Modelling Human Cognitive Development with Explanation-Based Learning in Soar

Abstract

Explanation-based learning has been shown to be an effective method for operationalising concepts implicit in a problem solver's knowledge base. Demonstrations have thus far used mainly deductive techniques over complete domain theories and with respect to limited tasks. This paper outlines some early work on augmenting EBL with a simple inductive capability and applying it to the real world domain of modelling the development of Piagetian number conservation concepts in children. The last few years has seen a profusion of work in Explanation-Based learning (EBL). The task of an EBL system is to accept a training instance and show that it is or it is not a member of a given concept, thereby automating the classification process for future instances. This is done by problem solving over a domain theory which supports a mapping between the predicates of the training instance and those of the concept definition. Resulting generalisations must comply with an operationality criterion which limits the operationalised concept description to one which is easily evaluated for new instances. (kr)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 02, 1990
Accession Number
ADA225612

Entities

People

  • Tony Simon

Organizations

  • Carnegie Mellon University

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Abstracts
  • Acquisition
  • Causal Reasoning
  • Cognitive Systems Engineering
  • Computer Science
  • Concept Formation
  • Demonstrations
  • Learning
  • Materials
  • Military Research
  • Procurement
  • Production
  • Psychological Phenomena And Processes
  • Psychology
  • Reasoning
  • Training
  • United States Government

Fields of Study

  • Computer science

Readers

  • Artificial Intelligence
  • Systems Analysis and Design