| The capacity to perceive depth is critical for any observer who wishes to meaningfully interact with their surroundings. Previous studies of the neural basis of depth perception have focused on how neurons represent binocular disparities, but the relative image motion that arises as a result of observer movement (motion parallax) provides an equally potent cue for depth perception. This thesis describes how single neurons in macaque cortical areas MT and MST are able to signal depth from motion parallax.;First, we used a virtual-reality setup to test the ability of single neurons to signal depth-sign (near or far relative to fixation) from motion parallax in the absence of other pictorial cues to depth. By fashioning monocular retinal stimuli that by themselves were ambiguous with respect to depth, we created a situation in which depth-sign could only be resolved by the combination of visual motion and an extra-retinal signal related to the observer's movement. Two-thirds of the neurons in MT (and a smaller percentage in MST) are able to do just that, and therefore may be a part of the heretofore undiscovered circuit that mediates depth perception from motion parallax.;We expanded on this finding with additional experiments that examine the source of the extra-retinal signal that allows these neurons to resolve ambiguous visual motion. While potentially vestibular or proprioceptive in origin, we show that the necessary extraretinal signal is in fact related to smooth eye movement.;Simultaneously, we assessed selectivity to depth-sign from binocular disparity in the same neurons, and showed that neurons that were selective for both cues individually often did not have the same depth-sign preferences from the single cues (e.g. near-tuned for motion parallax, but far-tuned for binocular disparity). However, such incongruent neurons may still participate in the representation of depth from multiple cues. On the other hand, neurons with consistent preferences may be the best candidates to subserve enhanced psychophysical performance when both cues to depth are presented together; we show that they exhibit significant improvement in selectivity during a combined cue condition.;This is the first description of a neural substrate for depth from motion parallax. By discovering components of this circuit in MT, we have shown that this visual area has access to extra-retinal information, and that this information could be sensibly used to go beyond simple motion processing to produce a more general representation of three dimensional space that is dependent on multiple cues. |