Decision making in phantom spaces (Q2256862)

This paper presents a general model of decisions under uncertainty with imprecise probabilities and imprecise utilities. Probabilities and utilities are given by so-called phantom numbers, which can be interpreted as directed intervals. For phantom numbers a ring-structure is defined; the defined additive and multiplicative operations allow to compute expected utilities as phantom numbers. These expected utilities are evaluated by a real-valued function. The authors give six axioms on the preferences of the decision maker, which imply the existence of a `phantom number representation' of the decision process. Different kinds of risk attitudes are defined and characterized; the degree of risk aversion and the degree of uncertainty can be measured. It is shown, that the Ellsberg paradox can be resolved within this approach.