Abstract
In everyday life, prediction plays a critical role in ocular motor target tracking. The ocular motor system employs a mixture of saccades and smooth pursuit across version and vergence eye movements to successfully follow objects as they move in the world. However, these eye movements are most often studied in isolation, saccades separate from smooth pursuit, version separate from vergence. Here we examined the perception/prediction of motion trajectories and how different types of eye movements are employed to coordinate the ocular motor tracking of those targets. Eye movements were recorded with an Eyelink 1000 (SR Research) at 500 Hz. The stimuli were presented using a PROPixx projector (VPixx Technologies) and an active circular polarizer, with subjects wearing passive circular polarizing glasses. The screen was set at a viewing distance of 70 cm. Cartoon images of angular size 2.2° moved with horizontal trajectories (initiating version eye movements) or motion-in-depth trajectories (initiating convergence and divergence). The motion trajectories in the predictable condition were sinusoids of varying amplitudes (5, 10 and 20 cm) and temporal frequencies (.25, .5, and 1 Hz). The unpredictable trajectories were smoothed Brownian random walks in position (sigma = 0.1 cm, 0.2 cm, and 0.3 cm). We measured the number of saccades across all conditions. In the horizontal motion condition, there were systematic increases in the number of saccades with increasing trajectory amplitudes and temporal frequency. For the motion-in-depth condition, participants made similar numbers of saccades regardless of the trajectory amplitude and frequency. There were no consistent differences between saccade behavior in the predictable vs. unpredictable conditions. A cross-correlogram analysis of the unpredictable trajectory condition revealed a longer latency (µ=77ms, st.d.=24ms) for vergence responses compared to version responses. These data demonstrate the complexity in predictive computation for tracking responses involving multiple types of eye movements in naturalistic 3D tasks.