Centre de Recherche Cerveau et Cognition seminar Toulouse, France
June 29, 2007
Scott Makeig , Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla, CA
Beyond averaging: EEG and neural computation
Neuroscience has traditionally dealt with the complexity of brain dynamic data by averaging recorded data epochs time locked to some class of events of interest, giving ERPs, ERFs, PSTHs and BOLD activity maps. Such averages are optimal for measuring stereotyped, passive responses to stimulation from data corrupted by unrelated measurement noise. They are not optimal, however, for modeling the active reactions to events of a complex agent driven by a heterarchy of often conflicting goals. Simply put, brains have evolved to deal optimally with the challenge and opportunity of the moment, taking into account all the current knowledge, goals, needs and priorities of the organism. The goal of cognitive neuroscience is to learn more about how the human brain accomplishes this. I will discuss new directions in information-based signal processing based on this model of the brain as an active agent, and will show some first results applied to EEG and fMRI data.