The cognitive side of probability learning.

paradigms. In the present approach, a common theoretical framework is sought in concepts of coding and organization in memory. Following the theoretical analysis, an observation-t ransfer paradigm is developed that permits the study of predictive behavior depending on categorical, as distinguished from episodic, memory. This paradigm yields evidence that probability learning and transfer derive from frequency learning. The individual categorizes events and forms representations in memory of relative frequencies of event categories. When the different cues in a multiple-cue, probability learning situation occur equally often, this process yields predictive behavior closely reflecting the probabilities that the alternative events associated with a cue will occur when the cue is present. But when cue frequencies are unequal, the categorical memory model implies (correctly) that predictive behavior may be grossly out of line with actual probabilities. In general, depending on task requirements, predictive responses are either direct reflections of relative frequency judgments or are governed by strategies involving an additional