Proceedings of the Annual Meeting of the Cognitive Science Society

Cognitive control refers to the ability to adjust our thoughts andbehaviors in order to achieve internalized goals. In the past,researchers have proposed several models of cognitive controlto account for the characteristic patterns of response timesobserved in the tasks (e.g., Botvinick, Braver, Barch, Carter, &Cohen, 2001). The goal of this study is to evaluate empiricalvalidity of such models in an experiment. To that end, wecompared two models of cognitive control, the conflictmonitoring model and the expectancy-based model. Eachmodel was implemented in two different modelingframeworks, neural networks and simple linear models. Therelative fits of the four models were then evaluated andcompared based on observed data from a flanker taskexperiment. The model comparison results showed thatperformance of the simple linear models was entirelycomparable to that of the neural network models. We alsoconstructed and fitted hierarchical Bayesian latent mixtureversions of the linear models to investigate individualdifferences. The result suggests that no single model ofcognitive control, whether conflict monitoring or expectancy-based, would be able to account for individual performance onthe task.