A rule for updating a probability given new evidence: P(A|B) = P(B|A)·P(A) / P(B). The posterior is proportional to the likelihood times the prior; a common error is flipping the conditioning in the numerator or omitting the total-probability denominator.