Advanced Topics in Cognitive Development (Psyc 4540)
(Neural Network Models of Cognitive Development)
TTh 11:00 - 12:50, Spring 1999
Frontier Hall 354, University of Denver
Yuko Munakata
munakata@kore.psy.du.edu
Frontier Hall 343, 871-4151

Course Goals The primary goal of this course is to understand cognitive development through the use of neural network (aka connectionist or parallel distributed processing) models. A secondary goal is to understand the role of neural network simulations in cognitive science more generally: their unique contributions, the controversies surrounding them, and so on. As we go through the course, you should keep in mind the following kinds of questions: What are the origins of our knowledge? What is the nature of developmental change? How does such change occur? How have neural network models succeeded and failed in exploring such questions? Where they have failed, could minor revisions lead to success? Or do the failures point to the need for a new or radically revised framework for modelling cognitive development?

You should come away from this course with responses to these questions for various aspects of cognitive development, including problem-solving, language, and representations of objects and number. Note that ``responses'' does not necessarily mean ``answers''. The goal is not to learn the answers to these questions (in many cases they don't yet exist!). Instead, the goal is to understand all of the relevant issues thoroughly enough that you can critically evaluate the course readings and provide well-informed opinions on the ongoing debate surrounding the questions above. If you succeed in this goal, you will be in a better position to evaluate new developments in the field and to create your own simulations to further enhance our understanding of cognitive development.

Prerequisites

There are no official course prerequisites. Familiarity with cognitive development and/or neural networks will help, but is not required. If you would like a refresher in or general introduction to cognitive development, see a solid introductory text such as Siegler (1998). If you would like a refresher in or general introduction to neural networks, see an introductory text such as Rumelhart & McClelland (1986), Elman et al. (1996), or O'Reilly & Munakata (in press). If you would like to be directed to the coverage of a specific topic, let me know.

Readings

We will read current and classic journal articles and book chapters. The readings for each week are in a file box in Frontier Hall 130 (near the department fax machine). To abide by copyright laws, students are responsible for making their own copies. You may speak with Nancy Pleiman about making copies with the Psychology Department copier. Please record your name, the course number, and the number of copies in the copier log book. The departmental charge is 6.5 cents per page.

Evaluation Your grade will consist of the following:

Reading Evaluations

You will be asked to evaluate each of the readings (unless you are leading the session, see below) by providing brief answers to the following:

1) Describe one significant contribution from the model/paper.

2) Describe one significant limitation of the model/paper.

Evaluations should be typed and either emailed to me before class or turned in at the start of class. One evaluation (with your lowest score) will be dropped from your final grade.

Participation

You will be expected to come to class prepared to ask questions and actively participate in discussion. Your participation and ideas will be crucial to our evaluation of the ideas in the field, and will help you to get the most out of this class. To support and encourage effective discussions, part of the first class meeting will be devoted to a discussion of ``What makes a discussion bad (and what we can do about it).'' Throughout the class term, we will evaluate the effectiveness of our discussions and welcome suggestions on improving them to help futher our understanding of issues in cognitive development.

Session Leading

You will be asked to lead discussion in several class sessions. Such session leading may include very brief summary presentations of the readings, posing of questions for discussion, and moderating of discussion/debate.

Grading Policy Grades are not curved; they are based on percentages:

Class Schedule

The class schedule includes three weeks on models of specific aspects of behavior (e.g., language), two weeks on models of general approaches to development (e.g., constructivism), and two weeks on commentary on neural network modelling. In addition, there is one uncommitted week, to be filled with a topic chosen by the class. We can swap this uncommitted week up or back as necessary.

Additional Materials of Interest Journals

Books

Carey, S., & Gelman, R. (Eds.). (1991). The epigenesis of mind. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Elman, J., Bates, E., Karmiloff-Smith, A., Johnson, M., Parisi, D., & Plunkett, K. (1996). Rethinking innateness: A connectionist perspective on development. Cambridge, MA: MIT Press.

Flavell, J. H., Miller, P. H., & Miller, S. A. (1992). Cognitive development. Englewood Cliffs, NJ: Prentice-Hall, 3rd edition.

Gazzaniga, M. S. (Ed.). (1995). The cognitive neurosciences. Cambridge, MA: MIT Press.

Gelman, R., & Au, T. K.-F. (1996). Perceptual and cognitive development. San Diego: Academic Press.

Johnson, M. H. (Ed.). (1993). Brain development and cognition: A reader. Oxford: Blackwell.

Karmiloff-Smith, A. (1992). Beyond modularity: A developmental perspective on cognitive science. Cambridge, MA: MIT Press.

O'Reilly, R.C. & Munakata, Y. (in press). Cognitive neuroscience: A computational exploration. Cambridge, MA: MIT Press.

Rumelhart, D. E., McClelland, J. L., & PDP Research Group (Eds.). (1986). Parallel distributed processing. Volume 1: Foundations. Cambridge, MA: MIT Press.

Siegler, R. (1998). Children's thinking. Englewood Cliffs, New Jersey: Prentice Hall, 3rd edition.