Three Analyses of Sour Grapes∗

The phenomenon of adaptive preferences – sometimes also known under the name of sour grapes – has long caused a stir in Social Theory. Among logicians, notably those in the dynamic logic or belief revision traditions, the question of preference change has recently seen a surge of interest. However, although the former question seems an instance of the latter, the theories of preference change proposed to date do not seem to give a firm handle on adaptive preferences, and certainly not the sort of deeper understanding which one might like. In this paper, the precise problem posed by adaptive preferences, as seen from the point of view of a theoretician who intends to model or understand the phenomenon, will be clarified, and three models of the phenomenon will be presented and compared. The general intention of the article is to sound out some of the wider consequences of the phenomenon for the project of modelling and understanding the relationship between decisions taken in different situations. Difficulties which arise when several decisions and several situations are involved shall be discussed, and an approach to these difficulties shall be suggested. This approach places particular demands on would-be models of the sour grapes phenomenon; these demands will shed light on the adequacy of the models proposed.