Acquiring Generalizations to Organize Human Databases
Abstract
This report describes a three-year program of research on category learning in unsupervised environments, and the role of learned categories in the processing and retention of individual instances. A computational model of unsupervised category learning is described, and the model's implications for the evaluation, comparison, and memorization of instances are explored in several experiments. We introduce a new index of unsupervised learning, referred to as attribute listing, and show that such learning tends to optimize the encoding of instance features and their organization in memory. The empirical techniques developed in this project appear to hold considerable promise for further research on conceptual knowledge and its role in cognitive performance.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 30, 1990
- Accession Number
- ADA230415
Entities
People
- Gordon H. Bower
- John P. Clapper
Organizations
- Stanford University