Dr. Xin Huang
Department of Neuroscience
Population neural coding of multiple visual stimuli
Neuroscientists have been investigating how neurons in the brain represent sensory information for decades. Previous studies in vision were often concerned with the neural coding of a single visual stimulus. However, natural environments are abundant with multiple entities that often co-occupy visual neurons’ receptive fields. Segmenting visual objects from each other and their background is a fundamental function of vision, but how the visual system encodes and decodes multiple visual stimuli to achieve segmentation is not well understood. Theoretical studies have proposed a framework that neurons are not coding a single stimulus value, but rather the full distribution of the stimulus in a sense of either multiplicity or probability distribution. However, neurophysiological evidence supporting this framework is limited. In this talk, I will present work in our laboratory that characterizes how neurons in the visual cortex represent multiple visual stimuli moving simultaneously at different speeds and/or in different directions. By applying decoding analysis to our neural data, we show that population neural response in the motion-sensitive, middle-temporal (MT) cortex carries information about multiple speeds and directions of overlapping stimuli. In the case of motion speed, we find it is possible to decode two speeds from MT population response in a way consistent with perception when the speed separation between visual stimuli is large, but not when it is small. In the case of motion direction, we found it is not only possible to decode multiple directions, but the extracted directions from MT population response also well matched with a well-known visual illusion that overestimates the angular separation between two overlapping stimuli moving in different directions. Our results provide strong support for the theoretical framework of coding multiplicity and probability distribution of visual features in neuronal populations and make predictions that can be tested through psychophysical and neurophysiological experiments.