Anderson, J. R., Boyle, C. F., & Reiser, B. J. (1985). Intelligent tutoring systemsScience228(4698), 456-462.


Intelligent tutoring systems are computer-assisted technology which has “programs that simulate an understanding of the domain they teach and can respond specifically to the student’s problem-solving strategies” (Anderson et al, 1985, 456). The authors are interested in exploring the aspects that make a successful intelligent tutoring system. Published in 1986, this is one of the earlier writings regarding ITSs.

The article provides a framework for designing intelligent tutoring systems according to the ACT, adaptive control of thought, theory. The ACT theory refers to cognitive psychology and the role of the types of memory in cognition. The focus of the ACT approach is on procedural memory and how that could be used by intelligent tutoring systems. The four aspects of the ACT theory approach that the authors highlight are knowledge compilation, use of production, working memory limits, and goal structures(Anderson et al, 1985, 457). By programming tutoring systems to address these aspects, the developers can provide personalized instruction for users.

The authors also discuss model-tracing as a framework for ITSs. Model-tracing involves programming an ideal model for problem-solving and possible paths a user could take to solve a problem. The model-tracing paradigm is based on ACT practices. The tutor does not provide instruction unless it sees the user has deviated from the model for solving the problem. Based on the type of error and what step the user is on in solving the problem the computer gives feedback on the user’s work(Anderson et al, 1986, 458). The authors provide two examples of tutoring systems, a geometry tutor and a LISP programming tutor,  they evaluated. The two examples showcase the ACT theory components and the model-tracing paradigm within intelligent tutoring systems.

Key Points:

  • ACT theory approach to intelligent tutoring systems
    • Knowledge compilation
    • Use of production
    • Working memory limits
    • Goal structure
    • Implications
      •  Explicit goal structure
      •  Minimizing working memory loads
      •  Instruction in problem-solving context
      •  Immediate feedback on errors
  • Model-tracing paradigm
    •  Specific production for the solution
    • Production of errors a student could make
  • Geometry tutor
    •  Cannot move to new concept without demonstrating mastery
    • User noted liking the subject after use
  • LISP tutor
    •  Attention directed to conceptual issues
    • Goals of the problem being solved displayed throughout process

Design Principles

  • One prominent feature of successful education is immediate feedback on errors. Both described tutoring systems employ immediate feedback. The immediate feedback allows users to adjust their problem-solving strategy and understanding of the problem.
  • An important design consideration for ITSs is to include instruction in a problem-solving context. This allows users to put concepts into practical use.
  • Goal setting is another design principle which the authors establish. By making problem-solving goals clear, the user can determine what steps they need to take to solve a problem. The authors suggest that goals be clearly displayed for each problem.
  • The final design principle in the article warns against stressing the user’s working memory, stating “many errors of learners are due to failures of working memory rather than failures of understanding” (Anderson et al, 457). The authors suggest that designers encode the computer with most of the information which the user is likely to forget to ease the working memory load. This allows users to focus more of their attention in solving the problem at hand.

Discussion Questions

  1. Is the ACT-based approach the most appropriate way to approach intelligent tutoring systems?  How would determine the best approach to intelligent tutoring systems?
  2. Reducing working-memory load is an important part of interface design. In what ways can we free up as much work-memory load as possible? Do current tutoring system designs reflect these needs?

Additional Resources:

  1. Explanation of ACT theory website: Interpersonal Communication and Relations | Act* Theory


2. Kodaganallur, V., Weitz, R. R., & Rosenthal, D. (2005). A comparison of model-tracing and constraint-based in telligent tutoring paradigmsInternational Journal of Artificial Intelligence in Education15(2), 117-144.




Icon for the Creative Commons Attribution 4.0 International License

Learning Environments Design Reading Series Copyright © by evrimb is licensed under a Creative Commons Attribution 4.0 International License, except where otherwise noted.

Share This Book