Predicting What People Learn From Examples.

Abstract

Students often memorize a set of steps from examples in domains such as probability and physics, without inducing what subgoals those steps achieve. Such students fail to solve novel problems with identical goal structures but which do not permit exactly the same set of steps as the examples. This final report contains three papers that examine whether examples can help people to learn relevant subgoals for solving problems in a particular domain, and if learning the subgoals helps them to solve novel problems that involve those subgoals but require new steps for achieving them. It was hypothesized that when learners are encouraged to group steps from example solutions then they will be more likely to learn subgoals, perhaps through a self-explanation process. The connection between grouping and subgoal formation was supported by transfer results as well as analyses of participants' descriptions of how to solve problems. More generally, the fact that subgoals can effectively conveyed by examples and that those subgoals can aid transfer, has important implications for the design of training materials and tutoring environments.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 11, 1995
Accession Number
ADA294290

Entities

People

  • Richard Catrambone

Organizations

  • Georgia Tech

Tags

Communities of Interest

  • Air Platforms
  • Human Systems

DTIC Thesaurus Topics

  • Acquisition
  • Artificial Intelligence
  • Cognition
  • Cognitive Science
  • Cognitive Systems Engineering
  • Computer Programming
  • Coordinate Systems
  • Educational Psychology
  • Instructors
  • Materials
  • Military Research
  • New York
  • Probability
  • Psychology
  • Random Variables
  • Task Performance And Analysis
  • United States Government

Readers

  • Artificial Intelligence
  • Systems Analysis and Design