Plantinga's unnatural naturalism

So, firstly, I'm back off holiday. Secondly, the rest of my life is starting to settle down. I currently work in a different country from the one I live in, which is causing some problems, but that still leaves me with the occasional snippet of time for blogging.

Thirdly, my sister just started a new module of her Philosophy degree: religious philosophy. Needless to say, this has resulted in many fun discussions. So far, though, it's all been window-dressing - I know the arguments inside out. Most of it stopped being interesting a while back, which is why these days I'm more focused on religious psychology.

One guy that did catch my attention, though, was Alvin Plantinga. This guy gets points for coming far closer than average to a reasonable summary of the skeptical atheist position. However, he still commits some howlers at time, which IMO betray a comparative ignorance of science and, in particular, evolutionary biology.

As an example, I draw your attention to his "paper" (actually more a lecture transcript) An Evolutionary Argument Against Naturalism. Eschewing all the philosophical language, the central point is that Godless rationality is self-defeating, since brains that evolved by natural selection don't give a damn about truth as long as they carry on surviving and breeding.

Anyone who's been in the skepticism trenches for a while will recognise this as a variant on the transcendental argument, aka the Argument from It's My Ball And You Can't Play With It. This is one of the more annoying arguments for God because any counter-argument you make can itself be interpreted by the theist as more self-refuting rationalism. Plantinga, however, lays the argument out clearly enough that the fault lines are visible, which is why I like him.

In his article, he summarises the evolutionary position as "beliefs are adaptive". He then uses a neat example to show why this could lead to "pathological beliefs" (beliefs that are adaptive but false) as easily as true ones. Imagine a critter that enjoyed petting vicious tigers, but thought that the best way to pet a tiger was to run very fast in the opposite direction. Then its beliefs would lead to the most survival-enhancing result (legging it) so would be selected for, despite being completely unreflective of reality.

Extending this logic further, the claim is made that a brain produced by unassisted evolution will not be particularly adept at picking true beliefs; rather, it will pick beliefs that cause survival. Hence, if we were produced by evolution, our cognitive systems would be so unreliable that we couldn't justifiably say we were produced by evolution. Catch-22.

There are two objections to this argument, one obvious and the other subtle. The obvious one is the classic "stopped clock" issue: although the critter's behaviour turned out for the best this one time, that doesn't mean it'll be effective in general. Plantinga's critter is going to spend far too much time running away from cute little bunny rabbits, which is a waste of time and resources. So its beliefs are still soundly beaten by the more reality-based position that tigers are scary and scary things should be run away from.

The subtle objection is to Plantinga's characterisation of the evolutionary position. What he is describing is not belief formation in humans. It is closer to belief formation (or the creation of equivalent neurochemical constructs) in nematode worms. Nematodes have only a few "beliefs", so it is possible for evolution to act on each of the worm's underlying rules-of-thumb in turn.

Humans operate by a different method. We are selected on the basis of our belief creation methodology - the generator of our beliefs - rather than the individual beliefs themselves. From evolution's perspective, this is massively more efficient because, rather than selecting for billions of different rules, you can just select for one generator and let it get on with it. The resulting creature will be able to adjust its beliefs on the fly when it meets new evidence, and will hence be more effective.

It's not immediately obvious whether there exist "pathological generators" that could reliably produce pathological beliefs, but I'd strongly suspect not. However, I'm open to informed argument.