Abstract
It is widely held that the targets of saccades (and perhaps of attention as well) are typically selected by a winner-take-all mechanism in which maximal neural activity on some brain map (perhaps in the superior colliculus or frontal eye field) converges on one location, inhibiting other locations. If circumstances permitted the brain regions in each hemisphere to separately process the candidate stimuli, might target selection for saccades occur more efficiently than if the alternatives were processed within the same hemisphere? In humans, there is some evidence that attention can be more easily split between the right and left hemifields than between two locations in the same hemifield (Sereno & Kosslyn, 1991; Awh & Pashler, 2000; McMains & Somers, 2004). We asked whether saccadic reaction times showed evidence of separate foci of attention.
Subjects were instructed to attend to two 5-deg stimulus patches 10 deg apart, and to make a saccade to the one that changed. The stimuli were a rotating colored grating, which changed from smooth rotation to rotation in 16 equal-angle steps, and a field of dots moving randomly, in which the lifetime of each dot changed from 75 msec to 375 msec. In different experimental blocks the two stimuli were the same or different, and were either located 3 deg below the fixation point, one to the left and one to the right, or 3 deg to the right of the fixation point, one above and one below. Across three subjects and four stimulus conditions, saccade latencies were shorter in 11 of 12 cases if the two stimulus patches were in opposite hemifields than if both stimuli were in the same hemifield, regardless of the nature of the stimuli (mean difference, 74 msec). This suggests that more attentional resources can be allocated to two locations if they are in different hemifields than if they are in the same hemifield.