Ed Vul: UC San Diego

Human behavior is robust, adaptable, and, human behavior often deviates from the utility maximizing “rational” agent. This is usually attributed to people relying on an assortment of cheap heuristics to make efficient, but frequently biased, decisions. While the heuristics and biases research program has highlighted the many deviations of human behavior from that of simplistic economical agents, it has also yielded a morass of idiosyncratic, unreliable, and often contradictory biases, with no method to decide which heuristics will play a role in a given situation. Here I will describe our recent progress on an alternate approach: accounting for the successes and foibles of human behavior by assuming that people are more sophisticated, robust, and probabilistic than simple economic agents, but must carry out these sophisticated inferences under cognitive resource constraints. This approach yields a single framework for human decision-making: resource-rational probabilistic inference. This parsimonious, predictive account reconciles economic and psychological models of decision making and behavior.