Physiological mechanisms of task-switching in human subjects.

Abstract

The nature, and timing, of the cerebral processes that are active when attention is switched between different tasks are not understood. The purpose of this research was to establish electrophysiological evidence for a two-stage, posterior-anterior cerebral processing model for the control of attention switching. Reaction Times (RTs) and Event-Related Potentials (ERPs) were recorded from 22 healthy young adults as attention was cued to switch between two visuomotor tasks. One task ("horizontal") involved determining whether a circle in one of the four boxes of a 2 x 2 grid was in the left or right half of the grid whereas the other task ("vertical") involved determining whether it was in the upper or lower half. Cues designating the appropriate task occurred 200, 1200 or 1500 ms before a target. Cues were either letters (H & V) at the center of ocular fixation, or arrows ( ⇕&⇔ ) at the periphery. The identity of the letters, and spatial location of the arrows, informed subjects what task to perform, what hand to respond with, and whether the task was the same (repeat) or different (switch) from the previous trial. The RT was longer for switch than repeat trials but only during short (200 ms) cue-target intervals. This demonstrates that subjects are able to completely switch attention prior to target stimuli when the cue-target interval was 1200 or 1500 ms. The cues evoked a sequence of potentials that were larger in the switch trials than in the repeat trials: an occipital N200, a parietal P390, and a mid-frontal negative wave with a latency between 400 and 800 ms. The N200 probably represents processing of the stimulus in the extrastriate cortex. The P390 peak was larger for arrow than letter cues at centroparietal electrodes (CP1, CP2), and the inverse was true at temporal-parietal sites (P7, P8). The P390 was also 55 to 59 ms earlier for the dorsal than ventral waveform. This demonstrates that at least two separate neurophysiological events contribute to the amplitude and latency characteristics of the scalp recorded P390. These processes are specific for the physical features of the cues and may correspond to the dorsal "where" and ventral "what" visual streams. Only the mid-frontal negative wave was found to be specifically with the switch in task. This wave may represent activity in the supplementary motor area or anterior cingulate as response rules are changed for the new task. The readiness potential (RP) showed complex relations to switching or repeating the task. In general, this potential was larger over the hemisphere contralateral to the hand that was being prepared for response. When there was some urgency in the task, the readiness potential was bilateral on switch trials, perhaps because the previous hand was automatically activated in case it might be needed. Left and right lateral pre-frontal slow waves occurred throughout each trial. These may represent task monitoring and/or working memory processes. These results suggest that both posterior and anterior brain regions participate in attention switching. Posterior ERPs seem to be associated with identifying the physical features of stimuli that signal the need to switch attention between tasks, whereas the mid-frontal negativity appears to be related to carrying out the switch once the need to switch has been identified. The RP indicates the preparation of a hand for response. The strategy of this preparation varies with the urgency of the switch. Other anterior processes may monitor task performance.