Hierarchy of visuospatial codes in the primate temporal lobe

Harvard Medical School, Boston MA

Investigators

Abstract

Project Summary The proposed project applies novel technologies to understand how the apparently disparate functions relate to each other in two structures of the primate temporal lobe: the inferior temporal cortex, a series of sensory areas that process detailed visual features, and the medial temporal lobe, a network of polymodal association areas that mediate representations of memory, space, and other abstract relations. These areas are closely interconnected, yet historical approaches to studying and understanding their functions have been separate. The proposed project will elucidate how visual and spatial representations relate across inferior and medial temporal areas. In contrast to the prevailing conception of these two structures as having discrete roles, we propose to test the overarching hypothesis that medial and inferior temporal functions are more continuous than thought. The project will ask, first, what visual features do space-encoding neurons represent? Second, how is this feature information, likely in part supplied by the retinotopic ventral stream, transformed into spatial representations? To answer these questions, the project will use innovative chronic multielectrode recordings and novel analysis methods to characterize simultaneously visual featural and spatial tuning in the same neurons to compare medial and inferior temporal areas. Aim 1 uses a natural image free-viewing task that engages feature-selective responses and gaze-based spatial codes with new analysis methods to compare both tuning types in the same neurons and across areas. Aim 2 uses rapid image presentation and a unique deep neural network-based method to characterize and compare the detailed visual feature selectivity across areas. Aim 3 studies freely moving animals to dissect how gaze-based visual codes transform into gaze-, head- and body-based spatial codes along the temporal cortex hierarchy. Overall, the results will contribute to a fuller understanding of both temporal-lobe structures and how they support a pivotal transformation in cortical information processing from sensory information to an allocentric world model.