Predictive elements in ocular interception and tracking of a moving target by untrained cats.

Abstract

We presented a mechanical target moving at constant velocity to awake, nontrained, head-restrained cats, in order to study how naive animals pursue objects moving at a high speed with their gaze. Eye movements were recorded while the target was moving in different directions at a constant velocity (20-80 degrees/s) through the center of the visual field. We observed two oculomotor strategies: cats either made an interception saccade (IS) toward the target but opposite to its motion, or tracked it in the direction of motion. They used the interception strategy more frequently when the gaze position error at the onset of target motion was large, and the tracking strategy when it was small. Interception was always achieved by single saccades, which were faster than tracking saccades (TS). During tracking, cats generated sequences of two to six saccades separated by "smooth" eye movements. Tracking quality varied considerably from trial to trial. When the level of motivation was high, cats would track the target at 80 degrees/s over up to 75% of the oculomotor range, with relatively small position errors. We compared ISs and TSs with respect to their metric properties and timing. The amplitudes of ISs positively correlated with position error existing 100 ms before saccade onset, but saccade vectors were directed to a point ahead of the target along the target's track. We conclude that, in programming the ISs, target motion is used to predict the future target position so as to assure a spatial lead of the gaze at the saccade end, instead of attempting a precise capture of the target. The amplitude of TSs did not depend on preceding position errors. TSs were usually small at the onset of the first saccade, as if cats would wait till the target arrived near the line of sight. A majority of primary TSs were initiated before the target arrived near the direction of gaze. Thus they had a direction, opposite to the position error sampled 100 ms before the saccade, but the same as the direction of target motion. Prediction of the future target position from its velocity vector should therefore contribute to the programming of TSs. In addition, we observed that TSs were faster when they were initiated with a spatial lag relative to the target and they were slower if there was a spatial lead or target velocity was reduced. Such a modulation appears to be analogous to the predictive correction of the saccade amplitude during smooth pursuit in primates. Considering strong visual motion sensitivity and motor properties of output neurons of the superior colliculus, it is likely that, in cats, the colliculus makes a major contribution to the integration of eye movement-related and target motion-related signals.