from evidence to
believe does that.
Perhaps Plantinga's proper function
requirement can be of service to the causal theory by explaining what it means
to say that Watson's belief is not caused in the right way whereas Sherlock's
belief is. Sherlock's belief results from
cognitive equipment functioning in the way it was designed to function, whereas
Watson's does not, and, we might claim, it is this difference that explains why
Sherlock's belief is properly based and Watson's is not.
I am not endorsing this proposal, but
only pointing it out as a proposal worthy of further investigation. If the proposal turns out to have merit, it
would explain some of the attractiveness of Plantinga's proposal to understand
warrant in terms of proper function. For
on the above proposal, there is a place for Plantinga's claims within a
complete account of the third condition for knowledge. The result would be that proper function has
an indirect connection with the fundamental concept of warrant, in that it
would be a necessary condition for propositional warrant to transfer to
belief.
Even if there is a place in a
complete theory of warrant or a complete theory of knowledge for Plantinga's
concept of proper function, the conclusion here nonetheless contravenes the
fundamental tenets of Plantingian epistemology.
For, strictly speaking, proper function has nothing to do with the
fundamental concept of warrant itself, for that concept is fundamentally a
concept that applies to propositional contents rather than beliefs. As a result, we ought to reject Plantinga's
proper function account of warrant.[xvii]
____________________
NOTES
Plantinga's Proper Function Account
of Warrant
Abstract
Plantinga says that sufficient warrant, together with true
belief, yields knowledge, and he holds that warrant is to be understood primarily
in terms of proper function. Plantinga
maintains that there is a design plan for various parts of our cognitive
apparatus, and that a belief has warrant only when our cognitive equipment to
be functioning as it was designed to function.
This account implies that the fundamental kind of warrant is that which
attaches to beliefs. I argue to the
contrary that the fundamental kind of warrant attaches to propositions, and
that an implication of this fact is that proper function is not relevant at all
to the concept of warrant.
[i].Alvin Plantinga, Warrant: The Current Debate and Warrant and
Proper Function, (Oxford, 1993).
[ii].Plantinga is not alone among
contemporary epistemologists in favoring an Aristotelian approach to
epistemology. Others include: D.M. Armstrong, Belief, Truth and
Knowledge, (London, 1973); Alvin Goldman, Epistemology and Cognition
(Cambridge, 1985); Robert Nozick, Philosophical Explanations,
(Cambridge, 1981); Ernest Sosa, Knowledge in Perspective: Selected Essays in Epistemology, (London,
1991). For a discussion of the Aristotelian approach of these and other authors, see Jonathan L.
Kvanvig, The Intellectual Virtues and the Life of the Mind: On the Place of the Virtues in Contemporary
Epistemology, (Savage, Maryland, 1992).
[iii].I borrow this terminology from
Roderick Firth. See his "Are
Epistemic Concepts Reducible to Ethical Concepts?", in Alvin I. Goldman
and Jaegwon Kim, eds., Values and Morals, Essays in Honor of William
Frankena, Charles Stevenson, and Richard Brandt, (Dordrecht, 1978).
[v].Alvin Goldman, "What is
Justified Belief?" in George Pappas, ed., Knowledge and Justification,
(Dordrecht, 1979), pp. 1-25.
[vi].Alvin Plantinga, Warrant: The
Current Debate, (Oxford, 1993), p. 3.
[vii].Plantinga, Warrant and Proper
Function, p. v.
[viii].I owe grasping this point to Trenton
Merrick.
[ix].Plantinga, Warrant: The Current
Debate, p. 4.
[x].One cannot always define Y in terms
of X when X is defined in terms of Y.
One can define mathematical equality in terms of the relations of being
strictly lesser than and greater than, but one cannot define either of these
latter two in terms of equality. In
general, if we define X in terms of Y plus other concepts, we have no guarantee
that we can define Y in terms of X.
[xi].Alvin Goldman, "What is
Justified Belief?", p. 21.
[xii].Richard Foley, "Evidence and
Reasons for Belief," Analysis 51.2 (March 1991), p. 99. Earl Conee discusses a similar case in
"Evident but Rationally Unacceptable," Australasian Journal of
Philosophy 65 (1987), pp. 316-26.
[xiii].Definition DW is not quite what is
needed. I have argued in “Coherentism:
Misconstrual and Missaprehension,” Southwest Philosophy Review 11
(1995), pp. 159-169, and in “Coherentists’ Distractions,” Philosophical
Topics, forthcoming 1995, that one can offer an account of the basing
relation which makes it out to be, not a relation between a belief and that
which warrants it, but rather between a belief and that which is epistemically
relevant to it (where x can be epistemically relevant to y without
propositionally warranting it, or increasing its warrant status at all). It is relatively easy to see that this notion
of epistemic relevance will itself be a concept of epistemic appraisal that
applies fundamentally to propositions and not to beliefs, and since the issues
raised by this complication would be far removed from the general issues of
this paper, I will stick with the simpler, even though slightly misleading,
formulation in the text.
[xiv].Warrant: The Current Debate, p. 111.
[xvi].Richard Feldman further develops
this complaint in "Proper Functionalism," Nos 72.1 (March 1993), pp. 34-50.
[xvii].I wish to express my thanks to those
who have commented on drafts of this paper.
They are: Colin Allen, Richard
Feldman, Michael Hand, Peter Markie, Hugh McCann, Paul McNamara, and Scott
Sturgeon.