Stories and Assessments of Plausibility

Perhaps the best known model of people's beliefs under uncertainty is that of subjective or personalistic probabilities. A decision theoretic justification for this model is provided by the appeal of Savage's (1954) axioms. In his theory, the objects of choice are acts, functions from a set of possible states to a set of consequences. The agent is assumed to possess preferences-formally a binary relation--over the set of acts. Savage shows that the agent's preference relation satisfies certain normative axioms if and only if she behaves as an expected utility maximizer. "Traditional" economics does not claim that someone who satisfies Savage's axioms necessarily possesses a subjective prior and uses the maximization of expected utility as a choice procedure. Rather, the claim is that she behaves as if she follows such a procedure. The problem with this position, however, is that it is difficult to find plausible alternative procedures that lead to behavior consistent with the axioms. (See Rubinstein 1998.) In light of this, it is important to examine the realism of the expected-utility hypothesis--not as a representation of preferences--but as a descriptive procedure of decision-making. Even if one were to accept subjective probabilities as a useful description of how agents view the world, questions remain. Where do subjective probabilities come from? How does the decision maker "construct" them? How is the support to subjective probabilities determined?