The use of expert pilot performance models to facilitate cockpit visual scan training

This work represents an extensive investigation into the nature of pilot expertise, specifically as it relates to the development of cockpit visual scan strategies. A review of the training literature reveals a variety of training strategies that could be employed to reduce the time to achieve expertise, by focusing on the qualitative nature of the novice-expert differences. We consider two of these: part task training (PTT) and guided training.; In Experiment 1, we asked whether the use of PTT would facilitate automatization of ADI data extraction and if so, whether this skill would transfer to full-cockpit panel flight (and in doing so, free attentional resources so as to facilitate cross-checking those secondary performance instruments taught using guided training). We also developed a computer-based part-task guided scan training protocol to teach instrument scan in which we employed salient characteristics of expert strategies. Three groups of novice pilots from the University of Illinois' Institute of Aviation were used as subjects in this study. One group received ADI PTT alone (ADI-Only) while a second received guided scan training in addition to ADI PTT (ADI+PP). A Control group received no special training. All subjects were tasked to ‘fly’ a 7-segment pattern whereby they had to make changes to airspeed, heading, altitude, and all combinations thereof. Tracking error and visual scan were measured.; The results suggested that (1) it was possible to facilitate scan by driving attention via computer-based guided practice. However, we failed to observe a transfer of improved flight control; the reason centering, we hypothesized, on the inability of our ADI training to foster more rapid data extraction and (2) the ADI part task training protocol failed to promote learning of pursuit tracking and may have contributed to the lack of transfer of improved flight skill to the other axes of flight and to the other (altitude and heading change) flight segments. These results strongly suggest that, although the importance of the ADI in cockpit scan strategy is understood, one cannot develop an effective ADI training protocol without first identifying (and exploiting) those characteristics of the task which best promote learning to extract and use cockpit attitude information.; In Experiment II, we therefore examined a range of ADI PTT training regimes that explicitly varied the extent to which this training is dynamic, and also which explicitly considered the training of pursuit movements. An additional feature of Experiment II, missing from Experiment I, was a systematic effort to create a mental model of flight dynamics for all subjects, which would therefore provide a context into which ADI information extraction could be incorporated. Five groups of flight-naïve subjects were first provided with a standardized flight and scan familiarization program. Three groups then received ADI PTT. One control group received no special training or practice, and another was afforded only practice time. They all ‘flew’ the same 7-segment task as in Experiment I. Tracking performance and scan data were analyzed. The results suggest that whereas PTT without mental model development training (bottom-up PTT) provided limited benefits to flight performance, top-down mental model building appears to rapidly foster expert-like performance.