An approach to and computational results on testing the form of a decision maker's utility function

In this paper we consider the problem of identifying the form of a decision maker's (DM) utility function. We consider a set of alternatives each defined by a number of criteria and we assume that the DM's preferences are consistent with some underlying utility function. We present some theory and develop approaches to test whether the DM's preferences are consistent with a linear, a quasi-concave, a quasi-convex or a general monotonic function. We present our computational experience showing that in almost all cases the approaches correctly identify the form of the underlying utility function and require very little preference information of the DM.