Dynamic Cognitive Networks in the Human Brain

Abstract

Even simple ognitive behaviors involve a complex spatiotemporal sequence of coordinated processes in many neural regions. Evoked potentials (EPs) have the subsecond temporal resolution needed to track the flow of information through the brain while subjects perform cognitive tasks. To take full advantage of the information provided by the EPs, both the spatial topography and the temporal relationships of responses measured at different scalp sites must be considered. Toward this end, we have been developing methods for statistically characterizing coordinated processing in distributed cortical networks by analyzing the spatial covariance of EPs recorded from multiple scalp locations. This method is called Evoked Potential Covariance (EPC) analysis.

This paper describes the results of a series of experiments in which sub-second EPC patterns were measured during simple but complete cognitive behaviors—from stimulus expectancy, through stimulus evaluation and decision processes, to response execution and evaluation of feedback, in situations that required responses to the current stimulus, or responses to previous stimuli (thus requiring working memory). Differences in the EPC patterns across the 4- to 6-second intervals spanning the trials demonstrated that different cortical areas form functional associations during each subsequent stage of processing. Subdural electrical recordings suggest that the EPC measure is a physiologically meaningful method of characterizing the association between different brain areas. We speculate that the EPC patterns associated with prestimulus preparation and feedback-evoked updating are manifestations of the internal models created by the brain which provide the context within which stimuli are interpreted.