Neural Model of Rivalry Between Occlusion and Disparity Depth Signals.

K. F. Arrington

Neuroscience Abstracts, 1995; 21, p.125.

Abstract

Stereo cells fall into tuned disparity populations (eg. Zero, Near, Far). Here it is conjectured that T-junction hypercomplex cells fall into tuned depth-relationship populations based on the depth tuning of the orthogonal complex cells that they sample for their respective post and lintel positions. The resulting T-junction populations would comprise: {Near lintel, Far post} = Occlusion, and {Far lintel, Near post} = Transparency. The case {Zero lintel, Zero post} does not seem to create a strong abutment percept, but rather is heavily biased toward an occlusion percept. Arrington & Held (1995) measured how rate and duty cycle of perceived depth oscillations vary as stereo disparity increasingly opposes the occlusion cue. Here a neural network circuit including a reciprocal inhibition oscillator is designed to test the above conjecture. The ability of the neural network to model the psychophysical data is demonstrated.

Return to CV.