Alvin Plantinga, answered from the texts and cited to the page.
Naturalism — the view that there are no supernatural beings, that nature is all there is — combined with contemporary evolutionary theory yields a self-defeating position.1 The argument: under naturalism plus evolution (N&E), the probability that our cognitive faculties are reliable (R) — that they produce mostly true beliefs — is either low or inscrutable.2
Either way, anyone who accepts N&E acquires a defeater for R. That defeater cannot itself be defeated, since any prospective defeater-defeater would involve beliefs produced by the very faculties under doubt.34 The defeater for R gives the believer a defeater for any belief her cognitive faculties produce — including N&E itself. So N&E is self-defeating; one cannot rationally accept it.
The conjunction of theism and evolution faces no such problem: a God who designed cognitive faculties for the acquisition of truth gives us reason to regard R as high.
Of course you also see that the very reflections that lead you to this position are also no more acceptable than their denials; you have a universal defeater for whatever it is you find yourself believing. This is a really crushing skepticism, and it is this skepticism to which the naturalist is committed.Where the Conflict Really Lies, pp. 264–265
It is instead an argument for the conclusion that (for one who is aware of the present argument) accepting naturalism is irrational. It is like the selfreferential argument against classical foundationalism: classical foundationalism is either false or such that I would be unjustified in accepting it; so (given that I am aware of this fact) I can't justifiably accept it.Warrant and Proper Function, pp. 235–236
For that belief, and for each of the premises, he has a reason for doubting it, a reason for being agnostic with respect to it. This reason, obviously, cannot be defeated by an ultimately undefeated defeater. For every defeater of this reason he might have, he knows that he has a defeater-defeater: the very undercutting defeater that attached itself to R and to N&E in the first place.The Analytic Theist, pp. 142–143
For that belief, and for each of the premises, he has a reason for doubting it, a reason for being agnostic with respect to it. This reason, obviously, cannot be defeated by an ultimately undefeated defeater. For every defeater of this reason he might have, he knows that he has a defeaterdefeater: the very undercutting defeater that attached itself to R and to N&E in the first place.Warrant and Proper Function, p. 234
EAAN is the most aggressive of my arguments and the one most actively contested. The Naturalism Defeated? volume (Beilby 2002) organized eleven challenges across four fronts: that evolution does produce reliable cognition (Ramsey, Fodor, Fales); that the skepticism move overreaches (Bergmann, Sosa, Van Cleve); that the conditional-probability apparatus is misused (O'Connor, Otte); and that inscrutability does not yield a defeater (Talbott, Merricks, Alston). 'Reply to Beilby's Cohorts' addresses each. I have acknowledged a technical error in the WPF formulation — the preliminary argument conflated the unconditional objective probability of R with its probability conditional on background knowledge — and the WCB reformulation repairs it. The main argument survives the repair. The defeater step remains the most actively contested; Sosa's two engagements (Naturalism Defeated? and the Cambridge volume) press it from different angles. Where the Conflict Really Lies (2011) embeds EAAN in the broader claim that the real conflict in science-and-religion is between science and naturalism, not between science and theism.