Cognitive Control in a Computational Model of the Predator Pilot

Abstract

This chapter describes four models of cognitive control in pilots of remotely piloted aircraft. The models vary in the knowledge available to them and in the aircraft maneuvering strategies that control the simulated pilot's interaction with the heads-up display. These models are Type 3 (knowledge/strategy) variants. The first two models are successive approximations toward a valid model of expert-level pilot cognitive control. The first model failed because of a naïve flight control strategy, and the second succeeded because of an effective flight control strategy that is taught to Air Force pilots. The last two models are investigations of the relative contributions of different major components of the more successful model of pilot cognitive control. This investigation of knowledge and strategy variants produces an anomalous result in relative model performance, which is explored and explained through a sensitivity analysis across a portion of the Type 3 parameter space. The lesson learned is that seemingly innocuous assumptions at the Type 3 level can have large impacts in the performance of models that simulate human cognition in complex, dynamic environments.