If, over the course of an experiment, a Bayesian observer reports one value of the stimulus estimate more often than another value, it means that the prior probability of the former is higher