The Origins and Development of Visual Categorization
Abstract
Forming categories is a core part of human cognition, allowing us to make quickly make inferences about our environment. This thesis investigated some of the major theoretical interpretations surrounding the neural basis of visual category development. In adults, there are category-selective regions (e.g. in ventral temporal cortex) and networks (which include regions outside traditional visual regions—e.g. the amygdala) that support visual categorization. While there has been extensive behavioural work investigating visual categorization in infants, the neural sequence of development remains poorly understood. Based on behavioral experiments, one view holds that infants are initially using subcortical structures to recognize faces. Indeed, it has been proposed that the subcortical pathway remains active for rapid face detection in adults. In order to test this in adults, I exploited the nasal-temporal asymmetry of the proposed retinocollicular pathway to see if preferentially presenting stimuli to the nasal hemiretina resulted in a fast face detection advantage when contrasted with presentations to the temporal hemiretina. Across four experiments, I failed to find any evidence of a subcortical advantage but still found that a rapid, coarse pathway exists. Therefore, I moved to investigate the development of the cortical visual categorization regions in the ventral temporal cortex (VTC). I characterised the maturity of the face, place and tool regions found in the VTC, looking at the long-range connectivity in 1-9 month-old infants using MRI tractography and a linear discriminant classifier. The face and place regions showed adult-like connectivity throughout infancy, but the tool-network underwent significant maturation until 9 months. Finally, given this maturity of face and place regions in early infancy, I decided to test whether the organization of the VTC was related to the sequence of categories infants acquire. I used language age of acquisition measurements, determining that infants produce significantly more animate than inanimate words up until 29-months, in line with the animacy distinction in the VTC. My work demonstrates the surprising role and maturity of the cortical regions and networks involved in visual categorization. My thesis develops new methods for studying the infant brain and underscores the utility of publicly available data when studying development.