Accuracy and Consistency in the Revision of Subjective Probabilities

Using Bayes' theorem as the normative model, four experiments examined the consistency with which subjective probabilities were revised in light of additional data. Consistency was found to be extremely high under all experimental conditions, and it was shown to be independent of the accuracy of Ss' subjective probabilities; Ss apparently apply a revision rule that is similar to Bayes' theorem to whatever subjective probabilities they possess at the moment regardless of the accuracy of their probabilities.