Co-Principal Investigator: David E. Meyer, Department of Psychology
David Meyer's EPIC website
A major bottleneck in obtaining high performance with human-machine systems is the design of the human interface. Even the highest-performance hardware and software can be seriously limited if the human operator must work slower than necessary. Thus designing human interfaces for systems such as airliner cockpits and computer user interfaces that maximize the total system performance is critical to the future success of our rapidly evolving technology.
Currently, the major method for designing such interfaces is an iterative process based on empirical testing in which humans use different versions of the design and their performance is carefully measured. This approach is slow and costly, limiting the number of design iterations that can be investigated. A more cost-effective approach is to use models of human performance to simulate the human-system interaction, so that the quality of an interface can be assessed rapidly. While modeling approaches have been advocated for many years, recent advances in cognitive science have made it possible to construct models quickly and effectively, and research has validated the predictions of a variety of such models.
The specific goal is to develop and validate a cognitive modeling architecture, called EPIC (Executive-Process/Interactive Control) for human information processing that accurately accounts for the detailed timing of human perceptual, cognitive, and motor activity. EPIC provides a framework for constructing models of human-system interaction that are accurate and detailed enough to be useful for practical design purposes. EPIC represents a state-of-the-art synthesis of results on human perceptual/motor performance, cognitive modeling techniques, and task analysis methodology, implemented in the form of computer simulation software.
Human performance in a task is simulated by programming the cognitive
processor with production rules organized as methods for accomplishing
task goals. The EPIC model then is run in interaction with a simulation
of the external system and performs the s ame task as the human operator
would. The model generates events (e.g. eye movements, key strokes, vocal
utterances) whose timing is accurately predictive of human performance.
The figure shows the overall architecture of an EPIC model interacting
with a simulated system.
An EPIC model interacting with an external system
The most important issue that we are studying with EPIC is the nature of human multiple-task performance: these are situations in which a person is executing more than one task simultaneously, such as tuning a radio while driving a car, or making tactica l decisions while tracking a specific target in a military fighter aircraft. The interesting and challenging phenomenon is that while humans can perform multiple simultaneous tasks, there are serious limitations on their ability to do so which depend on many specific aspects of the tasks that are poorly understood even after decades of empirical research.
Practically, these limitations on multiple-task performance are extremely important in tasks like airliner piloting and nuclear power plant operation. Scientifically, understanding the details of the limitations are especially informative about the human information-processing architecture. For example, humans can only see visual detail in a small portion of the surrounding space at a time. Thus the dynamics and control of eye movements become critical if the multiple tasks all require the use of detai led visual information. Likewise, humans can store and maintain only a small amount of verbal information in short-term memory, but apparently can maintain considerable amounts of other kinds of information; the extent to which the tasks demand different kinds of memory is a critical determinant of performance. In terms of motor control, some output modalities can be used simultaneously, such as the voice and the hands, but there appear to be certain bottlenecks, for example in the extent to which the two hands can be used independently.
Our goal in the EPIC architecture development is to abstract these limitations and abilities from the substantial research literature in human performance, and then realize them in the form of computational modules that represent the known properties of human information processing. By constructing models in EPIC for well-documented phenomena, we have been able to test the quantitative accuracy of the architectural components, and refine them appropriately.
Starting with a description of the system interface and a task analysis, we construct an EPIC model by writing the production rule program for performing the task using the proposed interface. We also program the Task Environment Module (see Fig. 1) that simulates the relevant behavior of the interface at an abstract, symbolic level. Then by running the EPIC model and Task Environment Module in a simulated interaction, we obtain performance statistics that predict human performance with the actual system. These predictors include the total time to perform the task, the time and sequence of individual actions, and various aspects of mental workload, such as how much information must be maintained in short-term memory. If the predicted performance is unsatisfactory, either in terms of overall system performance, or in comparison to another design, further examination of the simulated performance will reveal the cause of the performance limitation. For example, humans can overlap many activities, but a poor interface design will limit the extent to which overlapping can be done, such as requiring that the eyes be kept mostly on one location, thereby preventing a second, otherwise compatible task from being executed concurrently.
Applications of the EPIC architecture to system interface design is another case of "analytic" or "engineering" model approaches to user interface design, similar to the GOMS model. EPIC-based evaluations will be especially relevant for interfaces to systems such as telephone operator workstations and cockpit systems, in which operating speed is critical and multiple perceptual and motor modalities are involved.
Kieras, D. EPIC Architecture Principles of Operation. On-line publication describing the details of the EPIC architecture and software package.
EPIC tutorial materials. Lecture Slides, Handouts, and sample models.
For access to the source files, contact kieras at eecs.umich.edu.
Schumacher, E. H., Lauber, E. J., Glass, J. M. B., Zurbriggen, E. L., Gmeindl, L., Kieras, D. E., & Meyer, D. E. (1999). Concurrent response-selection processes in dual-task performance: Evidence for adaptive executive control of task scheduling. Journal of Experimental Psychology: Human Perception and Performance, 1999, 25, 1-24.
Kieras, D.E., Wood, S.D., & Meyer, D.E. (1997). Predictive engineering models based on the EPIC architecture for a multimodal high-performance human-computer interaction task. ACM Transactions on Computer-Human Interaction. 4, 230-275.
Kieras, D. & Meyer, D.E. (1997). An overview of the EPIC architecture for cognition and performance with application to human-computer interaction. Human-Computer Interaction., 12, 391-438.
Meyer, D. E., & Kieras, D. E. (1997). A computational theory of executive control processes and human multiple-task performance: Part 2. Accounts of Psychological Refractory-Period Phenomena. Psychological Review. 104, 749-791.
Meyer, D. E., & Kieras, D. E. (1997). A computational theory of executive control processes and human multiple-task performance: Part 1. Basic Mechanisms. Psychological Review, 104, 3-65.
Meyer, D. E., Kieras, D. E., Lauber, E., Schumacher, E., Glass, J., Zurbriggen, E., Gmeindl, L., & Apfelblat, D. (1995). Adaptive executive control: Flexible multiple-task performance without pervasive immutable response-selection bottlenecks. Acta Psychologica, 90, 163-190.
Kieras, D.E., & Meyer, D. E. (in press.) The role of cognitive task analysis in the application of prredictive models of human performance. In Schraagen, J.M.C., Chipman, S.E., & Shalin, V.L. (Eds). Cognitive Task Analysis. Mahwah, NJ: Lawrence Erlbaum Associates.
Kieras, D.E., Meyer, D.E., Ballas, J.A., Lauber, E.J. (in press) Modern computational perspectives on executive mental processes and cognitive control. Where to from here? In S. Monsell and J. Driver (Eds.), Control of cognitive processes: Attention and Performance XVIII. Cambridge, MA: MIT Press.
Kieras, D.E., Meyer, D.E., Mueller, S., & Seymour, T. (1999). Insights into working memory from the perspective of the EPIC architecture for modeling skilled perceptual-motor and cognitive human performance. In A. Miyake and P. Shah (Eds.), Models of Working Memory: Mechanisms of Active Maintenance and Executive Control. New York: Cambridge University Press. 183-223.
Meyer, D. E., & Kieras, D. E. (1999). Precis to a practical unified theory of cognition and action: Some lessons from computational modeling of human multiple-task performance. In D. Gopher & A. Koriat (Eds.), Attention and Performance XVII. Cognitive regulation of performance: Interation of theory and application (pp. 17 -88). Cambridge, MA: M.I.T. Press.
Ballas, J.R., Kieras, D.E., Meyer, D.E., Stroup, J., & Brock, D. (1999) How is tracking affected by actions on another task? Tenth Aviation Psychology Symposium, Columbus, Ohio, May 2-6, 1999.
Hornof, A. J., & Kieras, D. E. (1999). Cognitive modeling demonstrates how people use anticipated location knowledge of menu items. Proceedings of CHI 99, New York: ACM.
Hornof, A. J. & Kieras, D. E. (1997). Cognitive modeling reveals menu search is both random and systematic. Proceedings of the CHI'97 Conference on Human Factors in Computing Systems, 107-114. New York: ACM.
Kieras, D.E., & Meyer, D.E. (1995). Predicting performance in dual-task tracking and decision making with EPIC computational models. Proceedings of the First International Symposium on Command and Control Research and Technology, National Defense University, Washington, D.C., June 19-22. 314-325.
Kieras, D. E., Wood, S. D., & Meyer, D. E. (1995). Predictive engineering models using the EPIC architecture for a high-performance task. In Proceedings of CHI, 1995, Denver, Co, USA, May 7-11, 1995. New York: ACM.
Kieras, D.E., Meyer, D.E., Ballas, J.A., Lauber, E.J. Modern computational perspectives on executive mental processes and cognitive control. Where to from here? (EPIC Tech. Rep. No. 12, TR-98/ONR-EPIC-12). Ann Arbor, University of Michigan, Electrical Engineering and Computer Science Department. August 1, 1999.
Kieras, D. & Meyer, D.E. The role of cognitive task analysis in the application of predictive models of human performance. (EPIC Tech. Rep. No. 11, TR-98/ONR-EPIC-11). Ann Arbor, University of Michigan, Electrical Engineering and Computer Science Department. March 5, 1998.
Kieras, D.E., Meyer, D.E., Mueller, S., & Seymour, T. Insights into working memory from the perspective of the EPIC architecture for modeling skilled perceptual-motor and cognitive human performance. (EPIC Tech. Rep. No. 10, TR-97/ONR-EPIC-10). Ann Arbor, University of Michigan, Electrical Engineering and Computer Science Department. January 1, 1998.
Schumacher, E.H., Lauber, E.J., Glass, J.M.B, Zurbriggen, E.L., Gmeindl, L. Kieras, D.E., & Meyer, D.E. Concurrent response selection in dual-task performance: Evidence for adaptive executive control of task scheduling. (EPIC Tech. Rep. No. 9, TR-97/ONR-EPIC-9). Ann Arbor, University of Michigan, Psychology Department. July 1, 1997.
Meyer, D. E., & Kieras, D. E. (1997). Precis to a practical unified theory of cognition and action: Some lessons from EPIC computational models of human multiple-task performance. (EPIC Tech. Rep. No. 8, TR-97/ONR-EPIC-8). Ann Arbor, University of Michigan, Psychology Department. June 1, 1997.
Meyer, D. E., & Kieras, D. E. (1997). A computational theory of executive cognitive processes and multiple-task performance: Part 2. Accounts of psychological refractory-period phenomena. (EPIC Tech. Rep. No. 7, TR-97/ONR-EPIC-7). Ann Arbor, University of Michigan, Psychology Department. January 1, 1997.
Meyer, D. E., & Kieras, D. E. (1996). A computational theory of executive cognitive processes and multiple-task performance: Part 1. Basic mechanisms. (EPIC Tech. Rep. No. 6, TR-96/ONR-EPIC-6). Ann Arbor, University of Michigan, Psychology Department. December 1, 1996.
Kieras, D. & Meyer, D.E. An overview of the EPIC architecture for cognition and performance with application to human-computer interaction. (EPIC Tech. Rep. No. 5, TR-95/ONR-EPIC-5). Ann Arbor, University of Michigan, Electrical Engineering and Computer Science Department. December 5, 1995.
Kieras, D.E., Wood, S.D., & Meyer, D.E. Predictive engineering models based on the EPIC architecture for a multimodal high-performance human-computer interaction task. (EPIC Tech. Rep. No. 4, TR-95/ONR-EPIC-4). Ann Arbor, University of Michigan, Electrical Engineering and Computer Science Department. October 1, 1995.
Meyer, D. E., Kieras, D. E., Lauber, E., Schumacher, E., Glass, J., Zurbriggen, E., Gmeindl, L., & Dana Apfelblat, D. (1995). Another Brave New World: Human information processing without pervasive immutable response-selection bottlenecks. (EPIC Tech. Rep. No. 3, TR-94/ONR-EPIC-3). Ann Arbor, University of Michigan, Department of Psychology.
Meyer, D.E., & Kieras, D.E. (1994). EPIC computational models of psychological refractory-period effects in human multiple-task performance. (EPIC Tech. Rep. No. 2, TR-94/ONR-EPIC-2). Ann Arbor, University of Michigan, Department of Psychology.
Kieras, D.E., & Meyer, D.E. (1994). The EPIC architecture for modeling human information-processing: A brief introduction. (EPIC Tech. Rep. No. 1, TR-94/ONR-EPIC-1). Ann Arbor, University of Michigan, Department of Electrical Engineering and Computer Science.