Neural Basis for Stereoscopic Vision

Research Overview

With Hermann Wagner (Aachen University) and David Heeger (Stanford University), I am working toward a neural model for the processing of binocular disparity and 3d depth in visual cortex. Our goal is to explain the binocular interaction of cells in the primary visual cortex from a functional perspective, in terms of the computation and representation of binocular disparity.

The current model involves linear neurons and energy neurons, interocular position-shifts and/or phase-shifts, monocular and binocular normalization, pooling in local spatial neighbourhoods, and pooling across orientation- and scale-specific channels. The basic computational framework was derived as a modified form of {\sl phase-correlation}. This work reported a number of theoretical findings, including the fact that conventional energy models are not disparity detectors (as they regularly respond strongly to false matches), and some specific predictions concerning how one might measure the source of disparity selectivity in V1 neurons, and how one might construct disparity detectors from their outputs.

I am continuing this work in collaboration with Wagner and his students, with neurophysiological experiments on awake behaving owls underway to test predictons of the model.

Related Publications

Return to David Fleet's home page.