Research interests: Mid- and high-level visual object representation, attention, visual short-term memory, task performance, and human occipital/temporal/parietal cortex.
Yaoda Xu received her Ph.D. from MIT and did postdoctoral work at MIT, Harvard, and Yale University.
The human vision is fundamentally a reconstruction process. As reflected in the hierarchical structure of the human ventral visual system, complex visual inputs are broken down into simple features and then reassembled to form increasingly complex representations. Through this process, the human observers gain access to visual information at multiple distinctive levels, such as lines, edges, parts, and objects. Yet at any given moment, the human observers are able to navigate through these different levels of visual processing and extract information at the appropriate level for the task at hand. This highlights two important issues that are critical to understanding visual information processing in the human brain: (1) how are simple features combined to give rise to increasingly complex representations at multiple distinctive levels? And (2) how is the moment-to-moment goal-driven visual information processing accomplished? Using fMRI and multi-voxel pattern analysis, her research attempts to address these two questions by examining how parts, objects and ensembles are represented in the human ventral visual cortex and the role of the human parietal cortex in dynamic visual information representation.