Cultural group selection, co evolutionary processes and large-scale cooperation (by Joseph Henrich)

Abstract

Before addressing Joseph Henrich’s evolutionary explanation for altruism, it is important to ask whether the existing evidence taken in its entirety actually suggests that preferences are non-selfish in the standard game-theoretic (myopic) sense. Our own research corroborates Henrich’s claim that the amount offered in ultimatum and dictator games is greater than predicted by standard economic theory based on payoff dominance. We also find, however, that this fact alone is subject to over interpretation, and that its robustness should be tested in less restrictive interactions. Specifically, the interpretation of the results in these decision environments confound the following motives: (1) self-interest based on dominance; (2) positive reciprocity; (3) negative reciprocity; (4) various forms of other-regarding utility, altruism representing only one such form; and (5) the constant sum form of some games precludes effects based on gains from exchange. Even within this class of games, the altruistic utilitarian explanation for giving does not accord well with the fact that experimental instructions and procedures typically have a dramatic affect on outcomes (Hoffman et al., 1994, 1996). Thus, in the ultimatum game offers by the first mover vary from 44.4 percent of the available sum down to only 27.8 percent for subjects from the same American undergraduate culture. Moreover, similar large increases in ultimatum offers are observed by simply adding to the instructions two short sentences prompting the subjects to think about what they expect the other player to choose before making their choice (Hoffman et al., 2000). This “innocent” attempt by the experimenters to get the subjects to think about the strategic aspect of the interaction apparently had the effect of focusing the proposer’s attention on the possibility that his/her offer might be rejected by the responder. Hence, the critical importance that the experimenter not attempt any instructional or procedural variationswithout performing controlled comparisons. This maxim creates a nightmare of complication for important, but hazard-laden, programs designed to make cross-cultural comparisons.