Relationship between eye acceleration and retinal image velocity during foveal smooth pursuit in man and monkey.

Publisher Website
Google Scholar
Abstract
The basic input-output relationships underlying primate smooth-pursuit eye movements (SP) were studied in order to elucidate the neural mechanisms that generate SP. Humans and [rhesus] monkeys attempted to track periodic, nonperiodic and velocity-step target movements. Retinal error velocity (REV) was calculated as target velocity minus SP eye velocity. For sinusoidal target movements, SP gain was close to unity for target movements having low frequency and small peak-to-peak amplitude. Gain decreased when either the frequency or amplitude of target movement was increased. Three humans achieved maximum eye velocities of 115, 121 and 150.degree./s. Monkeys reached maximum eye velocities of 160.degree./s. The phase shift by which eye movement lagged target movement increased as a function of the frequency of target movement, but was independent of the amplitude of target movement. Two basic relationships accounted for SP at all frequencies (0.3-2.1 Hz) and amplitudes (.+-. 5 to .+-. 30.degree.) of target movement. SP gain was a consistent function only of maximum target acceleration, despite wide variations in the frequency and maximum velocity of target motion. Only maximum eye acceleration (and not maximum eye velocity) was a consistent function of REV. The same 2 basic relationships described the SP produced by nonperiodic target movements consisting of a random walk of sinusoids. However, SP gain at any given maximum target acceleration was much lower than for pure sinusoidal target movement. Eye acceleration associated with a given REV was lower than for pure sinusoidal target movement. For step changes in target velocity, the average eye acceleration in the first 120 ms of the response increased as a function of REV, but was independent of target velocity or final eye velocity. SP during head movement was tested by requiring subjects to fixate a small target that rotated exactly with them during sinusoidal vestibular stimulation. The gain of the visual contribution to the resulting eye movement was near 1.0 for low frequencies of head oscillation. Gain decreased when either the frequency or peak-to-peak amplitude of oscillation was increased. For each subject, the characteristics describing the gain and phase shift of the visual contribution to combined eye-head tracking were similar to those describing SP gain and phase during head-fixed tracking. SP operates in the same qualitative way, regardless of the trajectory or predictability of target movement and whether or not the head is turning. SP gain is determined by maximum target acceleration; eye acceleration is the direct result of target motion across the retina. These relationships must relfect the internal neural mechanism underlying SP and are compatible with the idea that the SP system contains a positive feedback pathway. Visual inputs would be expected to cause a change in eye velocity, which would be measured as eye acceleration.