Learning and Cognition in Humans, Animals, and Robots (B,C)
Connectionist theories of human and animal learning and cognition applied to robotics. Neural network theories of classical conditioning; concepts of models of the environment, prediction of future events, redundancy reduction, competition for limited capacity short-term memory, mismatch between predicted and observed events, stimulus configuration, inference generation, modulation of attention by novelty, and timing. Neural networks of operant conditioning; concepts of goal-seeking mechanisms, response-selection mechanisms, and cognitive mapping. How neural network models can be used to develop psychological theories, models of the brain, and robots. One course / 3 units.