A Reinforcement-and-Generalization Model of Sequential Effects in Identification Learning

Abstract

Responses in identification-learning tasks depend on events from recent trials. A model for these sequential effects is proposed, based on previous work in category learning and founded on theories of reinforcement learning and generalization. The model is compared to two other theories in their predictions of the influence of previous stimuli and previous feedback. Two experimental paradigms are introduced that allow separate assessment of these two effects. Results support the reinforcement-and-generalization model.


Back to Friday Papers