Anti-correlated Random-dot Stereograms Research Articles

Responses of primary visual cortical neurons to binocular disparity without depth perception.

The identification of brain regions that are associated with the conscious perception of visual stimuli is a major goal in neuroscience. Here we present a test of whether the signals on neurons in cortical area V1 correspond directly to our conscious perception of binocular stereoscopic depth. Depth perception requires that image features on one retina are first matched with appropriate features on the other retina. The mechanisms that perform this matching can be examined by using random-dot stereograms, in which the left and right eyes view randomly positioned but binocularly correlated dots. We exploit the fact that anticorrelated random-dot stereograms (in which dots in one eye are matched geometrically to dots of the opposite contrast in the other eye) do not give rise to the perception of depth because the matching process does not find a consistent solution. Anti-correlated random-dot stereograms contain binocular features that could excite neurons that have not solved the correspondence problem. We demonstrate that disparity-selective neurons in V1 signal the disparity of anticorrelated random-dot stereograms, indicating that they do not unambiguously signal stereoscopic depth. Hence single V1 neurons cannot account for the conscious perception of stereopsis, although combining the outputs of many V1 neurons could solve the matching problem. The accompanying paper suggests an additional function for disparity signals from V1: they may be important for the rapid involuntary control of vergence eye movements (eye movements that bring the images on the two foveae into register).

Depth in anticorrelated stereograms: Effects of spatial density and interocular delay

Disparity-based depth is not perceived in densely textured, anticorrelated random-dot stereograms (RDSs) whose elements carry opposite signs of brightness coatrast on corresponding loci, as extant data show. We observed global depth in anticorrelated RDSs flashed repetitively with an interocular delay. During the delay time, a dot array in one eye was paired with a gray frame hi the other eye and thus could interact with the negative afterimage of the contralateral dot array. A correlated RDS (e.g. 8 min arc dots, 50% density, 15-msec flash duration) lost depth with delays > 45 msec. An anticorrelated RDS, that was otherwise identical, showed robust depth when flashed with an interocular delay of some 60 msec. A delay was not always necessary to produce depth. At low dot density (1–2%), anticorrelated RDSs showed disparity-dependent local depth even when displayed continuously, or flashed simultaneously; as dot density alone was increased, depth was progressively lost. To make global depth visible in a dense RDS flashed with an interocular delay, the internal response had to be strongly biphasic. Our results support the generally held notion that cyclopean depth signals emerge exclusively from same-sign binocular cortical filters. However, the exclusionary rule may be invalid with respect to the processing of coarse local depth with figural stimuli. Relative depth between a pair of small dots was easily perceived when one of the dots was in opposite contrast, but the depth threshold was then about 0.5 log unit higher than with the same-contrast pair of dots indicating that the internal effects of contrast have not all lost their sign prior to binocular disparity processing. It remains to be determined whether depth can be perceived from edges of opposite contrast.