Savage's axioms show the rationality of maximizing expected utility when all uncertainties are explicitly modelled. But individuals actually make decisions in bounded contexts called small worlds. Savage's axioms do not imply the optimality of maximizing expected utility in small worlds unless...