Bayesian Contextual Choices Under Imperfect Perception of Attributes

Classic theories show that choices can be represented by a stable utility function when they satisfy consistency axioms, such as transitivity and the independence of irrelevant alternatives. Empirical choice data, however, display several contextual choice effects that violate these axioms. We study a choice model with a fixed underlying utility function and explain contextual choices with a novel type of information friction: the agent’s perception of the options is affected by attribute-specific uncertainty. Under this friction, the agent learns useful information when the agent sees more options. Therefore, the agent chooses contextually, exhibiting intransitivity, joint–separate evaluation reversal, the compromise effect, the phantom decoy effect, the attraction effect, and the similarity effect. Because the uncertainty is attribute-specific and common across alternatives, the classic axioms hold when the alternatives dominate one another in attributes.