IN DEFENSE OF A CONSTRUCTIVE, INFORMATION-BASED APPROACH TO DECISION THEORY

Since the middle of this century, the dominant prescriptive approach to decision theory has been a deductive viewpoint which is concerned with axioms of rational preference and their consequences. After summarizing important problems with the preference primitive, this paper argues for a constructive approach in which information is the foundation for decision-making. This approach poses comparability of uncertain acts as a question rather than an assumption. It is argued that, in general, neither preference nor subjective probability can be assumed given, and that these need to be generated by using the relevant information available to the decision-agent in a given situation. A specific constructive model is discussed and illustrated with a real example from this viewpoint.