Motion perceptual learning in noise improves neural sensitivity in human MT+ and IPS
43.541, Monday, May 13, 8:30 am - 12:30 pm, Vista Ballroom
Nihong Chen1, Hanyu Shao4, Xuchu Weng4, Fang Fang1,2,3; 1Department of Psychology and Key laboratory of Machine Perception (Ministry of Education), Peking University, 2Peking-Tsinghua Center for Life Sciences, 3IDG/McGovern Institute for Brain Research, Peking University, 4Center for Cognition and Brain Disorders, Hangzhou Normal University
Practice substantially improves motion direction discrimination. However, the neural mechanism of motion perceptual learning in visual noise is little understood. We studied this mechanism with functional magnetic resonance imaging (fMRI) before and after training. During training, subjects practiced six daily sessions (6,000 trials total) to discriminate motion directions of two successive random-dot kinematograms at 35% coherence, with the method of constant accuracy at 75% correct. Each subject was trained along one of the following eight directions that started from 22.5° and with an increment of 45°. Before, immediately after and two weeks after training, subjectsÂ’ motion direction discrimination thresholds were measured along the directions that were 0°, 30°, 60°, and 90° away from the trained direction. Blood-oxygenation-level-dependent (BOLD) signals were also measured in a 3T GE magnet while subjects performed the same motion discrimination task along these motion directions at 75% correct. Behaviorally, practice gave rise to 48% improvement along the trained direction immediately after training and 45% improvement two weeks later, which transferred little to the untrained directions. BOLD signals were analyzed in V1, V2, V3, V3a, V5/MT+, and intra-parietal sulci (IPS), all of which were sensitive to visual motion. Compared to stimuli along the untrained directions, no significant amplitude change of BOLD signal responding to stimuli along the trained direction was found in the above ROIs. Using multi-voxel pattern analysis method, we discovered a persistent improvement of decoding accuracy in V5/MT+ and IPS for the trained direction, which lasted up to two weeks. These findings suggests that, after practice, the trained motion signal in noise could be represented and read out more accurately in the visual cortex and in the higher-level cortex respectively.