As is well known to readers of this blog, I do not have much sympathy for philosophers like Alvin Plantinga. That’s not because the guy’s not smart (he certainly is), nor because he hasn’t published interesting philosophical arguments (he certainly has). But people like Plantinga still stride what should by now be an impossibly uncomfortable divide and ever widening gap between serious philosophy and theology.

Not to put too fine a point on it, Plantinga is engaging in what Ladyman and Ross referred to as “neo-scholasticism” (not a compliment), the sort of philosophy that was perfectly acceptable at the time of Thomas Aquinas, but that should have by now securely been consigned to the dustbin of intellectual dead ends.

Nevertheless, one issue that keeps coming up whenever I mention Plantinga on my Twitter feed is his famous Evolutionary Argument Against Naturalism (EAAN), and so I feel compelled to comment on it, though plenty of other people have already done so in the technical philosophical literature.

The EAAN (which can actually be traced back to C.S. Lewis) goes something like this:

1. Our beliefs about the world can only have evolutionary consequences if they affect our behaviors (otherwise they are invisible to natural selection);
2. Natural selection favors advantageous behaviors, not directly the ability to form true beliefs;
3. Natural selection has no way to favor true non-adaptive beliefs over false but adaptive beliefs.

Therefore, (C1) the probability that our minds consistently deliver true beliefs if both philosophical naturalism and naturalistic evolution are true is low or inscrutable. Also, (C2) it is more likely that God created us in a way that resembles him, specifically in a way that makes it possible for us to reliably hold true beliefs.

Plantinga accompanied the above with a lot of corollary discussions of Bayesian probability (his argument gets off the ground because of his assignment of priors to certain events, such as the outcome of natural selective pressures and the reliability of human beliefs about the world) as well as of different theories of mind-body interactions (since his argument depends on the causal properties of beliefs, and even desires). Regardless, the sketch I presented provides the bare bones of the EAAN.

Now, the first reaction of skeptics and atheists is likely to be: bullshit (or, in my case, nonsense on stilts). But we need to resist easy dismissals here. The EAAN is a serious argument advanced by a serious philosopher (despite my comments above about theology and neo-scholasticism), and it is intellectually honest to actually engage with it and refute it on its own grounds.

Generally speaking, there are two broad strategies for refuting a formal argument in logic: we can show either (a) that the argument is invalid (i.e., the conclusions do not logically follow from the premises) or (b) that it is not sound (i.e., one or more of the premises is false).

As far as I can tell, premises 1-3 are indeed true, so the problem — if any — is with the structure of the argument. Let’s start with C2: it simply does not follow even if premises 1-3 were true. For instance, there is no non-arbitrary reason to think that God created us in his image (what, just because it says so in a book written by unknown human beings thousands of years ago?), nor that “in his image” ought to include the ability to form reliable beliefs about the world. After all, there are a number of respects in which God did not bother to make us similar to himself (omnipotence, for instance; not to mention that he likely doesn’t have nipples), so why arbitrarily assume that reliability of beliefs is one such aspect? It certainly doesn’t follow from the premises of the EAAN. Indeed, nothing about gods follows from those premises, as we could have been created by a Big Simulator and exist only as virtual characters in his Big Video Game. So there is no reason for us to accept C2, which — make no mistake about it — is the whole reason why Plantinga is bothering with all this to begin with.

What about C1? That one also doesn’t follow from the argument as stated, unless we add an additional, hidden premise: that natural selection is the only way for us to evolve the ability to form (largely) reliable beliefs about the world. But biologists agree that natural selection is just one evolutionary mechanism, and that a number of things come into existence in the biological world as byproducts of evolution. No serious biologist, for instance, would argue that our ability to solve Fermat’s Last Theorem, or — more prosaically — to, say, read trashy novels, is the result of evolution by natural selection. We are capable of both (and many other) feats as a byproduct of having large brains capable of sophisticated thinking. Those large brains evolved for reasons of survival and reproduction (e.g., the ability to coordinate large game hunting, or to advantageously interact in socially large groups, etc.) that have nothing directly to do with Fermat’s theorem or trashy novels.

There are more subtle problems with Plantinga’s argument. For instance, C1 states that the probability of humans being capable of generating reliably true beliefs about the world is “low or inscrutable.” Let’s parse the two possibilities: low from inscrutable probability. The first one is arrived at by Plantinga through the assignment of arbitrary Bayesian priors, without regard to actual empirical evidence. But when based on subjective priors Bayesian inference only converges to the true priors after repeated evidence-based rounds, without which the initial priors simply reflect personal prejudice. In this case Plantinga completely ignores further empirical evidence and does not adjust his priors accordingly.

If, however, we agree that such priors are in fact “inscrutable” (and I am inclined to agree that they are) then precisely nothing can be derived from them in terms of the truth or falsity of naturalism. To see why we need to understand that Plantinga’s conclusion here is derived in a way similar to the infamous argument from low probability of the evolution of adaptive proteins advanced by creationists and intelligent design proponents such as William Dembski. Dembski confidently “estimates” probabilities of unique historical events to which we cannot actually attach probabilities  at all, for the simple reason that we have no idea from which statistical distribution they are drawn. Lacking such information, again, talk of “probabilities” is simply a thinly disguised way to attempt to impose one’s own subjective (and arbitrary) beliefs on the problem at hand. It is intellectually naive, if not downright dishonest.

There is another obvious problem with Plantinga’s argument: the definition of naturalism. To begin with, as Michael Ruse has pointed out in his own response to the EAAN [1], Plantinga fails to make the crucial distinction between methodological and philosophical naturalism. Science, and in particular the theory of evolution, is committed only to the former, not to the latter. More broadly, naturalism is actually surprisingly difficult to define, and what it logically entails is even more subject to debate (as became clear during the Moving Naturalism Forward workshop that Sean Carroll organized and in which I participated). Plantinga gets around this by defining naturalism as “the idea that there is no such person as God or anything like God.” But which god is he talking about? Would the above mentioned Big Simulator count as a god? Not in Plantinga’s book, by why not? Indeed, naturalism doesn’t even need to be limited to physicalism, as for instance in the case of naturalist philosophers who entertain some versions of mathematical Platonism or structural realism. This doesn’t make naturalism incoherent, but it certainly makes it much less of the clearly defined target that Plantinga’s attempts to deploy.

Finally, let’s look at the fundamental points underlying the whole EAAN argument: are human beliefs generally reliable or not? And what is the best explanation for such reliability or lack thereof?

Well, the first thing to do is to break down the category “human beliefs” (which Plantinga conveniently treats as homogeneous) into subgroups. Our beliefs about everyday experience, for instance, are pretty darn reliable, and the most likely explanation for that fact is indeed the theory of evolution by natural selection. If we held systematically wrong beliefs about things that are dangerous (predators, poisonous fruits) or good for us (animals we can use or hunt, edible stuff) we would have long ago gone the way of the Dodo (which went extinct, ironically, precisely because it held systematically mistaken “beliefs” about the friendliness of a certain species of bipeds newly present in its natural environment).

Second, the fact that we have systematic cognitive biases is actually pretty well explained by (or at least consistent with) evolutionary theory (again, understood as natural selection plus other mechanisms, including the likelihood of evolving behavioral byproducts). Moreover, it poses a significant threat to any theological account: why, exactly, would god build us in a way that makes us prone to confirmation bias, or to “discover” patterns where there are none? Is that also a consequence of being made in his image? Does god have an in-born tendency to believe in conspiracy theories and in making stuff up?

In the end, Plantinga’s argument is little more than a throwback to Descartes’ famous invocation of god as guarantor of the reliability of our knowledge. The difference is that Descartes can be (somewhat) excused because he was writing at a time when there really was no reasonable alternative to the invocation of divine intelligence to explain certain facts about human existence. Plantinga — writing almost a century and a half after Darwin — has no such excuse, and we are right in rejecting his EAAN as a clever but inevitably flawed example of neo-scholasticism.


[1] In: “The New Creationism: Its Philosophical Dimension,” a chapter of The Cultures of Creationism, Ashgate, 2004.

Originally on Rationally Speaking, July 15th, 2013