Designing Experimental Protocols to Elicit Vigilance Variations in Controlled Laboratory Settings

Abstract

Vigilance plays a vital role in numerous high-risk professions, where real-time monitoring of vigilance levels is highly beneficial. Brain-Computer Interfaces (BCIs) utilizing EEG signals, along with machine learning and deep learning algorithms, present a promising solution for the classification and monitoring of vigilance levels. This study aimed to create a dataset for training such models by evaluating four experimental paradigms: the Hitchcock Air Traffic Controller Task (ATC), the simultaneous line task, the successive line task, and the Oddball task. Subjective reports and behavioral performance were analyzed to determine the effectiveness of these tasks in inducing vigilance decrements over time. The findings reveal that both the ATC and simultaneous line tasks effectively induced significant declines in subjective vigilance ratings and behavioral performance. In contrast, the Oddball task was less successful in generating a noticeable vigilance decrement. This research demonstrates the potential of the ATC and simultaneous line tasks to induce vigilance variations, providing valuable datasets for training vigilance detection algorithms. Additionally, it highlights the importance of considering non-linear fluctuations in vigilance and the need for more advanced techniques to accurately classify different vigilance states. Such improvements in vigilance monitoring could substantially enhance safety and well-being in critical work environments.