Binocular mechanisms underlying the processing of three-dimensional visual motion.
Abstract
In this dissertation, I examine binocular 3D motion processing through a series of psychophysical and neuroimaging experiments aimed at uncovering the neural computations involved and their interaction with the known hierarchy of visual motion processing. Two primary binocular cues could be used to compute 3D motion: one based on changing disparities over time (CD), the other based on interocular velocity differences (IOVD). Under normal viewing conditions, both cues coexist and (potentially) provide the same 3D direction information, yet whether CD, IOVD, or both mechanisms exist has distinct implications for how 3D motion is processed along the visual stream. First, I measured 3D direction discrimination sensitivity is measured for isolated binocular cues under a range of 3D motion speeds and visual eccentricities. Comparison of isolated-cue sensitivity to corresponding combined cue sensitivity (i.e. concurrent IOVD & CD cue stimuli) provided an estimate of relative cue contributions under normal viewing conditions. Second, I conducted a series of motion adaptation experiments to differentiate the neural representation of 2D and 3D directions of motion, and examine the degree to which IOVD or CD mechanisms can account for 3D motion adaptation. Third, I examined the neural locus of 3D motion processing by measuring 3D direction- selectivity throughout a range of visual cortical areas using functional neuroimaging in an event-related paradigm that parallels psychophysical adaptation experiments. Finally, I discuss the broader implications for the neural mechanisms of binocular 3D motion processing and future experimental directions. Together, these results reveal that: (1) the IOVD cue is the dominant cue to 3D motion processing across the majority of natural speeds & eccentricities, (2) neural tuning for 3D motion is distinct from 2D motion and can be fully explained by an IOVD mechanism, and (3) the IOVD cue is computed relatively late in the visual processing stream, in areas MT & MST— cortical areas primarily associated with 2D/retinal motion and thought to be beyond the point of binocular combination. The significance of IOVD —but not CD—cues to 3D motion perception motivates a drastic modification to canonical models of motion processing to include the late-stage comparison of eye- specific motion signals.