Abstract
In the laboratory stereoscopic perception is often slow, and perhaps for this reason its contribution to early visual processing has been neglected. It has long been known that stereoscopic depth can be perceived in briefly presented displays, but early work did not employ masking procedures or measure response times, and thus could not address how quickly disparity signals become available. We conducted experiments in which the task was to choose which side of a stereo LCD display contained a disk with 5 arcmin crossed disparity relative to background and fixation (dense RDS, disk diameter 2 deg, eccentricity 5 deg). The other location had a disk with 5 arcmin vertical disparity. The disks were displayed for 17, 34, 50, or 67 ms, and were followed immediately by a mask that remained until response. The disks were presented against a zero-disparity background. The mask was flat with 0 or 5 arcmin disparity, or had two disks in crossed or uncrossed disparity, or was binocularly uncorrelated. Accuracy was above chance at 17 ms for some conditions and increased with display duration. Accuracy was similar in most conditions; it improved when no mask was used. Accuracy was degraded when relative disparity was impoverished by removing the background. Accuracy was at chance for the crossed-disks mask. Mean response times were 300-400 ms; RT was lower (with accuracy still above chance) when the task was speeded. We conclude that relative disparity is made available quickly, while the process of constructing perceived depth has a long integration window. We measured this window by introducing a variable duration blank screen between the stimulus and the crossed-disks mask, with stimulus duration fixed at 50 ms; the integration window was more than 150 ms in rough agreement with published estimates measured using sinusoidal depth modulation in time.
Meeting abstract presented at VSS 2012