Determining working memory from ERP topography.

Event-related potentials were recorded during a delayed matching-to-sample design from 17 volunteers (5 f) using high-resolution (65 channels) EEG-recordings. In the two-stimulus paradigm, the 500-ms stimulus S1 comprised a visual pattern of two diamonds differing in size, angular rotation and location; in the delay period, Working Memory (WM) load was varied in the following way: a stimulus-free interval of 1 s was followed by a 6-s presentation either of a pattern identical to the S1 (low WM load) or of a pattern differing from S1 (high WM load). The 500-ms stimulus S2 comprised one diamond; the subject's task was to indicate by left- or right-hand (respectively) button press, whether the S2 matched the (a) left- or (b) right-positioned S1-diamond, or (c) did not match at all (NoGo). The topographical distribution of activity in the time intervals (a) following S1-offset, (b) during the WM manipulation interval and (c) prior to S2 were evaluated in the signal (scalp potential) and source (Minimum Norm) space. Following S1-offset the ERP pattern was characterised by negativity over posterior areas, slightly more so over the right hemisphere. In the subsequent 6-s interval high WM load elicited a larger negative slow ERP than low WM load, the negativity increase due to high WM load being larger over frontal than central areas. Source modelling indicated activity in anterior areas under high, and posterior activity under low WM load. Asymmetry of activity, although indicating a shift to left-hemispheric activity under high compared to low WM load, varied considerably between subjects. Results suggest that high-resolution ERP recordings allow to examine cortical activity during WM challenge.