I agree with Wright that there’s a time and a place for “tossing out a hypothesis on the basis of little evidence and then letting people argue about it.” What is vital, though, is that in subsequent discussion the conjectural nature of the hypothesis is not forgotten. This is the trap that sociobiologists of yore all too often fell into: They offered a purely speculative hypothesis about the evolutionary history of some piece of human psychology (which is okay) and then proceeded to build all subsequent discussion on the assumption that the hypothesis is true (which is not okay). Thoughts of how the hypothesis might be tested were often far from their minds, leading Stephen Jay Gould to admonish them for creating a pseudo-science in which “virtuosity in invention replaces testability as the criterion for acceptance” (1978, “Sociobiology: The art of storytelling” New Scientist 80, p. 530).
In the present instance, Wright argues that there is solid evidence for the hypothesis that human psychology contains a mechanism for distinguishing non-zero-sum from zero-sum games and responding accordingly. He cites “a large empirical literature on reciprocal altruism” and also some primatological work. I am familiar with and impressed with much of this research, but I should like to urge caution in deciding just which hypothesis the data supports. In particular, I’d like to draw attention to the problem of identifying at which level of generality it is appropriate to describe the putative adaptive mechanism.
An imaginary example. Suppose we observe that in the course of its natural development a type of monkey will reliably manifest a fear of leopards. Noting that this fear seems to come on-line in advance of any learning from its kin or peers, we might think it reasonable to conclude that natural selection has endowed the monkey with an adaptive mechanism for dealing with this kind of threat. But what, precisely, is the “kind of threat” that the monkey needs to deal with? Does it have a mechanism devoted to fear of leopards, or is the mechanism devoted to fear of big cats, or perhaps devoted just to fear of predators, or maybe just large animate objects? It would be difficult to say without having more detailed data. If, for example, the monkey’s fear response is triggered not just by leopards but also by lions or tigers, but not by canines or bears, then it might be reasonable to call it a mechanism for dealing with big cats. If the response is triggered by leopards but not by lions or tigers, then it might be more reasonable to call it a mechanism designed to cope with leopards in particular. (That’s too quick, but you take my point.)
This observation problematizes Wright’s citation of evidence in support of his favored hypothesis. It may be granted (if only for the sake of argument) that human psychology contains a suite of mechanisms for dealing with certain kinds of interpersonal reciprocal relations. Perhaps we are designed for engaging in the trade of concrete goods, for example. Trade is a ubiquitous and truly ancient human practice, stretching back at least to the Upper Paleolithic, making it reasonable to suspect that the human brain comes with some design features dedicated to governing trade relations. A sense of distributive fairness in exchanges, emotions of anger at unfair exchanges, a sense of ownership (“This is mine and that is yours”) all might be expected to emerge in the course of the evolution of the human mind in order to enable and enhance trade. The few grand social experiments that have attempted to expunge the notion of ownership from the human psyche—such as in the Soviet Union or the kibbutzim of Israel—have encountered an extremely stubborn opponent, suggesting that these utopia-builders were up against a human trait entrenched by natural selection.
But then one faces the problem of how, precisely, to describe the mechanism. Assuming that there is an adaptive mechanism in play, is it devoted to governing trade of concrete goods or just trade (which may include exchange of favors, of information, of access to sexual partners, etc.)? Or perhaps we should describe the mechanism as devoted to reciprocity, or perhaps to non-zero-sum games. Leaping to the very last of these descriptions would be hasty—though it may be the correct conclusion to come to after careful consideration of the data. In short, the fact that humans may have an inbuilt mechanism for dealing with one kind of non-zero-sum game doesn’t mean that we have a mechanism designed to deal with non-zero-sum games in some general sense. Therefore I remain unconvinced that the empirical evidence at which Wright gestures in his response should be interpreted as supporting the hypothesis that the human mind contains a mechanism for distinguishing non-zero-sum from zero-sum games and responding accordingly.