Seminar: 4/3 - Bob Rehder
11:00 to 12:30 PM at: Tolman 5101
Department of Psychology, New York University
"A Generative Model of Conceptual Representation."
In the talk I will introduce a model of conceptual representation that incorporates not only categories' typical features but also their interfeature causal relationships. According to this generative model of classification, items likely to be generated by a category's causal model are considered good category members and those unlikely to be generated are poor category members. Four experiments tested the generative model's predictions regarding the causal status effect (cause features are more important to categorization than effect features) and the coherence effect (good category members exhibit combinations of features likely to be generated by the causal model). They also tested a novel prediction regarding how a feature's category validity is affected by interfeature causal links. Results confirmed the predictions of the generative model and disconfirmed those of a competing model. Applications of the generative model to other key category-based judgments (property generalization and prediction) are discussed.