What is Bayesian updating?
Bayesian updating is the process of revising beliefs or probability estimates in light of new evidence, using Bayes’ theorem. It describes how a rational agent should change their mind when confronted with data.
How it works
Starting from a prior belief (initial estimate), new evidence shifts the belief toward a posterior (updated estimate). The magnitude of the shift depends on how surprising the evidence is relative to the prior. Strong priors require more evidence to shift. In behavioral science, people often deviate from Bayesian updating: they overweight vivid anecdotal evidence, underweight base rates, and show confirmation bias by seeking evidence that supports existing beliefs.
Applied example
A doctor who initially believes a patient has a 10% chance of a rare disease (prior) receives a positive test result with 90% sensitivity and 5% false positive rate. Bayesian updating yields a posterior probability of about 67%, far lower than most doctors intuitively estimate because they neglect the base rate.
Why it matters
Bayesian updating provides the normative standard against which human belief revision can be measured, revealing systematic biases in how people process new information.
