Hierarchical Bayesian Modeling for Physiological Data in Small-N Aviation Human Factors Research

Monitoring pilot cognitive state in real time is becoming increasingly important as automation plays a larger role in aviation. Traditional workload assessments, such as questionnaires or task-based performance metrics, provide useful insights but can be limited in rapidly changing flight environments. Physiological measures, including heart rate, respiration, and electroencephalogram (EEG), offer continuous data streams, yet their variability and complexity present challenges for analysis. This study explores the use of a hierarchical Bayesian framework to quantify patterns from physiological signals recorded during high-fidelity flight simulations. Five certified pilots flew scenarios that varied in automation level and working memory demand while heart rate, respiration rate, and EEG-derived workload estimates were monitored. The model generated individualized and condition-specific estimates, quantified uncertainty, and remained stable with a small participant pool. Heart rate appeared to be the most consistent indicator, followed by EEG-derived workload, while respiration rate was less reliable across conditions. These results suggest that Bayesian inference may provide a promising way to interpret physiological data in aviation settings and could support the development of adaptive automation that responds to pilot workload. The approach emphasizes transparency and efficiency, offering complementary value to existing modeling techniques for aerospace human factors and flight deck applications.