Understanding classification is a major challenge for formulating theories of both human cognition and artificial intelligence. Classification is a human mental activity which covers such diverse things as forming concepts, categorizing medical patients, recognising an acquaintance, or discriminating phonetic components of a language. This book considers both previous theories and new research and comes up with a new core model which it proposes as the basis of allforms of classification. Two versions of this model are developed, one based on symbol-processing, and one on connectionist architecture, and it is suggested that modules of each type of model coexist in the human cognitive system, competing for control of behavioural output.