CHI '95 ProceedingsTopIndexes

Cognitive Architectures & HCI

Susan S. Kirschenbaum Code 2212, Bldg 1171/1
Naval Undersea Warfare
Center, Division Newport
Newport, RI 02841-1780
Tel: (401) 841-3354
Fax: (401) 841-4749

Wayne D. Gray George Mason University
m/s 3f5
4400 University Drive
Fairfax, VA 22030-4444
Tel: (703) 993-1344
Fax: (703) 993-1359

Richard M. Young MRC APU
15 Chaucer Road
Cambridge CB2 2EF
Tel: +44 (223) 355 294 ext 310
Fax: +44 (223) 359 062



This workshop will focus on appropriate use of cognitive models for the analysis and solution of HCI problems.


Cognitive modeling, User modeling, Simulation


The use of cognitive architectures for modeling is beginning to have an interesting, but little noted, impact on the types of modeling applied to HCI issues. In the past, the software used to construct models was either very generic (as in the use of OPS5) or handcrafted by the individual modeler. Recently this has changed. At first generic connectionist architectures and, more recently, symbol manipulation architectures (Soar and ACT-R) have entered the public domain as systems that are centrally supported but whose use has spread beyond their developers. Most of the models presented at recent CHI conferences have been developed and presented as an instance of a modeling architecture (e.g., Soar-PUMS or Soar-GOMS).

The latest generation of models incorporate large amounts of cognitive theory and enable the modeler to build more generative and realistic models that take into account more features of human cognition, the artifact (software design) being modeled, and the task to which the artifact is being applied. Controlling this C-A-T (cognition-artifact-task) triad promises to be the key to more realistic, powerful, and productive models.

The purpose of this workshop is to bring together three overlapping groups of HCI researchers: those who have used more than one of the architectures, those who have used just one major architecture, and those who have developed cognitive models using either generic or handcrafted software. The aim is to identify ways in which the existence and choice of cognitive architectures bears on the practice and prospects for modeling in HCI.


To provide depth and perspective on the topic, participants will be individuals who (1) have developed a computer-based, cognitive model, (2) have a history of research in HCI issues, and (3) are interested in the effect of their choice of architecture on the way their models have developed.

Three pre-workshop activities are designed to provide a common frame of reference. First, participants will provide a written description of one of the models that they have developed to be shared with other workshop participants. Second, three articles (from a list nominated by the participants) were read by all, before coming to the workshop. Third, participants will be asked to consider how their approach to modeling would be applied to model one of a small number of benchmark tasks.

During the workshop, presentations and exercises will use these models and readings to focus discussions on appropriate use of cognitive models for analysis and solution of HCI problems. Specific topics include:

Results of this workshop will serve as a foundation for the scientifically valid and appropriate use of cognitive modeling and cognitive architectures for future generations of HCI practitioners. One product of the workshop will be a report that will be published in the SIGCHI Bulletin. An expected product is the development of a support-group of researchers interested in modeling HCI phenomena.