The perceptual consequences and neural basis of monocular occlusions

Occluded areas are abundant in natural scenes and play an important role in stereopsis. However, due to the treatment of occlusions as noise by early researchers of stereopsis, this field of study has not seen much development until the last two decades. Consequently, many aspects of depth perception from occlusions are not well understood. The goal of this thesis was to study several such aspects in order to advance the current understanding of monocular occlusions and their neural underpinnings. The psychophysical and computational studies described in this thesis have demonstrated that: 1) occlusions play an important role in defining the shape and depth of occluding surfaces, 2) depth signals from monocular occlusions and disparity interact in complex ways, 3) there is a single mechanism underlying depth perception from monocular occlusions and 4) this mechanism is likely to rely on monocular occlusion geometry. A unified theory of depth computation from monocular occlusions and disparity was proposed based on these findings. A biologically-plausible computational model based on this theory produced results close to observer percepts for a variety of monocular occlusion phenomena.