Related Experiment Video
Updated: Jul 14, 2026
Eye Tracking During A Complex Aviation Task For Insights Into Information Processing
Published on: April 4, 2025
Pilots' monitoring strategies and performance on automated flight decks: an empirical study combining behavioral and
Nadine B Sarter1, Randall J Mumaw, Christopher D Wickens
1Department of Industrial and Operations Engineering, University of Michigan, 1205 Beal Ave., Ann Arbor, MI 48109, USA. sarter@umich.edu
Objective:
The objective of the study was to examine pilots' automation monitoring strategies and performance on highly automated commercial flight decks.
Background:
A considerable body of research and operational experience has documented breakdowns in pilot-automation coordination on modern flight decks. These breakdowns are often considered symptoms of monitoring failures even though, to date, only limited and mostly anecdotal data exist concerning pilots' monitoring strategies and performance.
Method:
Twenty experienced B-747-400 airline pilots flew a 1-hr scenario involving challenging automation-related events on a full-mission simulator. Behavioral, mental model, and eye-tracking data were collected.
Results:
The findings from this study confirm that pilots monitor basic flight parameters to a much greater extent than visual indications of the automation configuration. More specifically, they frequently fail to verify manual mode selections or notice automatic mode changes. In other cases, they do not process mode annunciations in sufficient depth to understand their implications for aircraft behavior. Low system observability and gaps in pilots' understanding of complex automation modes were shown to contribute to these problems.
Conclusion:
Our findings describe and explain shortcomings in pilot's automation monitoring strategies and performance based on converging behavioral, eye-tracking, and mental model data. They confirm that monitoring failures are one major contributor to breakdowns in pilot-automation interaction.
Application:
The findings from this research can inform the design of improved training programs and automation interfaces that support more effective system monitoring.
