【主題】A Parsimonious Theory of Subjective Probability
【時間】11月17日 星期二 12:00—13:30
【摘要】 In contrast with prior treatments, we deduce the existence of a subjective probability on a finite collection of states without requiring an enriched outcome space nor the existence of utility. Our finite-state axiomatization of subjective probability relies on Chew and Sagi's (2006) definition of event exchangeability and delivers subjective expected utility along with cumulative utility which contains Choquet expected utility as a special case. Our approach enables the modeling of source preference over finite domains coresponding to instances where individuals do not exhibit probabilistic sophistication globally (e.g., Ellsbergian behavior and familiarity bias) yet act as if they possess distinct subjective probabilities over different finite sub-collections of events.