It is proposed that memory schemata connect information about a single individual's mannerisms, traits, and other characteristics to the likelihood of that individual's acting in a particular fashion. Such schemata could account for certain categorization errors in which inappropriate generalizations are made in the course of predicting the likelihood of a person's action. The study examines learning of such personal-action schemata, and evidence supportive of such schemata are obtained. Theoretical implications for clinical transferee are considered.
