As 2025 comes to a close, it has been a busy and rewarding year at ASTG. With close to 450 members, we hope you enjoyed the events, insights, and initiatives we shared with you throughout the year, and that our work continues to support your practice in meaningful ways. As we look ahead to 2026, it's still time to renew your ASTG membership so you don't lose access to the newsletter and all the resources we provide. 
Wishing all our members a warm and restful holiday season!

Recent Publications

By Anika Fairooz Shafi, from the University of Waterloo

Operator Decision-Making in the Deployment of Lethal Autonomous Weapon Systems: A Scoping Review
Shaheen Shekh, Mark Wiggins, Ben Morrison
Journal of Cognitive Engineering and Decision Making

Uncrewed vehicles capable of delivering lethal outcomes and equipped with AI are referred to as Lethal Autonomous Weapon Systems (LAWs). As their deployment in operational contexts become more likely, there are nevertheless concerns around ensuring meaningful human control. The authors conducted a scoping review to identify research on factors that influence operators' decisions to deploy LAWs. Of the 313 publications reviewed, most focused on governance, design, and development - covering topics such as ethical design, morality, legality, policy, human factors, and public opinion. However, none examined the factors that shape an operator's actual decision to deploy LAWs. As a result, we currently lack empirical understanding of how legal, moral, and cognitive considerations interact during such decisions. Using a Socio-Technical Framework (SOTEF), this study identified significant gaps in research related to operators' decision-making around LAWs deployment. Findings suggest that organizational culture will likely influence the acceptance and the use-case for LAWs, together with the threshold for deployment in the field. Training and experience will also play a crucial role. A training package using hypothetical scenarios could assist current and prospective operators' decision-making in actual future battlefield scenarios. System characteristics such as Level of Automation (LOA), reliability will impact trust and decisions of deployment. Additionally, operational demands such as temporal pressure may impair decision-making, underscoring the importance of error management versus error avoidance training. Finally, further research with operators in realistic scenarios can identify human factors (such as mental workload, stress, fatigue) and physiological signatures associated with risky deployment of LAWs. 

Inverse Counterfactual for AI-Assisted Decision Support: Enhancing Knowledge Elicitation for Capturing Aircraft Pilot Decisions
Jonay Ramon Alaman, Daniel Lafond, Alexandre Marois, Sébastien Tremblay
Proceedings of the Human Factors and Ergonomics Society Annual Meeting

Can our understanding of shared mental models in human teams be leveraged to enhance AI-assisted Decision Support Systems (DSSs)? The authors explored this question by recruiting 44 students and employees from Université Laval to perform a series of weather-avoidance cases while having them interact with Cognitive Shadow, a DSS designed to model operator behavior and report decision discrepancies. In each weather-avoidance case, participants chose from four options: staying on track or selecting one of three diversion maneuvers (lateral or vertical). A custom-interface provided all necessary information including primary flight display, weather, fuel information; it also allowed participants to communicate with the DSS and generate inverse counterfactuals (modifying a single factor of the case scenario such that the second-choice becomes more preferable). The study comprised three phases. In the first phase, participants completed 130 cases, selecting the best avoidance option for each case. In the second phase, for 20 new cases, participants first chose the best option, then created an inverse counterfactual. In the third and final phase, participants were assigned either to a control (DSS trained only on first-choice decisions) or to an experimental condition (DSS trained on both first-choices and inverse counterfactuals). In this 20-case phase, the DSS provided recommendations whenever its prediction differed from the participant's choice, which participants either accepted or rejected. Results indicate that the DSS aligned more closely with participants' decisions in the control condition than in the experimental one. However, inverse counterfactual quality impacted results such that larger Euclidean distances between original cases and their inverse counterfactuals were associated with lower agreement between participants and the DSS. Overall, findings suggest that while inverse counterfactuals hold promise as a knowledge-elicitation technique for AI-assisted DSSs, their effectiveness depends on quality as well as nuanced model refinement.

Advancing vision standards in aviation: Embracing evidence-based approaches
Allison Lynch, Ewa Niechwiej-Szwedo, Shi Cao, Suzanne Kearns, Elizabeth Irving
Journal of Vision

Aviation licensing relies on visual acuity ("distance" visual acuity), which reflects the ability to recognize small details at a distance, with 20/20 or 0.0 logMAR representing normal visual acuity. However, some literature suggests that contrast sensitivity may be a more sensitive predictor of pilot performance. In this study, the authors examined the effects of vision degradation on flight performance under challenging weather conditions. Participants' vision was degraded either by using Cambridge Simulation Glasses, which affect both visual acuity and contrast sensitivity, or by defocusing lenses, which primarily affect visual acuity. Twenty-six individuals (14 male, aged 18–31 years, 0-400 flight hours) performed simulated landings under visual flight rules and three different weather conditions: a clear day with high winds (220° at 22 knots), heavy rain with no wind, and heavy rain with high winds (220° at 22 knots). The landings involved five randomized visual degradation conditions: 0, 1.0, 1.3, 1.6, and 1.8 logMAR, for a total of 15 trials. Flight performance was evaluated using both a competency-based assessment by a flight instructor and objective flight simulator data. Analysis revealed significant differences in the variability (standard deviation) of aircraft control parameters – vertical speed, pitch, roll, and the slope of altitude descent – as a function of both vision degradation level and weather condition. There was also a significant main effect of vision degradation type (scattering versus defocus) on the slope of altitude descent. Post hoc analyses indicated that flight performance was first affected at 1.0 logMAR for scattering lenses and 1.3 logMAR for defocusing lenses. These findings highlight the potential of contrast sensitivity as a sensitive indicator of pilot performance and underscore the need for further research to inform evidence-based vision standards for aviation.

Cognition in the cockpit: assessing instructional modalities in pilot training simulations
Laurie-Jade Rochon, Alexander John Karran, Thadde Rolon-Merette, François Courtemanche, Constantinos Coursaris, Sylvain Senecal, Pierre-Majorique Léger
Frontiers in Psychology

Given the growing use of automated instructional modules in flight simulator training, the authors investigated how different instruction modalities influence learning. In a between-subjects design, 30 flight-school students from Quebec, Canada, were randomly assigned to one of three conditions: audio-only, text-only, or combined audio-text instruction. Each participant completed two flights in Microsoft Flight Simulator: a guided instructional flight (navigating from Airport A to Airport B) and a solo evaluation flight (navigating from Airport B to Airport A), both conducted under visual flight rules. Eye movements were recorded using Tobii Pro Nano. Flight performance was evaluated post-hoc by instructors using video recordings. Perceived cognitive load was assessed with NASA-TLX after each flight segment, while experienced cognitive load was assessed through Percentage Change in Pupil Diameter (PCPD), calculated from changes in pupil diameter relative to a pre-stimulus baseline. Visual attention was measured via eye-tracking metrics of visual transition entropy and attention dispersion. Affective state (emotional valence) was assessed in real time using facial-expression analysis software. Motivation and immersion were measured once at the end of the experiment using Situation Motivation Scale (SIMS) and Immersive Experience Questionnaire (IEQ), respectively. During the evaluation flight, the text-only and combined audio-text groups showed significantly lower visual transition entropy, indicating more structured visual scanning. The text-only group also displayed higher emotional valence, suggesting greater engagement. No significant differences emerged in overall flight performance or cognitive load, however, trends pointed toward higher motivation and immersion in the text-only condition. These findings highlight concerns that observable behavior alone may mask underlying overload or motivational decline. Thus, instructional modalities should be matched to task demands - text-based instructions may better support concurrent in-flight tasks, whereas bimodal instructions could be more suitable for pre-flight learning phases.

ASTG Election