Spectral characterization of ongoing and auditory event-related brain processes

Brain processes phase-locked to stimuli can be readily observed with electro- and magnetoencephalography (EEG & MEG, respectively) using stimulus-triggered averaging of the measured signal. The detection of non-phase-locked brain processes depends on the method used for analyzing the unaveraged data. Here we introduce a technique, partition-referenced moment (PRM) power spectrum, which uses established spectral estimation algorithms but yields a power spectrum with sharp, easily distinguishable peaks in an otherwise level spectrum even when the signal (such as EEG & MEG) is of the one-over-frequency-slope type. Employing this method and wavelet transforms, we show that transient auditory brain responses are followed by dispersed small-magnitude power reductions. Power reductions occurred around 400-600 ms and were specific to ongoing 10 Hz-oscillations. The PRM-method also indicated ongoing oscillations in the 15-30 Hz frequency range where power reductions occurred at around 200-400 ms. Thus, the presented methods enable the straightforward detection of ongoing brain oscillations and their association with event-related power changes.