Tuesday, December 23, 2008

Razib is looking at a recent science headline that brings out the worst of evolutionary psychology:
One of the most tiresome aspects of evolutionary psychology is the paradigmatic straitjacket which many of the practitioners operate under; the only type of evolution that exists is unidirectional. Deviations from expectation are explained away. The importance of human universals mean that variation can not exist. These sorts of evolutionary psychologists resemble the caricature of the economist who holds to rational choice so that behavior which deviates from the model is explained by ad hoc contingencies.
He goes on:

In Demonic Males Richard Wrangham reports that orangutan males come in two morphs. A very large one, and a small one which seems to be a case of paedomorphism. Female orangutans prefer the normal large male orangutans as sexual partners. But the small ones do reproduce. How so? They ambush and chase the females and rape them. This is a behavioral strategy which can work well if the small variant morph is not extant as too high a frequency, because females will then not be "on alert." How these sorts of variations emerge is clear to anyone who is cursorily familiar with evolutionary game theory.

I use the orangutan "raper" strategy as an example for a reason: some of the press will no doubt spin the new research as a victory for feminism and a rebuke to heteronormative males or something of that sort, at least implicitly. But the intersection of biology and behavior is fundamentally value neutral; humans are the ones interjecting norms. The old paradigm of evolutionary psychology, which I must admit is justifiably "Just So" and rooted in an coarse and outdated understanding of biological science, should be rejected on scientific grounds. Not only are there many areas where it offers little insight, but it is no longer compelling on theoretical grounds.

There are certainly a core of human behavioral traits which exhibit little variance, and are extremely constrained by purifying selection which fixes the trait so that only one morph is extant at appreciable frequencies. But there are many which no doubt exhibit continuities, while a fair number likely can be modeled as discrete strategies across the adaptive landscape, buffeted by stochastic & frequency dependent dynamics as well as exogenous parameters. As simple as sufficient, but as complex as necessary, should be the maxim.

No comments: