Exploring Latent Biometric Constructs in a Model Predicting Mental States of Aviators

Single-crew aircraft persistently have a high accident rate; these accidents are associated with high mental workload (MWL). The aviation industry would benefit from a passive MWL monitoring system that would predict flight performance. Passive biosensors offer an economical and non-intrusive method for indexing MWL. Many studies have overemphasized tonic data while ignoring phasic data. The present study explores the viability of a phasic data centered model in indexing MWL to predict flight performance. The study had non-pilots fly a simulator. Cardiovascular and epidermal data, objective and subjective MWL states, subjective reports of simulator sickness, and a variety of flight performance indicators were measured. The data were decomposed into several components to build formative latent variables that were pruned based on an objective MWL measure to then predict flight performance measures. The results indicate that phasic components explain more variance in flight performance than objective and subjective MWL and tonic data.