REVEALED LIKELIHOOD AND KNIGHTIAN UNCERTAINTY

Nonadditive expected utility models were developed for explaining pref erences in settings where probabilities cannot be assigned to events. In the absence of probabilities, difficulties arise in the interpretat ion of likelihoods of events. In this paper we introduce a notion of r evealed likelihood that is defined entirely in terms of preferences an d that does not require the existence of (subjective) probabilities. O ur proposal is that decision weights rather than capacities are more s uitable measures of revealed likelihood in rank-dependent expected uti lity models and prospect theory. Applications of our proposal to the u pdating of beliefs and to the description of attitudes towards ambigui ty are presented.