ONR-supported researchers at Carnegie Mellon University have developed a computer model that mimics the cognitive ability of humans to perform multiple tasks simultaneously. In a recent competition, the Carnegie model -- developed by Drs. John Anderson and Christian Lebiere -- simulated the performance of an air traffic controller.
The model performed as well as a human undergoing the same test. The test included managing seven tasks simultaneously, while coping with frequent interruptions. "The model performed strongly in a very realistic environment," said Dr. Harold Hawkins, the program officer who funds the research. "It handled the same cognitive workload as a human operator and experienced 'overload' at the same point a human would have been overwhelmed." The training advantages of the Carnegie model are enormous, Hawkins said. For operators to obtain realistic experience in environments such as air traffic control rooms or military command and control centers, many other players must take part in the exercise.
By substituting computer models for humans, operators can gain valuable experience at much less cost and disruption to the training organization. Eventually, operators in command and control environments may receive the majority of their training in simulated environments. "We'll continue to fund development of the model to simulate team behavior and give it learning capability," Hawkins said. The general model of human cognitive functioning was developed under the sponsorship of ONR Program Officer Susan Chipman and her predecessors and has been used to model human performance in a wide variety of tasks. More recently, the model has been used to build computerized tutors for LISP programming, high school algebra and geometry and Pascal programming. The tutors are used in a classroom setting that permits Anderson to study human learning behavior on a much larger scale than what was previously possible.