A large number of accidents have always been attributed to operator error. While automation has eliminated many types of operator error, it has also resulted in new types of operator error. Many of these new errors are the result of what human factors experts have labeled technology-centered automation, where designers focus most of their attention on the mapping from software inputs to outputs, on mathematical models of required functionality, and on the technical details and problems internal to the computer. Little attention is usually given to designing and evaluating software with respect to whether it provides transparent and consistent behavior that supports operators in their monitoring and control tasks. Human factors experts have tried to overcome what they call clumsy automation by changing the human interface to the automation, by changing operator training, or by designing new operational procedures to eliminate the human errors resulting from poor software design. These efforts have had only limited success. Some human factors researchers have concluded that training cannot and should not be the fix for bad design, and have called for more human-centered automation. If we can identify software design features that lead to human error, then we should be able to design software in such a way that errors are reduced without sacrificing the goals for which computers are being introduced, such as increased productivity and efficiency. Building on previous preliminary work by the PI, the goal of this research is to identify problematic software behavior (design features) that can lead to human operator error, to define algorithms for analyzing formal software specifications for these features, and to create new human-centered software design practices that will lead to reduced operator errors and safer systems.
Broader Impacts: Most researchers in human-computer interaction have focused on the design of the interface between the human and the computer. Software design, however, is starting to play an important role in accidents by confusing human operators and contributing to new types of human error. Very little is currently known about the relationship between software design and operator error. The potential benefits of developing this relatively under-researched area in the science of software design are obvious in terms of saving lives and property. In addition, students will get the opportunity to work in and learn about an interdisciplinary field that requires knowledge from both computer science and from psychology and human factors. The results of this research will be disseminated widely and incorporated into both undergraduate and graduate classes.
