Anomalies de la théorie des préférences : Une interprétation et une proposition de formalisation

Based on experimental evidence, it is often said that one should give up the concept of preference. In this paper, we show that the validity of this inference relies on an implicit assumption, the invariance principle (ip), according to which certain features of the objects of choice are irrelevant to the decision to be made. This assumption is usually confounded with a necessary assumption of any economic model, the faithfulness assumption, (f), which states that the language of the model is rich enough to describe the diversity of empirical observations. In order to clarify the situation, we claim that one must disentangle (f) and (ip). We show that this can be done thanks to the structure of decision problem with normative equivalence (dpne) that we introduce. We study the decision process associated to it and show that it involves the salience of attributes as the decisive element. Classification JEL : A12, B40, C0