Michael Freed and James C. Johnston
This paper describes first steps towards the production of a broad, psychologically plausible model of human cognition, emphasizing those cognitive subsystems of greatest interest in understanding expert air traffic controller performance. At least three problems must be addressed in constructing such a model. First, perceptual, motor and cognitive models known to the cognitive psychology community and described in the literature must be redescribed at a level of detail and in terms suitable for computer simulation. Second, existing models, which typically describe particular cognitive functions, must be pieced together to form a broad "cognitive architecture," thus allowing simulation of a broad range of mental phenomena. Third, new models have to be constructed where the cognitive psychology literature is incomplete. In general, empirical human research provides progressively fewer constraints on models of cognitive phenomena that take place over longer periods of time (e.g. goal-selection, problem-solving) [Newell1990]. To fill in these gaps, artificial intelligence techniques have been introduced and integrated into our model.