Cognitive Science & Policy Hub // Vol. 12

Behavioral Economics Lab

The Nudge & Bias Lab — an interactive, citable directory of cognitive biases, peer-reviewed journals, and empirical nudge findings.

Belief Updating

Errors in Probabilistic Reasoning and Judgment Biases

Daniel J. Benjamin

Handbook of Behavioral Economics / NBER · 2019

Abstract

A comprehensive survey of how human belief formation departs from Bayes' rule, cataloguing the systematic errors people make when reasoning about probabilities and updating on evidence.

Methodology

Review and organization of the experimental literature on probabilistic judgment, formalizing biases such as base-rate neglect, conservatism, the law of small numbers, and overconfidence within a common framework.

Findings

Documents a stable catalogue of deviations: people neglect base rates yet are conservative in updating, over-infer from small samples (the gambler's and hot-hand fallacies), and are systematically overconfident. These errors are predictable and interact, so real belief-updating is neither Bayesian nor purely random.

Applied nudge

Do not assume users integrate evidence correctly—present base rates explicitly, pre-aggregate noisy signals, and design feedback that counteracts over- and under-inference.

Citation

Benjamin, D. J. (2019). Errors in probabilistic reasoning and judgment biases. NBER Working Paper No. 25200. In Handbook of Behavioral Economics.

Download PDF