Pumping up your predictive power for cognitive state detection with the proper GAINS

Detecting cognitive states and impairments through EEG signals is crucial for applications in aviation and medicine and has broad applications in the field of human-machine interaction. However, existing methods often fail to capture the fine-grained neural dynamics of critical brain processes due to limited temporal resolution and inadequate signal decomposition techniques. To address this, we introduce the Spectral Intensity Stability (SIS) algorithm, a novel technique that analyzes the stability and competition of dominant brain frequency oscillations across granular timescales (≈4 ms). Unlike traditional spectral methods, SIS captures rapid neural transitions and hierarchical frequency dynamics, enabling more accurate characterization of task-specific cognitive processes. Our study focuses on EEG data from pilots performing multitasking simulations under hypoxic and non-hypoxic conditions, a high-stakes scenario where cognitive performance is crucial. We divided this multitasking scenario into specific cognitive states, such as task precursor, interruption, execution, and recovery. Our algorithm SIS achieved a 29.8% improvement in cognitive state classification compared to conventional methods, demonstrating superior accuracy in distinguishing both task states and hypoxic impairments. This work is novel because it bridges gaps left by traditional methods by revealing the role of hierarchical spectral dynamics in maintaining cognitive performance. Through the Granular Analysis Informing Neural Stability (GAINS) framework, we reveal how neuronal groups self-organize across fine-grained time scales, providing new understanding of task-switching, neural communication, and criticality. The findings highlight the potential for developing real-time cognitive monitoring systems to enhance safety and performance in environments where cognitive impairments can have serious consequences. Future research should extend these insights by incorporating transient behaviors and spatial dynamics to achieve a more comprehensive framework for characterizing cognitive states.