{"pk":33059,"title":"Belief Revision in Models of Category Learning","subtitle":null,"abstract":"In an experiment, subjects learned about new categories for \nwbich tbey had prior beliefs, and made probability \njudgments at various points during the course of learning. \nThe responses were analyzed in terms of bias due to prior \nbeliefs and in terms of sensitivity to the content of the new \ncategories. These results were compared to the predictions \nof four models of belief revision or categorization: (1) a \nBayesian estimation procedure (Raiffa & Schlaifer, 1961); \n(2) the integration model (Heit, 1993, 1994), a \ncategorization model that is a generalization of the \nBayesian model; (3) a linear operator model that performs \nserial averaging (Bush & Mosteller, 1955); and (4) a \nsimple adaptive network model of categorization (Gluck & \nBower, 1988) that is a generalization of the hnear operator \nmodel. Subjects were conservative in terms of sensitivity \nto new information, compared to the predictions of the \nBayesian model and the linear operator model. The \nnetwork model was able to account for this conservatism, \nhowever this model predicted an extreme degree of \nforgetting of prior beliefs compared to that shown by \nhuman subjects. Of the four models, the integration model \nprovided the closest account of bias due to prior beliefs and \nsensitivity to new information over the course of category \nlearning.","language":"eng","license":{"name":"","short_name":"","text":null,"url":""},"keywords":[],"section":"17","is_remote":true,"remote_url":"https://escholarship.org/uc/item/4359909r","frozenauthors":[{"first_name":"Evan","middle_name":"","last_name":"Heit","name_suffix":"","institution":"Northwestern University","department":""}],"date_submitted":null,"date_accepted":null,"date_published":"1995-01-01T18:00:00Z","render_galley":null,"galleys":[{"label":"PDF","type":"pdf","path":"https://journalpub.escholarship.org/cognitivesciencesociety/article/33059/galley/24120/download/"}]}