CHOICE UNDER PARTIAL UNCERTAINTY

This paper analyses problems of choice under uncertainty where a decis ion-maker does not use subjective probabilities. He has a set of belie fs about which states are more likely than others, but his beliefs can not be represented as subjective probabilities. Three main kinds of de cision rules are possible in this framework. These are maximin-type, m aximax-type and choosing that action which gives the highest payoff in the state, which the decision maker believes to be most likely. We re place the commonly used ''merger of states'' axiom with a version of t he sure-thing principle.