Temporal Dynamics of Binocular Display Processing with Corticogeniculate Interactions
MetadataShow full item record
A neural model of binocular vision is developed to simulate psychophysical and neurobiological data concerning the dynamics of binocular disparity processing. The model shows how feedforward and feedback interactions among LGN ON and OFF cells and cortical simple, complex, and hypercomplex cells can simulate binocular summation, the Pulfrich effect, and the fusion of delayed anticorrelated stereograms. Model retinal ON and OFF cells are linked by an opponent process capable of generating antagonistic rebounds from OFF cells after offset of an ON cell input. Spatially displaced ON and OFF cells excite simple cells. Opposite polarity simple cells compete before their half-wave rectified outputs excite complex cells. Complex cells binocularly match like-polarity simple cell outputs before pooling half-wave rectified signals frorn opposite polarities. Competitive feedback among complex cells leads to sharpening of disparity selectivity and normalizes cell activity. Slow inhibitory interneurons help to reset complex cells after input offset. The Pulfrich effect occurs because the delayed input from the one eye fuses with the present input from the other eye to create a disparity. Binocular summation occurs for stimuli of brief duration or of low contrast because competitive normalization takes time, and cannot occur for very brief or weak stimuli. At brief SOAs, anticorrelatecd stereograms can be fused because the rebound mechanism ensures that the present image to one eye can fuse with the afterimage from a previous image to the other eye. Corticogeniculate feedback embodies a matching process that enhances the speed and temporal accuracy of complex cell disparity tuning. Model mechanisms interact to control the stable development of sharp disparity tuning.