| In a situation of decision under uncertainty, a decision maker wishes to choose according to the maxmin expected utility rule, and he can observe the preferences of a set of experts who all share his utility function and all use the maxmin EU rule. This paper considers rules for aggregating the experts' sets of priors into a set that the decision maker can use. It is shown that, in a multi profile setting, among the rules that allow the decision maker's evaluation of an act to depend only on the experts' evaluations of that act, the only rule satisfying the standard unanimity or Pareto condition on preferences is the "set of weights" aggregation rule, according to which the decision maker's set of priors is the set of weighted averages of the priors in the experts' sets, with the weights taken from a set of probability vectors over the experts. An analogous characterisation is obtained for variational preferences. |
Récupérer au format
