Light and dark adaptation.

The high-frequency cutoff of the semicircular canals is unknown but probably also lies above 5 Hz. The high-frequency cutoff of the extraocular muscles is 1 Hz. so that central mechanisms (that introduce phase lead) must compensate from 1 to 5 Hz. The pursuit system begins to cut off between 0.5 to 1 Hz. so that it can compensate for slow head movements while the vestibular system compensates for rapid head movements. RATLIFF: In some cases your integrator seemed to be ideal and in other cases leaky. ROBINSON: The integrator is not likely to be ideal. Its time constant (of the leak) need only be larger than a typical slow phase duration of nystagmus, say 2 to 5 sec. Its time constant may be revealed by the fact that, on eccentric fixation in the dark, the eyes tend to drift back toward the primary position at a velocity which suggests a leak time constant of 10 sec. RODIECK: Isn't it curious that in an animal with a fovea one finds less directional selectivity in colliculus and lower visual centers than one finds in afoveate animals? ROBINSON: Yes. In lower animals it may be that directionally selective units form a special subset of ganglion cells that are used to stabilize the eye with respect to the visual world but do not participate in form discrimination. Lack of such cells in foveate animals would suggest that this system has been abandoned in favor of an equivalent system which utilizes the more complex processing of the visual cortex. This observation is a serious problem in trying to draw an analogy between Collewijn's model for the rabbit's ocular stabilizing system and the smooth pursuit system of man. LEVICK: The output of these directional selective units is a signal which corresponds to the velocity of the retinal image. So that if you really want the system to stay put you can't use these units because they don't go down to dc. Couldn't you consider the concentric units which do give you absolute position? ROBINSON: The directionally selective on-units in the rabbit detect velocities of 0.01 degrees per second and stabilize the eye drift to comparable values. If the drift disturbance were unidirectional (it usually isn't) the eye would travel six minutes of arc in ten seconds or about 0.5 degree in a minute. Since the drift disturbance is not unidirectional, the long-term drift of mean eye position is so slow that it would be difficult to distinguish it from a true dc mechanism. All one can say from Collewijn's experiments is that the directionally selective units appear to be capable of explaining the observed behavior. This, of course, does not exclude the possibility of an absolute position system based on concentric units.