Seeing is believing: Priors trust and base rate neglect

Abstract

Tversky and Kahneman (1974) described an effect they called `insensitivity to prior probability of outcomes', better known as base rate neglect (Bar-Hillel, 1980). This describes people's tendency to underweight prior information in favor of new data. Probability theory requires these prior probabilities to be taken into account, via Bayes' theorem, when determining an event's posterior probability. The fact that most people fail to do so has been taken as evidence of human irrationality and, by other authors, of a mismatch between our cognitive processes and the questions being asked (Cosmides & Tooby, 1996; Gigerenzer & Hoffrage, 1995). In contrast to both views, we suggest that simplistic Bayesian updating using given base rates is not always a rational strategy. Instead, we reconsider Bar-Hillel's original relevance theory, and argue that, since base rates differ in their perceived degree of trustworthiness they are, accordingly, rationally discounted by people.