Cognitive Control: Anterior Computational Modelling

Much of CCP research involves the use of neuroscience-based computational models of cognitive processes, both to test the explanatory power and sufficiency of our theories, and as a means for generating new hypotheses and predictions that can be tested in subsequent empirical studies. A critical component of the modeling endeavor is to provide an explicit account of the computational principles and neural mechanisms that underlie specific functions related to cognitive control.

Our work so far has resulted in the development of new theories regarding the functional contributions of three distinct neural systems to normal cognitive control: the dorsolateral prefrontal cortex (DL-PFC), the dopamine (DA) neurotransmitter system, and the anterior cingulate cortex (ACC). We have suggested that each of these systems is specialized to subserve a unique aspect of cognitive control: 1) DL-PFC represents and maintains goal/context representations (Braver, Cohen and Servan-Schreiber, 1995; Cohen, Braver and O'Reilly, 1996; Braver and Cohen, 2001; Braver, Barch and Cohen, in press); 2) the DA system updates these goal/context representations at appropriate junctures (Braver, Barch, and Cohen, 1999; Braver and Cohen, 1999; Braver and Cohen, 2000); and 3) the ACC detects processing conflicts, which index the demand for cognitive control and regulate the strength of goal/context representation (Carter, Braver, et al., 1998; Botvinick, Braver et al., 2001; Braver, Barch, et al., 2001; Barch, Braver, et al., 2001).

Our current work involves extending these models in various directions. One line of research examines how anterior cingulate function in conflict detection might contribute to performance during response inhbition, task-switching and other speeded response tasks. A second line of research examines computational principles that may underlie prefrontal cortex organization and specialization. A third line of research extends ideas regarding how dopamine modulation of prefrontal cortex may influence working memory. A fourth, just beginning line examines how neuromodulatory effects might underlie emotion-cognition interactions. Part of the research process involves the development of new software tools for conducting sophisticated neurally-based simulations. In particular, Researchers in our lab have recently developed such a simulator called RNS++ .