The purpose of this week's lecture is to briefly introduce mathematical models, and attempt to relate some of their properties to connectionist networks. In particular, we will be looking at Rescorla-Wagner learning, its relationship to perceptrons, and will talk a bit some recent work of mine that challenges the accepted view of this relationship. Our activity will also explore using perceptrons to model a different aspect of animal learning, probability matching.
Lecture slides, in handout format or web format, are available from the menu on the right. There is also a link to a monograph that provides lots of additional details on this material if you are interested (Dawson, M.R.W. (2008). Connectionism and classical conditioning. Comparative Cognition and Behavior
Reviews, 3 (monograph), 1-115. Retrieved from http://psyc.queensu.ca/ccbr/index.html)