USING NEURAL NETWORKS TO ASSESS HUMAN-AUTOMATION INTERACTION

This paper presents the utility of using a neural network to model a human-automation interaction taxonomy. Automation, context, and operator features which are believed to influence human-automation interaction are identified, and the effect of changing these features on human-automation interaction are transformed from a conceptual model linkage to a computational model in the form of a neural network. The theoretical requirements of transforming the model into a computational neural network capable of analysis are discussed, and ongoing efforts to collect the required data are outlined. Additionally, the various analyses which the computational modeling enables are described. As a case study, the work uses pilots and their use of automation in the flight deck.