Nonlinear Utility Models Arising from Unmodelled Small World Intercorrelations
证明，当不同小世界中的彩票存在概率依赖时，Savage公理不再导出期望效用最大化，而是导出非线性效用模型，包括Chew加权线性效用、Bell兴奋/失望模型和Allais均值方差模型。
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 lotteries in different small worlds are probabilistically independent. Relaxing this independence assumption causes Savage's axioms to imply the optimality of maximizing a nonlinear utility model which includes, as special cases, the Chew weighted linear utility model, the Bell elation/disappointment model and the Allais mean/variance model in utility-independent small worlds.