How does one choose among competing explanations of scientific data? When the explanations are mathematical models, researchers frequently apply quantitative methods to answer this question. The purpose of this research program is to develop such methods for selecting among quantitative models of cognition. Three lines of inquiry will be undertaken. In the first, theoretical work is proposed to increase our understanding of a central issues in model selection, model complexity, by exploring the use of one measure of complexity (Geometric Complexity) in new model-testing situations, and also by comparing it with other measures of complexity. In the second line of work, simulation experiments will be carried out to extend the application of geometric complexity and a particularly successful model selection method, Minimum Description Length, to more complex testing situations and to other types of models in cognitive psychology (e.g., connectionist, random walk). In a third line of research, simulation experiments will explore the application of Geometric Complexity and another quantitative tool (Response Surface Analysis) in other stages of the research process, such as in the analysis of a model's components and in guiding the design of future experiments to test models. This work should advance our understanding of model selection and provide researchers with new tools with which to test models of cognition. The fruits of this research should also extend into other areas of psychology and other social sciences.

Agency
National Institute of Health (NIH)
Institute
National Institute of Mental Health (NIMH)
Type
Research Project (R01)
Project #
2R01MH057472-04
Application #
6370862
Study Section
Special Emphasis Panel (ZRG1-BBBP-4 (01))
Program Officer
Kurtzman, Howard S
Project Start
1998-08-10
Project End
2006-06-30
Budget Start
2001-09-24
Budget End
2002-06-30
Support Year
4
Fiscal Year
2001
Total Cost
$147,000
Indirect Cost
Name
Ohio State University
Department
Psychology
Type
Schools of Arts and Sciences
DUNS #
098987217
City
Columbus
State
OH
Country
United States
Zip Code
43210
Wu, Hao; Myung, Jay I; Batchelder, William H (2010) Minimum description length model selection of multinomial processing tree models. Psychon Bull Rev 17:275-86
Wu, Hao; Myung, Jay I; Batchelder, William H (2010) On the Minimum Description Length Complexity of Multinomial Processing Tree Models. J Math Psychol 54:291-303
Cavagnaro, Daniel R; Myung, Jay I; Pitt, Mark A et al. (2010) Adaptive design optimization: a mutual information-based approach to model discrimination in cognitive science. Neural Comput 22:887-905
Myung, Jay I; Tang, Yun; Pitt, Mark A (2009) Evaluation and comparison of computational models. Methods Enzymol 454:287-304
Myung, Jay I; Pitt, Mark A (2009) Optimal experimental design for model discrimination. Psychol Rev 116:499-518
Myung, Jay I; Montenegro, Maximiliano; Pitt, Mark A (2007) Analytic Expressions for the BCDMEM Model of Recognition Memory. J Math Psychol 51:198-204
Myung, Jay I; Pitt, Mark A; Navarro, Daniel J (2007) Does response scaling cause the generalized context model to mimic a prototype model? Psychon Bull Rev 14:1043-50
Pitt, Mark A; Myung, Jay I; Altieri, Nicholas (2007) Modeling the word recognition data of Vitevitch and Luce (1998): is it ARTful? Psychon Bull Rev 14:442-8
Pitt, Mark A; Kim, Woojae; Navarro, Daniel J et al. (2006) Global model analysis by parameter space partitioning. Psychol Rev 113:57-83
Navarro, Daniel J (2004) A note on the applied use of MDL approximations. Neural Comput 16:1763-8

Showing the most recent 10 out of 16 publications