Human smooth pursuit: stimulus-dependent responses

Abstract

We studied pursuit eye movements in seven normal human subjects with the scleral search-coil technique. The initial eye movements in response to unpredictable changes in target motion were analyzed to determine the effect of target velocity and position on the latency and acceleration of the response. By restricting our analysis to the presaccadic portion of the response we were able to eliminate any saccadic interactions, and the randomized stimulus presentation minimized anticipatory responses. This approach has allowed us to characterize a part of the smooth-pursuit system that is dependent primarily on retinal image properties. The latency of the smooth-pursuit response was very consistent, with a mean of 100 +/- 5 ms to targets moving 5 degrees/s or faster. The responses were the same whether the velocity step was presented when the target was initially stationary or after tracking was established. The latency did increase for lower velocity targets; this increase was well described by a latency model requiring a minimum target movement of 0.028 degrees, in addition to a fixed processing time of 98 ms. The presaccadic accelerations were fairly low, and increased with target velocity until an acceleration of about 50 degrees/s2 was reached for target velocities of 10 degrees/s. Higher velocities produced only a slight increase in eye acceleration. When the target motion was adjusted so that the retinal image slip occurred at increasing distances from the fovea, the accelerations declined until no presaccadic response was measurable when the image slip started 15 degrees from the fovea. The smooth-pursuit response to a step of target position was a brief acceleration; this response occurred even when an oppositely directed velocity stimulus was present. The latency of the pursuit response to such a step was also approximately 100 ms. This result seems consistent with the idea that sensory pathways act as a low-pass spatiotemporal filter of the retinal input, effectively converting position steps into briefly moving stimuli. There was a large asymmetry in the responses to position steps: the accelerations were much greater when the position step of the target was away from the direction of tracking, compared with steps in the direction of tracking. The asymmetry may be due to the addition of a fixed slowing of the eyes whenever the target image disappears from the foveal region. When saccades were delayed by step-ramp stimuli, eye accelerations increased markedly approximately 200 ms after stimulus onset.(ABSTRACT TRUNCATED AT 400 WORDS)