Seminar on Cognitive Modeling for UI Design

Instructor Guest Lecturer

Moderator

Professor David E. Kieras Professor Bonnie E. John Lisa Neal

Electrical Engineering and Computer Science Department
University of Michigan
138 Advanced Technology Laboratory Building
1101 Beal Avenue
Ann Arbor, MI 48109-2110

Email: kieras@eecs.umich.edu

Human-Computer Interaction Institute
Carnegie Mellon University
3521 Newell-Simon Hall
5000 Forbes Avenue
Pittsburgh, PA 15213

Email:
bej@cs.cmu.edu

Editor-in-Chief, eLearn Magazine
3 Valley Road
Lexington, MA 02421

Email: lisa@acm.org

 

Overview: This seminar is based on the archives from the online seminar presented on February 28 - March 2, 2005. The seminar was offered using phone conferencing and streaming audio, with participants following pdf files. There were opportunities for Q&A by text chat, email, and phone. Using the audio and pdf files below, you can follow the entire seminar, only missing the opportunity to ask questions live.

Seminar description : This seminar presents the current state of the art of evaluating user interface designs using models of human performance that are based on cognitive architectures. Such models can yield usability results without the delay and expense of user testing of prototypes, but because they are new and still under development, whether and how to apply them is a challenge. This seminar will survey current theory and practice; no "how-to" of actual model construction will be presented; rather the goal is to enable a good choice of whether a modeling approach will be useful, and which type of model would be best to pursue.

Seminar Objectives : In this seminar, you will become familiar with

*

The basic approaches to model-based evaluation of user interfaces, and their advantages and disadvantages.
*

The important features of cognitive architectures that can be applied to evaluation, and specific properties of important current architectures.
* What type of model is likely to work best in a particular application.
*
How to deal with practical issues of model construction, validation, and usage, using some current tools.

Target audience: This seminar will be of interest to human-computer interaction researchers or designers who want an current overview of the modeling approach, or who are considering applying this approach to future projects. Prior background in psychology or user interface design is desirable; no prior knowledge of programming or modeling is required.

Recommendations: The seminar was offered over three days, for 2.5 hours each day, and the material below follows that format. You may wish to listen to the seminar similarly spread over three days. Also, when you open the pdf files, the easiest way to follow along is to open them in full-screen mode and use the arrow keys at the bottom to advance along with the presenters.

Topics and Materials

Day 1     
  Audio file (instructor: David Kieras)
  Presentation materials:
    1a. Introduction to the Seminar
    1b. Model-based Evaluation of User Interfaces
      Motivation, basic approaches.
      Modeling system must provide psychological constraints.
      Three approaches - differ in constraints, detail, when to use:
        Task Networks - before detailed design.
        Cognitive architectures - packaged constraints.
        GOMS Models - in the "sweet spot."
    1c. An Example Cognitive Architecture
      Use EPIC to illustrate cognitive architecture.
      Should help clarify later comparisons.
    1d. Walk-through of EPIC doing an example task
Day 2    
  Audio file (instructor: David Kieras)
  Presentation materials:
    2a. GOMS Models as Simplified Cognitive Architectures
      Keystroke-Level Model, CPM-GOMS, NGOMSL, GOMSL/GLEAN.
      Basic rationale and approach, with some examples.
      Use in design of functionality.
    2b. Survey of Cognitive Architecture Systems
      Suitable for use in UI design, considered broadly.
      Summary, pros & cons.
      Compared on degree of psychological constraint.
Day 3    
  Audio file (instructors: David Kieras and Bonnie John)
  Presentation materials:
    3a. Practical Issues in Cognitive Modeling for UI Design
      Task analysis, Generativity, Detail, Validity, Simulating the System.
      Major obstacles and issues in using Cognitive Modeling.
    3b. Making simple models easy: A new tool for Keystroke-Level Models (Bonnie John)
      CogTool Project Website

Supplementary Materials

  Further Reading on Cognitive Modeling, compiled by David Kieras and Bonnie John.
   

This .pdf file is an annotated bibliography of readings, including papers refered to during the seminar and additional materials.

 

Chipman, S. F. & Kieras, D. E. (2004) Operator centered design of ship systems. Proceedings of Engineering the Total Ship Symposium, 2004   Held at NIST, Gaithersburg, MD March 17-18, 2004, organized by the American Society of Naval Engineers.

    Abstract: Cognitive science research, much of it supported by the Office of Naval Research, is bringing about a scientific revolution in our understanding of the human operator.   It is yielding computational theories of human cognition and perceptual/motor activity that provide precise quantitative predictions of important variables such as the times required to complete tasks or to learn them in training.   Although the scope of coverage of these theories is limited and basic research aimed at expanding them is on-going, they already have much to offer in aiding the design of ship systems that will optimize the combined effectiveness of human operators and the systems they will be using.   This presentation discusses 1) what can be done now, 2) tools under development that will facilitate the use of these theories, reducing the labor involved, and 3) a long-term vision for what might be achieved in this area.
 
Some additional information, e.g., some email discussing issues brought up during the live seminar, is available at the CognitiveModeling Yahoo Group.

Acknowledgements

This seminar is supported by a grant from the Cognitive Science program of the Office of Naval Research.   Basic research on computational modeling of human cognitive architecture has been a major emphasis of ONR's Cognitive Science program.   These theories have now reached a sufficient level of maturity that many practical applications in the design of human system interaction, as well as training applications, are now feasible.   This seminar is intended to inform participants of these new developments.

For further information, please contact Dr. Susan Chipman, CHIPMAS@ONR.NAVY.MIL

Back to top