Spontaneous Discovery and Use of Categorical Structure

Abstract

These experiments investigated unsupervised category learning using tasks in which subjects attempted to memorize the features of training instances from two contrasting categories. On each trial, subjects studied a verbal feature list (training instance) for 24 seconds, after which they were given multiple choice recognition tests to evaluate their memory for each list item. The amount of time spent looking at each feature during the study phase, and the accuracy of recognition during the test phase, provided two separate indices of unsupervised learning on each trial. The main independent variable in these experiments was the specific sequence in which instances from the two categories were presented. The effects of these sequence manipulations on learning provided strong evidence for the use of an explicit, non-incremental, category invention process to capture the consistent structure of the stimulus domain. The present experiments also showed the selective encoding process and enhanced memory for instances predicted by standard, schema-based, theories of learning.... Unsupervised learning, Category invention, Attribute, Feature, Default.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 15, 1993
Accession Number
ADA261658

Entities

People

  • Gordon H. Bower
  • John P. Clapper

Organizations

  • Stanford University

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Abstracts
  • Accuracy
  • Artificial Intelligence
  • Autocorrelation
  • Classification
  • Coding
  • Cognitive Science
  • Computer Programs
  • Computers
  • Information Science
  • Machine Learning
  • Psychology
  • Recognition
  • Reliability
  • Security
  • Self Organizing Systems
  • Unsupervised Machine Learning

Fields of Study

  • Psychology

Readers

  • Artificial Intelligence
  • Brain and Cognitive Science; Experimental Psychology; Cognitive Neuroscience
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks