A study of human smooth pursuit eye movements has been performed in order to clarify the velocity information processing capabilities of the visual system. A set of stimuli were designed which, when presented in motion to the visual system, contained no position information. Thus, the velocity sensitive pursuit system was stimulated in isolation from the saccadic system, which is position sensitive. The smooth eye movements which were elicited by step increases in target velocity from zero velocity were analyzed in detail by a nonlinear least squares curve fitting procedure. Eye velocity was found not to exactly match stimulus velocity, the differences being unsystematic with velocity amplitude or direction. Response latency and the duration of eye acceleration were found to vary about average values of 150 and 300 msec, respectively. Averaged velocity and acceleration responses were computed from first and second derivatives of the position responses to repeated presentations of the step velocity target motion. Response average acceleration pulse amplitude did not vary linearly with stimulus acceleration pulse amplitude. Increases in stimulus pulse amplitude were met with changes in amplitude, shape, and duration of the response pulse. Averaged acceleration responses to two consecutive stimulus velocity steps were shown not to correspond with the response predicted by the linear superposition to two single pulse responses. The departure of the actual double pulse response from the predicted response was discussed in terms of possible sources of a such a nonlinearity. The results of the single and double pulse experiments were used to design a pseudo-random stimulus acceleration signal which was subsequently used in a white-noise type of nonlinear analysis of the pursuit system. The first and second order pursuit system kernels which were obtained from the analysis showed that the technique could be used to study interactions within the pursuit-perceptual system. The results were not reliable to the extent that a mathematical model could be constructed. However, several suggestions for improving the pseudo-random input analysis technique were included in the conclusion chapter. Finally, the experimental results were discussed in terms of the relationship between visual perception and the control of smooth pursuit eye n1ovements.
R. Stanley Williams (Thu,) studied this question.