The Irrational Economist and the Quest for Rationality

Research in behavioral economics should have consequences of three types. First, policy analyses should take account of the weaknesses and strengths of human inferential and decision processes. Second, general education should strive toward improving these same processes. For example, good courses in probabilistic reasoning should perhaps be central in high-school and college education. Third, however, the concept of rationality also should change, to attain reflective equilibrium with the inferences and decisions people actually make. Much of Howard Kunreuther’s work has focused on the policy side. Some of my past research has gauged possible effects of education. In this contribution, however, I focus on the third issue: How should our concept of rationality be modified? What inferences, what decisions deserve to be called irrational? With specific reference to Howard Kunreuther’s oeuvre: Are “peace of mind” and “consolation” irrational? Is it irrational to opt for low insurance deductibles? Is a short time horizon irrational? Is probability neglect irrational? Is following one’s neighbor’s lead (without any change in awareness of options or in expectations) irrational? One revisionist approach is evolutionary in flavor: rationality is whatever works best in a given ecological niche. I reject this approach, for two reasons. First, we lack sufficient about the ecological niches that homo sapiens has filled, about the rapidity of human evolution, and about human specialization to acquire and deploy new culturespecific intellectual tools. Second, the adaptive value of human inference and decision processes has to be evaluated in the social context in which these processes are deployed. We are just beginning to learn about the effects of social context. My approach is based on the central finding of behavioral economics, that goals and decision processes are context dependent. On the one hand, this finding undercuts the sweeping axiomatic approaches to rationality. For example, intransitivity of pairwise choice cannot necessarily be viewed as irrational: A may be chosen over B in one context, B over C in another, yet C may be chosen over A in still a third context. Nor can one say that transitivity should hold in a constant context, because the pair of options itself can activate goals differentially and thus must be considered to be part of the context. If goal activation is labile, then transitivity can be violated. (The traditional money-pump argument also does not stand up to scrutiny.) Context dependence not only undercuts sweeping axiomatic definitions; it also provides a route toward more positive statements about what is rational. Choices made in an unusual context, designed to activate particular goals, are often unstable. Reframing may lead to different choices. Sometimes two alternative frames seem equally valid, e.g., gain versus loss framing in the famous Asian Disease Problem of Kahneman and Tversky. But sometimes, reframing leads to the insightful adoption of a preferred, stable frame; in such cases, one criticizes the choice made prior to reframing. For example, adoption of a long-duration temporal frame may be stable, leading to the conclusion that short time horizons often lead to choices that can be legitimately criticized as myopic. Similarly, framing insurance in terms of negative expected value traded for protection against catastrophe may be stable and may lead to criticism of low deductibles. Advertising often attempts to create frames in which unusual goals are active. Does one really want to purchase a car as an instrument of sexual seduction? If not, then one might view the emotions evoked by images of flirtation as irrational. (Even if yes, one may question whether paying a much higher price for the car with sexy image is an efficient approach to seduction.) There is nothing irrational about “peace of mind” as a goal for insurance purchase, but in a context where a wide range of insurance plans is available; there may be relatively inexpensive ways to attain this goal. Some of the most difficult questions for a theory of rationality concern the activation of social goals when decisions are made in a social context. Is fairness rational? Reciprocation? Altruistic punishment? Status seeking? Role-seeking? Adherence to or enforcement of group, status-dependent, or role-related norms? All these questions require nuanced answers.