Oxyopia Abstract

January 30, 2004
Friday, 4 PM
489 Minor Hall

Bruce Cumming, MD, PhD
Laboratory of Sensorimotor Research, National Eye Institute
Host: Dennis Levi

Title

"The neural substrate of the Pulfrich effect: Perhaps he was right."

Abstract

If a moving object is viewed while the input from one eye is delayed by placing a filter over that eye, a sensation of depth results (the Pulfrich effect). Modern explanations of this phenomenon depend upon the idea that motion and depth (binocular disparity) are jointly encoded early in visual processing. In the primate visual pathway, the earliest point at which selectivity occurs for either direction of motion or disparity is the primary visual cortex (V1). We measured the responses of disparity selective neurons to random dot stereograms over a range of interocular delays. The great majority of neurons showed no evidence of joint motion-depth encoding.

This led us to reevaluate the evidence in favor of joint encoding. The strongest evidence has come from studies in which the Pulfrich effect is produced under stroboscopic illumination. Here the image in one eye appears a few ms after the other eye, but at an identical location. The spatial disparities are therefore zero in this display (despite the time delay). That depth changes are still experienced was explained by invoking joint encoding of motion and depth. We now show that the same phenomenon can be explained by early mechanisms that are insensitive to the direction of motion, provided that they integrate input signals over finite times. This explanation makes quantitative predictions of the magnitude of depth induced by stroboscopic displays. For small time delays stroboscopic presentation should produce smaller depth percepts than traditional displays. As the time delay gets larger (up to half the strobe period), the induced depth should become more similar in the two displays. We measured perceived depth in computer generated displays with a nulling procedure, and found exactly this pattern. Data from a large number of configurations (with different delays and different strobe periods) can be explained by a simple model assuming an integration time of around 20ms. This is similar to the integration time observed in cortical neurons. Taken together, these observations indicate that the Pulfrich effect, even in its stroboscopic form, is largely the result of the geometrical consequences of the interocular delay f or disparity detectors, rather than a consequence of joint motion-disparity encoding.