Neo-classical microeconomic theory in its conventional presentation has scarcely changed in nearly a hundred years, although economists have several times shifted their interpretation of its claims, its subject matter and their own intentions in elaborating the theory. In this paper I suggest that the conventional explanations of these shifts bely their real causes, and that the real causes render seriously questionable the viability of the sort of account of economic behavior which is ubiquitous in contemporary economic theory. I do so by offering a sketch of the history of the theory of consumer behavior, and more generally, rational choice theory. This sketch exhibits the shifts in interpretation of the causal variables of the theory, and the changes in the explanatory scope and presumptive subject matter of the theory as largely ad hoc qualifications and restrictions that have preserved the theory against a series of failures to empirically substantiate it. I claim the changes to be ad hoc in the sense that they purport to be "solutions of all and only those empirical problems which were ... refuting instances for earlier interpretations. ''1 Naturally economists themselves did not advertise these changes as ad hoc, but more often than not as changes reflecting improvements in scientific method, especially as reflected in the operationalist imperatives of twentieth century empiricism.
My purpose in offering this piece of interpretative history is not destructive, for I consider the theory under discussion to be the most impressive edifice in social science yet erected. My aim is rather to pose a question: In the light of the facts as I allege them, why do economists continue to commit themselves to this theory and its successors? Surely their aim is not to elaborate a well confirmed systematic nomological account of economic (or other sorts of intentional) action. What then is their aim? What is contemporary neo-classical economics all about?
The marginalist economists built a theory of economic behavior on three distinct general claims: That all agents have complete information about all