I haven't read the entirety of this piece of stupidity in (Pseudo)Scientific-American, it was just too painful:
Now, I guess I shouldn't expect much from someone who uses the words "patternicity" and "agenticity", but this was linked over at RichardDawkins.net, and people there seemed to like it. So I think there's some merit in spelling out why armchair evolutionary psychology is woo, plain and simple. Here's a quote:
"[W]e make two types of errors: a type I error, or false positive, is believing a pattern is real when it is not; a type II error, or false negative, is not believing a pattern is real when it is. If you believe that the rustle in the grass is a dangerous predator when it is just the wind (a type I error), you are more likely to survive than if you believe that the rustle in the grass is just the wind when it is a dangerous predator (a type II error). Because the cost of making a type I error is less than the cost of making a type II error and because there is no time for careful deliberation between patternicities in the split-second world of predator-prey interactions, natural selection would have favored those animals most likely to assume that all patterns are real."
The reasoning is silly. Suppose I said: "If you believe the rustle in the grass is an indication of food when it is just the wind (a type I error), you are LESS likely to survive than if you believe the rustle in the grass is just the wind, when it's actually food (A type II error). Both times you get no food, but the in the first case you spent energy and effort that could've been used to look elsewhere. Because the cost of making a type II error is less than the cost of making a type I error, natural selection would have favored those animals most likely to assume that NO patterns are real."
Or, a parody more to the point: "Because I can think of one case where thinking X is better than thinking Y, and the thought that X has property F and the thought that Y has property not-F, it follows that evolution makes it the case that having thoughts with property F is always better and evolution made us prefer them in all cases."
Yes, people say this sort of crap. And get published in Scientific American. And fawned over at RichardDawkins.net. It's unsettling.
Let's set the record straight. The argument assumes (in particular, provides *no* evidence) that pattern-forming is in general better or less costly. We know this to be false, because Bayesianism is the true account of ideal rationality. If you don't set your present credences equal to your prior subjective probability distribution conditional on your evidence, and you don't act so as to maximize expected utility, you will be more likely to die. That's a theorem. Sometimes your evidence confirms a correlation, sometimes it doesn't. But picking "all patterns, all the time" as your update rule is not optimal, and therefore we've no reason to *a priori* expect it.
Now maybe we do form patterns more than it is rational to. It would be nice to have an explanation for that. Presumably it's because cobbling together agents that can survive for more than a few seconds is difficult, and evolution had to jury-rig some aspects of our psychology from non-optimal other bits. But pursuing any particular theory of what went on requires looking at the damn facts. We need to know what psychological traits are heritable (if any), we need to know which among them were available in the ancestral population to be selected for, which competed with which, what the selection pressures were, etc. And we don't freaking know any of that.
All right. End of rant. FYI I plan to update the blog on Fridays this summer. And I do genuinely plan on blogging, so check in, say, every Saturday