Category Learning Systems (2008)
Dagmar Zeithamova
Category learning is an essential cognitive function. Empirical evidence and theoretical reasons suggest existence of multiple dissociable category learning systems. Here, a proposal is made that different category learning tasks are dominated by different category learning systems. A dual system theory of category learning COVIS proposes dissociation between an explicit, hypothesis-testing system, and an implicit, procedural learning system. Two studies testing this dissociation are presented, supporting the notion that hypothesis testing, utilizing working memory and explicit reasoning, mediates learning in rule-based tasks, while gradual and automatic S-R learning mediates information-integration tasks. Inconsistent findings in the literature regarding a prototype learning task suggest that two versions of this task, the A/nonA, single prototype task and the A/B, two prototype task, are mediated by distinct category learning mechanisms. A novel methodology for studying the A/nonA task and the A/B task is proposed and utilized in a functional magnetic resonance imaging study. The study reveals that the A/B task is mediated by declarative memory while the A/nonA task is mediated by perceptual learning. We conclude that at least four category learning systems exist, based on four memory systems of the brain: working memory, procedural memory, declarative memory and perceptual memory. The four category learning systems compete or cooperate during learning, each system dominating in a different category learning task. Category learning tasks provide a useful tool to understand learning and memory systems of the brain.
PhD Thesis, Institute for Neuroscience, The University of Texas at Austin.

Dasa Zeithamova Formerly affiliated Collaborator zeithamova [at] mail utexas edu