But I want to make two points.

I. The relation between evolution and biology, and the relation between rational choice theory and social science, is very different.

Evolution describes an end-result (adaptation), the way things got there (natural selection, reproduction, and mutation), and the situational tendencies which made the end-result probable among others (climate, food supply, etc). Present and past biological facts may be impressively explained by this theory, once the admitted relevant peripheral facts are plugged in. In a sense, evolutionary theory is like a skeleton, a truly grand design.

Meanwhile, RCT describes a model of behavior (rational choice), the kinds of consequences we can predict from rational behavior (i.e., the tragedy of the commons), and contextually sparse model situations (for instance, with RCT’s close allies, the game-theoretic ‘Prisoner’s Dilemma’, the ‘Monty Hall Problem’, etc). As models, they are fine enough; but whatever they explain, it isn’t any impressive bulk of the social sciences. It isn’t even a skeleton — it is, perhaps, a hip-bone, or a torso. Whether or not it succeeds as a “theory”, depends on how it defines the domain of phenomena it would like to explain. If it is social behavior with respect to goal-achievement, then it fails. If it is goal-achievement for a great part of the time, then it succeeds.

II. To make a few more philosophical remarks. (Two quick, one longer.)
a) What’s so rational about (let’s face it) dumb, selfish, short-sighted goal seeking?
b) Is the estimation of probabilities and possibilities — of possible worlds — really so rational? These “a prioristic” arguments smack of Pascal’s wager, along with all of its pitfalls.
c) As admitted, rational choice theory seems to fail to describe successful group behavior. And if true, this would be devastating in its own right: for if it were to fail to describe group behavior, then strictly speaking, it wouldn’t even enter the social sciences in the first place, except in a threadbare way as the study of other-influenced psychology, or the study of aggregate behavior.

But first we should double-check to see that it really is quite so embedded into the view of homo economicus, and quite so distant from the homo sociologicus. Take the following two major situations, which are in the neighborhood of RCT’s stomping grounds, require explanation (borrowed from Jared Diamond’s opus, ‘Collapse’):

i) Using the language of RCT, how do you explain the bottom-up collective behavior of small, enduring, and isolated societies which succeeded to thwart long-term disasters, seemingly with the gift of foresight?
ii) Using the language of RCT, how do you explain the top-down collective behavior of large societies which succeeded to thwart similar disasters?

The answer to i) lies in your analogy: through both rational choice and evolution (of a kind). On the one hand, the lives of individual members of a small, isolated society seem to be intertwined deeply to such an extent that the interests of a single member are identical with the interests of the rest. With enough nags around, and no place to escape to, the pressure is simply there to stay. On the other hand, their techniques were perfected through the trials and errors of passed generations, to such an extent that their technology is merely habitual, not resting on any genuine knowledge-base. Long-term goals are satisfied by induction, which falls to habit.

But the answer to ii), of course, is “we don’t”. Not unless we’re prepared to engage in a study of power relationships — but this is something that, in most cases, RCT isn’t cut out to do.