Divergent eye-brain-heart fatigue signatures during sustained simulated agricultural operations

Decoding the dynamic physiological responses of the human central and autonomic nervous systems to sustained cognitive loads is essential for occupational health and neuroergonomics. This study investigates the multidimensional neurophysiological mechanisms of fatigue during complex human-machine interactions, utilizing a simulated agricultural operation paradigm. We developed a multimodal "eye-brain-heart" physio-logging framework, synchronously recording 21-channel EEG, ECG, and pupillometry in 12 experienced subjects over a continuous 5-hour protocol. Given the controlled small-sample and male-only cohort, the present study was positioned as an exploratory within-subject investigation rather than a definitive validation study. Cross-channel coupling analysis suggested task-associated divergence in multimodal physiological fatigue signatures, which was not fully consistent with a simple monolithic accumulation pattern. High-demand interactive tasks induced an "active overload" state, associated with sustained prefrontal beta synchronization, continuous pupillary dilation consistent with elevated cognitive arousal, and sympathetic dominance. Conversely, monotonous monitoring tasks triggered "passive deprivation, " marked by physiological patterns consistent with task disengagement, prominent parieto-occipital alpha bursts, and a transition to parasympathetic vagal control. These findings provide preliminary evidence that, under the present experimental conditions, occupational cognitive fatigue may be better described as a context-specific neurophysiological divergence rather than a uniform vigilance decrement. The proposed multimodal fingerprint may help reduce the diagnostic ambiguity of single-channel assessments, providing an exploratory neuroergonomic basis for adaptive interventions to preserve optimal physiological states in high-demand operational environments. Further validation in larger, gender-diverse, and more heterogeneous operator cohorts is required before generalizing these findings to broader agricultural populations.