The Neural Representation of Perceived Color
Past Event Date:
Speaker:Steven K. Shevell Eliakim Hastings Moore Distinguished Service Professor Departments of Psychology and Ophthalmology & Visual Science Director, Institute for Mind and Biology University of Chicago
Host:Arvind Chandna and Co-host Nick Nichiporuk
Meeting room:Room 204 - Main Conference Room
Abstract - The distinction between stimulus chromaticity and perception of hue is critical for understanding color vision. A chromatic stimulus has spectral properties that characterize light entering the eye. The colors we experience, on the other hand, are perceptual constructs from neural processing within hierarchically organized cortical visual areas. Previous research, however, often has confounded stimulus chromaticity with perceptual color experience, leaving unsolved the fundamental question of whether the neural responses at each stage of processing represent a physical stimulus or instead a color we see. This question is addressed by dissociating the physical domain of chromatic stimulation from the perceptual domain of color experience, using a novel approach that causes sustained color percepts to vary over time without altering a rapidly changing temporal stream of retinal stimulation. fMRI measurements show a hierarchical progression of color processing in the ascending human ventral visual pathway, from a representation of the chromatic stimulus in V1 to perceptually experienced color in V4.
[METHODS: Stimulus chromaticity was dissociated from color appearance using dichoptically rivalrous chromaticities that were swapped between the eyes every 120 msec. The surprising percept is a single sustained color for about 2 seconds (during which there are 16 eye swaps) followed by a different sustained color (Christiansen, D’Antona & Shevell, J. Vision, 2017). In the fMRI experiments, observers reported their sustained color percept while BOLD responses were measured at each level of the visual cortical hierarchy. A chromatic representation was reconstructed from BOLD responses separately for each stage (V1, V2, V3, V4v, VO1) by transforming the fMRI measurements using an inverted encoding model (Brouwer & Heeger, J. Neurosci., 2009).]