Proceedings of the Annual Meeting of the Cognitive Science Society

A standard view in cognitive psychology is that people esti-mate probabilities using heuristics that do not follow proba-bility theory. We describe a model of probability estimationwhere people do follow probability theory in estimation, butare subject to random error or noise. This model predicts thatpeople’s conditional probability estimates will agree closelywith probability theory for certain noise-cancelling expres-sions, but deviate from probability theory for other expres-sions. We describe an experiment which strongly confirmsthese predictions, suggesting that people estimate conditionalprobabilities in a way that follows standard probability theory,but is subject to the biasing effects of random noise.