The cognitive and neural mechanisms of visual expertise training

Visual categorization is a remarkable ability that allows us to effortlessly identify objects and efficiently respond to our environment. The neural mechanisms of how visual categories are learned and become well-established are largely unknown. Here, in three experiments, subjects were trained to learn new visual categories and functional magnetic resonance imaging (fMRI) was used to delineate the neural mechanisms of category learning. The first experiment characterized the neural mechanisms involved in retrieving recently learned visual categories. Subjects were trained to categorize stimuli for 1260 trials and then performed the same task in the fMRI scanner. Activity was compared between stimuli close to and far from the category boundary during categorization and a perceptual matching task. Results demonstrate that the hippocampus and left superior frontal sulcus are more responsive to stimuli that are efficiently categorized whereas inferotemporal cortex and prefrontal cortex are more responsive that are most difficult to categorize. The second experiment assessed the neural changes involved in the transition from initial learning to expert categorization. Subjects were trained to categorize stimuli for 4250 trials and then were scanned while performing the well-learned categorization task and a novel categorization task. Activity and connectivity changes between the well-learned and novel categorization tasks were assessed. The results suggest that novel and well-learned visual categorization utilize a similar network, and that learning involves an increased coordination between inferior temporal, hippocampal, and premotor regions that support the retrieval and representation of categories. In the third experiment, visual category training was used to enhance face processing in a congenital prosopagnosic (MZ), a condition in which face identification abilities are not properly developed. MZ was extensively trained to categorize face stimuli (from experiment 2) and face recognition abilities were assessed before and after training. Additionally, using fMRI, we measured activity and connectivity changes with training. The results show improvements in face identification and these improvements are accompanied by activity and connectivity changes in inferior temporal cortex.