Plantinga's
argument begins with an articulation of doubts about the
reliability of our cognitive faculties (namely our memory,
perception, and reason), given metaphysical naturalism and
the typical mechanisms of evolution (such as natural
selection, genetic drift, etc.).
According to Plantinga a cognitive faculty is reliable if
"the great bulk of its deliverances are true" 3
Plantinga's first main claim can be expressed as
follows:
(PT): Pr(R/N&E) is low or inscrutable
where 'R" is the
proposition that our cognitive faculties are reliable, 'N'
the proposition that naturalism is true, and 'E" the
proposition that we have evolved according to the
suggestions of contemporary evolutionary theory.
But why should one believe PT?
Plantinga follows evolutionists 4 who believe
that natural selection is directly interested only in
behavior, not belief, and that it is interested in belief,
if at all, only indirectly, by virtue of the relation
between behavior and belief. If adaptive behavior guarantees
or makes probable reliable faculties, then perhaps Pr(R/N&E)
will be fairly high. On the other hand, if having reliable
faculties isn't guaranteed by or even particularly probable
with respect to adaptive behavior, then presumably Pr(R/N&E)
will be rather low.
Therefore, according to this view, the nature of the
relationship between belief and behavior is integral to
determining Pr(R/N&E).
Plantinga produces four "mutually exclusive and jointly
exhaustive" accounts of the relationship between belief and
behavior. 5
But before that, he suggests that we think, not about
ourselves, and our behavior, but about a hypothetical
population a lot like us. These creatures are rational, that
is, they form beliefs, reason, change beliefs, and the like;
and also their cognitive faculties evolved by way of the
mechanisms to which contemporary evolutionary theory directs
our attention. Now we estimate Pr(R/N&E), specified to them,
not to us.
There are four possible relationships between belief and
behavior, according to Plantinga:
(A1) Beliefs do not causally influence
behavior.(Epiphenomenalism)
(A2) Beliefs causally influence behavior syntactically, but not
semantically. (And here by "semantic" he means the content of
belief, and by "syntax" he means some neurophysiological or
electrochemical properties that a belief as a , for example,
long-term pattern of neural activity might have.) Plantinga
calls this account "semantic epiphenomenalism".
(A3) Beliefs causally influence behavior semantically as well as
syntactically, but are maladaptive.
(A4) Beliefs causally influence behavior semantically as well as
syntactically, and are adaptive.
Now
Pr(R/N&E) will be the weighted average of Pr(R/N&E&Ai) for each of
these four possibilities Ai- weighted by the probabilities , on N&E,
of those possibilities. It can be calculated according to the
following formula:
(F1)Pr(R/N&E)=[Pr(R/N&E&A1)×Pr(A1/N&E)]+[Pr(R/N&E&A2)×Pr(A2/N&E)]+
[Pr(R/N&E&A3)×Pr(A3/N&E)]+[Pr(R/N&E&A4)×Pr(A4/N&E)]
To
simplify this formula, note that:
(1) Among these four alternative options, A3 seems very unlikely,
and so it can be ignored safely.
(2) A1 and A2 unite in declaring or implying that the content of
belief lacks causal efficacy with respect to behavior. So these two
possibilities can be reduced to one: the possibility that the
content of belief has no causal efficacy. Let's call it '-C'. Now we
have two major possibilities A4 ( for the sake of uniformity ,let's
call A4, hereafter 'C') and '-C'.
Now we can reduce (F1) to the following form:
(F2)Pr(R/N&E)=Pr(R/N&E&C)×Pr(C/N&E)+ Pr(R/N&E&-C) ×Pr(-C/N&E)
Here
Plantinga claims that:
(a) Pr(R/N&E&C) prima facie has a fairly high probability, but in
fact it is not as high as one might think. For, he believes that
"beliefs do not causally produce behavior by themselves; it is
beliefs, desires, and other things that do so together." 6
Assuming
this fact, lantinga offers two scenarios in which the basic
cognitive systems could be adaptive while generating falsehood.
First scenario: According to Plantinga, natral selection can
generate suitable desires and goals that, when conjoined with
appropriate false belief, give rise to fitness-enhancing behavior.
For example, consider Paul, a prehistoric hominid who likes the idea
of being eaten by a tiger, but when he sees a tiger, always runs off
looking for a better prospect, because he think it unlikely that the
tiger he sees will eat him. This will get his body parts in the
right place so far as survival is concerned, without involving much
by way of true belief. 7
But here one might say no one denies that
there could exist false beliefs that, when coupled with certain
desires, would produce adaptive behavior in a certain way. But the
critical question at this point is whether or not a process of
natural selection is likely to generate cognitive mechanisms that
are systematically unreliable but somehow prove adaptive. It is easy
to see, for just one of Paul's actions, that there are many
different belief5 desire combinations that yield it; but it is less
easy to see how it could be that most of all his beliefs could be
false but nonetheless adaptive or fitness-enhancing. In other words,
could it be the case that one holds a mostly false belief system,
and nonetheless survives over time?
Here Plantinga offers his second scenario.
Second scenario: In the second scenario, Plantinga supposes that the
organism suffers from some sort of profound misconception that
undermines the veracity of its beliefs without undermining the
belief's usefulness. He says: "Perhaps Paul is a sort of early
Leibnizian and thinks everything is conscious (and suppose that is
false); furthermore, his ways of referring to things all involve
definite descriptions that entail consciousness, so that all of his
beliefs are of the form That so-and-so conscious being is
such-and-such. Perhaps he thinks all the plants and animals in his
vicinity are witches, and his ways of referring to them all involve
definite descriptions entailing witchhood. But this would be
entirely compatible with his belief's being adaptive; so it is
clear, I think, that there would be many ways in which Paul's
beliefs could be for the most part false, but adaptive
nonetheless." 8
Even though this scenario is quite conceivable, but
there is no reason to think it was an actual or available option on
which natural selection can act. Here Plantinga seems to ignore the
fact that the probability of a trait's evolving depends not just on
its fitness, but on its availability. "The reason zebras don't have
machine guns with which to repel lion attacks is not that firing
machine guns would have been less adaptive than simply running
away", but it is because that option was not available as a
variation on which selection could act ancestrally. 9
But can we find some actual cases in which (at least from the
perspective of a naturalist) we have a system of mainly false but
adaptive beliefs? Plantinga offers a third scenario to show an
actual instance of such system (specified to the naturalist's way of
looking at the world):
Third scenario: Suppose naturalism is true and in fact there is no
such person as God. But some naturalists believe that belief in God,
while false, is nonetheless adaptive. Now suppose a tribe of
cognitively gifted creatures believe that everything (except God
Himself) has been created by God. Suppose therefore that their only
way of referring to the various things in their environment is by
way of such definite descriptions as 'the tree creature before me '
or 'the tiger creature approaching me'. Suppose still further that
all their beliefs are properly expressed by singular sentences whose
subjects are definite descriptions expressing properties that
entails the property of creaturehood- such sentences as 'The tiger
creature approaching me is dangerous' or 'The tree creature before
me is full of apple creatures'. Suppose, finally, that their
definite descriptions work the way Bertrand Russell thought definite
descriptions work. 'The tallest man in Boston is wise', for example,
abbreviates 'There is exactly one tallest man in Boston, and it is
wise'. Then from the naturalist perspective all their beliefs are
false. Yet these can still be adaptive: all they have to do is
ascribe the right properties to the right 'creatures' 10 that would yield adaptive
behavior, but are unreliable, one does not quite know what to say
about Pr(R/N&E&C). But if one thinks this probability is high, then
perhaps it would be reasonable to estimate it as somewhat more than
1/2.
(b) Plantinga also argues that Pr(-C/N&E) is high. His argument for
this claim can be stated as follows:
(1) One who accepts metaphysical naturalism will be a
materialist with respect to human beings.
(2) From the perspective of a materialist –C is hard to avoid.
(3) Therefore, the probability that -C is the case, given N&E,
is high.
The
reason Plantinga thinks (2) is true, is that if there are such
things as beliefs, then, according to materialism, they will have to
be material processes or events- maybe something like long-standing
neural events of some kind in the nervous system. But a belief has
to be the belief that P for some proposition P. In other words, P is
the content of that belief. Now he believes that "it is far from
clear how a neural structure can acquire a content….Granted that
such a structure can acquire content, however, it still seems
extremely difficult to see how it can enter the causal chain leading
to behavior by virtue of its content." 11
(c) Now Plantinga claims that Pr(R/N&E&-C) is low.
According to the evolutionary theory natural selection can eliminate
certain structures (those that come at the cost of reproductive
fitness) and encourage others (those that enhance reproductive
fitness). Now if false belief caused maladaptive action, therefore,
natural selection could presumably modify belief-producing
structures in the direction of greater reliability i.e., a greater
proportion of true as opposed to false belief. But if content does
not enter the causal chain that leads to behavior, then of course it
will not be the case that a belief causes maladaptive behavior by
virtue of its being false, and it will not be the case that a true
belief causes the behavior it does by virtue of being true. And then
it is hard to see how natural selection can promote or enhance or
reward true belief and penalize false belief. As Plantinga puts it,
in this situation the hypothetical population's beliefs would be
"invisible to evolution". 12
Now we are in position to estimate Pr(R/N&E) according to (F2):
Pr(-C/N&E) is high. Suppose it is 0.8. Then Pr(C/N&E) will be 0.2.
Pr(R/N&E&-C) is low.
Suppose it is around 0.2. And Pr(R/N&E&C) is high, but not as high
as one might think. Suppose it is 0.7. Then Pr(R/N&E) will be less
than 1/2 (under these assignments, 0.3). So far, what Plantinga
shows (if any ) is that Pr(R/N&E), specified to the hypothetical
population, is low. Now at this stage, Plantinga quickly derives the
same conclusion about "us" i.e., human beings. He says:" The
reasoning that applies to these hypothetical creatures, of course
applies to us; so if we think the probability of R with respect to
them is relatively low on N&E[…], we should think the same thing
about the probability of R with respect to us." 13 This is because he
believes we are relevantly like them in that our cognitive faculties
have the same kind of origin and provenance as theirs are
hypothesized to have. 14
Of course, Plantinga confesses that we cannot assign specific real
numbers to the probabilities involved in estimating Pr(R/N&E); the
best we can do are vague estimates. Therefore, his estimates of the
various probabilities involved in (F2) is "imprecise and also poorly
grounded" and so "the right course here is simple agnosticism: one
just does not know what that probability is. … Then this probability
is inscrutable for you." 15 Therefore, Pr(R (specified to us)/N&E) is
either low or inscrutable.
II. The Preliminary Argument
Plantinga's second argument (which he called the "preliminary
argument") was intended to be an argument for the falsehood of
naturalism. The first version of the argument was based on two major
premises: 16
(1) The probability of Pr(R/N&E) is fairly low;
(2) Our cognitive faculties are reliable, i.e., Pr(R)≈1.
Now
by using Bayes's Theorem, we seem to have a straightforward argument
against naturalism. Because according to Bayes's Theorem, we have:
Pr(N&E/R) = Pr(N&E) × Pr(R/N&E) / Pr(R)
Plantinga already argued that Pr(R/N&E) is not more than 1/2. And
according to the second premise, since we believe R, we assigned it
a probability of 1 or nearly 1. It means Pr(N&E/R) will be no
greater than 1/2 times Pr(N&E). Now even if we assign a high
probability to N&E independent of the consideration of R ( a value
less than one any way), Pr(N&E/R) will be less than 1/2. Therefore,
if we think R is true, then we have evidence against evolutionary
naturalism. We do think R is true; therefore, we do have evidence
against evolutionary naturalism (N&E). 17
As Plantinga confessed this argument is "straightforwardly
incorrect" 18 The second premise is false. The fact that we believe
R, does not make its probability 1. Of course if we conditionalize R
on our background knowledge (B), then Pr(R/B) will be high, or even
close to unity. But then the relevant application of Bayes's Theorem
would be as follows:
Pr(N/R&B) = Pr(N/B) × Pr(R/N&B)/ Pr(R/B) 19 ( Here for
simplicity I just dropped E)
But
what Plantinga argued for is the claim that Pr(R/N) is low, and what
he needs here is the claim that the probability of R given N&B is
low. He did not provide any reason for the latter claim. On the
other hand one might say here what Plantinga means by Pr(R), is the
absolute probability of R (conditioned only on necessary truths). If
so, then there is no reason to assume that Pr(R) is one, or nearly
one. The fact that R is true in the actual world is no reason to
believe that in a large proportion of all possible worlds, R is also
true. 20
Therefore, the first version of the preliminary argument fails. In
the second version, in order to avoid those difficulties, Plantinga
discharged Pr(R). But how? Since after all Plantinga's argument is
supposed to provide a ground for preferring theism (T) to naturalism
(N), he compared these two alternative "comparable" positions as
follows:
Pr(N/R)/Pr(T/R) = Pr(N) × Pr(R/N)/ Pr(T) × Pr(R/T) 21
Pr(R/N), according to Plantinga, is low. On the other hand,
Plantinga claims that Pr(R/T) is high. According to traditional
theism, we human beings have been created in the image of God. This
means, among other things, God created us with the capacity for
achieving knowledge. Therefore, given T, R is just what we would
expect. So Pr(R/T) is not low, or at least we have no reason for
thinking so. Now given that we do not assign N a considerably higher
absolute probability than T( since initially we seem not to have any
reason to prefer one to the other), we can conclude that
Pr(R/N)/Pr(R/T) is less than 1, and therefore, Pr(N/R)/Pr(T/R) < 1.
It means Pr(N/R) is lower than Pr(T/R). If R is the only available
evidence, then we have a reason to prefer T to N. 22
III. The main argument
Finally, Plantinga's major argument is an argument against
rationality of naturalism. He calls it "the main argument". The main
argument is as follows:
(1): Pr(R/N&E) (specified to us) is low or inscrutable. (The
probability thesis (PT))
(2): For a devotee of naturalism, PT is a defeater for R. In
other words, N&E&PT (Let's call it 'Q') defeats R.
(3): For any naturalist (S), if S has a defeater for R, then S
has an undercutting defeater for every belief B she holds,
including N&E itself.
(4): Therefore, evolutionary naturalism (N&E) is self-defeating.
(5): Therefore, S cannot hold N&E rationally.
In
this argument there are two sets of rationality defeaters and
defeatees at work. In (2), Q functions as an undercutting defeater
for R, and in (3), the denial of R functions as an undercutting
defeater for N. So it seems clear that the notion of "rationality
defeater" plays an important role in Plantinga's argument. What does
he mean by "rationality defeater"? Plantinga tried initially to
explain the notion of rationality defeater in his argument as
follows:
(D) D is a defeater of B for S at t if and only if (1) S's noetic
structure N(i.e., S's beliefs and experiences and salient relations
among them) at t includes B, and S comes to believe D at t, and (2)
any person (a) whose cognitive faculties are functioning properly in
the relevant respects, (b) whose noetic structure is N and includes
B, and (c) who comes at t to believe D but nothing else independent
of or stronger than D would withhold B(or believe it less
strongly). 23
Defeaters, including defeaters for defeaters, can be divided into
rebutters and undercutters. If a defeater gives us evidence against
belief B (but not strong enough to require the belief that not-B),
we may call it a "rebutting defeater". An "undercutting defeater" is
a belief that indicates not that the defeatee is false but that
one's grounds for it are inadequate or neutralized.
How does Plantinga support the main claims in this argument?
1. To move from (1) to (2) is a very crucial step in this argument.
The crucial claim(CC) at this step seems as follows:
(CC): If Pr(R/N&E) is low or inscrutable, then one who believes N&E
has a defeater for R. But what is the reason for (CC)?
There might be a general principle behind this claim as follows:
Principle (1): If Pr (B/A) is low or inscrutable, then one who
believes A has a defeater for B. But this principle is obviously
false. The fact that Pr (Plantinga's evolutionary argument against
naturalism is sound/UCSB has a philosophy department) is low or
inscrutable, does not seem a defeater to Plantinga's argument for
someone who believes that UCSB has a philosophy department.
Plantinga believes even though this principle does not work for all
cases, but it might be sufficient at least for some cases. 24 And in
fact he claims in the case of naturalist there is such connection
between PT and R. But how is he going to support this claim?
Plantinga's argument for (CC), in principle, is an argument from
analogy. He does not appeal to any general principles to establish
(CC), instead he suggests cases where low or inscrutable probability
seems to lead to a defeater, and then suggests that the evolutionary
naturalist's position is, regarding R, relevantly analogous to those
cases.
First analogous case: Suppose I believe that I have been created by
a Cartesian demon who takes delight in fashioning creatures who have
mainly false beliefs (but think of themselves as paradigms of
cognitive excellence). Or the contemporary version of this scenario:
suppose I come to believe that I have been captured by
Alpha-Centaurian superscientists who have made me the subject of a
cognitive experiment in which I have been given mostly false
beliefs.
In both cases, I have a defeater for my natural belief that my
faculties are reliable (i.e., R).
Second analogous case: Suppose I come to believe that I am a brain
in a vat, and also
believe that the probability of my cognitive faculties are reliable,
given the belief that I am a
brain in a vat, is low or inscrutable. Again I have a defeater for
R.
Third analogous case: Suppose you believe that XX is a drug that
causes widespread
cognitive disorder: an hour after XX is ingested, the subject's
cognitive function is no longer
reliable. And this drug also prevents its victim from detecting her
unreliability. Now
suppose someone believes that she has taken this drug, and suppose
she also believes that 90
percent of those who take the drug are no longer reliable, although
of course unable to detect
their unreliability. The person in this condition will continue to
form perceptual beliefs,
memory beliefs, and the like; and she will also continue, in some
sense, to assume that her
cognitive faculties are reliable. Again here Plantinga thinks that
if the person believes that
Pr(R/I ingested XX) is low, and also she believes that she ingested
XX, then she has a
defeater for R.
In the UCSB case, the soundness of Plantinga's evolutionary argument
against
naturalism is conditionalized on a fact which is irrelevant to it.
But in Plantinga's cases, R is
conditionalized on the facts which are directly relevant to R. But
is there any general
principle to capture this relevancy? It seems in all these cases,
including the naturalist case,
the crucial point is that, for example, N&E, or creation by
Cartesian demon, involves a claim
about the origin of our cognitive faculties. And because of that, if
there is a low, or
inscrutable, probability, given such an origin, that those faculties
are reliable, then this does
provide a defeater for R. Plantinga believes that it is hard to find
or establish any principles
along this line. But if a principle is wanted, he suggests a very
limited one, something about
beliefs specifying the origin and provenance of cognitive faculties.
The antecedent of such
principle would capture what is in common to the situation in which
(a) I believe that I have
taken XX, or am a brain in vat, or a victim of a Cartesian evil
genius, and (b) I also believe
that the probability of R on the condition in question is low or
inscrutable; the consequent
would say that in those situation I have a rationality defeater for
R. 25
As we see, in all cases Plantinga tries to show that the
naturalist's situation is
relevantly like the situation of the person who is created by
Cartesian demon, or of the brain
in a vat, or of the person who digested XX. Since they have a
rationality defeater for R,
therefore the reflective naturalist (who is aware of PT and
Plantinga's argument) also has a
defeater for R. And of course in all these cases, the defeater of R
is an undercutting defeater
(as we will see).
2.
(3), in the main argument, at least prima facie does not seem
problematic. In general, if we have considered the question whether
a given source of information or belief is reliable, and have a
defeater for the belief that it is, then we have a defeater for any
belief such that we think it originates(solely) from that source.
Here we have a defeater for R, and it means we get a defeater for
all of our beliefs. But this conclusion and also the final
conclusion of the main argument will follow only if some other
conditions are met:
C1: D is a defeater of B only if D is a defeater of B for someone; D
is a defeater for someone S only if S believes both D and B. (C1
simply follows from the definition of "rationally defeater")
C2: The defeater must be itself undefeated. If we get a defeater for
the defeater of R, then we can hold R (and therefore, all
deliverances of our cognitive faculties, including N&E) rationally.
C3: There must be no sufficient independent evidence for R. Because
the naturalist (who accepts N&E&PT) might have such strong support
for R as to outweigh the negative impact of N&E. 26
In the case of the reflective naturalist, C1 is obviously met.
Plantinga believes that in the case of the reflective naturalist
,both C2 and C3, are also met, i.e., the devotee of N&E has no
defeater for this defeater, and no independent evidence for R. He
argues that:
(a)One cannot have non-question- begging support for R. For example,
suppose one argues inductively for R: she might consider any of her
beliefs. She believes that A1; that is she believes that A1 is true;
A2 is also one of her beliefs and A2 is true, and so on. By
induction, she argues that all or nearly all of her beliefs are
true, and therefore, her faculties are probably reliable. Here, when
the person claims that, for example, A1 is true, and when she relies
on induction to conclude the reliability of her faculties, she has
to rely on one or other of her cognitive faculties (i.e.,
perception, memory, and so on) , and she cannot rely on them unless
she assumes they are reliable. In other words, in so doing she
assumes R. So any argument for R is in this context delicately
circular or question-begging. Of course it is not formally circular,
i.e., its conclusion does not appear among its premises. It purports
to give a reason for trusting our cognitive faculties, but is itself
trustworthy only if those faculties (at least the ones involved in
its production) are indeed trustworthy. Once we come to doubt the
reliability of our cognitive faculties, we cannot properly try to
allay that doubt by producing an argument; for in so doing we rely
on the very same faculties we are doubting. 27
(b) The same goes for the defeater of defeater of R. The defeater of
R cannot be defeated because any defeater would arise from the very
faculties or belief-producing processes in question. For example,
the defeater might take the form of an argument, perhaps for the
conclusion that those belief-producing processes are reliable after
all. But then we would have the same defeater for each of the
premises of this argument, as well as for our belief that if the
premises are true, then so is the conclusion. 28
3.
Now suppose R has an undefeated undercutting defeater. It mean the
reflective naturalist has a good reason to believe –R. If –R is the
case (i.e., if reflective naturalist’ cognitive faculties are not
reliable), then she has an undercutting reason for doubting every
belief she has, including N&E, unless she has other evidence for it;
but any purported other evidence will be subject to the same
defeater as N&E. Therefore, she is not rational in holding N&E any
more.
But there seems to be a problem here: As soon as N&Eer comes to
doubt R, she should also come to doubt her defeater for R; for that
defeater, after all, depends upon her beliefs, which are a product
of her cognitive faculties. So her defeater for R is also a defeater
for that defeater, that is, for itself. But then when she notes
that, and doubts her defeater for R, she no longer has a defeater
(undefeated or otherwise) for N&E; and so in spite of Plantinga,
N&Eer seems to be rational to hold her belief in N&E.
To meet this objection, Plantinga offers the following scheme:
Imagine a person who comes to believe –R, that is, that her
faculties are not reliable. Then she can see that there is an
infinite series of propositions related in an interesting way. At
the first level, there is –R, which she believes. But there is a
connection between –R and any belief she has, including –R itself:
if –R is true, then any belief she has, including –R itself, will be
unreliably formed. But that belief, i.e., the belief that –R is
unreliably formed, gives him a potential defeater for –R. Suppose we
let '-R (p)' express the proposition that p is unreliably formed.
Then at the second level we have –R and –R (-R), which is a defeater
of –R. But then at the third level we have –Rand –R (-R (-R)), which
is a potential defeater for –R (-R), the defeater of –R; and so on.
We can schematically represent the structure as follows:
Level 0: -R
Level 1: -R and –R (0) (i.e., -R (-R)) which is a potential
defeater for level 0, i.e., -R)
Level 2: -R and –R (1) (which is a potential defeater for level
1 and a potential defeaterdefeater for level 0)
Level 3: -R and –R (2) (which is a potential defeater for level
2 and a potential (defeaterdefeater)-defeater for level 0)
………………………
Level n: -R and –R (n-1) (which is a potential defeater for
level n-1)
………………………
There
is an infinite series of propositions here, but not an infinite
series of defeaters and defeatees. Because, according to C1 (see,
p.14), D is a defeater of B for S only if S believes both D and B.
But no one could believe all the propositions in this infinite
series. So even though each member is a potential defeater for the
preceding member, nevertheless at some level, it is possible for S
not to have any actual defeater for her belief B.
In this scheme, the original defeatee (-R) shows up at every
subsequent level. When that happens, the defeater-defeater does not
nullify the defeater. The defeater gets defeated, but the defeatee
remains defeated too. Accordingly, any time at which the person
believes –R, she has a defeater for –R, even if she also has, at the
time, a defeater for that defeater.
This position, then, really is self-defeating, even if it is also
the case that the person has a defeater for his defeater. In the
case of N&Eer, Q ( i.e., PT &N&E) is a defeater for R, and the
reflective naturalist (i.e., the naturalist who believes Q) has a
defeater for N&E at any time which she accepts Q. She has such a
defeater for N&E even if at that time she also has a defeater for
that defeater. Her problem, after all , is that Q gives her a
defeater for everything she believes, including themselves. 29
Therefore, N&E is a self-defeating position, and it is not rational
to be held by the reflective naturalist.font>
4. In
all Plantinga's analogous cases, the probability in question is low.
Now suppose we are unable to determine the value of Pr(R/N&E). In
other words, it is not low, but inscrutable. Is that sufficient,
along with other relevant features of the situation, to make N&E a
defeater of R? Plantinga's response is again, yes. And his strategy
to support this answer is again to argue from analogy: imaging a
person who comes into a factory, sees an assembly line carrying
apparently red widgets. Suppose she is agnostic about the
probability of a widget's being red, given that it looks red. Now
the shop superintendent tells her that those widgets are being
irradiated by red light; but then a vice-president comes along and
tells her that the shop superintendent suffers from a highly
resilient but fortunately specific hallucination, so that he is
reliable on other topics even if totally unreliable on red lights
and widgets. Still, the vice president himself does not look wholly
reliable. Then she does not know what to believe about those alleged
red lights. She will presumably be agnostic about the probability of
a widget's being red, given that it looks red; she won't know what
that probability might be; for all she knows it could be very low,
but also, for all she knows, it could be high. Therefore, Plantinga
believes that the rational course for her is to be agnostic about
the deliverances of her visual perception (so far as color detection
is concerned) in this situation. 30
Part Two: Objections
I.
About the Preliminary Argument
The
conclusion of the preliminary argument (hereafter PA), is a
conditional statement:
If R is the only available evidence for us, and we believe PT, then
we have a reason to prefer theism to naturalism.
(1) But it seems clear that R is not the only relevant evidence
available to us. Therefore, if we consider all relevant evidence
(O), we may get a different result. In other words, it seems quite
possible to find some additional relevant information (O) in such a
way that Pr (T/R&O) comes out less than Pr (N/R&O). The argument
does not rule out such possibility.
But there seems to be a problem here: It might be difficult to say
what O could be. For according to one version of Bayes' Theorem, we
have:
Pr(T/R&O)/Pr(N/R&O)=Pr(R/T&O)Pr(O/T)Pr(T)/Pr(R/N&O)Pr(O/N)Pr(N) 31
Suppose in this case, Pr(R/N&O) is greater than Pr(R/N), and suppose
it also makes Pr(R/N&O) greater than Pr(R/T&O)(even though the
latter clearly needs some reason to be acceptable), but still we
cannot say that Pr(N/R&O) comes out greater than Pr(T/N&O). This is
because we have to look at Pr (O/N) and Pr (O/T), and see how likely
that information O is, given naturalism and given theism. Even if O
raises the probability of R given naturalism, it may be improbable
given N. Therefore, whatever O is, the probability of O given N must
be higher than Pr (O/T). 32
And now, whatever O is, one may argue
for the claim that Pr(O/T) is high, in the same way that Plantinga
argued for the claim that Pr(R/T) is high.
(see, p.9)
(2) Suppose we admit that it is not easy to find some additional
information such that Pr (O/N) is greater than Pr (O/T). But still
Plantinga's way of reasoning in PA provides a recipe for replacing
any nondeterministic theory in the natural sciences. For example,
suppose quantum mechanics (QM) predicts that a certain experimental
outcome (E) was merely very probable. But theist may claim that this
outcome was inevitable outcome of God's will (T). Presumably, Pr
(E/T) is much greater than Pr (E/QM), and if we compare Pr(T/E) and
Pr(QM/E) based on Bayes' Theorem, we will see that Pr(T/E) is
greater than Pr(QM/E), and so we have a reason to prefer T to QM. 33
This example makes it clear that the pattern of Plantinga's argument
does not seem good at all.
(3) Suppose atheist accepts the conditional conclusion of PA. But
then he can provide the same type of argument to prefer atheism
(which is implied by naturalism) to theism. For example, the atheist
might appeal to the existence of evil in the world (or some
particular cases of evil, or the amount of evil in the world), and
provide a parallel argument in favor of atheism as follows:
Suppose O is the only available evidence we have:
(O): There is a great amount of evil in this world.
And suppose we have only two alternative hypotheses to compare with
respect to O:
(T): The nature and the condition of sentient beings on earth is the
result of benevolence actions performed by God.
(N) The nature and the condition of sentient beings on earth is not
the result of the action of theistic God as an omnipotent
omniscient, and perfectly good being. If O is the only available
evidence, it is reasonable to say that Pr (O/N) is greater than
Pr(O/T). According to Bayes' Theorem we will have:
Pr (N/O)/Pr (T/O) =Pr (N) ×Pr (O/N)/ Pr (T) ×Pr (O/T)
Now
given that we do not assign N a considerably higher absolute
probability than T, and Pr (O/N) is greater than Pr (O/T, Pr (N/O)
comes out greater than Pr (T/O), and so we have a reason to prefer N
to T. 34
And again it shows that the pattern of Plantinga's argument
is not good.
(4) And finally, as we saw (p. 9), the claim that Pr (T/R) is
greater than Pr (N/R) depends upon the claim that Pr(R/T) is greater
than Pr(R/N). But if Pr(R/N) is inscrutable, then no one can figure
out what value it should be assigned: it could be very high or very
low, or somewhere in between. And so we are in no position to make
any judgment about the relation between Pr(R/T) and Pr(R/N). And it
means if Pr(R/N) is inscrutable, the preliminary argument will fail.
Therefore, either PA fails, or the conditional conclusion of it does
not seem a great achievement: because at best it only provides some
reason to prefer T to N.
II. About the Probability Thesis
There
are two important types of objection to the Probability Thesis(PT):
One type is about PT per se – regardless its role in Plantinga's
main argument; and the second type is concerned with the role that
PT plays in the main argument as a defeater for R (as a part of Q
i.e., N&E&PT).
Objections to the Probability Thesis per se:
1. The Belief-Behavior Problem:
As we
saw, Plantinga's crucial point to estimate Pr(R/N&E) was that
Pr(R/N&E&C) is not as high as it might seem (where 'C' is the claim
that beliefs causally influence semantically as well as
syntactically, and are adaptive( i.e., A4 –see, p.3 ).) It is
because, as he claimed, false adaptive beliefs are just as likely to
evolve as true adaptive beliefs.
But there are some points regarding this claim:
(1) According to the commonsense view, though some of our actions on
false beliefs have succeeded from time to time, nevertheless in
general, our actions on false beliefs have failed; and they failed
precisely because they were actions on false beliefs. So we have
lots of inductive evidence that the holder of a mainly false
belief-system will get into trouble
sooner or later, even though the system might work in the short
run. 35
(2) False beliefs contain some truths. 36 For example, the claim that
that appletree witch is blooming is false, but this belief contains
some truths, like, that is blooming; that's tree; something is
blooming; something is there; appletree witch is not tiger witch….
And these true beliefs can explain that person's behavioral
successes.
(3) As we saw, Plantinga's most promising reason for a mainly false
but adaptive belief-system is the scenario of Paul - a prehistoric
hominid, who suffers from some sort of "deep misconception" i.e., a
person who has false beliefs about essential or definitional
properties, as opposed to merely contingent features. Paul is a
pan-psychist or an animist and, given that this position is false,
most of Paul's beliefs are false. It seems Plantinga has an implicit
argument here:
(a) Paul suffers from a deep misconception of some subject;
Therefore,
(b) All (or most) of Paul's beliefs about that subject are false.
It is not obvious how we can go from (a) to (b). Plantinga's
strategy to defend this move (in the third scenario- p.5) explicitly
assumes that the only way of referring to the various things is by
way of definite descriptions, and these descriptions work the way
Russell thought they do. Suppose Paul is a prehistoric hominid who
believes every thing is a kind of witch. Therefore, when Paul says:
"The tiger before me is dangerous", he means something like: "There
is exactly one tiger-witch before me, and it is dangerous". This
belief is false.
But suppose we follow the work of Kripke, and say that in such an
utterance the reference of he term "tiger" is due in large measure
to causal connections that exist between the speaker and actual
instance of tiger. In that case, it seems that the speaker can refer
to a tiger, and can predicate truths about the tiger, even if he
suffers from profound ignorance and/or false beliefs about the
nature of tigers themselves. It seems plausible that a person can
truthfully say some tiger is dangerous- and really be referring to a
tiger- even though the person mistakenly thinks that that tiger is,
say, essentially a creature of a certain witch kind. 37
Now suppose all these critiques are true. This line of reasoning, at
best, proves that Pr(R/N&E&C) is high. But it should not be taken as
a serious threat to PT. Because, even if Pr(R/N&E&C) comes out high,
Pr(R/N&E) still has a great chance to be low (as long as we accept
Plantinga's argument for the claim that Pr(-C/N&E) is high). It is
because, as we saw, Pr(R/N&E) can be calculated as follows:
(F2): Pr(R/N&E)=Pr(R/N&E&C)×Pr(C/N&E)+ Pr(R/N&E&-C) ×Pr(-C/N&E)
It
shows that Pr(R/N&E) is not entirely dependent on Pr(R/N&E&C). Now
following Plantinga, suppose that Pr (-C/N&E) is high, and
Pr(R/N&E&-C) is low (the latter is the case if we assume that
Pr(R/N&E&C) is high). Let's say the first one is 0.8, and the second
one is 0.2. Then Pr(C/N&E) will be 0.2; and then even if we assign
Pr(R/N&E&C) a value of 1, Pr(R/N&E) will be less than 1/2(0.36, to
be exact).
2. The Inscrutability Problems:
There
are some features in Plantinga's argument which make us to say that
Pr(R/N&E) is inscrutable rather than low. Some of the reasons are as
follows:
(4) Belief and action might be related in a way which is not among
Plantinga's four possibilities: Beliefs and actions might have
neural event as common cause. In this scenario even though beliefs
do not cause action, but they are, in an important and relevant way,
related.(This scenario, in fact, appeared in the first version of
Plantinga's argument, but in the
recent versions, he dropped it.) Plantinga believes that the
probability of R given this common-cause scenario is
"inscrutable". 38 Since Pr(R/N&E& common-cause scenario) is
inscrutable, and that Pr(R/N&E) is the weighted average of a set of
probabilities, including Pr(R/N&E&, common-cause scenario),
therefore, Pr(R/N&E) must be inscrutable as well.
(5) F2 will do the job only if (i) all possible relevant cases are
taken into account, (ii) those cases are mutually exclusive, and
(iii) all possible cases are equally probable (i.e., every case
which is conceivably possible is essentially treated as being
equally probable until additional information indicates otherwise-
it is a version of the Principle of Indifference.)
As long as Plantinga worked with F1, we had no reason to think (i)
and (ii) are met, and he never provided any reasons to think so.
Maybe he switched to F2 to fix this problem. However, F2 or F1
cannot do the job unless they assume (iii) or the Principle of
Indifference.
But it is not clear at all why we should assume A1- A4 are initially
(and regardless additional information we may receive later on)
equally probable. Besides that the Principle of Indifference itself
is philosophically a matter of controversy: it claims to obtain
probabilities from ignorance, and it is not clear at all that the
ignorance of probabilities is a good guide to
belief. 39
Anyway, whether or not we are legitimate to use F2 , Plantinga
himself confesses that the assignment of specific real numbers to
the various probabilities involved in (F2) is "laughable", and his
estimates of those probabilities are "both imprecise and poorly
grounded". And therefore, "the right course here is simple
agnosticism: one just doesn't know what that probability is…. Then
this probability is inscrutable to you". 40 After all, if we are
legitimate to use F2, then this claim seems more plausible than the
claim that Pr(R/N&E) is low.
Objection to the Probability Thesis as a premise in the main
argument:
3. The Problem of Conditionalization:
According to the main argument the fact that the probability of R
conditionalized on N&E is low or inscrutable, is a defeater for R
and hence for every thing else he believes. But here there is an
important question: Is the conditional probability of R given N&E
the appropriate probability to work with? What is so special about
this particular conditional probability? Suppose it is true that the
probability of R given N&E is low. But there are many propositions P
that naturalist believes (or are consistent with his naturalistic
position) such that Pr(R/N&E&P) is high. Why can't naturalist just
add P to N&E, and raise the probability of R given this body of
propositions?
We may call it "the problem of conditionalization". This is the
problem of determining which conditional probabilities should be
used in the main argument. 41
Plantinga's argument seems to ignore other information we have that
would make R likely. There might be some additional relevant
evidence such that if we take it into consideration, we find that
the probability of R conditional on it is high. What could that
piece of information be?
One possible option could be R itself: even though Pr(R/N&E) is low,
but we know that R is true, and if we add this piece of knowledge to
the conditional probability in question, we raised the value of it
drastically (in fact, by adding R, the probability comes out
1). Then why isn't it rational for a naturalist to do so? 42
Another option could be a proposition which either entails R or in
some way raises the probability of R given the new set of evidence.
For example, we might add the proposition O to N&E, where O is
"simply a general proposition to the effect that the initial
conditions of the development of organic life and the sum total of
evolutionary process (including ones as yet known or only dimly
understood) were and are such as to render Pr(R/N&E&O) rather
high." 43
But there is a problem here: In the main argument, Q (i.e., PT&N&E)
is a defeater for R. In other words, in this context the debate is
over whether a certain source or origin of our beliefs or faculties
would be reliable. Because of this we cannot conditionalize on
anything that presupposes or is based upon that source being
reliable. Such a probability statement would not be useful in this
context, because it would presuppose what is being questioned. If we
were to conditionalize on R itself, or any proposition such as O,
then we could come up with a high conditional probability, but in
doing so we would have begged the question in an important sense,
because it assumes the source in question is reliable.
To avoid the problem of begging the question, we can conditionalize
R on (N&E&R), or (N&E&O) only if we have some independent evidence
for R or O, and Plantinga, as we saw before, argued that it cannot
be the case. But is he right? Here we need to ask two important
questions:
(1) Why cannot R be supported by its deliverances?
(2) What if we accept R in the basic way, i.e., we accept it,
but do not accept it on the evidential basis of other
propositions? It is true that all our propositional evidence are
dependent upon R, but it seems quite plausible to say that we
have also nonpropositional evidence for R, and this sort of
evidence is not based on R.
The
first question is the question of "epistemic circularity": you
cannot come to believe that a belief source is trustworthy by
relying on the trustworthiness of that very source. But epistemic
circularity at least in some cases seems acceptable. For example,
suppose one gives a simple inductive argument for the reliability of
sense perception. She lays out a large, carefully chosen sample of
perceptual beliefs, and report in each case that the belief is true.
Thus the argument looks like this:
1. At t1, S1 formed the perceptual belief that p1, and p1.
2. At t2, S2 formed the perceptual belief that p2, and p2.
....................
Therefore, sense experience is a reliable source of belief.
Or
perhaps I have never considered the question whether my perceptual
faculties are reliable; I immediately conclude that they are, on the
grounds that, as far as I can remember, they have produced beliefs
of which the vast majority were true. Alston has argued in detail
that epistemically circular argument of these kinds can be
acceptable. 44 Plantinga also seems willing to concede that
sometimes epistemically circular arguments are acceptable. For
Plantinga, it could be the case only if there has been no reason to
question R. But as he pointed out, sometimes one has a serious
question as to whether R. If I have thought about whether R is true
and have a serious doubt as to whether it is, according to
Plantinga, I can't properly assuage that doubt by giving myself an
epistemically circular argument. 45 Consider the unreliability drug
XX example: suppose I take a good dose of XX. I believe that 95
percent of the population has the blocking gene; but I have no
belief as to whether I myself have that gene. Since Pr (one's
cognitive faculties are reliable/one has taken XX) is low or
inscrutable, and I am aware of this fact, according to Plantinga, I
have a defeater for R. Now suppose I came to believe that my
physician has telephoned me and told me that I am among the lucky 5
percent whose reliability is unimpaired by ingesting XX. Can I rely
on this new piece of information? Plantinga's answer is "No". 46 (We
will get back to this point later in our discussion of "Rationality
Defeaters".)
Suppose Plantinga is right to think that when we have a serious
reason to question R, R cannot be supported by its deliverances. But
now suppose R has a great deal of warrant in the basic way. It may
be that the general reliability of our cognitive faculties is a
fundamental presupposition that we are all warranted in accepting in
the absence of any reasons or
evidence whatever. 47 Or, if we employ Plantinga’s language, one
might say R has a special status: it might be a part of our design
plan that we cannot abandon R. In fact, Plantinga himself explicitly
says: "we do indeed, just as Reid said, believe R noninferentially,
and this belief, held in that basic way, has a good deal of warrant
for us." 48
If it is so, then (a) it means we do have some independent evidence
for R; and (b) R has a good deal of warrant for S, and that might
commit S to believing tacitly that the probability of R on all the
relevant evidence S possesses is high, and in so doing, she does not
seem to be the question.
III. About the Main Argument
We
may classify objections to the main argument into three major
categories:
(1) Objections to "The Crucial Claim", (2) Questions about "The
Rationality of Defeated R", and (3) The problem of "Atheist's
Knowledge".
1. Objections to "The Crucial Claim"
The
heart of the main argument is the move from PT to the defeater
claim. This crucial step can be stated in terms of the following
claim:
(CC) If Pr(R specified to oneself/N&E) is low or inscrutable, then
one who believes N&E has a defeater for R specified to herself or
himself.
(1) As we saw, Plantinga's argument for (CC) is the argument from
analogy. His basic strategy takes the following form: X is
relevantly like Y; Y has F, therefore X has F.
One strategy for disarming such arguments is to show that X is not
relevantly like Y. The first analogy Plantinga suggested was the
analogy between the case of being created by evil demon (a demon
whose creations have mainly false beliefs), and the situation of the
reflective naturalist. But there is an important difference between
these cases. In evil demon case, the claim is that if the person is
created by evil demon, her cognitive faculties are actually
unreliable. But Plantinga does not suggest that his arguments show
that the evolutionary naturalist must believe that the products of
N&E are not just probably but actually unreliable. In other words,
in evil demon scenario, Pr(R/S's being created by evil demon) is
zero, but in naturalist case, Pr(R/N&E), at worst, is low.
Therefore, these two cases are not analogous. 49
In the first scenario, the conditional probability clearly functions as a
defeater for R, but it is clear because the value of its conditional
probability is zero. We may change the evil demon scenario in a way
that the conditional probability comes out low, but by doing so, it
won’t be clear any more that the conditional probability does
function as a defeater for R. The same point applies to the second
scenario, i.e., the brain in a vat case.
Another strategy for disarming arguments from analogy is to suggest
that X is more relevantly like Z than X is relevantly like Y, that Z
lacks F, and therefore there is no reason to suppose that X has F,
and indeed the more reasonable conclusion is that X lacks F. Thus if
we can suggest an alternative to Plantinga's third analogy (i.e.,
the unreliability drug XX case), which is more relevantly like the
reflective naturalist's situation, in fact, we may get a very
different result. For example, imagine a pharmacologist who
discovers that he was given a sizable dose of XX in utero. Whatever
effects XX had in his cognitive faculties have been present for his
entire life. Little is known about XX, so the pharmacologist sets
about to study it. He studies its chemical composition and concludes
that, when taken, XX will cause some sort of process involving the
subject's cognitive faculties to occur- possibly resulting in total
and permanent cognitive unreliability. He understands the basic
principles of this process, and suspects that, based on what he
knows about the process, Pr(one's cognitive faculties are
reliable/one has taken XX) is low or inscrutable. But he also
believes that the precise effects of XX depend on certain unknown
factors. He realizes he needs more information about the effects of
XX. One good way to gain information about the effects of a drug is
to administer the drug to test subjects and observe what happens to
them.
Suppose the pharmacologist sets up the appropriate drug trials and
observes the vast majority of the test subjects in fact remain
perfectly reliable. 50 Given these results, it is surely appropriate
for him to revise his estimation of Pr(one's cognitive faculties are
reliable/ one has taken XX)-in light of the results of his trails,
this probability turns out to be quite high. Of course, in all this
process the pharmacologist relied on her cognitive faculties, but
there seems to be a strong intuition behind the claim that the
epistemic circularity in this case is harmless, and the
pharmacologist has no sufficient reason to doubt the reliability of
her cognitive faculties in this scenario.
The reflective naturalist's situation seems more like this scenario
than Plantinga's third case. If it is so, it is plausible to say
that the naturalist, like the pharmacologist, has no reason to doubt
R specified to herself.
(2) And finally, as it was mentioned before, N&E will function as a
defeater for R only if it meets some other conditions (C1-C3) as
well (p.14). Among those conditions, C3 requires that R has no
sufficient independent evidence. But as we discussed before(p.25 &
26), C3 is not met, i.e., R , when it is accepted in the basic way,
has a great deal of independent warrant. But if it so, then we
cannot say N&E defeats R.
2. Questions about the Rationality of Defeated R:
According to the main argument, if one believes that the conditional
statement of the (CC) is true, and also she affirms the antecedent
of the conditional (i.e., PT), then she has to take N&E as a
rationality defeater for R. And therefore, R is defeated for her,
and it is no longer rational for her to hold R.
(3) But it is not clear what Plantinga means by "rational" in this
context. There are some cases in which it seems rational to continue
to believe the defeated belief, especially when the belief is R. For
example, suppose again I take a good dose of XX. And I believe that
XX is a drug that causes widespread cognitive disorder, and it
prevents the victim from detecting her unreliability. I also believe
that 90 percent of those who take XX are no longer reliable,
although of course unable to detect their unreliability. In this
condition, I will continue to form perceptual beliefs, memory
beliefs, and the like; and I also continue to assume that my
cognitive faculties are reliable. Am I irrational in doing so? It
doesn't seem so. 51
Even though I have a defeater for R, but I can't help forming belief
in the ordinary way, and it might be a part of my design plan (using
Plantinga's terminology) that I cannot abandon R. Abandoning R might
lead to an epistemic/cognitive disaster in my every day life. 52
Therefore, on some occasion a person might acquire powerful evidence
against a given belief B but fail to change belief, due, not to
dysfunction, but to the fact that the processes producing B are
aimed not at the production of true belief (that is, not at the
maximization of true beliefs and the minimization of false beliefs)
but at the production of belief with some other property. In such a
case, the design plan may dictate no change of belief in face of
some sort of defeaters.
Plantinga believes there is a point here. In the XX case, in a
sense, I am rational to continue to believe R, and in some other
sense, I am not. In order to make this point clear, Plantinga
modifies his notion of "rationality defeater". He distinguishes
between two different kinds of "rationality defeater": "proper
function rationality defeater", and "purely epistemic or purely
alethic defeater". The proper function defeater is the kind of
defeater he defined and employed in the main argument. But now he
thinks it is not the appropriate notion of defeater we should employ
in the main argument. The purely alethic defeater is the one we need
to make the argument work. He defines "purely alethic defeater as
follows": 53
(D*) D is a purely epistemic [alethic] defeater of B for S at t
iff (1) S's noetic structure N at t includes B and S comes to
believe D at t, and (2) any person S* (a) whose cognitive
faculties are functioning properly in the relevant respects, (b)
who is such that the bit of the design plan governing the
sustaining of B in her noetic structure is successfully aimed at
truth (i.e., at the maximization of true belief and minimization
of false belief) and nothing more, (c) whose noetic structure is
N and includes B, and (d) who comes to believe D but nothing
else independent of or stronger than D, would withhold B (or
believe it less strongly).
The
person who believes she has ingested XX has a purely alethic
defeater for R. And now Plantinga claims that the naturalist who
comes to see PT also has a purely alethic rationality defeater. She
might continue to assume R; nevertheless she has a defeater of this
sort. She won't be able to help believing or at least assuming R;
but if she reflects on the matter, she will also think that what she
can't help believing is unlikely to be true. 54
(4) We cannot simply conclude that N&E defeats R unless N&E is
significantly better supported than R. If N&E is no better supported
for S than R, then there is no rational basis for a preference for
withholding R, rather than withholding N&E. So the question is "What
defeats what?" Why does Plantinga think in this context N&E defeats
R, and not visa versa? 55 At any rate, it seems quite clear that we
believe that R is supported much more than N&E, therefore the
natural thing to say is that N&E cannot defeat R, and if anything
defeats anything, it must be R which defeats N&E.
Of course, it shouldn’t be a serious objection to Plantinga’s
position. It is because either N&E is not supported by evidence
enough to defeat R, and therefore it would be defeated by R, or it
can defeat R, and so it would be a self-defeating position. In
either case, it is irrational to hold N&E.
(5) But there is a problem here. Plantinga claims that even if
Pr(R/N&E) is inscrutable, N&E still will be a defeater for R. But
what is the reason for this claim? Here again, the heart of
Plantinga’s argument is the argument from analogy (see, for example,
his case of “the widget under the red light, p. 17). But here again
we may disarm or weaken his argument from analogy by suggesting
another analogy: Suppose I take a good does of XX by mistake. I run
to the pharmacy to ask the pharmacist about XX. The pharmacist tells
me that XX causes sever hallucinations within one hour of taking it.
But then the manager of the store comes along, and tells me that the
man in the pharmacy is not the pharmacist but a new technician, and
he is not reliable on drug information. Still the so-called manager
himself does not look wholly reliable. So I do not know what to
believe about the alleged hallucinogen drug. It means I am agnostic
about Pr( R/ I ingested a good dose of XX). Does it mean as soon as
I found out that the conditional probability in question is
inscrutable I have a defeater for my cognitive faculties being
reliable, and I am not rational in believing that I have hands and
so on? It seems we have very strong intuition that it is not the
case. In this scenario, it is surely rational for me to hold R, and
rely on my cognitive faculties. If it so, then we have good reason
to doubt about Plantinga's claim that if Pr(R/N&E) is inscrutable,
N&E will be a defeater for R.
3. The Problem of "Atheist's Knowledge":
Now
suppose the reflective naturalist (who is also, by definition, an
atheist) has indeed an undercutting defeater for R. Does it mean she
has a defeater for everything else she believes? Plantinga's answer
is clearly "yes": An atheist, if she is a reflective naturalist, has
no knowledge. Of course Plantinga believes that "a naturalist
[/atheist] who has never raised the question of reliability, or seen
its connection with the Probability Thesis…will have no defeater for
R or anything else he believes." 56 Presumably, we can say that a
naturalist who has not been exposed to Plantinga's argument can know
all sorts of things. She can further know that she knows that she
knows those things, and even she can know that her cognitive
faculties are reliable. But as soon as she is exposed to Plantinga's
argument, she knows nothing and even if she wants, she is in no
position to abandon naturalism in favor of theism, and return to
God!
(6) More precisely, Plantinga's response to the question of the
knowledge of the atheist is based on two assumptions:
(i) If S has a defeater for R, then she has a defeater for
everything else she believes.
(ii)If S has a defeater for B, then she has no knowledge of B.
Suppose we granted (ii), but why should one believe (i)? One
possible reason is that (i) is based on the following implicit
premise:
(IP) Whenever S accepts any deliverances of a faculty, S's doing so
rests on the implicit premise that that faculty is reliable. Since R
is an implicit premise behind any belief, if S has a defeater for R,
she has a defeater for every thing else she believes. But (IP) is
false. For example, a child may be innocent of any belief in the
reliability for her faculties, yet it is hard to say she is
irrational in her simple beliefs about food and shelter. 57
Conclusion
Plantinga's argument against naturalism, whether nor not sound, may
provide important insight into the epistemic shortcomings of
naturalism. It seems plausible to me to make the following claims
about Plantinga's argument against naturalism:
(1) The Probability Thesis is plausible only if Pr(R/N&E) is
considered inscrutable.
(2) But if Pr(R/N&E) is inscrutable, the Preliminary Argument
will fail. If Pr(R/N&E) is low, then the second version of the
Preliminary argument will survive, but the conclusion is not an
important achievement.
(3) The strength of the Main Argument to a large extent depends
upon the intuition behind Plantinga's analogies. For many, it is
not easy to share the same intuition Plantinga has, and there
are some ways to defend other intuitions which might lead to
entirely different results. In particular, if we assume that
Pr(R/N&E) is inscrutable, it is hard to see how N&E can be a
defeater for R. The Crucial Claim, when Pr(R/N&E) is
inscrutable, does not seem true at all.
Notes
* I am indebted to Professor William Forgie, and Professor Anthony
Brueckner not only for many helpful discussions
of the issues involved in this paper, but also for their helpful
criticisms of earlier draft of it.
1) 1Plantinga sometimes calls it "Darwin's Doubt", because this doubt seems to have (at least occasionally) crept
into the mind of Darwin himself. See, Plantinga (1993) p.219
2 Plantinga (1994),p.1
3 Plantinga (2002,1) P.2
4 For example, Patricia Churchland(1987), p.548.
5 Plantinga , (1993), pp223-225, introduced five
accounts, but later on he reduced them into four
accounts, see
Plantinga (2002,1) Pp.6-9.
6 Plantinga (1993) p.225. Plantinga, hereafter, for
simplicity, assumes that our behavior is a causal
product just
of our beliefs and desires.
7 Plantinga (1993),p225.
8 Plantinga (1994), p.9
9 Branden Fitelson and Elliot Sober, (1998), p.417. And
Fodor also seems to make a similar point; see, Fodor,
(2002),p.39
10 Plantinga, (2002,2), p.260
11 Ibid,p.213; also Plantinga (2000), p.236
12 Plantinga (2002,1), p.6.
13 Plantinga (1993),p.228
14 Plantinga (1994),p.10
15 Ibid. ,p.10
16 The first version appeared in Plantinga
(1991),pp.38-39, and reappeared in Plantinga (1993),
p228-229
17 Plantinga (1993),p.228
18 Plantinga, (2000),p.229
19 This formula is the Bayes's Theorem we already had,
but this time we are concerned with the probability of
each item ( in the previous standard formula)given our
background knowledge(B).
20 For some other difficulties of the second premise,
see; Brand Fitelson and Elliott Sober,(1998),pp414-416
21 Pr(N/R)=Pr(N)Pr(R/N)/Pr(R)
and also
Pr(T/R)=Pr(T)Pr(R/T)/Pr(R)
Therefore;
Pr(N/R)/Pr(T/R)=Pr(N)Pr(R/N)/Pr(T)Pr(R/T)
22 The second version appeared in Plantinga (2000),
p230-231
23 Plantinga (1994), p.29; and later on ,Plantinga
(2000),p.362
24 Plantinga (1994), p.37
25 Plantinga (2002,2), p.240
26 For these conditions, see: Plantinga (1993),p. 233
27 Ibid. p.234
28 Ibid. p.233
29 Plantinga (2002), Pp.269-271
30 Plantinga (1993),pp. 230-231
31 We can get this formula as follows:
According to the General Conjunction Rule of
Probability(GCR), we have:
(1) Pr(A&B)=Pr(A/B)Pr(B); or,
(2) Pr(A/B)=Pr(A&B)/Pr(B)
Now we want to compare the following probabilities:
(F): Pr(T/R&O)/Pr(N/R&O)
If we apply GCR(2) to (F), we will get:
Pr(T/R&O)/Pr(N/R&O)=[Pr(T&R&O/Pr(R&O)]/[Pr(N&R&O)/Pr(R&O)]
=Pr(T&R&O)/Pr(N&R&O)
=Pr(R&T&O)/Pr(R&N&O)
=[Pr(R/T&O)Pr(T&O)]/[Pr(R/N&O)/Pr(N&O)] ---{GCR(1)}
=[Pr(R/T&O)Pr(O&T)]/[Pr(R/N&O)/Pr(O&N)]
=[Pr(R/T&O)Pr(O/T)Pr(T)]/[Pr(R/N&O)Pr(O/N)Pr(N)]---{GCR(1)}
32 See, Richard Otte (2002),pp.138-141
33 Fitelson and Sober, (1998),p.416
34 For a similar kind of reasoning , see, Paul Draper,
(1996),pp. 12-29
35 Fodor,(2002),p.38
36 Ibid., p.34
37 William Ramsey made this point; see, his
article(2002), p.27
38 Plantinga, (1994), p.8, fn. (16).
39 William Alston, (2002),p183. Fitelson and Sober also
have some critical points on the use of the Principle of
Indifference in Plantinga's argument, see their article,
(1998), p. 422-423.
40 Plantinga, (1994) p.10
41 See; Richard Otte (2002), pp.141-142. Also, Fales
(2002), p.53. Also, O'Connor (1994),pp.534-538. Also
the similar point, even though in the different context,
can be found in Fitelson and Sober (1998),p.418.
42 See, Carl Ginet(1995),p.407.
43 O'Connor, (1994),p.535
44 For a detailed discussion on "epistemic circularity",
see, Alston (1986),pp.1-30. Also the following authors
raised some questions on epistemic circularity in the
main argument: Bergmann (2002),pp.76-82;Van Cleve
(2002),pp.109-111.; Talbott (2002),pp153-165.
45 Plantinga (2002,2),p.242
46 Ibid., pp.226-227.
47 The following philosophers believe that one can
believe R in a "basic way" (if we use Plantinga's
jargon),
and it can provide a reason to deny Plantinga's
conclusion in the main argument: Bergmann (2002),esp.
pp.66-68, Bergmann tries to give a Reidian response to
Plantinga, and his main point is based on the basicality
of R.
O'Connor even claims that "strictly speaking, there
couldn't be a defeater for R, for any creatures in any
possible
world", see O'Connor (2002),p.131. Also see, Alston
(2002), p. 200-201.
48 Plantinga (2002,2),p. 230.
49 Merricks (2002),p.172
50 This scenario is suggested by Wielenberg (2002),p.94.
51 Talbott (2002), esp.155-159. The case of "Purloined
Letter" in O'Connor's article is employed to make the
same point, see his article(2002),p.130-131. Also
Bergmann made the similar point, see his
article(2002),p.71-72.
52 O'Connor (2002),p.131
53 This definition appears for the first time in
Plantinga (2000),p. 363; and then in Plantinga (2002,2),
p. 209
54 See: Plantinga(2002,2), p.205-211
55 Alston (2002), pp.196-199
56 Plantinga (2002,2), p.243.
57 Sosa (2002),p.99; and also see, Van Cleve (2002),esp.
pp.121-123.
References
Alston,
William (1986), "Epistemic Circularity", Philosophy
and Phenomenological Research, Vol. 47, Issue 1, Sep,
1-30 _____________ (2002), "Plantinga, Naturalism, and
Defeat", in Naturalism Defeated?, 176-203
Beilby, James (ed.) (2002) Naturalism Defeated?
(Essays on Plantinga's Evolutionary argument against
Naturalism. Cornell University Press.
Bergmann, Michael (2002), "Commonsense
Naturalism", in Naturalism Defeated? Pp. 61-90
Churchland, Patricia (1987), "Epistemology in the
Age of Neuroscience", Journal of Philosophy, 84, Oct.
Draper, Paul (1996), "Pain and Pleasure: An
Evidential Problem for Theists", in The Evidential
Argument from Evil, ed. by Daniel Howard-Snyder, Indiana
University Press, pp.12-29.
Fales, Evan (2002), "Darwin's Doubt, Calvin's
Calvary" in Naturalism Defeated?, pp.43-58
Fitelson, Branden & Sober, Elliot, (1998)
"Plantinga's Probability Argument against Evolutionary
Naturalism", Pacific Philosophical Quarterly (1998,
Vol.79, pp.115-129), republished in: Pennock, T. Robert
Intelligent Design Creationism and Its Critics
(Philosophical, Theological and scientific
Perspectives), Massachusetts Institute of Technology,
2000, pp.411-427.
Fodor, Jerry (2002) "Is Science
Biologically Possible?", in Naturalism Defeated?,
pp.30-42
Ginet, Carl (1995), "Comments on Plantinga's
Two-Volume Work on Warrant", Philosophy and
Phenomenological Research, Vol.55, Issue 2, Jun.,
pp.403-408.
Merricks, Trenton (2002), "Conditional
Probability and Defeat", in Naturalism Defeated?,
163-275.
O'Connor, Timothy (1994), "An Evolutionary
Argument against Naturalism?", Canadian Journal of
Philosophy, Vol.24, No.4, Dec., pp.527-540.
________________ (2002), "A House Divided against Itself
Cannot Stand", in Naturalism Defeated?, pp.129-134.
Otte, Richard (2002), "Conditional Probabilities
in Plantinga's Argument", pp.135-149
Plantinga, Alvin (1991), "An Evolutionary
Argument against Naturalism", Logos: Philosophic issues
in Christian Perspective, Vol. 12, pp.27-49.
_____________ (1993) Warrant and Proper Function.
Oxford: Oxford University Press.
_____________ (1994) "Naturalism Defeated", unpublished
paper. See the following link:
www.homestead.com/philofreligion/files/alspaper.htm
_____________ (2000) Warranted Christian Belief. Oxford:
Oxford University Press.
______________ (2002,1), "The Evolutionary Argument
against Naturalism: An Initial Statement of the
Argument", in Naturalism Defeated?, 1-12.
_____________ (2002,2), "Reply to Beilby's Cohorts", in
Naturalism Defeated?, pp.204-275
Ramsey, William (2002), "Naturalism Defended", in
Naturalism Defeated? Pp.15-29.36
Reiter, David (2000), "Plantinga on the epistemic
implications of naturalism", Journal of Philosophical
research, Vol. XXV, pp.141-147.
Sosa, Ernest (2002), "Plantinga's Evolutionary
Meditations", in Naturalism Defeated?, pp.91-102.
Talbott, W.J. (2002), "The Illusion of Defeat",
in Naturalism Defeated?, pp.153-164.
Van Cleve, James (2002), "Can Atheists Know
Anything?" in Naturalism Defeated?, pp.103-125.
Wielenberg, Erik J. (2002), "How to be an
alethically rational naturalist?", Synthese, 131:81-98.
