This project is concerned with how organisms learn. The mechanisms uncovered by this work should have applications to education and methods for job training, aid in the design of artificially intelligent systems, and guide research on the neurobiological basis for learning. Theorists have distinguished three kinds of learning: supervised, unsupervised, and reinforcement learning. This project deals with learning through reinforcement, which is one of the most widely observed types of learning in people and animals. Reinforcement has both short-term and long-term effects. The tendency to repeat a rewarded action is a short-term effect. The tendency to seek out situations that have been associated with reward in the past is a long-term effect. The previous phase of this project focused on short-term effects. This phase focuses on long-term effects: how memory affects choice. Dr. Staddon is looking at this problem in two kinds of situation: a simple choice situation in which hungry animals repeatedly choose between two probabilistically rewarded alternatives, and situations in which rewards occur at predictable time intervals. The former situation allows researchers to study how the animal's reward history (over a period of days) determines its present choice performance; the latter situation allows researchers to study how the recent distribution of rewards in time (over a period of seconds and minutes) affects current behavior. By repeated construction and testing of theoretical models for learning and memory, Dr. Staddon hopes to arrive at a real-time theory of the learning process that will make contact with its neurophysiological basis.
