Abstract
Neurons in primate MSTd exhibit selectivity to a continuum of spiral optic flow patterns (Graziano et al. 1994, J Neurosci), but the function of this sensitivity is unknown. MSTd is believed to encode the direction of self-motion (heading) because many neurons in MSTd respond to radially-expansive optic flow, which is experienced by an observer traveling along a straight path without body rotations. Yet, humans often travel along curvilinear paths and due to the rotation introduced by the path curvature, the optic flow patterns appear distinct from those produced by travel along straight paths. The neural mechanisms underlying the perception of path are unknown. Froehler & Duffy (2002, Science) discovered "path-selective cells" in MSTd that elicited differential activity when a monkey traveled clockwise and counterclockwise around a circle. The sequence of optic flow fields experienced by the monkey were identical in either case, but in reverse-order. Our analysis indicates that the temporally-accumulated optic flow yields distinct spiral patterns for clockwise and counterclockwise movements, for which we predict the "path-selective cells" are selective. We introduce a model that clarifies the role of primate MSTd in heading and path perception. Model neurons compete in visuotopic and pattern selectivity space, defined by a continuum spanning radial, spiral, and center optic flow. The distribution of activity peaks across MSTd cell subpopulations that are sensitive to different spiral optic flow patterns at the same spatial location corresponds to perceived path curvature judgements in human subjects. Humans produce systematic errors in their judgments of future curvilinear path depending on gaze and eye movements (Li & Cheng 2011, Journal of Vision). In conditions whereby human subjects under-(over-)estimated path curvature, MSTd subpopulations sensitive to radial (spiral) optic flow were most active. The model simultaneously represents heading and perceived path across the population activity in MSTd.
Meeting abstract presented at VSS 2013