I analyze a model in which agents choose whether to undertake a task with an individual-specific probability of success of which they only have a noisy perception. I show how, when agents do not have the tools to correct for noise as a Bayesian would, they can use statistics about the prevalence of their social group among the successful individuals in the task to bias their noisy perception in a direction contingent on their social type and limit the adverse effects of the noise on decision making.

We ask how enforcement can endogenously emerge in a landscape in which only raw power, iron fists, govern the interaction of agents. If two agents are ranked in terms of power, the more powerful one can expropriate, at a cost, the less powerful one. Alternatively, both agents can engage in surplus-augmenting cooperation (e.g. trade). If expropriation is not too costly and cooperation is not overwhelmingly productive, for any pair of ranked agents the possibility of expropriation prevents cooperation. The more powerful agent finds it profitable to expropriate the less powerful one.