The Neural Representation of Multiple Visual Objects in Human Visual Cortex

Abstract

In everyday visual environments, our visual systems process multiple objects embedded in complex scenes and extract useful information to complete challenging tasks. Although the neural mechanisms mediating single visual object perception have been extensively studied in the past few decades, little is known about how the neural representation of multiple objects and the interactions among them. This dissertation investigates how contextually associated object pairs, and the layout information of multiple visual objects may be represented in the human visual cortex. In particular, we examined functional defined high-level visual object processing regions in lateral occipital (LO) cortex and posterior fusiform gyrus (pFs), topographically defined early visual areas, and functionally defined parietal regions involved in processing multiple objects. In Chapter 1, using functional magnetic resonance imaging (fMRI) and multi-voxel pattern analysis (MVPA), I found that contextual association significantly affected how an object pair may be represented in pFs, such that objects were more integrated when they were related than unrelated. In Chapter 2, using the method established in Chapter 1, I found that related object pairs exhibited a reliable effect of contextual association irrespective of the amount of implied action between the two objects. In Chapter 3, I found that the layout information of a multi-object display was explicitly encoded in LO, but not in pFs, early visual areas, or parietal regions involved in object processing. Taken together, these results suggest that LO and pFs, which are critical for single object representation, are differentially involved in processing the interactions among multiple objects.