Open AccessThe role of structural consistency between categories and attributes in hierarchical category learning
Author(s) -
Saiki Jun
Publication year - 1998
Publication title -
japanese psychological research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.392
H-Index - 30
eISSN - 1468-5884
pISSN - 0021-5368
DOI - 10.1111/1468-5884.00086
Subject(s) - consistency (knowledge bases) , selection (genetic algorithm) , task (project management) , mathematics , concept learning , artificial intelligence , hierarchy , natural language processing , psychology , pattern recognition (psychology) , machine learning , cognitive psychology , computer science , management , economics , market economy
This study investigated how consistency between categories and attributes determines attribute selection in hierarchical category learning. Participants learned six categories for which number and color were equally relevant attributes, followed by a transfer task, to test which attribute was used. Before that, half of them learned embedding higher‐level categories for which numbers were likely to be used. Orthogonal to this factor, the hierarchical structure was made explicit for half of them by category labels. The results showed that participants used numbers in the prior learning, but that the use of numbers was inhibited in the subsequent six‐category learning task. However, this inhibitory effect was reduced when the hierarchical structure was explicit. The pattern of results suggests that attribute selection is determined by structural consistency between categories and attributes, not by a prior use of an attribute.