Observations on Human Performance in Air Traffic Control Operations: Preliminaries to a Cognitive Model

We have previously described a computational approach to modeling human performance, built using the Apex architecture [1],in which extended behavioral sequences are automatically constructed from simpler CPMGOMS templates [2]. Here we report on efforts to extend that method to the more complex domain of air traffic control operations. We describe the overall approach and our initial analysis of patterns of human performance observed in a simulation of air traffic control operations conducted by the Federal Aviation Administration. Of particular interest are patterns that characterize how operators manage multiple tasks, and distribute their attention over items. We show how templates in our compositional approach might be structured to accommodate these patterns.