In a lecture he gave at Stanford many years ago, Richard Rorty recounted the following story: A friend, an eminent decision theorist, had to decide between competing job offers from two universities. Unable to choose, he called up Rorty for advice. “Why don’t you make one of those fancy decision trees you’re always writing about?” suggested Rorty. His friend’s response: “Oh, come on, Dick, this is serious.”

The story came to mind in reading through the papers in behavioral economics presented at the Seventh Annual Conference on Empirical Legal Studies (CELS). One of the most exciting bodies of work to come out of the social sciences over the last fifty years, the heuristics and biases (H&B) wing of behavioral economics has identified robust patterns in human decisionmaking that undermine many of the core assumptions of rational choice theory (RCT): that we have stable preferences, that we act “rationally” to optimize those preferences, and that utility depends on end states (e.g., total wealth), rather than gains and losses off of a reference point (e.g., of prior wealth).

But RCT and much of the H&B research on consumer behavior share one presupposition, which is arguably more important than all of the ones they disagree about: that the ultimate carriers of utility in consumer transactions are commodity bundles, that consumers’ “true” preference is to optimize on those bundles, and therefore that prospect theory and other violations of expected utility theory documented by H&B scholars “must lead to normatively unacceptable consequences.” The question I want to raise here is: what if that supposition is wrong? Suppose, for example, that in the typical purchase decision, the consumer’s “true” preference is not to maximize some function of wealth (absolute, relative, or changes in wealth measured against a reference point), but instead to minimize the time and mental energy spent on trades, because once she has reached an acceptable level of material consumption at a fairly low level of demandingness, she gets little if any additional utility from optimizing, compared to the other things she could be doing with that time. What difference would that make in how we interpret experimental findings in behavioral economics and psychology?