EEG-Based Neural Mechanisms of Confidence and Metacognition in Construction Safety Hazard Recognition

Hazard recognition is essential for preventing construction accidents. However, existing behavioral interventions often fail to effectively enhance workers’ self-monitoring abilities and decision-making performance. Furthermore, the cognitive mechanisms underlying hazard recognition—particularly confidence as a metacognitive indicator of judgment accuracy—remain insufficiently explored. To explore the neural mechanisms of confidence and metacognition in high-risk construction, a safety hazard recognition task was designed using electroencephalography (EEG) to record brain signals and confidence ratings from 38 participants. A metacognitive model based on signal detection theory was then applied to assess the alignment between recognition accuracy and confidence. The results show the following: (1) Although participants achieved relatively high hazard recognition accuracy, their confidence did not consistently reflect actual performance, revealing metacognitive inefficiency. (2) Most participants showed a conservative bias in hazard recognition, tending to judge situations as hazardous, and displayed overconfidence in their decisions. (3) Neural analysis revealed that prefrontal event-related potential (ERP) amplitudes were positively correlated with confidence levels within 1.0 to 1.4 s poststimulus. (4) Time-frequency analysis showed that confidence was positively associated with high beta power (20–30 Hz), negatively associated with alpha power, and positively associated with delta power. The formation of confidence relies on the integrative functions of the prefrontal cortex and the synchronized modulation of multifrequency brain activity. This study elucidates the mechanisms of confidence in hazard recognition from both behavioral and neural perspectives. Based on the identified behavioral metrics and EEG features, this study proposes an application framework for a real-time cognitive monitoring and intervention system utilizing smart helmets integrated with active dry electrode technology. The system is designed to trigger targeted management measures based on real-time cognitive risk levels. Consequently, this approach optimizes workers’ hazard recognition performance and decision quality, offering intelligent and personalized support for construction safety management.