Abstract

1. In general we agree to recognize the existence, if not the methodological fertility or epistemological legitimacy, of a "rationalist model," at least when we refer to what economists do when they offer explanations.1 However two remarks must be made about this. First, it must be emphasized that this model is not unique, but generic: in fact, it is more a family of models of which the fundamental theoretical suppositions are susceptible to large variations. There are here, as it were, several possible axiomatic bases. One formulation is, for example and to use von Mises's term, praxeological (axed on means/end relation). Another emphasizes the preference/constraint relation (Harsanyi). Other contrasting models are those putting forward the maximization of expected utility (von Neumann-Morgenstern) and those making the satisficing hypothesis (Simon). Second, however, it must also be said that philosophers of science do not really agree on what is asserted by the "principle of rationality" or on what is its status. For example, Popper (1982) made it a minimal empirical hypothesis, or practically empty, and denied, furthermore, that it was a universally valid psychological law. Hempel (1965) did not hesitate, in contrast, to make it a law of experimental psychology which would play an implicit or explicit role in any adequate and complete explanation of behaviour or action in social sciences or history. Dray, to cite another classic author (cf. Dray 1957), made this principle the base for the explanations of historians, but denied it the status of a nomological proposition. In consequence, it is difficult to proceed as if there were a theoretical consensus among researchers when they base their explanatory arguments on the rationality of agents. This is clearly not the case. However, Raymond Boudon is right to consider the "rationalist model" an extension to all the social sciences of the model of explanation which is found at the base of what has been called the "marginalist revolution" in economics..