On Error Classification from Physiological Signals within Airborne Environment Conference: Extended Abstracts of the CHI Conference on Human Factors in Computing Systems; April 26-May 1, 2025; Yokohama, Japan Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25)

Niall McGuire, Yashar Moshfeghi · 2025

Read On Error Classification from Physiological Signals within Airborne Environment Conference: Extended Abstracts of the CHI Conference on Human Factors in Computing Systems; April 26-May 1, 2025; Yokohama, Japan Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25) by Niall McGuire, Yashar Moshfeghi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-12769.txt.

Introduction

Niall McGuire and Yashar Moshfeghi test whether body signals can flag pilot errors during real flight, in this 2025 paper. Human error drives most aviation accidents, yet lab detection tools may fail in the air. Nine airline pilots flew set tasks across lab, calm flight, and hard-turn settings while sensors recorded brain, eye, and heart signals.

Brain-based sorting held near 88 percent accuracy in hard turns, close to its lab score. Eye tracking landed in the low eighties. Heart signals sat near chance. Accuracy barely dipped as flight grew rougher. The authors claim the first proof that such detection survives live flight.

Catalog record

CREATORNiall McGuire, Yashar Moshfeghi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-12769

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On Error Classification from Physiological Signals within Airborne Environment Conference: Extended Abstracts of the CHI Conference on Human Factors in Computing Systems; April 26-May 1, 2025; Yokohama, Japan Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25), April 26-May 1, 2025, Yokohama, Japan DOI: 10.1145/3706599.3719995 ISBN: 979-8-4007-1395-8/2025/04 CCS: Human-centered computing Collaborative interaction CCS: Human-centered computing Empirical studies in HCI Niall McGuire, Yashar Moshfeghi Abstract . Human error remains a critical concern in aviation safety, contributing to 70-80% of accidents despite technological advancements. While physiological measures show promise for error detection in laboratory settings, their effectiveness in dynamic flight environments remains underexplored. Through live flight trials with nine commercial pilots, we investigated whether established error-detection approaches maintain accuracy during actual flight operations. Participants completed standardized multi-tasking scenarios across conditions ranging from laboratory settings to straight-and-level flight and 2G manoeuvres while we collected synchronized physiological data. Our findings demonstrate that EEG-based classification maintains high accuracy (87.83%) during complex flight manoeuvres, comparable to laboratory performance (89.23%). Eye-tracking showed moderate performance (82.50%), while ECG performed near chance level (51.50%). Classification accuracy remained stable across flight conditions, with minimal degradation during 2G manoeuvres. These results provide the first evidence that physiological error detection can translate effectively to operational aviation environments. 1. Introduction Despite decades of technological progress and regulatory improvements in aviation, human error continues to be implicated in 70-80% of aircraft accidents and incidents ( Wiegmann and Shappell 2017 ) . While current approaches rely on post-incident analysis and self-reporting, these methods cannot enable real-time intervention when errors occur ( Reason 1990 ) . Prior advances in wearable physiological sensors offer a promising new direction – the potential to detect errors as they emerge, enabling proactive intervention ( Dehais et al. 2020 ) . Laboratory studies have demonstrated that physiological signals like EEG, ECG and eye movements can indicate error states with high accuracy ( Gehring et al. 1993 ) . However, transitioning these findings from controlled settings to the dynamic aviation environment presents significant challenges. Aircraft motion, electromagnetic interference, and operational stresses could all impact signal quality and reliability ( Wilson et al. 2005 ) . To the best of our knowledge, no prior work has validated error detection from physiological signals in airborne scenarios. We conducted a novel investigation examining whether established error-detection approaches from laboratory studies can translate effectively to the airborne environment. Through a series of live flight trials with nine participants, we collected synchronized physiological data (EEG, ECG, eye tracking) while participants completed controlled multi-tasking scenarios across different flight conditions. Our preliminary results suggest EEG-based detection achieves 87.8% accuracy even during complex manoeuvres, comparable to laboratory performance ( Chavarriaga and Millán 2010 ) . These findings lay the groundwork for developing adaptive cockpit interfaces that could enhance aviation safety through early error detection and intervention ( Dehais et al. 2017 ) . This work makes three key contributions to human-computer interaction research: 1) a methodology for eliciting and measuring human error across varying environmental conditions; 2) a framework for assessing the translation of laboratory-derived physiological measurement approaches to dynamic operational settings; and 3) quantitative evidence examining the feasibility of multi-modal physiological monitoring in aviation contexts. 2. Related Work Traditional approaches to error detection in aviation have primarily relied on observer ratings, self-reporting, and post-flight analysis ( Wiegmann and Shappell 2017 ) . While valuable, these methods suffer from significant limitations - they are retrospective, subject to perception bias, and crucially, cannot enable real-time intervention ( Yeh and Wickens 1988 ; Li et al. 2001 ) . Prior work has explored automated detection systems based on behavioural metrics ( Wickens et al. 2017 ) , but these approaches often fail to capture the cognitive states that precede error commission. This gap between detection and intervention capabilities has motivated the exploration of physiological approaches in human-computer interaction research ( Fairclough 2009 ; Dehais et al. 2017 ) . Laboratory studies have identified several promising physiological indicators of error states. EEG research has revealed specific error-related potentials (ERPs) that occur within milliseconds of error commission ( Gehring et al. 1993 ; Falkenstein et al. 1991 ) . These neural signatures have shown particular promise for human-computer interaction, with prior work demonstrating their potential for adaptive interface design ( Zander and Kothe 2011 ; Chavarriaga and Millán 2010 ) . Previous work by Vi and Subramanian ( Vi and Subramanian 2012 ) demonstrated that error-related potentials can be reliably detected using consumer-grade EEG headsets in controlled settings, while Vi et al. ( Vi et al. 2014 ) showed these signals can even be detected when observing others perform tasks. Eye-tracking studies have revealed characteristic patterns in gaze behaviour and pupil dilation associated with error awareness ( Holmqvist et al. 2011 ; Martinez-Marquez et al. 2021 ) , while heart rate variability measures derived from ECG have shown sensitivity to cognitive states that may predict error likelihood ( Stuiver et al. 2012 ; Malik 1996 ) . Figure 1 . Overview of the Task Procedure This figure outlines from left to right the process of the task procedure. Starting from the left is the initial setup and familiarisation with the task that each participant must go through. Once happy that they can complete the tasks they are then equipped with the neurophysiological sensors i.e. a wet EEG cap, a heart rate monitor, and eye-tracking glasses. The task then begins on the baseline i.e. ground trial, during this trial each of the low, medium, and high task settings are performed in a random order to limit mitigating factors. The process is then repeated for each of the airborne trials i.e. Straight \& Level as well as the 2G sustained spiral. However, implementing physiological monitoring in operational environments presents significant challenges for human-computer interaction design. Aviation contexts introduce unique complications including motion artefacts, electromagnetic interference, and the need for unobtrusive sensor placement ( Dehais et al. 2020 ; Wilson et al. 2005 ) . Previous attempts to translate laboratory-based monitoring to real-world settings have highlighted the importance of robust signal processing and careful consideration of environmental factors ( Kingphai and Moshfeghi 2021 ; Han et al. 2020 ; McGuire and Moshfeghi 2024a ; McGuire and Moshfeghi 2024b ) . This transition requires careful validation of both signal quality and classification reliability under actual operational conditions. Previous work in human-computer interaction has explored various approaches to error prevention in complex operational environments, from adaptive automation systems ( Parasuraman et al. 2000 ) to context-aware interfaces ( Riley et al. 2016 ) . However, these approaches typically rely on behavioural or performance metrics rather than direct physiological indicators of error states. The integration of physiological monitoring could enable more proactive and personalised error prevention strategies ( Sheridan 2016 ; Hoc 2000 ) , but this requires understanding how well these measures perform in dynamic environments. Our work addresses this crucial gap by examining how physiological error detection methods translate to airborne environments, providing essential insights for the development of physiologically-adaptive interfaces in safety-critical systems. Figure 2 . IMPACT Tool Visualisation Figure \ref{IMPACT Tool Overview} IMPACT Tool Overview presents the IMPACT Tool Overview within a cockpit setting. The image shows a pilot's view

Read the complete text