Implicit concept learning: Multiple adaptive processes of unsupervised category construction

A distinction is drawn between an explicit mode of concept learning that is characterized by external, corrective feedback and an implicit mode that is completely untutored. Several current models of implicit concept learning are discussed, including models of incremental partitioning and auto-correlators. Based on recent research a tentative framework for studying processes of implicit concept learning is developed. This framework comprises two closely intertwined components. The first component aims at cumulative encoding and flexible usage of exemplar-specific information, the second component builds abstract, prototyp-like representations of implicitly constructed categories. Finally, several currently unsolved issues concerning the content, process, and structure of the implicit formation of concepts are pointed out.