Goals

This course provides an introduction to PDP (or neural network or connectionist) models of cognition.

Because I expect that there will be students with a wide range of backgrounds in the class I assume no prior knowledge--for the sake of those who are new to connectionist models--but will try to proceed at a brisk enough pace that that those who do will still find the course useful.

The emphasis will be on the relevance of connectionist models for cognition. I hope to accomplish three sorts of things in the class.

First, I want people to emerge with a good overview of the field and a first-hand acquaintance with the literature.

Second (and most important), I want us to spend time trying to understand the basic principles and computational properties which characterize connectionist models. Although there is no orthodoxy and there is considerable diversity within the field, I think it is still possible—and very important—to identify the fundamental characteristics of neural networks and to contrast them with other computational mechanisms.

Finally, because I believe intuitions and understanding grow out of hands-on experience, those students who do not have prior experience with modeling will work through a set of 8 simulation exercises. Alternative options are provided for those students who already have such experience (see Course Requirements, below).

Format:

The class will be given once a week. The format will be a mix of informal lectures and discussion. I strongly encourage interaction and questions.

Discussion board:

There is a discussion board available for discussion about any of the readings or issues raised in class. Feel free to post questions, comments, or responses. The board can be accessed through the link on the class homepage, or by clicking on this link. Submissions can either be sent through the link on the discussion page or by sending email to cog202@crl.ucsd.edu

Course requirements:

Readings:

Required readings should be read prior to the class when they will be discussed (see Syllabus). In order to encourage critical reading (and to give me feedback on how effective the readings are), students should turn in 2 questions/comments for each reading assignment. These can either reflect questions about material that was unclear, or your own reactions (positive or negative) about the reading. Reading comments will be graded and will make up approximately half the final grade.

Simulations:

The grade on this requirement will make up the other half of the final grade. There will be 8 simulation exercises using the TLEARN simulator and the exercises from "Exercises in Rethinking Innateness". No prior experience with simulations is required, and no programming knowledge is necessary. Both the simulator and the exercises are available on-line. Exercises will be turned in over the course of the quarter on the dates indicated in the syllabus.