Abstract
The use of visual motion signals in the control of eye movements can be investigated by “opening the loop” linking eye movements and retinal image motion. Goldreich, Krauzlis and Lisberger (1992) introduced artificial delays in visual feedback during pursuit in monkeys, and observed oscillations in velocity, whose frequency varied systematically with delay. Using a dual-Purkinje image eye tracker equipped with stimulus deflectors, we have applied the delayed feedback paradigm to humans. By feeding back the eye movement signals to the stimulus deflectors, we are able to transiently stabilize a target, delivering the effect of the eye movement after a delay period. For moderate delays (greater than 100 msec), this results in sinusoidal oscillations of the eyes, with the period of oscillation varying linearly with the artificial delay. Although feedback is applied in all four dimensions (vertical and horizontal vergence and version), the oscillations are attracted to horizontal version, presumably because this type of movement has the highest pursuit gain. Extrapolating the period-vs-delay function to its x-intercept allows estimation of the internal delay, while the slope of the function is indicative of the control mechanism. Goldreich et al. observed slopes between 2 and 4, which are consistent with their model in which eye acceleration is controlled by a combination of retinal velocity and acceleration signals. In our data, on the other hand, the period of oscillation increases much more slowly with delay, with slopes around 1.5. The results can be explained by a model in which eye acceleration is controlled by a mixture of velocity and position errors.
Supported by NASA Aviation Safety Program and NEI RO1-EY12986