讲座:Visual perception as retrospective decoding from high- to low-level features
19
2016-09
张鸣沙老师课题组邀请美国哥伦比亚大学--钱宁教授，来实验室做学术报告，欢迎感兴趣的老师和同学参加。以下为报告信息：
报告题目：Visual perception as retrospective decoding from high- to low-level features.
报告人：Professor Qian Ning
时 间： 9月22日（周四）下午2:30
地 点： 英东楼422会议室
Abstract：When a stimulus is presented, its neural encoding progresses from simple, low-level features to complex, high-level features. How these features are decoded to produce perception is less clear and proposed decoding hierarchies are not quantitatively compared with perception. Moreover, observers often inspect different parts of a scene sequentially yet no decoding model considers working memory. We probed decoding hierarchy by comparing absolute judgments of single orientations and relative/ordinal judgments between two sequentially presented orientations. We found that lower-level, absolute judgments failed to account for higher-level, relative/ordinal judgments. However, when ordinal judgment was used as a Bayesian prior to retrospectively decode memory representations of absolute orientations, striking aspects of absolute judgments, including correlation and forward/backward aftereffects between two orientations in a trial, were explained. We suggest that the brain prioritizes decoding of more useful, higher-level features, which are also more invariant and categorical and thus easier to specify and maintain in noisy working memory, and that more-reliable higher-level decoding constrains less-reliable lower-level decoding.
