Quantitative assessment of the components of the integral cognitive load of aircraft pilots

Context and relevance. Flight safety critically depends on the pilot's ability to effectively manage cognitive load under multitasking conditions. Existing assessment methods (resource models, reserve measurement techniques) allow for predicting general workload or stating its presence; however, they do not enable the identification and quantitative assessment of specific latent ergonomic vulnerabilities in the human-machine system. Objective: development and testing of a method for the quantitative assessment of pilot integral cognitive load components. The method should not only detect the fact of overload but also determine the contribution of various professional activity modalities to its occurrence. Hypothesis. The combined use of Kohonen self-organizing maps for detecting anomalous states based on oculomotor activity patterns and solving a linear programming problem for load decomposition allows for obtaining unambiguous quantitative estimates applicable for vulnerability localization in near-real-world operational conditions. Methods and materials. The method is based on a two-stage procedure: 1) detecting the exceeding of critical load using a Kohonen self-organizing map trained on individual OMA indicators (average fixation duration, gaze movement entropy, saccade and blink frequency) in the normal state; 2) assessing the contribution of activity modalities (piloting, navigation, communication, etc.) by solving a linear programming problem that uses a system of constraints formed during experiments with different task combinations. Results. The method's operability was demonstrated using examples of assessing three activity modalities and analyzing landing phases. Quantitative estimates of load components were obtained, confirming the possibility of decomposing integral load and identifying the most resource-intensive activities. Conclusions. Unlike classical resource and diagnostic approaches, the proposed method provides a direct link between the fact of overload, its quantitative structure, and specific activity modalities, enabling vulnerability localization. The results open prospects for the method's application in ergonomic interface evaluation, risk prediction, and individualization of simulator-based training.