Prediction of response speed by anticipatory high-frequency (gamma band) oscillations in the human brain.

Response to a stimulus is faster when a subject is attending and knows beforehand how to respond. It has been suggested recently that this occurs because ongoing neuronal activity is spatially and temporally structured during states of expectancy preceding a stimulus. This mechanism is believed to mediate top-down processing, facilitating the early grouping and selection of distributed neuronal ensembles implicated in ensuing sensory-motor processing. To validate this model, it must be shown that some features of this early ongoing neural activity are correlated with subsequent perceptual decisions or behavioral events. We investigated this hypothesis in an electrophysiologic study in 12 subjects carrying out a simple visuomotor reaction-time task. Local field potentials (LFP) at each brain voxel were estimated using a linear distributed inverse solution termed "ELECTRA" for each single trial of each subject. The energy of oscillations for different frequency bands was computed for the period between the warning cue and visual stimuli by applying a time-frequency decomposition to the estimated LFP. A nonparametric correlation coefficient was then calculated between energy of oscillations and reaction times for each single sweep. Gamma band oscillatory activity in a frontoparietal network before stimulus onset significantly correlated with reaction time for a significant amount of subjects. These results provide direct evidence for the role of neural oscillations as a top-down attentional control mechanism that mediates the speed of motor actions. (c) 2004 Wiley-Liss, Inc.