Abstract
We have shown that the latency of small saccades depends on the size of the attended object, being long for steps so small that the attended objects remain within the attentional window, and short for larger steps. We now use this oculomotor difference to measure the time it takes for attention to change its spatial scale, by having a stimulus change in size at different times before or after it steps right or left by 2 deg whilst we record the latency of the saccades elicited. The stimulus consisted of a rotating segmented ring either 1 or 4 deg in diameter. At the time of the size-change, the number of segments briefly changed. The subject's task was to identify the new number of segments, a task greatly facilitated by having attention at the appropriate spatial scale. To verify that the subjects set the size of their attentional field appropriately, in half the trials the size of the stimulus did not change, and we observed that the saccadic latencies were short when subjects attended to the small ring and large when attending to the large ring, regardless of the time between size-change and stimulus-step. We found that if the size-change occurred more than 200 msec before the saccade, the latencies were appropriate to the final stimulus-size, implying that the change in the size of the attentional “spotlight” had fully adjusted before the saccade was programmed. If the size-change occurred less than 100 msec before the saccade, the latencies were appropriate to the initial stimulus-size, implying that it was too late to alter the programming of the saccade. In between these values, the saccadic latencies were intermediate. We conclude that it takes approximately 100 msec for zooming of the attentional spotlight to fully affect saccade programming. Also, even after the target step occurs, there is sufficient time to adjust the attentional scale and, consequently, the saccade latency.