Notes on Tyler Wunder’s “Anti-Naturalism and Proper Function”, Religious Studies 44 (2008), pp. 209-224.
Plantinga's first argument: The Problem of Artifactual Paradigms
We can express the argument as follows: There is a set of inter-definable terms related to the notion of proper function, e.g., purpose, function, dysfunction, damage, (non-statistical) normality, and design. The literal, primary meaning of such terms is fixed by the set of clear, unambiguous, paradigm cases to which the term applies. But such cases are all of intelligently designed things – artifacts--, and not natural organisms or their parts. Therefore, the notion of intelligent design is constitutive of the literal, primary meaning of such terms. But naturalism entails that organisms and their parts are not intelligently designed. Therefore, if naturalism is true, then ‘proper function’ and related terms are not literally applicable to organisms or their parts. But organisms and their parts cannot be adequately described without reference to such terms. Therefore, naturalism is inherently incapable of adequately describing the world. But any worldview that is inherently incapable of adequately describing the world is false. Therefore, naturalism is false.
I. Grant, arguendo, that the primary notion of proper function and related terms are fixed by the set of clear and unambiguous cases to which such terms apply. Still, it’s not obvious that only artifacts are among the set of paradigm cases of things to which such notions apply. For Plantinga’s criteria as to what constitutes a paradigm case appears to be that it is clear and unambiguous, or obvious to all or most people, that the term or terms at issue apply to it. But examples of natural organisms and their parts with proper function are at least as clear and unambiguous cases of properly functioning things as cases of artifacts, as all or most people think such terms apply to them. So it appears that Plantinga will need to do more work to rule out the reply of just standing this argument on its head and asserting that since prima facie non-designed organisms and their parts are among the paradigm cases that fix our notion of proper function and related terms, naturalists can appropriately and literally apply the notion of proper function to organisms and their parts.
To put the point another way: there appears to be a tension in Plantinga's argument. For the evidence for premises  and  is the same: they are both supported by clear cases of proper function (artifacts for premise , organisms and their parts for premise ). Now organisms are either clear cases of things capable of proper function or they aren't. If they are, then they are among the set of things that fix the meaning of our notion of proper function, in which case premise  is called into doubt. But if organisms and their parts are not among the clear cases of things capable of proper function, then it's no longer clear that talk of proper function is ineliminable when applied to organisms and their parts, in which case premise  is called into doubt. Therefore, either way, the argument from artifactual paradigms is in trouble.
II. But in any case, grant, arguendo, that all paradigm cases of properly functioning things are artifacts. Still, it’s not clear that the naturalist needs to refer to natural organisms and their parts in terms of proper function and related terms. For there is another set of concepts that will do even better: the notion of health and related notions, e.g., disease, injury, pathology, pain, life, and death (examples: "a properly functioning heart pumps blood" is more naturally and literally put as "a healthy heart pumps blood"; "That bird's heart is malfunctioning" is more naturally and literally put as "that bird’s heart is diseased", etc.). The naturalist can then say that the notion of proper function and related notions are mere analogical extensions of the more primary, literal notions of health and related concepts. And if that's right, then it looks as though "proper function" talk, when applied to organisms and their parts, is eliminable, in which case premise  of the argument from artifactual paradigms is defeated.
Plantinga anticipates this sort of reply, and asserts that the notion of health and related notions are derivative of the notion of proper function and related notions, and thus that such a reply fails: the artifactual notion of proper function is ineliminable. But this is unconvincing for the following reasons:
First, we’ve seen that it’s not clear that the paradigm cases are all artifacts, in which case there is no problem here. But let's waive this objection for now, as we are currently assuming arguendo that all the paradigm cases are of artifacts.
Second, if inter-definable terms refer to identical paradigm cases, then it’s not true that ‘health’ and related terms reduce to talk about proper function. For the clear and unambiguous, paradigm cases of health and related terms are organism and their parts, not artifacts. So, for example, it's odd to say that:
-My car is healthy.
-My friend’s leaky tire is injured.
-The pump in my car is diseased or dead (if it stops working).
Thus, it looks as though the paradigm cases of properly functioning things and the paradigm cases of healthy things are distinct. And if that’s right, then by Plantinga’s paradigm case method of analysis, the notions of proper function and health are likewise distinct.
Third, the cases of organic proper function Plantinga appeals to can be re-described via the organic notion of health and related terms:
-“The hawk’s heart that only beats 25 times per minute is not functioning properly” can be replaced with “the hawk’s heart that only beats 25 times per minute is diseased, or pathological, or suffering a stroke, or a heart attack, or is on its way to death”
-“AIDS damages the immune system and makes it function poorly” can be replaced with “AIDS interferes with the body’s ability to withstand disease”.
-“Multiple sclerosis causes the immune system to malfunction in such a way that the white blood cells attack the nervous system” can be replaced with “multiple sclerosis is a disease that inhibits the body’s immune system so that white blood cells attack the nervous system”.
-“The purpose or function of the heart is to pump blood, not to make that thumpa-thumpa sound” can be replaced with "healthy hearts pump blood, but it is irrelevant to the health of their organism that they make that thumpa-thumpa sound insofar as that sound is independent of pumping blood”.
Fourth, the organic family of terms makes better sense than the proper function family of terms when they are applied to whole organisms (as opposed to just their parts). For example, it’s odd to say that a hawk has a function or purpose, but it’s not odd to say that a hawk is healthy.
Thus, for all these reasons, we have good reason to think that the organic family of terms (health, etc.) applies primarily to organisms and their parts, and that the proper function family of terms only applies analogously or derivatively to organisms and their parts. And if that's right, then we can eliminate literal use of proper function language when referring to organisms and their parts, in which case premise  of the argument is defeated, and the argument is a failure.
Plantinga's Second Argument: Against Naturalistic Analyses of Proper Function
Plantinga picks what he takes to be a representative sampling of the most promising analyses of proper function: Pollock’s, Millikan’s, and Bigelow & Pargetter’s accounts. But, argues Plantinga, all such analyses are hopelessly flawed and inadequate: they fail to provide a set of logically necessary and sufficient conditions that apply to all cases of the analysandum. This is because one can find, for each such proposed analysis, a logically possible scenario according to which the conditions stated in the analysis are either not necessary or not sufficient. Therefore, probably, naturalistic analyses of proper function are inherently inadequate. But proper function is a real feature of natural organisms and their parts. Therefore, probably, naturalism is false.
I. Plantinga’s standards for a satisfactory analysis are too stringent
First, Philosophy's track record for providing such analyses is abysmal. From Plato to the present, philosophers have been trying to offer such analyses, and few if any have succeed. So, if we were to apply such standards of a satisfactory philosophical analysis universally, we’d have to say that virtually no analysis is satisfactory.
Second, Plantinga’s own analysis of proper functionalist epistemology fails such standards. First, it’s possible for there to be a world just like ours, and with people just like us, but which popped into being five minutes ago without a designer. So, it’s possible to have proper function without a designer. (see also Sosa’s variation on Davidson’s famous Swampman thought experiment for a variation on this. Here's my own counterexample: Suppose theism is true, and suppose God wants to create a multiverse, and uses a randomizing device to alter the fundamental constants of nature for each universe that arises. In such a multiverse, there is a universe that includes a planet with conditions relevantly similar to ours, with people very much like us in terms of their cognitive faculties, except such faculties arose by unguided mutation and natural selection, and without the intention of an intelligent designer (recall the stipulation about the randomizing device). So their cognitive faculties are just like ours, except no intentional design was involved in the creation of their faculties. Intuitively, however, many of the beliefs of such beings have warrant.). So, designed cognitive faculties aren't necessary for warrant. Nor is design sufficient for warrant. (Here I refer to Richard Feldman's more intuitive counter-example, which can be found here.) So if we are to apply the standards of analyses Plantinga applies to naturalistic analyses of proper function consistently and universally, then not even Plantinga's analyses pass muster.
Third, such an approach to analysis, while perhaps appropriate for mathematical and logical concepts, seem inappropriate for the “messy” realm of biology. For we lack satisfactory analyses (in Plantinga’s sense here) of other biological concepts, such as “life”, “species”, and “gene”. It therefore seems we should expect the same is true of “(biological) function”.
II. Plantinga double-standards for satisfactory analyses
Plantinga is applying two standards of analysis here:
(i) The strict one of logically necessary and sufficient conditions for concept application.
(ii) A comparatively lax one according to which the logically necessary and sufficient conditions need only apply to a small set of "core", paradigm cases to which the target concept applies. The "penumbral" cases -- i.e., cases where it's at least somewhat less clear that the target concept applies in the most literal sense -- need not satisfy the stricter standards of satisfying the logically necessary and sufficient conditions spelled out in the relevant analysis.
Plantinga applies the strict type-(i) analysis to views he doesn't like, but applies the relaxed type-(ii) analysis to his own accounts of warrant and of proper function. Pending some principled basis for applying this double-standard, such a method of evaluating his own views vis-a-vis others appears to be illegitimate. Let's see how Plantinga's use of this double-standard plays out with the two key notions at issue here: the concept of warrant and the concept of proper function.
A. Double-standards for his concept of warrant
As alluded to above, Plantinga himself rejects the strict, type-(i) standards for satisfactory philosophical analyses when it comes to his own account of warrant. As he puts it:
"I must acknowledge a complication with respect to my way of thinking of warrant. I aim at something in the neighborhood of an analysis of warrant: an account or exploration of our concept of warrant, a concept nearly all of us have and regularly employ. (As we all know, desperate difficulties best any attempt to say precisely what analysis is.) Thus at the least I should be looking for necessary and sufficient conditions. But I very much doubt that there is any short and elegant list of conditions at once severally necessary and jointly sufficient for warrant. This is a way in which philosophy differs from mathematics; and epistemology differs more from mathematics, along these lines, than, for example, philosophy of logic or the metaphysics of modality. Our concept of warrant is too complex to yield to analysis by way of a couple of austerely elegant clauses."(Warrant and Proper Function, ix)
In the place of the strict, type-(i) standards for adequate analysis, Plantinga states the conditions for the more lax, type-(ii) standards for adequate analysis mentioned above:
"The structure of [warrant], I believe, involves a central picture, a group of central paradigms – clear and unambiguous cases of knowledge – surrounded by a penumbral belt of analogically related concepts, concepts related by different analogies and standing in different degrees of closeness to the aboriginal paradigms. Between the central core area and this penumbral belt there is a more shadowy area of borderline possible cases, cases where it isn’t really clear whether what we have is a case of warrant in the central sense, or a case of one of the analogically extended concepts, or neither of the above; and beyond the penumbral belt we have another area of borderline cases. Hence perhaps a good way to characterize our system of analogically related concepts of warrant is to give first, the conditions necessary and sufficient for the paradigmatic core. (Even here, as we shall see, there is no stylishly sparse set of necessary and sufficient conditions: various qualifications, additions and subtractions are necessary.) Second, what is needed is an exploration of some of the analogical extensions, with an explanation of the analogical bases of the extensions. This way of proceeding is less elegant and pleasing and more messy than the analysis that we learned at our mother’s knee: it is also more realistic. (Warrant and Proper Function, ix)
Thus, as we've pointed out above, Plantinga thinks the necessary and sufficient conditions of a proper analysis need only apply to the “core” or paradigm cases; proposed counter-examples that only apply to the “penumbral” cases are not successful counter-examples to the proposed analysis of a given concept. But Plantinga’s proposed counter-examples to naturalistic analyses of function and proper function appear to involve just the penumbral cases. But if so, then if we apply such relaxed standards to naturalistic analyses of function and proper function, then Plantinga’s proposed counter-examples appear to be unsuccessful.
B. Double-standards for his concept of proper function
Again, Plantinga himself rejects the strict, type-(i) standard of analysis when it comes to his own “intelligent design" account of proper function:
“Perhaps it is true that our concept of proper function doesn’t have non-trivial necessary and sufficient conditions, but if that’s true, it is not in my opinion a defect in the concept; in particular, it isn’t grounds for supposing the concept unintelligible. (Many magnificently intelligible concepts do not have nontrivial necessary and sufficient conditions – for example, red, horse, and belief). (Plantinga, “Reliabilism, Analyses, and Defeaters”, Philosophy and Phenomenological Research 55 (1995), pp. 454-455.)
But if Plantinga accepts such relaxed standards of analysis for his own accounts, then it’s not clear what principled grounds he could offer against the naturalist applying the same standards to naturalistic accounts of proper function. But if not, then since Plantinga's proposed counter-examples to naturalistic accounts of proper function all appear to involve only "penumbral" cases, it looks as though his proposed counter-examples against such accounts are unsuccessful.