Cognitive Neuroscience Society San Francisco, CA
March 21-24, 2009
Scott Makeig and Julie Onton , Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla, CA
Associating event-related brain dynamics with event context
Active human agents both create and respond to events 'in real time,' often without the luxury of a self-chosen delay. Ongoing EEG source signals and their event-related perturbations largely index the actions of processes that maintain or adjust the distribution of attention between sensory, mnemonic, and imaginative processes in response to the perceived significance of events --- which may be heavily influenced by the context in which they occur. To determine from the data themselves which event contexts are linked to which brain dynamics changes, we decomposed event-related log spectrograms from maximally independent EEG components (IC) processes time locked to delivery of auditory feedback signals in a 'Two-back with feedback' visual working memory paradigm. Before decomposition of the [frequencies*latencies by trials] matrix for each IC, we appended a matrix of 'answers' to 19 questions about the trial context, in the form of a [answers by trials] matrix of ('yes|no') 1's and -1's. Maximally independent components of the joint data matrix gave independent factors (IXs) comprising an IX log spectral time/frequency modulation template and an IX loading on each of the questions, plus a vector of weights specifying the relative effect of the IX on each trial. Sorting the individual trial context vectors by their IX trial weights revealed significant trial sub-groupings, even for IXs linked to relatively simple event contrasts. Context ICA decomposition allows new insights in the connection between events in context and the complex spatiotemporal patterns of local cortical field synchrony that produce the ongoing EEG.