Collapse of perceptual space during saccades

Purpose To determine where observers locate targets flashed briefly before, during and after saccades Methods Observers made voluntary horizontal saccades (monitored by eye-tracker) from a fixation point on the left of the screen to a target 5, 10 or 20° to the right. They were asked to locate vertical bars of luminance or colour-contrast, flashed briefly (8 ms) before or after the saccade onset, with the aid of a ruler that appeared one second later. Time 0 is defined as the beginning of the saccade. Results (1) Bars before -50 ms or after 70 ms were located veridically and accurately. (2) Bars flashed in the interval of -25 to 0 ms were located at one of two points, depending on its actual spatial position: bars that were to the left of the initial fixation point were seen at that fixation point, while all other were seen at the target. (3) There was a systematic trend towards veridicality as the saccade progressed. (4) Two bars (one upper, one lower) flashed simultaneously at different positions during the interval of -25 to 0 ms were seen as collinear if both were to the right or to the left of fixation. (5) Temporal separation of the two bars could induce apparent spatial separation. Conclusions. Saccades can cause an apparent displacement either in the direction of the saccade or in the opposite direction, depending on the spatial position of the stimulus. These results cannot be explained by a general coordinate shift, continuous or otherwise, since the amount and direction of error in locating a bar depends on its spatial as well as its temporal position. They indicate a collapse of perceptual space along the direction of the saccade towards two attractors, the pre- and post-saccadic fixation points, starting about 25 ms before saccades begin. An attention driven model, incorporating a spatial collapse and a progressive recovery during saccades will be presented.