The ability of an animal to respond to sensory events is limited by reaction time, or latency. The latency of smooth pursuit is approximately 130 msec, shorter than button press latency (~250 msec), or that of rapid, saccadic eye movements (~180 msec). Nevertheless, a lot can happen in this time. Consider a baseball thrown by a good pitcher. At a speed of 90 miles/hour, in 130 msec the ball will travel almost one third of the distance from pitcher to batter. A rattlesnake can strike, and even withdraw its fangs in 130 msec (9). Our laboratory is interested in how movement systems compensate for latency constraints. The pursuit system uses the strategy of anticipating predictable motion. Note above that anticipatory eye velocity is high when the direction the target moves is held constant (green traces). However, some anticipatory eye velocity still occurs when target direction is randomized (orange traces). These residual eye movements are due to cognitive expectations about the direction the object will move or memory of target motion from recent trials (10; 11; 12).

Our laboratory also studies factors that influence when movement is initiated, i.e., movement initiation timing. A starting hypothesis is that smooth pursuit initiation is limited simply by neuronal conduction delays between the retina and the oculomotor control mechanism. However, anticipatory eye velocity occurs before the target moves. Since anticipatory pursuit is guided by cognitive factors, it could be initiated by a different neural substrate than visually-guided pursuit. However, we have evidence that a single internal timing mechanism can limit pursuit initiation even when a visual motion signal is available. Our experiments were done by randomizing fixation duration, and hence the timing of target motion onset. This initial goal of this work was to determine whether the internal timing mechanism is disengaged under random timing conditions. In fact this mechanism is very active despite randomization, as can be seen below.