Theism and the Genetic Fallacy, Part II

(Redated post originally published on 5 March 2009)

A few weeks ago I engaged in an exchange with Victor Reppert on theism and the genetic fallacy. I had meant to get back to him right away, but administrative b.s. of the sort always imposed on university faculties slowed me down. Anyway, our conversation made me think of ways to employ some of the recent biological belief theories (BBT’s) of Boyer, Atran, Dennett and others in constructing a more rigorous atheological argument:

My argument is simple. I think that Alvin Plantinga is right. If God exists, humans will very likely possess a sensus divinitatis, a God-detecting faculty, which, when functioning properly and in the appropriate circumstances, will present us with warrant-basic (both warranted and epistemologically basic) awareness of his existence. If this is so, and if God does exist, then humans, provided that their sinfulness has not impaired the proper functioning of their sensus, will have a warrant-basic awareness of God’s existence. On the other hand, if there is no God, it is extremely unlikely that humans would possess a cognitive faculty that would produce the warranted (but false) belief that God exists. In this case, evidence that belief in God is not caused by a warrant-conferring cognitive faculty, but rather is generated by a noncognitive process that does not confer warrant on that belief, will, ipso facto, constitute evidence against the existence of God. An atheological argument can therefore be set out semi-formally like this:

1) If God exists, then humans very likely possess a sensus divinitatis, a cognitive faculty which, when functioning properly and in the appropriate circumstances, produces the warrant-basic belief that God exists.

2) If there is no sensus divinitatis, then God probably does not exist, unless the background probability of his existence is very high.

3) It is not the case that the background probability of God’s existence is very high.

4) There is no sensus divinitatis.

5) Therefore, God probably does not exist.

Plantinga and others argue cogently for (1). The God of orthodox theism is not an aloof Lawgiver or Watchmaker. A God that loves his children will not allow them to stumble about in the dark, vainly searching for light. Surely nothing could be more important for God’s sentient creatures than to have sure knowledge of his existence, his nature, and his will for them. Therefore, he will reveal himself to them in some clear, unmistakable, and authoritative way. That is, God will reveal his existence, nature, and will in so clear and forceful a manner that no honest and rational person will mistake it. God very probably will not make himself knowable only through a process of complex inference (modal ontological arguments, say) since very many persons lack the education or the intellect to follow such complex and recondite inferences. Therefore, God will make himself knowable either through a very simple and direct inference, or immediately and noninferentially as a basic belief. The surest way would be to implant a universal sensus divinitatis that imparts a warrant-basic knowledge of God to all humans, or at least to those in whom sin has not fatally corrupted their God-detecting faculty.

(2) follows from the probabilities we have assigned. If G = God exists and S = humans possess a sensus divinitatis, and we say, with Plantinga, that p(S/G) is very high, say, .99, then p(~S/G) = .01. Also, p(~S/~G), the probability that there is no sensus divinitatis if God does not exist, effectively equals one. By Bayes’ Theorem:

                p(~S/G) X p(G)

p(G/~S) = --------------------------------------------

          p(~S/G) X p(G) + p(~S/~G) X p (~G)

Since we assume that p(~S/G) is .01, and that p(~S/~G) is effectively 1, then, the only way for p(G/~S) not to be quite low is for p(G), the background probability that God exists, to be very high. For instance, if p(G) = .8, then

             .01 X .8

p(G/~S) = ----------------------- = .038

          (.01 X .8) + (1 X .2)

Even if we say that p(~S/G) is only .1, rather than .01, and we keep p(G) =.8, that would still give us only p(G/~S) = .29. So, if we say, with Plantinga, that p(~S/G) is quite low, then the only way to keep p(G/~S) above .5, is to put p(G) very high.

With respect to premise (3): That it has not been established that p(G) is very high, I take for granted. The numerous arguments of natural theology have been extensively, and, in my view, cogently debunked by critics like Mackie, Matson, Martin, Everitt, Oppy, Gale, Le Poidevin, Sinnott-Armstrong, Fales, Drange, Edis, Carrier, Parsons (ahem), and many, many others. For the record, I think that, among the many arguments of natural theology, Victor Reppert’s argument from reason is about the best of the lot, though I have argued copiously elsewhere why I find it wholly unconvincing.

Naturalistic explanations of religious belief, such as the various Biological Belief Theories (BBT’s) of Wilson, Boyer, Atran, Dennett, Guthrie, et al., come in to support premise (4). Theistic belief is warranted, in Plantinga’s sense, only if it is the product of the proper functioning of a cognitive faculty designed to produce true belief. Now one may wish, following Richard Dawkins, to speak of highly adapted phenotypic features as “designed” by evolution, but even so, no one would say that the belief-producing mechanisms postulated by the various BBT’s were designed to generate theistic belief if and only if that belief is true. They are not God-detecting faculties. On the contrary, these postulated mechanisms would work just as well in generating theistic belief whether or not God exists. Clearly, if theistic belief is generated in a manner like those proposed by BBT’s, then theistic belief is not warranted in Plantinga’s sense. Belief is the result of an epistemologically unreliable belief-forming process, not the consequence of the proper functioning of a cognitive faculty in the appropriate circumstances, and so belief cannot be warrant-basic. If humans possess a sensus divinitatis, then theistic belief cannot be explained in terms of an unreliable, noncognitive, non-warrant-conferring process, such as a BBT. Therefore if any BBT is sound, there is no sensus divinitatis, and the fourth premise of the above atheological argument is supported.

The upshot is that appeal to BBT’s or other naturalistic explanations of theistic belief need not involve the genetic fallacy. On the contrary, when added to other highly plausible claims, they provide support for the atheological argument I have given.