I'm currently working up some notes on Adrian Bardon's “Reliabilism, Proper Function, and Serendipitous Malfunction”, Philosophical Investigations 30:1 (Jan. 2007), pp. 47-64. Here's a draft of what I have so far (which is a bit rough and incomplete). What is primarily of interest to me is his nice reply to Plantinga's arguments against the prospects of a naturalistic analysis of function.
1. Plantinga’s Case for Preferring Proper Functionalism Over Straight Reliabilism
1.1 Setup: Goldmanian reliabilism explicated
1.1.1 first condition: the relevant cognitive processes must yield a high ratio of true beliefs
1.1.2 second condition: local reliability: the faculties must also be able to discriminate: they wouldn’t cause one to believe in counterfactual situations where the belief is false
1.2 Plantinga’s criticism:
1.2.1 mere reliability isn’t sufficient for warrant
1.2.2 this is because a cognitive processes can be merely accidentally reliable, due to a serendipitous malfunction. (see this post for one of Plantinga's thought experiments involving such a case).
1.2.3 therefore, a proper function condition must be added to a pure reliabilist account to rule out such cases
2. Plantinga’s Proper Functionalism Explicated: a belief B is warranted if and only if:
2.1 it is produced by a properly functioning cognitive faculty
2.2 the relevant cognitive faculty is aimed at producing true beliefs
2.3 the belief is produced in an epistemically congenial environment
2.4 the objective probability that the relevant cognitive faculty will produce a true belief is high
2.5 the belief produced by the relevant cognitive faculty is held with a sufficiently high degree of firmness (this condition is added to account for the fact that warrant admits of degrees).
3. Bardon’s First Criticism of Plantinga’s Proper Functionalism: Concerns about Generality and Applicability
3.1 Plantinga argues that the notion of function applies most fundamentally to artifacts, and only derivatively to natural objects like organisms and their parts
3.2 This comes out strongly in his replies to certain criticisms of his view
3.2.1 Taylor, Feldman, and Sosa have each offered counterexamples to Plantinga’s account, according to which a person acquires a reliable cognitive faculty by means of a serendipitous accident or malfunction.
3.2.2 In these counterexamples, then, it appears that the person has a warranted belief in the absence of proper function
3.2.3 One of Plantinga’s main replies to these counterexamples:
188.8.131.52 distinguish between “core” and “penumbral” (or analogical) uses of the term “proper function”, where the core cases are those involving artifacts produced by intelligent agents and the penumbral cases are those produced by nature or accident.
184.108.40.206 say that his analysis of warrant only involves the “core”, intelligence-based notion of proper function
220.127.116.11 conclude that since the alleged counterexamples are all cases involving just the penumbral use of “proper function”, they aren’t successful counterexamples to his account
3.3 Bardon’s reply: whatever one makes of Plantinga’s replies, they entail that his supernaturalistic version of proper functionalism is inapplicable to the beliefs of actual human beings if naturalistic evolution is true. For if it is true, then all cognitive faculties are the result of the serendipitous malfunction of previous organs -- via the creative process of mutation and natural selection – and not by intelligent design. Thus, if his analysis of warrant is correct, then it’s inapplicable to humans in the actual world(!).
4. Bardon’s Second Criticism: Plantinga’s Case Against a Naturalistic Version of Proper Functionalism is Undercut
4.1 Setup, part I: a representative naturalistic analysis of function and proper function: Bigelow and Pargetter’s (B&P's) account
4.1.1 it doesn’t appeal back to the design plan of an intelligent agent
4.1.2 it doesn’t appeal back to the organ's or organism's natural etiology
4.1.3 rather, it appeals to the propensities of the organ, organism, or character thereof
4.1.4 the account: (SEP) an organ or character thereof has a (naturally produced) function if and only if it confers a survival-enhancing propensity on the organism in its natural habitat.
4.2 Setup, part II: Plantinga’s three criticisms of B&P's naturalistic analysis of proper function
4.2.1 it's circular
18.104.22.168 in their description of the relevant sort of environment that produces a character that has a function (where, in the case of an organ, the environment is a larger system of organs) they say that it is one that is a "healthy, natural,properly functioning environment"
22.214.171.124 thus, they appeal to proper function while defining the notion of function(!)
4.2.2 their conditions aren’t sufficient: The Nazi example
126.96.36.199 imagine a Nazi-like regime that kills off all infants of a people group who have non-defective vision
188.8.131.52 then the people group’s environment confers survival enhancing propensities in those with defective vision
184.108.40.206 yet, intuitively, their visual faculties don’t function properly
4.2.3 their conditions aren’t necessary: the Immune Cells example
220.127.116.11 some systems only function when an organ or organism is damaged or under attack
18.104.22.168 but if so, then if such systems were activated in normal conditions, they would not confer survival enhancing propensities; rather, they would tend to kill or maim the organisms who have them
4.2.4 Plantinga’s verdict: all such naturalistic analyses of function are doomed to fail
4.3 Bardon’s revision of B&P’s account
4.3.1 The account: (SEP*) A character of an organ or organism has an n-function (i.e., a naturally-produced function) iff: the character generates survival-enhancing propensities for the organism in circumstances that have been common (i.e., of sufficient frequency and duration) to the activity of the mechanism that produces the character and to the species to which the character belongs.
4.3.2 it’s like B&P’s account in that it takes the organ’s or organism’s environment to be what is relevant to the production of the function of an evolved character
4.3.3 it differs from their account by appealing only to the portion of the environment that constitutes and manifests the mutation and natural selection pressures that produce the character in question, whereas B&P appeal to a “healthy, natural, properly functioning environment”.
4.4 How Bardon’s account avoids Plantinga’s criticisms of B&P’s account
4.4.1 how it avoids the “Circularity” objection: unlike B&P’s account, Bardon’s account doesn’t appeal to proper function in his definition of the environment that generates survival-enhancing propensities. Nor does his account illicitly appeal to function in any other way.
4.4.2 how it avoids the “Not Sufficient” objection: since Bardon’s account stipulates that the portion of the environment relevant to creating a character or organ with a function is only that which involves the natural pressures of mutation and natural selection, Plantinga’s scenario involving a Nazi-like regime that selects humans with impaired vision is not a true counterexample. For the environment in that scenario involved artificial selection, appealing as it did to the intelligent, intentional action of rational agents. (Indeed, Bardon could adopt Plantinga’s “core vs. penumbral case” strategy here and say that Plantinga’s alleged counterexample only involves a penumbral case of a character-creating environment).
4.4.3 how it avoids the “Not Necessary” objection:
22.214.171.124 recall that Plantinga’s alleged counterexamples involve systems and mechanisms that operate only when an organ or organism is damaged or under attack (the blood-clotting cascade system, the anti-body system, etc.).
126.96.36.199 but we’ve seen that Bardon’s account stipulates that the relevant environment is the one involving the processes of mutation and natural selection that give rise to a given character or organ.
188.8.131.52 so on Bardon’s account, the relevant environments for such organs and characters are only those within the relevant organism that involve damage or attack.
184.108.40.206 but of course the operation of those systems during damage or attack will not normally lead to the death of the relevant individual or species.
4.5 Other virtues of Bardon’s account (fill in later)
4.6 Conclusion: Plantinga’s case against naturalistic accounts of function is undercut.
5. Why a Revised Version of Goldmanian Reliabilism is to be Preferred to Plantinga’s Proper Functionalism (fill in later)