Individuals may repeatedly face a choice of whether to obey a legal rule. Conventional legal scholarship has long assumed that whether such a choice is made repeatedly or is a one-time event has little or no effect on individuals’ decisions. Following models of rational-choice theory, traditional legal analysis predicts that, in either case, individuals would behave in a way that maximizes their payoffs. A large body of experimental literature, however, suggests that individuals facing a recurring choice, in contrast to individuals making the choice only once, do not behave as maximizers. Instead, individuals facing the choice repeatedly apply the strategy of “probability matching.” Individuals, for example, failed to maximize when presented with a die with 4 red faces and 2 white faces, and asked to predict the colors of a series of rolls. Although maximization demands consistently betting on the red, individuals preferred a “mixed” approach. Bets were divided so that red was chosen in 2/3 of the rolls and white in 1/3 of the rolls. This article presents several normative and descriptive applications that probability matching has in the legal context. Normatively, the analysis shows how probability matching affects the optimal level of investment in law enforcement. Descriptively, it suggests that probability matching provides a new rationale for existing legal doctrines, such as the rule allowing the imposition of punitive damages on repeated wrongdoers, the imposition of harsher sanctions on recidivists, and the increasing application of rules attributing liability for the mere infliction of risks. This article thus shows that experimental findings corroborate the intuition that law ought to, and in fact does, differentiate sharply between repeated and single-instance behavior.
