Proceedings of the Annual Meeting of the Cognitive Science Society

Research has shown that 1D category representations acquired through supervision change after unsupervised exposuresthat suggest a different boundary. However, it is unclear whether this effect generalizes to categories in which multipledimensions are relevant. To address this question, we trained participants on a 2D information integration structure (adiagonal boundary) under supervision. Participants then classified unsupervised items that implied either a steeper orflatter boundary than that established by supervision creating a conflict region where items should switch membership.Participants classified a grid of the stimulus space both immediately before (pretest) and after (posttest) unsupervisedlearning to assess for differences. We found that conflict-region items were more likely to be classified as members ofthe opposite class on the posttest, relative to pretest in a manner consistent with the unsupervised learning condition.Implications of these findings for semi-supervised learning research and theories of category learning are discussed.