Submitted for presentation at 2005 meeting of the Cognitive Neuroscience Society

Evidence suggests that differing processes underlie verbal versus non-verbal working memory, with verbal memory depending specifically on subvocal rehearsal of items. We recorded EEG at 60 scalp electrodes while subjects performed a Sternberg working memory task for lists that varied in rehearsability.

In each trial, subjects judged whether a probe item was one of the three items in a just-presented list. Lists within each block of trials were composed of stimuli from one of five stimulus pools that were chosen to be either verbally rehearsable (consonants, words, meaningful pictures) or non-rehearsable (sinusoidal gratings, single dots).

The ERP-FN400 differentiated targets and lures for rehearsable, but not for non-rehearsable, stimuli. Oscillatory power in the 4-8Hz theta range rose following each item presentation. Power at 9-30Hz increased at fixation cue onset, declined with each stimulus presentation, and increased as each stimulus disappeared. Power decreased during the probe interval reaching a minimum at the end of each trial. Power at higher frequencies (30-50Hz) declined during trials.

Overall, trials using non-rehearsable stimuli produced weaker oscillatory power than trials using rehearsable stimuli below 30 Hz. This effect was most pronounced in the 14-28Hz beta range. Power differences below 30 Hz were significant during study, retention, and probe intervals, with the largest effect seen in the parietal area. In summary, oscillatory effects in the theta and beta ranges appear to distinguish rehearsable and non-rehearsable stimuli during a working memory task.