Predicting Behavioral Reliance on AI-Based Depression Treatment Recommendations Among Engineering Graduate Students: A Pilot Study Using EEG Signals.

Abstract

OCCUPATIONAL APPLICATIONSThis pilot study demonstrates the feasibility of using EEG-derived features to characterize behavioral reliance among engineering graduate students interacting with AI-labeled recommendations. Engineering professionals frequently engage with AI-supported decision systems in safety-critical and cognitively demanding contexts. In such occupational environments, inappropriate reliance, either over-reliance or unwarranted rejection, may compromise performance, safety, and decision quality. Although predictive performance was modest, the use of conservative participant-wise validation underscores the importance of rigorous evaluation when developing neurophysiological models for occupational human-AI interaction. These findings support EEG as a complementary tool for studying reliance behavior in professional settings while highlighting the need for multimodal and context-sensitive approaches in real-world engineering applications.