Les nouveaux modèles de décision dans le risque et l’incertain : quel apport ?[The new models of decision under risk or uncertainty: What approach?]

The decision theory under risk or uncertainty has object to describe the behaviour of agents facing several uncertainty perspectives, waited that every agent is characterized by his own preferences. As it is difficult to describe these preferences exhaustively, we try to represent them: thus, while associating a numeric value to each uncertain perspective, we can order an agent's preferences as merely that one orders some real numbers. The recourse to a representative function of preferences (called as function value) constitutes a long time since the usual method of behaviour description in uncertainty. The interest obvious of this method is to permit to integrate these data directly in a formalized model and, by extension, to understand the underlying optimization process to all decision. The determination of the representative function of preferences must rest on an axiomatic foundation. One hears by there that a certain number of rules or general behaviours (called axioms) are reputed common to all human beings. Of these axioms, one will drift a precise specification of the function value. The objective of this paper is to examine the historic of theories having looked for to determine satisfactory criteria to answer to the problem of decision under risk or uncertainty and to analyze the approach of these models.