Application of a Novel Neural Approach to 3D Gaze Tracking: Vergence Eye-Movements in Autostereograms

Vergence eye-movements occur not only in real environ- ments, but also in virtual 3D environments. Autostere- ograms can cover large visual angles without requiring vergence beyond natural parameters and are thus well- suited for the investigation of vergence movements in virtual 3D images. We developed an anaglyph-based 3D calibration procedure and used a parametrized self- organizing map (PSOM) to approximate the 3D gaze point from a subject's binocular eye-movement data. Besides analyzing the general pattern of vergence eye- movements in autostereogram images, the present study examined the in∞uence of image granularity on these movements. Unlike previous research on random-dot stereograms, we found substantially overshooting con- vergence eye-movements, especially for medium gran- ularities. Moreover, divergence movements were com- pleted more quickly for coarse than for One granularities. Results are discussed in the context of granularity eAEects on autostereogram perception and the dissociation be- tween convergence and divergence eye-movements.