Skip to content

Belief Updating

Summary

The process of revising probabilities assigned to hypotheses as new evidence becomes available — the core mechanism of Bayesian reasoning.

The Update Cycle

Prior belief → New evidence → Likelihood calculation → Posterior belief → New prior

Each posterior becomes the prior for the next round — beliefs evolve iteratively.

How Much Should Beliefs Change?

The amount of change depends on two factors: 1. Strength of prior — Strong priors require more evidence to shift 2. Diagnosticity of evidence — Evidence that's much more likely under one hypothesis than another causes bigger shifts

Principles

  • Irrelevant evidence doesn't change beliefs — If P(E|H) = P(E|¬H), posterior = prior
  • Strong priors resist weak evidence — Extraordinary claims require extraordinary evidence
  • Multiple weak evidence can combine — Each piece shifts the posterior incrementally

See Also