Theory development processes in the social sciences: The case of stochastic choice theory

Abstract

Almost three hundred years ago, Pascal and de Fermat produced a quantitative model for choice under uncertainty, based on expected payoffs. This event triggered a theory development process fueled by contributions like the logarithmic utility model of Bernoulli, the expected utility theorem (EUT) of von Neumann and Morgenstern, the non-expected utility (NEU) theories of Allais, Bell, and Machina, as well as a long series of empirical studies. This article uses the paradigm-oriented framework of Kuhn and Lakatos to describe and evaluate the history of ideas in stochastic choice theory. Particular attention is paid to the rivalry between the EUT and various NEU models, which has been called one of this century's most important disputes in the social sciences. Specific issues include the quick acceptance of the EUT, the large export of the EUT to several decision-oriented disciplines, and the periods of progress, stagnation and revitalization of NEU research. The article also contrasts the EU and NEU schools by analyzing the remarkable consensus on subject and methodology, the ambiguous use of the rationality concept, the difference in basic implicit assumptions, and the heterogeneous response to given empirical findings.