Résumé (EN)

Expected utility theory can be viewed in two different ways. First, as a formal mathematical theory providing representation and uniqueness theorems for preference structures. Secondly, as a decision-aid model, seemingly resting on the formal theory and which emerged in the late 50’s : what I will call decision analysis. It is not easy from a historical point of view to understand how a purely formal theory designed to handle mixed strategies in Game Theory gave rise to a decision-aid model. Nevertheless the best-known argument in favour of decision analysis is that it is logically sound and axiomatically based. The aim of this paper is to investigate the links between these two levels, in order to clarify the way the decision-aid model works in practice. After a brief description of the main features of the formal model, I will try to see to what extent the decision-aid model is based on it and how it actually works. In the last section, I will review empirical studies dealing with the decision-aid model.