I bring you a weekly bite-sized chunk of the science behind helicopter human factors and CRM in practice, simplifying the complex and distilling a helicopter related study into a summary of less than 500 words.
TITLE:
On error classification from physiological signals in an airborne environment.
WHAT?
Study examining whether physiological signals (brain activity (EEG), eye movements, and heart rate) can be used to reliably detect pilot errors during real flight operations.
WHERE?
University of Strathclyde, Scotland, with live flight trials performed in a Diamond DA42 aircraft.
WHEN?
Published 2025.
WHY?
Physiological monitoring has shown promise in controlled environments, but its practical application in real flight environments is still limited by uncertainty over data quality and classification accuracy due to factors such as flight motion, vibration, and operational workload.
HOW?
Nine pilots completed structured multi-tasking scenarios while flying, designed to generate measurable performance errors under controlled conditions. Testing progressed from laboratory baseline conditions to straight-and-level flight and then to higher workload manoeuvres, allowing comparison across increasing levels of operational stress.
Participants interacted with a tablet-based task interface, which simulated cognitive demands such as radio communication, threat detection, and system monitoring. An error was defined objectively by task performance, rather than judgement or observer interpretation. The errors aligned into three categories relevant to aviation performance: Incorrect actions; omissions; and timing errors.
Each task had clear error criteria:
- In communications task, an error occurred when pilots incorrectly tracked or identified positions from RT updates.
- In the ground threats task, errors were logged when targets were misidentified or missed.
- In the warnings panel task, an error occurred if a system alert was not acknowledged within a time limit.
All errors were automatically time-stamped, and aligned with the physiological data such that analysis captured snapshots of physiological data (approx 1 sec) as ‘error data’ if it immediately coincided with an error.
FINDINGS:
EEG-based data maintained accuracy in flight (approximately 88%) similar to laboratory conditions by correctly identifying error vs non-error 88 times out of 100. Eye-tracking also demonstrated consistent performance by identifying errors around 82% of the time. In contrast, ECG data correlated with error insignificantly (no better than chance), probably due to confounding effects of physical workload.
SO WHAT?
The study is significant because it represents the first successful demonstration of using physiological measure to detect errors in a live airborne environment. This opens a credible pathway towards real-time monitoring of pilot cognitive state in flight, with the potential to detect error precursors before they manifest behaviourally.
The strong performance of EEG and eye-tracking using relatively accessible sensor technologies could underpin the development of adaptive cockpit systems, enhanced decision-support tools, and new forms of in-flight safety monitoring. It also creates opportunities for true “in the field” human factors research, enabling data collection in real operational contexts rather than relying solely on simulators. However, the poor performance of ECG highlights the need for careful sensor selection and reinforces that not all proxies for workload or stress translate into useful operational data.
REFERENCE:
McGuire, N. G., & Moshfeghi, Y. (2025). On error classification from physiological signals within airborne environment. In Extended abstracts of the CHI conference on human factors in computing systems (CHI EA ’25). ACM. https://doi.org/10.1145/3706599.3719995
competentaviators.com 