Abstract
The eye-movement latency to a randomly moving target varies with target color and contrast (ARVO '98). Here we confirm the results for achromatic stimuli, and extend them to binocular viewing. Subjects viewed a display through the optics of a binocular dual-Purkinje image eye tracker, and independent motion trajectories were presented to each eye. Four targets were used: 1) a simple spot; 2) a slowly-changing textured field; 3) binocularly correlated dynamic random noise; and 4) a second-order stimulus consisting of binocularly uncorrelated dynamic random noise that was contrast-modulated with a binocularly correlated bullseye pattern. Independent temporal responses for version and vergence in the horizontal (H) and vertical (V) directions were obtained by reverse correlating the measured eye velocity with the corresponding stimulus velocity, and response latency was estimated using the time of peak correlation. The H version response is generally the fastest (100 msec), followed within 10–20 msec by V version. Vergence responses are delayed by an additional 50–70 msec, with V vergence 10–20 msec faster than H vergence. Responses to the spot are delayed 20 msec with respect to the response for an extended texture, while the pure disparity stimulation provided by the dynamic noise (vergence only) produces an additional delay of 50–80 msec. Varying texture contrast produces a delay of approximately 50 msec / log unit, for all types of eye movement. The pure second-order stimulus produces a response delayed approx. 100 msec (compared to the same stimulus with a small amount of luminance contrast added to the modulator), except for the case of vertical vergence, which does not respond at all to the second-order stimulus. The results are consistent with a neural architecture in which the contrast-dependent delay is introduced before the segregation of binocular visual signals, and multiple visual mechanisms contribute differentially to each type of movement.