Abstract
Analyzing and predicting visual object motion is important for action and interactions. To successfully track a moving target, the oculomotor system has to take into account internal processing delays of 100 ms: the target will be already at a different location when the eyes start to move. To investigate predictive mechanisms of the oculomotor system under different conditions we measured tracking responses to vertical 10 deg target steps to the center, followed by linear ramp movements into one of the four cardinal directions (speed 10, 15, 20 deg/s). A single blob target appeared on a neutral gray background; it was white, gray or isoluminant red for testing under high and low luminance or color contrast conditions. We compared the dynamics of initial directions and curvatures of saccades for the different conditions and found that for high and low luminance targets initial saccade directions changed with their latencies, suggesting a continuous access to updated target movement predictions. For isoluminant targets, this directional updating was much weaker presumable due to poor motion signals. The comparison of the initial directions of saccades with the optimal direction based on the target position at saccade offset revealed that saccades to high and low luminance targets tended to initially overestimate target speeds. However, saccadic end points were quite accurate because saccadic curvatures scaled with initial directional errors for correction. These dynamic adjustments of saccadic movements suggest that the oculomotor system continuously updates predictions about target movements. In contrast to the classical view of saccade programming with a saccadic deadtime with limited correction possibilities from 80 ms before saccade onset, we found that corrections for the initial direction are still possible at least until saccade onset. Based on refined target movement predictions initial direction errors of saccades can still be corrected by adjusting their movement curvatures.
Meeting abstract presented at VSS 2018