Influences of prior knowledge on selective weighting of category members

Three experiments addressed how prior theories affect categorization, comparing the influence of theory-congruent versus theory-incongruent category members. Subjects observed descriptions of persons, some congruent with prior knowledge and some incongruent, then made transfer judgments. In Experiment 1, subjects were given a relatively long time to study each description, whereas in Experiment 2 study time was manipulated between subjects. In Experiment 3, learning was self-paced by each subject. It was found that, with enough study time, prior knowledge had 2 distinct influences. First, prior knowledge provided an initial representation, subsequently revised in light of new observations. Second, incongruent observations had more impact than congruent observations on categorization. In comparison, when study time was more limited, revision proceeded in a Bayesian manner, in that congruent and incongruent observations had equal impacts.