Latency and amplitude of catch-up saccades to accelerating targets

To track moving targets, humans move their eyes using both saccades and smooth pursuit. If pursuit eye movements fail to accurately track the moving target, catch-up saccades are initiated to rectify the tracking error. It is well known that retinal position and velocity errors determine saccade latency and amplitude, but the extent to which retinal acceleration error influences these aspects is not well quantified. To test this, 13 adult human participants performed an experiment where they pursued accelerating/decelerating targets. During the ongoing pursuit, we introduced a randomly sized target step to evoke a catch-up saccade and analyzed its latency and amplitude. We observed that retinal acceleration error (computed over a 200 ms range centered 100 ms before the saccade) was a statistically significant predictor of saccade amplitude and latency. A multiple linear regression supported our hypothesis that retinal acceleration errors influence saccade amplitude in addition to the influence of retinal position and velocity errors. We also found that saccade latencies were shorter when retinal acceleration error increased the tracking error and vice versa. In summary, our findings support a model in which retinal acceleration error is used to compute a predicted position error ∼100 ms into the future to trigger saccades and determine saccade amplitude.NEW & NOTEWORTHY When visually tracking object motion, humans combine smooth pursuit and saccadic eye movements to maintain the target image on the fovea. Retinal position and velocity errors are known to determine catch-up saccade amplitude and latency, however, it is unknown if retinal acceleration error is also used to predict future target position. This study provides evidence of a small but statistically significant contribution of retinal acceleration error in determining saccade amplitude and latency.