Anderson Colloquium Speaker Series

The notion of domain-generality of cognitive control is used widely and loosely in many areas of psychology. Researchers correlate performance on a task of interest with performance on tasks such as Flanker, Stroop, and Simon, to determine the involvement of “domain-general” cognitive control. Some even go further and use these and similar tasks to train cognitive control, hoping for wide-ranging transfer. While these assumptions seem reasonable from the perspective of partially shared neural correlates among control-demanding tasks, they do not fit well with the modern view of cognitive control rooted in learning.

In this talk, I will use language as a tool to test the predictions of the “learning view” of cognitive control. I will implement this view in a computational model and show that the model makes nuanced predictions regarding moment-by-moment fluctuation in control, otherwise known as the congruency sequence effect, both within-task and during task-switching. I will then present behavioral, electrophysiological, and neuroimaging data verifying the predictions of the model. I will conclude by showing that the learning view of cognitive control accounts for the entirety of the findings. This, in turn, questions some of the common assumptions and practices in psychological research.