On the evolution of preferences

A common feature of the literature on the evolution of preferences is that evolution favors nonmaterialistic preferences only if preference types are observable at least to some degree. We argue that this result is due to the assumption that in each state of the evolutionary dynamics some Bayesian Nash equilibrium is played. We show that under unobservability of preference types, conditional on selecting some self-confirming equilibrium as a rule for mapping preference into behavior, non-selfish preferences may be evolutionarily successful.