PROBABILISTIC CHOICE AS A CONSEQUENCE OF NONLINEAR (SUB) OPTIMIZATION

Several researchers have convincingly argued that much probabilistic c hoice could be the result of (attempted) maximization of some utility function that is nonlinear in the choice probabilities. Others researc hers have developed descriptive utility theories that are nonlinear in event probabilities. The present work combines these two viewpoints a nd motivates utility theories that can be nonlinear in either one or b oth of choice and event probabilities. The presented theories are base d on the idea that most choice situations involve multiple incompatibl e goals which have to somehow be combined in reaching a decision. In p articular, I assume that the decision maker is faced with a vector opt imization problem which is converted to a scalar optimization problem, the latter being solved to yield a noninferior solution to the origin al vector optimization problem. I show that in certain special cases t his approach yields the same utility representations as certain other recent nonlinear utility theories. (C) 1997 Academic Press.