Representing individual differences in computer models: Theoretical considerations and parameter estimation

It is argued that the deterministic approach is more conducive to the development of detailed process models that account for individual differences.

Abstract

Computer models of behavior are more powerful and precise when explicit provision is made for individual differences. The most common method of representing individual differences, involving stochastic processes, is described and evaluated. The advantages and disadvantages of this method are compared to those of an alternative method, involving the estimation of parameter values for individual subjects in deterministic models. We argue that the deterministic approach is more conducive to the development of detailed process models that account for individual differences. Some of the problems encountered in estimating individual difference parameters in computer models are discussed and tentative solutions are offered.