MMN2012: The 6th Conference on Mismatch Negativity (MMN) and its clinical and scientifc application

New York City



May 1-4, 2012

Scott Makeig , Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla, CA

Time-frequency and independent component analysis in MMN research

Electroencephalography (EEG), the recording and measurement of scalp electric potentials produced by emergent local synchrony of electrical field activity in cortical neuropile, was the first and is today the most widely known, most temporally acute, most non-invasive, most wearable, and least expensive brain imaging modality. It is now well established that spatiotemporal EEG activity patterns correlate with changes in 'top-down' cognitive arousal, attention, intention, event valuation, and the like, thereby providing a 'window on the mind.' However, the biological mechanisms that link EEG patterns to these or other aspects of cognition are not yet understood in enough detail.

Still today, most neurologists and EEG researchers still typically review EEG 'squiggles' by visual inspection of highly visible features of the channel signals or their event-related averages. In recent years, however, researchers with background knowledge of physics and engineering have increasingly considered the possible deployment of EEG as 1) a 3-D functional imaging modality for trait and state diagnosis and monitoring, and 2) as a potent active feedback modality for various training and retraining purposes. A relative weak point of most EEG uses today remains the application of adequate signal processing to undo the effects of mixing of the source signals during volume conduction, a phenomenon whose extent and implications have not been enough appreciated by many researchers. Because of the biophysical and mathematical difficulty of this EEG inverse problem, direct biophysical approaches to separating and locating EEG brain (and non-brain) source signals need to be married to statistical signal processing techniques that can reveal the abundant, time-varying information about cortical processes contained in high-density EEG data and about their interactions.

In coming decades, more adequate and intensive signal processing for EEG feature extraction and cognitive state and response recognition, combined with new, non-invasive, wearable sensor technologies, will I believe elevate EEG imaging to a true 3-D functional brain imaging modality and will support real-time brain-computer interface (BCI) applications in a wide range of diagnostic, therapeutic, and workplace applications. I will survey these ongoing advances and present some applications of recent modeling approaches to the study of detection in awake brain detection of unexpected auditory and other sensory events.

Return to Current Abstracts

Back to Makeig Home Page