Plantinga argues that internalist accounts of warrant are inherently inadequate, and thus that some form of externalism must be correct. However, he also argues that standard versions of externalism -- e.g. Goldmanian reliabilism -- are inadequate as well, on the grounds that they can't account for the fact that a belief can't be warranted if it's formed by a mechanism or process that is only accidentally reliable:
"Suppose I am struck by a burst of cosmic rays, resulting in the
following unfortunate malfunction. Whenever I hear the word 'prime' in
any context, I form a belief, with respect to a randomly chosen natural
number less than 100,000, that it is not prime. So you say "Pacific
Palisades is prime residential area" or "Prime ribs is my favorite"...; I
form a belief, with respect to a randomly selected natural number
between 1 and 100,000 that it is not prime. The process or mechanism in
question is indeed reliable (given the vast preponderance of
non-primes...) but my belief -- that, say, 41 is not prime -- has little
or no positive epistemic status. The problem isn't simply that the
belief is false; the same goes for my (true) belief that 631 is not
prime, if it is formed in this fashion. So reliable belief formation is
not sufficient for positive epistemic status." (Warrant: The Current Debate (OUP, 1993), p. 210.).
Plantinga then goes on to argue that an externalist about warrant can avoid the problem of accidental reliability if they go proper functionalist, as a properly functioning truth-aimed cognitive process or faculty precludes accidental reliability. Or so argues Plantinga.
However, Richard Feldman (“Proper Functionalism,” Nous 27 (1993), pp. 34-50) has constructed a variation on Plantinga's "primes" counterexample above to show that the problem of reliably-formed-yet-unwarranted beliefs arises for Plantinga's proper functionalist version of externalism as well. Thus, Feldman changes the counterexample
so that the cause of the beliefs about primes is not a burst of cosmic
rays, but rather a cognitive faculty formed by an intelligent designer
who designs the person to naturally and spontaneously form such beliefs
about prime numbers whenever they hear the word 'prime'. In such a case,
all of Plantinga's conditions of warrant are satisfied: according to
the thought experiment, we have a cognitive faculty that, when
functioning properly, reliably produces sufficiently firmly-held true
beliefs when in epistemically congenial environments. However, such
beliefs have little by way of warrant.
Therefore, the conditions laid
out in Plantinga's proper functionalism are not sufficient for warrant. And since his proper functionalist amendment to straight reliabilist externalism fails to rule out the problem cases he raises for the latter, the former appears to be unmotivated.