Optimal Decision Making

A decision maker is said to behave optimally with respect to a given set of preferences or objectives whenever he or she chooses an alternative that is weakly preferred to (i.e., at least as good as) the other alternatives on the menu of available options. When the relevant objective is easily understood from context—as in the case of a manufacturer who wishes to minimize production costs—the decision maker is often said simply to behave optimally.

Scholarly work on optimal decision making can be usefully divided into two broad and overlapping areas of study. The engineering discipline known variously as decision analysis, management science, or operations research focuses on well-specified problems involving a clear objective and usually adopts an explicitly prescriptive outlook. Within economics, on the other hand, the principle of consumer sovereignty, which states that the social scientist should not impose his or her own tastes or values on the individuals under investigation, has led to a tradition of axiomatic work that links assumptions about preferences with conclusions about the associated optimal behavior, but that, for the most part, remains in the descriptive mode. (Also in this mode there is an enormous amount of literature in psychology documenting and seeking to explain the observed suboptimality of actual human choice behavior in precisely the types of scenarios in which the techniques of decision analysis might be profitably applied.)

As an illustration, consider the problem of choosing whether or not to proceed with a particular business or public policy proposal, such as building a new factory in Malaysia or setting aside land for a state park. Here, management science would explain how to construct the appropriate “decision tree” diagram showing the chains of events that might lead to different possible consequences of the two available options; it would suggest how to assess the costs, benefits, and probabilities relevant to the choice; it would show how to combine all of this information to form estimates of the expected values of the alternatives; and it would recommend proceeding only if the [Page 637]estimated value of doing so exceeds that of the status quo. Such an analysis would draw upon results from axiomatic choice theory that characterizes normatively justifiable behavior over time and in the presence of uncertainty. And in a “rational choice” model in which the fate of the proposal in question were to be decided by a particular corporate executive or public servant, this agent might be presumed to act according to the decision-analytic prescriptions previously described.

In strategic settings, the appeal of an alternative to a given agent will depend upon the actions taken by one or more other agents. Optimal decision making in situations of this sort is the province of game theory.