The spatial extent of attention for saccades: Attentional facilitation compared to inhibition of return in humans and monkeys

The aim of the current work was to investigate the spatial extent of attentional facilitation and inhibition of return (IOR) in response to an exogenous cue over time. We used saccade latencies as a behavioral correlate of attentional allocation. Two humans and two monkeys made saccades to visual targets at 136 locations spread across the visual field (9° up/down/left/right) from a central fixation target. In 80% of the trials, a behaviorally irrelevant cue was flashed at one of 4 oblique locations (7° eccentricity) either 50 or 200 ms before the target. In the remaining 20% of the trials, no cue was presented but the same timing was maintained. We calculated average saccade latencies as a function of distance from the cue across all four cue locations normalized relative to saccade latencies when no cue was present. Attentional facilitation was seen when the cue preceded the target by 50 ms. The facilitation was confined to targets presented within the same quadrant as the cued location. In contrast, when the cue preceded the target by 200 ms, we found an inhibition that extended from the quadrant with the cued location to the two adjacent quadrants. The greatest increase in latencies occurred in the cued quadrant with a smaller increase in the two orthogonal quadrants. These differences were similar for both humans and monkeys. These findings suggest two separate mechanisms for attentional facilitation and IOR, with the cue resulting in a much more localized spread during attentional facilitation than during IOR. This is consistent with a low-level local network for attentional facilitation with smaller receptive fields (e.g. superior colliculus) vs. a higher-level network for IOR involving large receptive fields (e.g. parietal or frontal areas).