Deborah A. Boehm-Davis is currently Professor of Psychology in the Human Factors and Applied Cognition Program at George Mason University in Fairfax, Virginia. She holds an A.B. (1975) in psychology from Rutgers the State University (Douglass College) and an M.A. (1977) and Ph.D. (1980) in cognitive psychology from the University of California, Berkeley. She worked on applied cognitive research at General Electric, NASA Ames and Bell Laboratories prior to joining George Mason University in 1984.

Dr. Boehm-Davis is the President-Elect of Division 21 (Applied Experimental and Engineering Psychology) of the American Psychological Association. In the past, she has served as the President and the Secretary-Treasurer of the Human Factors and Ergonomics Society. She is on the editorial boards of Human Factors, the International Journal of Human-Computer Studies, the International Journal of Human-Computer Interaction, and Theoretical Issues in Ergonomic Sciences.

In 1985, she received the Franklin Taylor Award from the IEEE Systems, Man, and Cybernetics Society; in 1994, she received the Washington Academy of Sciences Award for Scientific Achievement in the Behavioral and Social Sciences and in 2002, she was selected as a member of the Douglass Society. In 2003, she received the Franklin V. Taylor Award from Division 21 of the American Psychological Association. She is a Fellow of both the American Psychological Association and the Human Factors and Ergonomics Society.

Research Interests

I am interested in how human performance is helped or hindered by the design of tools that help us accomplish everyday tasks and I have a particular interest in how improved display of information can improve human performance. My current research falls into four categories: transportation, the influence of interruptions on performance, interpretation of graphical displays, and cognitive workload.

Transportation Research

Project: Analysis of Pilot Procedures and Practices for Automated Flight Decks (1998 – current)

The relationship between Crew Resource Management (CRM) and flight deck automation management has been evident for some time. A number of carriers include specific automation skills under CRM, but the research community has yet to connect research in these two related areas. This effort integrates and extends what has been learned from prior research on proceduralized CRM to flight deck automation procedures and practices. The objectives of this grant are to: 1) develop advanced multivariate techniques for analyzing AQP data and answering more complex operational questions about pilot/crew performance; 2) model pilot/crew procedures and processes for two or three focal automation problems; and 4) use the model to develop new and better assessments of automation use in a fleet based on automation procedures and practices. The objectives build on a sequence from examining the current database information on crew automation performance, to isolating frequent and important automation-related problems, to computationally modeling the processes underlying those problems, and finally, to designing and evaluating model-based performance assessment tools. The results of this effort will be improved methods for collecting and analyzing automation performance data for an Advanced Qualification Program (AQP) setting, a specific cognitive model of particular problems in crew automation performance, and a prototype for the model development process.

Project: Abatement of Automation Errors by Training based on Cognitive Methods (1999-2003)

Cockpit automation has changed the roles, responsibilities, and activities of pilots, leading to new types of errors on the flight deck. This research is focused on understanding those errors through the development of a computational cognitive model that describes how pilots interact with automated systems. The cognitive model under development is based on a cognitive task analysis supplemented with eye tracking data collected from commercial pilots flying a low-fidelity simulator. These data informed our design decisions about what information pilots are acquiring from the flight deck while working with automated systems during climb or descent. We have developed a working computational cognitive model, which is built in ACT-R. The model allowed us to develop potential interventions to improve pilot performance, which are being tested empirically.

Project Funded by: National Aeronautics and Space Administration

Project: Analysis of Cockpit Management System in Multiple Carrier Environments (1994-1999)

The long-term goal of transforming Crew Resource Management (CRM) from a set of attitudes to CRM-specific knowledge and skills is increasingly realistic. Knowledge of CRM is generally transferred through training in the classroom. Skills are usually trained in the flight simulator environment through Line-Oriented Flight Training (LOFT) and related sessions. Typically, a pilot, once qualified, undergoes LOFT once or twice a year. This six to twelve month time lag between training sessions has raised questions regarding the effectiveness of this training plan. One approach to increasing a crew's practice of CRM skills is to develop procedures that allow pilots to exercise CRM skills every time they fly on the line. Could CRM procedures be implemented in such a way that crews could systematically improve their CRM skills? This research grant addressed these questions through the design and implementation of prototype CRM training programs based on CRM procedures at a regional air carrier.

The grant evaluated the effects of CRM procedures training in Line Check and Line Operation Evaluation (LOE) performance of flight crews. The study employed procedure-based CRM that integrates CRM training and requirements into the standard operating procedures of the air carrier. Traditionally, CRM training has been treated as a separate training issue, but in this project, CRM was closely integrated with technical procedures. The grant team, with the strong cooperation of the regional air carrier, developed a set of CRM procedures that were trained under a carefully designed program called Advanced Crew Resource Management (ACRM). The empirical data collected over a 3-year period showed that the integration of CRM into procedures is effective in improving pilot performance.

Project Funded by: Federal Aviation Administration, AAR-100, sponsored by AFS-230

Interruptions Research

Our research in the area of interruptions focuses on understanding the components of interruptions and the cognitive processes that occur when someone is interrupted during task performance.

Graphical Displays

This research focuses on the development of a framework that allows us to understand how global or trend information is extracted from graphs. Existing theories of graph comprehension focus primarily on the extraction of “local” information from simple and moderately complex graphs. However, less is known about what happens when people are asked to extract global or trend information from a graph. Further, the processes that might underlie these differences between local and global extractions have not been elaborated. Our research is focusing on the qualitative differences between the way people answer global and local questions and on understanding how different questions activate different cognitive operations.

Cognitive Workload Research

Project: Understanding and Measuring Cognitive Workload: A Coordinated Multidiscipline Approach (1997 – current)

This research program is designed to develop predictive (based upon cognitive modeling) and descriptive (based upon physiological data) measures of cognitive workload that are highly correlated. Such measures should be theoretically grounded and empirically verified. Our main engineering goals in this project are to show (1) how the predictive measures (cognitive modeling) can be applied to guide the design of novel interfaces and communication protocols for decision making tasks, and (2) how the descriptive measures (physiological) may be used to measure workload during real-time task performance.

In the area of interface design, we applied cognitive modeling techniques to the development and description of microstrategies. Empirical data were collected to describe how these microstrategies develop and contribute to workload using Argus Prime, a synthetic task that allows us to swap minor and/or major interface components while holding the task itself constant. In these studies, we have explored the impact of making subtle changes in the design of the interface on performance and on the strategies (and microstrategies) selected by participants.

In parallel, we have been modeling this task using ACT-R/PM, which allows us to represent and study embodied cognition in performing this task. Here again, our goal is to understand how subtle aspects of an interface may lead to large increases in cognitive workload. The modeling activity is based on the ACT-R/PM architecture, which combines ACT-R's theory of cognition with modal theories of attention and motor movement. This level of modeling has allowed us to represent the microstrategies that we observed our participants using in the Argus Prime task into a computational cognitive model. The models demonstrated that interactive behavior in complex tasks is constrained not only by cognition but also by perception and motor processes. Although these constraints exist at the millisecond level, the milliseconds added to an interaction matter when the task requires thousands of interactions over an extended period of time. Further work on the modeling will include expanding the modal models of visual attention and motor movement as well as working to incorporate a modal model of eye movements. These expansions are necessary to build models that respond adaptively to subtle differences in interface design.

We have also examined potential physiological indicators of workload. For this portion of our research, we focused our efforts on examining eye blinks within the Argus Prime task. Although Stern (1988) had argued that “...blinking serves as a kind of mental punctuation of mental processing”, others had argued that reductions in eye blink rates are due to "visual busyness". In our data, we found that eye blinks were: inhibited during period of highest reliance on visual processing and visual attention and elevated where there was “slack time” predicted by our cognitive task analysis. We took this as evidence for the position that eye blinks are better understood as a measure of visual workload than of cognitive workload.

Project Funded by: Department of Defense through the Air Force Office of Scientific Research