2003/11/20 by Jacob Feldman · 2 citations
Psychology · Social Sciences · #Child and Animal Learning Development #Language and cultural evolution #Categorization, perception, and language #Simplicity #Automatic summarization #Generalization #Abstraction #Set (abstract data type) #Cognitive science #Psychology #Epistemology #Concept learning #Learning theory #Computer science #Artificial intelligence #Cognitive psychology
paper · doi:10.1046/j.0963-7214.2003.01267.x
openalex publication_date 2003/11/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/06/03
How do we learn concepts and categories from examples? Part of the answer might be that we induce the simplest category consistent with a given set of example objects. This seemingly obvious idea, akin to simplicity principles in many fields, plays surprisingly little role in contemporary theories of concept learning, which are mostly based on the storage of exemplars, and avoid summarization or overt abstraction of any kind. This article reviews some evidence that complexity minimization does indeed play a central role in human concept learning. The chief finding is that subjects' ability to learn concepts depends heavily on the concepts' intrinsic complexity; more complex concepts are more difficult to learn. This pervasive effect suggests, contrary to exemplar theories, that concept learning critically involves the extraction of a simplified or abstracted generalization from examples.