« Weekend Reading | Main | Journal Rankings | DEVITT'S LGSCD-INDEX & DEVITT'S GSCD-INDEX »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Will Thomas

Christopher, thanks for this post. This is a deep question that I've had to deal with in my work on the history of operations research, decision theory, and their relations to economics and other areas. My book on this, Rational Action, will be coming out with the MIT Press this winter.

I think the key in this case is to understand that rational choice theory is a heuristic enterprise. If ambiguity-averse preferences are irrational, then there is nothing more to be said about them: they simply constitute an arbitrary preference between alternatives. However, if the preference is non-arbitrary, then there is a hidden rationale that must be explored using theory.

I've never looked at ambiguity-averse preferences explicitly, but I expect that the preference would be founded on a treatment of uncertainties suspected to exist in the more ambiguous option. We may surmise that the Nikkei stock index is as likely to go up as down as a Bayesian prior, but our lack of knowledge suggests a risk in the assumption. For instance, we might suspect that stock indexes are actually slightly more likely to go up than down on a given day. That treatment of risk constitutes a rational calculation, which can only be assessed by presuming its rationality.

The crucial question in establishing whether a theory of rational choice is positive or normative is to establish how well it captures the (explicit or implicit) rationality of extant decision makers. If it fails to do so, then it is a heuristic attempt to describe that rationality. If it does so and exceeds it, then it is prescriptive. The crucial event in this history is the development of sequential analysis during World War II (see Judy Klein's "Economics for a Client"), where unsuspected, and newly powerful quality control tests were developed by explicitly formulating the decisions made by experienced quality control engineers to truncate a testing sequence.

The comments to this entry are closed.