It is widely believed multiple cognitive processes and distinct brain areas mediate categorization. Furthermore, the ability to categorize is amenable to experience, ultimately leading to a state of categorization expertise. The goal of this dissertation is to examine how expertise mediates the neuronal correlates of categorization. The neural correlates of the type of complex visual categorization studied herein is consistent with memory or representation-based categorization processes, and expert, relative to novice performance suggest (1) decreased working memory, visual attention, and rule-based processing, as well as (2) individual subject-based expertise-dependent changes in perceptual processing areas.

Categorization performance during novice and relatively expert categorization was compared using a novel set of highly complex visual stimuli and a novel adaptive categorization task. Participants were trained to make category judgments on a pair of concurrently presented stimuli loosely resembling snowflakes under magnification. Unbeknownst to the participants, all stimuli are exemplars from one of four category prototypes. Furthermore, the difficulty of the categorization task, operationally defined by the degree of distortion of the exemplars from their prototypes, adapts to the participant's performance, thus preserving a predetermined response accuracy of 75%.

Experiment 1 validates the novel stimuli and paradigm. Experiment 2 introduces a transfer test, that is consistent with the interpretation that snowflake categorization is not mediated by knowledge of a generalizable rule or rule-based process, and thus is probably preferentially dependent on memory or representation-based processes.

Lastly, Experiment 3 recapitulates Experiment 1 behaviorally, including functional magnetic resonance imaging (fMRI) data during novice (Day 1) and relatively expert (Day 5) snowflake categorization. These data indicate that successful snowflake categorization is preferentially mediated by increased activity in the medial temporal lobes, as well as decreased activation in frontal and parietal networks, respectively associated with memory or representation-based categorization processes and attention. Expertise in snowflake categorization is associated with decreased frontal and parietal activation, which is consistent with decreased working memory, visual attention, and rule-based processing. At the individual-subject level of analysis, distributed patterns of neuronal activity in occipital and inferior temporal cortices, are interpreted as expertise-dependent changes in perceptual processing.