Syllabus for Q750:

Recent Advances in Connectionist Models

(listed as "Neural Networks as Models of Cognition")

Spring 1996, Section 1972, Tu & Th 2:30-3:45, PY 230

Instructor: Prof. John K. Kruschke
Office: 855-3192, PY 336 by appt.
E-mail: kruschke@indiana.edu

Contents:

Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer

Course Description and Goals:

This course selectively surveys some recent applications of connectionist models to cognitive phenomena. The emphasis is on models that attempt to address detailed empirical data. We will not survey "generic" algorithms or architectures that are not directed at specific cognitive phenomena and empirical data. The survey will be accomplished by reading and discussing various journal articles. Each student will lead (or co-lead) discussion of selected articles.

Each student will also program, from scratch, a connectionist model applied to cognitive phenomena of his or her choice. The model could be a novel architecture or an existing model applied to new phenomena. The last few weeks of the course will be devoted to presentations of these projects.

The course has three main goals. First, students should gain understanding of a variety of connectionist models applied to empirical data. Second, students should gain experience applying a connectionist model to specific cognitive phenomena. Third, students should gain a better understanding of the role of models in explaining empirical phenomena.

[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]

Prerequisites:

  1. You must have strong background knowledge in connectionist models. This can be achieved by previously completing the course Q550. If you are self-taught or have taken some other introductory course in connectionist models, please discuss the matter with Prof. Kruschke.
  2. You must be adept at computer programming. We will not have time to discuss programming methods much in class, so you are expected to be able to program independently. The preferred programming language is C, but any language is permitted. For all languages, the code turned in with your final project must be thoroughly commented.
[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]

Schedule:

The schedule below is approximate, and may flex and bend as the course progresses. The order of topics will probably remain the same, but the exact dates for each topic might change. There might be changes to the readings as the course progresses. Full reference information for each reading is provided at the end of the syllabus.

Schedule of Project Due Dates:
A brief project proposal is due on the Thursday of the fourth week of class. The proposal must specify (1) the empirical phenomena to be addressed, (2) the type of model architecture you intend to use, (3) the programming language you will use, and (4) which computer you will use. The final write-up of your project is due on the Thursday of the 15th week of class, except for those people who present their projects on that day. For these people, the final write-up is due the next day, Friday. There is no final exam.

Schedule of Readings:
Week 1: Introduction and scheduling of presentations.
Week 2: Meta-Theory.
Week 3: Category Learning.
Week 4: Development.
Week 5: Sequence Learning.
Week 6: Similarity.
Week 7: Strength of Processing.
Week 8: Memory.
Week 9: Neural Dynamics of Vision.
Week 10: Word Recognition.
Week 11: Lesions.
Week 12: Reprise of Meta-Theory.
Weeks 13-15: Presentations of Student Projects.
[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]

Grading Method:

Grades will be assigned on the basis of (a) quality in leading discussion of articles, (b) thoughtful participation in discussion even when not formally leading discussion, and (c) quality of the modeling project. For leading discussion, quality is judged by accuracy of content and by clarity of presentation. For the modeling project, quality is judged by many criteria, including but not limited to the following: Appropriateness of the model mechanisms to the phenomena being modeled; breadth and detail of empirical phenomena addressed; clarity of presenting the model and the program code; thoughtfulness in discussion of how the model helps explain the cognitive phenomena, based on successes and/or failures of the model.

[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]

Computers:

There will be two main uses of computers in this course. First, we will use the World Wide Web and a Newsgroup (ac.cogs.q750) for posting announcements and discussion. Our class web page is

http://www.indiana.edu/~jkkteach/Q750/q750.html
                                 ^    ^
                        upper case    lower case

Second, you will be using a computer to program your own connectionist models from scratch. You may use any computer that is convenient for you. Connectionist models are typically "computationally intensive," so you will want to locate a fast machine. If you have access in a lab to a Unix machine or a fast PC or Mac, that would be fine. Otherwise, there are various UCS Unix machines on which you can get accounts. For information regarding how to get accounts, see the Web page ... and get additional information from the UCS Knowledge Base at http://kb.indiana.edu/ .

[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]

Required Readings:

We will read the articles listed below. A package of the collected readings is available at the Indiana Memorial Union (IMU) Bookstore. There might be changes to the readings as the course progresses.

Chappell, M. & Humphreys, M. S. (1994).
An auto-associative neural network for sparse representations: Analysis and application to models of recognition and cued recall.
Psychological Review, 101, 103-128.

Cleeremans, A. & McClelland, J. L. (1991).
Learning the structure of event sequences.
Journal of Experimental Psychology: General, 120, 235-253.

Cohen, J. D. & Servan-Schreiber, D. (1992).
Context, cortex, and dopamine: A connectionist approach to behavior and biology in schizophrenia.
Psychological Review, 99, 45-77.

Cohen, J, D., Servan-Schreiber, D. & McClelland, J. L. (1992).
A parallel distributed processing approach to automaticity.
American Journal of Psychology, 105, 239-269.

Dienes, Z. (1992).
Connectionist and memory-array models of artificial grammar learning.
Cognitive Science, 16, 41-79.

Goldstone, R. L. (1994).
Similarity, interactive activation, and mapping.
Journal of Experimental Psychology: Learning, Memory & Cognition, 20, 3-28.

Goldstone, R. L. & Medin, D. L. (1994).
Time course of comparison
Journal of Experimental Psychology: Learning, Memory & Cognition, 20, 29-50.

Grossberg, S. (1994).
3-D vision and figure-ground separation by visual cortex.
Perception & Psychophysics, 55, 48-120.

Hinton, G. E. & Shallice, T. (1991).
Lesioning an attractor network: Investigations of acquired dyslexia.
Psychological Review, 98, 74-95.

Humphreys, G. W.; Freeman, T. A.; Muller, H. J. (1992).
Lesioning a connectionist model of visual search: Selective effects on distractor grouping.
Canadian Journal of Psychology, 46, 417-460.

Karmiloff-Smith, A. (1992)
Nature, nurture and PDP: Preposterous Developmental Postulates?
Connection Science, 4, 253-269.

Kruschke, J. K. (1992).
ALCOVE: An exemplar-based connectionist model of category learning.
Psychological Review, 99, 22-44.

Kruschke, J. K. (1996).
Base rates in category learning.
Journal of Experimental Psychology: Learning, Memory and Cognition, 22, 1-24.

McClelland, J. L., McNaughton, B. L. & O'Reilly, R. C. (1995).
Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory.
Psychological Review, 102, 419-457.

McCloskey, M. (1991).
Networks and theories: The place of connectionism in cognitive science.
Psychological Science, 2, 387-395.

Plunkett, K., Sinha, C., Moller, M. F. & Strandsby, O. (1992).
Symbol grounding or the emergence of symbols? Vocabulary growth in children and a connectionist net.
Connection Science, 4, 293-312.

Seidenberg, M. S. (1993a).
Connectionist models and cognitive theory.
Psychological Science, 4, 228-235.

Seidenberg, M. S. (1993b).
A connectionist modeling approach to word recognition and dyslexia.
Psychological Science, 4, 299-304.

Seidenberg, M. S., Plaut, D. C., Petersen, A. S., McClelland, J. L. et al. (1994).
Nonword pronunciation and models of word recognition.
Journal of Experimental Psychology: Human Perception and Performance, 20, 1177-1196.

Slezak, P., Latimer, C., Coltheart, M., Andrews, S., Oliphant, G., Bakker, P., Heath, R. A. & Watson, E. (1994).
Symposium on Connectionist Models and Psychology: The Rationale for Psychologists Using (Connectionist) Models, January 1994, University of Queensland, Australia.
http://www.cs.indiana.edu/Noetica/OpenForum.html
[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]
Disclaimer: This syllabus is meant to be suggestive, not absolute. Any and all of the information on this syllabus is subject to change at any time, including grading policies, office hours, etc. Changes will be announced in class.
[ Top | Description and Goals | Prerequisites | Schedule | Grading Method | Computers | Readings | Disclaimer ]