The dynamics of smooth pursuit eye movements in normal human subjects were evaluated with the search coil method. The influence of the position and velocity of a target were investigated by using a computer-controlled spot of light to present these stimulus features separately and in various combinations. The combination of search coil recording and computer control of the experiments allowed us to obtain far higher resolution of the early part of the pursuit response than has previously been reported. The responses to all stimuli were surprisingly machine-like, with a latency of only 100 plus/minus 4 (SD) ms. The accelerations of the initial part of the response were low (40-50 degrees/s2) and nearly independent of stimulus velocity in the range 5-40 degrees/s. Furthermore, a change in position alone, without an attendant change in velocity, produced comparable accelerations and latencies. Accelerations of the later parts of the response were much higher and depends strongly on stimulus velocity. When position and velocity information were in conflict, the response was briefly toward the position component and then reversed toward the velocity component. These observations, along with data collected under an open-loop condition, lead us to surmise that there are several distinct phases--and probably mechanisms--present in the pursuit system. We are currently extending these experiments to stimuli that cover most of the field of vision, and also to stimuli that require adaptation of the system in order to accurately track the target. These findings have allowed us to model the pursuit system in humans in more detail and to make predictions about the tracking deficits that might be present with failure of the various sub-systems. We plan to evaluate these individual components in patients with pursuit deficits to identify the nature and possible location of their abnormality.
