| The focus of this research is on the role of depth perception in motion processing during visually guided navigation. In seven psychophysical experiments, we investigated the use of depth cues in the segmentation of the visual scene, detection of object trajectories and the perception of self-motion.;Chapters 1 & 2 provide the motivation and background for this work. In Chapter 3, we determine the behavioral and computational implications of the disparity tuning properties of neurons in the middle temporal area (MT), an area critical for visual motion processing, on disparity segmentation, motion transparency detection and center-surround modulation. We compared psychophysical performance to an ideal observer model and showed that when presented with multiple motion components with small differences in direction and binocular disparity, the visual system implements a smoothness constraint, leading to the grouping of similar features into a single percept. Two biologically constrained models, one based on the population statistics of MT neurons and one using actual neuronal recordings (courtesy of G. DeAngelis), demonstrate that these neurons can segment motion using binocular disparity, and that through dynamic read-out the function of the network varies with the task demands.;In Chapters 4 and 5, we investigate the perception of motion in depth via angular expansion and scale change cues. First, we showed that time-to-collision estimation based on angular expansion and two-dimensional, gap closure, cues use functionally separate mechanisms. Second, we showed that scale changes---the change in spatial frequency associated with changes in relative depth---can be used for the detection of heading, but not of object motion trajectories. Results from an adaptation study and from a patient blind to radial motion support the hypothesis that scale changes are processed independently from optic flow and may constitute an alternative mechanism for the perception of self-motion.;Taken together, these results demonstrate a functional specialization of depth cues during motion perception. We show that the specific depth cue used by the visual system depends on the task, and that a number of functionally separate mechanisms exist to accurately process the depth information readily available in the visual world. |
