It seems to me that, at least in a certain sense, we must take ourselves to have reliable cognitive faculties since we simply have no rational choice but to do so. Even to entertain the question we must rely on our own judgment. This is not, of course, to say that we must take our cognitive faculties to be reliable in the broadest sense, as if we should place as much confidence in our judgment on matters of metaphysics as we do in our judgment of whether or not the front door is closed. However we must be able to trust at least some significant subset of our beliefs if we are to at all aspire to rational thought. This is what I mean by having reliable cognitive faculties, and I think it's what Plantinga means as well when he discusses the proposition he labels R. So, while it is true in another sense that we might have unreliable faculties---that event R might be false---if it turns out that they really are unreliable, we cannot ever hope to know it. All we can do is be humble about our unusual situation, and acknowledge that we can be mistaken even about that which seems to us obviously true.
Now, Plantinga argues that P(R/N&E), where N and E are, respectively, the events that naturalism and biological evolution are true, is low, and that, since naturalists are bound by the evidence to accept E, this constitutes a defeater for R. However he admits that his arguments for taking P(R/N&E) to be low are rather weak, and so we might instead take the value to be inscrutable. According to Plantinga, either conclusion gives the naturalist a defeater for R, and he supports this claim with the following argument: Let F denote a conjunction of "the relevant facts about their origin, purpose, and provenance," that is, the origin, purpose and provenance of our cognitive faculties (WCB, p224). If P(R/F) is low, he argues---or even just inscrutable---then believers in F should come to doubt that R is true.
To support the step from concluding P(R/F) is low or inscrutable to doubting R, Plantinga offers two kinds of analogy. Both of these, I think, fail. In the first place, he suggests that if we consider not ourselves but rather a hypothetical alien race, then learning that P(R/F) is low or inscrutable should give us to doubt R as applied to those creatures. Similarly, if we discover that P(R/F) is low or inscrutable when applied to ourselves, then we should likewise come to doubt our own cognitive faculties.
In a second type of analogy, Plantinga describes particular situations where he thinks we would be inclined to doubt R. For instance, if we believe that we have been created by a malevolent intelligence which aims to trick us into having mostly false beliefs, say a Cartesian demon or Alpha-Centaurian mad scientist, then we have a defeater for R. He elaborates:
But to have a defeater for R it isn't necessary that I believe that in fact I have been created by a Cartesian demon or been captured by those Alpha-Centaurian superscientists. It suffices for me to have such a defeater if I have considered those scenarios, and the probability that one of those scenarios is true, is inscrutable for me---if I can't make any estimate of it, do not have an opinion as to what that probability is. It suffices if I have considered those scenarios, and for all I know or believe one of them is true. In these cases too I have a reason for doubting, a reason for withholding my natural belief that my cognitive faculties are in fact reliable. ("Naturalism Defeated," 1994, p12)
At least in the first analogy, I think he's quite right that we should doubt R as applied to the hypothetical alien population. However, the fact that we are doubting the cognitive faculties of some other collection of creatures is key to the analogy. When we turn our attention to ourselves, we find that we believe R in a basic way. Moreover, we hold this belief before ever forming any other beliefs about the origin of our cognitive faculties. So any story we tell about the origin of those faculties will always be incomplete so long as we neglect to stipulate that it ends with R being true. In other words, if we run into a situation where P(R/F) is low or inscrutable (or indeed anything other than exactly 1), then we can be assured F does not contain all the relevant information about the origin of our cognitive faculties. In that case, we ought not be bothered that P(R/F) is low or inscrutable.
In the case of the second analogy, we can offer essentially the same objection as described above. Since to begin with we all believe R in a basic way, then I have no business saying that, "for all I know, it could be the case that a malevolent intelligence has ensured that R is false"---at least not without some significant qualification.
Nevertheless, this approach may seem too cheap to satisfy. As Plantinga points out, we ought to be able to defeat even basic beliefs. Personally, I don't think that's true in this case---I cannot imagine a situation where R could ever be defeated. We could lose confidence in ourselves, and be persuaded to minimize our trust in our own judgment, but we can never escape our own point of view. As much as we might try to outsource rational judgment, our decisions ultimately remain our own. So to suggest that we ought to stop trusting our cognitive faculties is about as realistic as suggesting that we should stop being nonomniscient. However suppose for the sake of argument that we could have some kind of rational obligation to stop trusting our own beliefs. Then the question is, have Plantinga's analogies demonstrated that in fact we do have such obligation?
The answer appears to me a firm no. Consider that by a defeater for R, Plantinga evidently means a rational obligation to stop believing R. We can interpret this unusual situation in one of two ways: First, we can take unbelief in R to rationally obligate us to stop trusting our cognitive faculties. However, to the extent that it is not in our power to do this, we cannot ever be rationally obligated any more than we are rationally obligated to become omniscient. Second, we can take unbelief in R leave us free to rationally trust our cognitive faculties. In that case, we face no dire consequences in rejecting belief in R---no radical skepticism endangers us.
However Plantinga faces a second problem: Suppose his argument against naturalism goes through, and this throws naturalists into an irreconcilable skeptical trap. As far as I can see, this is no epistemic advantage for the theist. We have two cases to consider why this is so: First, suppose that theism has warrant independent of the EAAN. In that case we have no need of additional warrant through the EAAN. And especially given that the EAAN is designed to undermine the confidence of naturalists instead of directly strengthening the confidence of theists, I don't see how it lends any additional warrant to theism. Yet even if the EAAN really is found to strengthen a pre-existing, independent warrant for theism, the fact remains that we will first need to stake out that independent warrant.
Consider instead the case where theism is not warranted independent of the EAAN. Then since the EAAN does not directly argue for the existence of God, how else might it serve as justification for theism? One obvious possibility is that it offers us a pragmatic justification to that effect. In other words, it's not that we have a reason to think that God really does exist; but rather---on this view---we find that if we fail to assume in advance that God exists, then we fall victim to a self-defeating skeptical trap. At least, I can see no other plausible way to argue for theism based on Plantinga's conclusions about naturalism. On this view, the apologist ends up not concluding that theism is probably true, but rather he finds---or so he thinks---that theism is an inescapable pragmatic assumption.
Whether we should actually believe our pragmatic assumptions is an open concern, but suppose for a moment that indeed we ought to do so. Even then, though, the apologist encounters a serious difficulty: It is simply not true that theism is the only alternative to naturalism which avoids the skeptical trap of doubting our cognitive faculties. In fact theism by itself is insufficient for this. Instead we require theism "plus a little something extra" (Naturalism Defeated, p46). In particular, Plantinga needs to presuppose a certain variety of theism whereby God creates us with reliable cognitive faculties. We might call this epistemically good theism (Plantinga calls it "theism simpliciter," in contrast to ordinary, or "austere," theism). So given this observation, it's easy to see how naturalists can make a similar move---instead of simply accepting a basic, stripped-down naturalism, we can assume (again motivated by pragmatic concerns) that if naturalism is true then it must be a kind of epistemically good naturalism. In other words, we can assume R itself, as originally suggested, and take a naturalist position only under the umbrella of that crucial assumption.
A variation of this approach seems to have been first suggested by Carl Ginet in 1993-4 (though I cannot find it published until 1995), whereby he advises the naturalist to assume R in the same way that the good theist assumes his good theism. In response, Plantinga objects by claiming that
the warrant austere theism has for the [good] theist is derivative from the warrant [good] theism has for her... (Naturalism Defeated, p47)
In other words, Plantinga thinks that good theism has warrant independent of the EAAN. However, remember that we are considering the case where theism has no such independent warrant. So Plantinga's response to Ginet cannot be adapted as a response to my objection here. Thus it seems his only recourse is to take the EAAN to drive us to a pragmatic assumption which entails R; yet in that case, why not simply assume R itself? After all, we appear to have independent motivation to do so, and even if we hadn't any such motivation, it seems less arbitrary to assume R than to assume something unnecessarily complicated (good theism) for the sole purpose of entailing R. At the very least, the EAAN itself offers us no reason to prefer good theism to good naturalism.
Indeed Plantinga seems only to have added to his burden. For when he depends on the fact that his warrant for ordinary, "austere" theism is derivative of his warrant for good theism, he cuts off those avenues to theism which are unspecific to good theism. So, for instance (and for the present purpose), he couldn't use the William Lane Craig's Kalam cosmological argument, nor James Anderson's and Greg Welty's presuppositionalist argument from laws of logic, since these do not move us specifically to an epistemically good God.
In sum, it seems that Plantinga needs to argue more persuasively that, first, we really can be rationally obligated to reject belief in R in such a way as to produce a skeptical trap, and second, such skepticism moves us towards theism and away from good naturalism. As things stand, I can find no reason at all to embrace radical skepticism, reject good naturalism, or accept theism.