Fast Perception of Binocular Disparity

Seminar | April 17 | 12-1:30 p.m. | 489 Minor Hall

 Benjamin Backus, State University of New York

 Neuroscience Institute, Helen Wills

How quickly can binocular disparities be converted into perceived depth? Classic experiments by Barbara Gillam and colleagues showed cases in which stereopsis took many seconds, but in retrospect their task and stimulus design were problematic. Rapidly changing disparities are perceptually difficult to track, which also suggests that stereopsis is generally slow, but waiting for integration might not be obligatory for a freshly presented stimulus. We directly compared speed-accuracy tradeoff functions (SATFs) between two forced-choice discriminations: one based on stereoscopic depth, and one based on luminance. Unexpectedly, both SATFs deviated from chance levels of accuracy at the same response time—approximately 200 ms. Thus, the initial processing of disparity for perceived depth did not take any longer than the initial processing of luminance for perceived brightness. However, for some observers stereo accuracy increased more slowly than luminance accuracy after this initial delay. Stereoscopically defined surface slants were also quickly discriminable as compared to slant discrimination based on texture cues, especially for slants of small magnitude. These findings show that that binocular disparities are perceived as depth very quickly, so that disparity may be more widely important for everyday visual function than previously thought. The search is now on for experiments that can reveal uses to which stereo is put in everyday situations. We tested one such use: the recognition of objects in peripheral vision. Stereo facilitated object recognition both by attracting covert attention to the object’s location, and, to a lesser extent, by specifying its contour.

 CA, nrterranova@berkeley.edu