Flexible visual information representation in human parietal cortex

Abstract

In many everyday activities, we must visually process multiple objects embedded in complex real world scenes. Our visual system can flexibly extract behaviorally relevant visual information from such scenes, even though it has a severely limited processing capacity. This dissertation proposes that human superior intra-parietal sulcus (IPS) plays a central role in this flexible visual information processing. In Chapter 1, using functional magnetic resonance imaging (fMRI) with univariate analysis, I found that distractor processing in superior IPS was attenuated when target locations were known in advance. In Chapter 2, using multi-voxel pattern analysis (MVPA), I showed that superior IPS encoded object shapes, but only when such information was required by task. In Chapter 3, I showed that, given a set of perceptually distinct, but semantically grouped visual inputs, superior IPS could represent abstract object identity. The neural similarity of identities in superior IPS significantly correlated with perceived similarity between identities, confirming the representation in this region indeed reflected identity. Taken together, these results suggest that human superior IPS encodes a wide range of visual information, from simple features to abstract identities, in a task-dependent manner, enabling flexible goal-directed visual information processing in the human brain.