Little is known about the neural mechanisms involved in feature classification, or where it occurs, or how it influences other aspects of vision. Evidence suggests that image context is the key to finding the solution of how attributes of the visual scene are derived from retinal image features. There are several contextual cues such as T- and X- junctions which are important in helping us determine the retinal-image- to-visual-scene transformation. These contextual cues leading to feature classification and surface segmentation can be studied both statically and dynamically; the motion processing system serves as a valuable model in the dynamic case because there exists a tight relationship between motion and surface segmentation. I hope to further understand the nature of this relationship by (i) identifying and characterizing neurons sensitive to cues that allow feature classification, and (ii) understanding how surfaces and feature classes are encoded by neurons.
