Selecting among Computational Models of Cognition:
A Minimum Description Length Approach

Dr. In-Jae Myung

Ohio State University

June 19, 2000 4:00 pm

The question of how one should decide among competing explanations of data is at
the heart of the scientific enterprise. Computational models of cognition and ot
her processes are increasingly being advanced as explanations of behavior.
The success of this line of inquiry depends on the development of sound and robu
st methods to guide the  evaluation and selection of these  models. 
In this paper, we introduce a method of selecting among mathematical models of c
ognition, known an Minimum Description Length. It provides an intuitive and theo
retically well-grounded understanding of why one model should be chosen over ano
ther.
A central, but elusive, concept in model selection, complexity, can also be deri
ved with the method as formulated in the space of probability distributions.
The adequacy of the method is demonstrated in two areas of cognitive modeling: p
sychophysics and information integration.

Brief CV:
1975-80 B.S. Physics, Seoul National University
1980-82 M.S. Biological Science, KAIST
1985-90 M.S. & Ph.D Psychology, Purdue University
1990-91 Post doc, University of Virginia
1991- present: Assistant & Associate Professor, Department of Psychology, 
	Ohio State University
1999-2000: Visiting Professor, KAIST CS Department



This page is maintained by InYoung Kim (iykim@scai.snu.ac.kr).
Last update: October 2, 2000