Dual-task costs and benefits in anti-saccade performance.

It has been reported that anti-saccade performance is facilitated by diverting attention through a secondary task (Kristjánsson et al. in Nat Neurosci 4:1037-1042, 2001). This finding supports the idea that the withdrawal of resources that would be taken up by the erroneous movement plan makes it easier to overcome the tendency to look towards the imperative stimulus. We first report an attempt to replicate this finding. Four observers were extensively tested in an anti-saccade paradigm. The luminance of the fixation point or peripheral target was briefly increased or decreased. In the dual-task condition observers signalled the direction of the luminance change. In the single-task condition the discrimination stimulus was presented, but could be ignored as it required no response. We found an overall dual-task cost in anti-saccade latency, although some facilitation was observed in the accuracy. The discrepancy between the two studies was attributed to performance in the single-task condition. For latency facilitation to occur, performance should not be affected by the discrimination stimulus when it is task-irrelevant. We show that naive, untrained observers could not ignore this irrelevant visual event. If it occurred before the imperative movement signal, the event acted as a warning signal, speeding up anti-saccade generation. If it occurred after the imperative movement stimulus, it acted as a remote distractor and interfered with the generation of the correct movement. Under normal circumstances, these basic oculomotor effects operate in both single- and dual-task conditions. An overall dual-task cost rides on top of this latency modulation. This overall cost is best accounted for by an increase in the response criterion for saccade generation in the more demanding dual-task condition.