This study investigates the neurocognitive mechanisms underlying smoking urges by combining event-related potential analysis with behavioral measures in a controlled experimental setting. Building on prior research into the influence of social and smoking cues, we implemented a 3 (social cues) \(\times\) 2 (smoking cues) within-subject design to examine their effects on subjective smoking urge and neural indices of cognitive control and evaluative processing. Electroencephalogram (EEG) data were analyzed for N2 and P3 amplitudes, revealing that social exclusion significantly reduced P3 amplitudes and decision confidence under supportive contexts, whereas smoking cues increased P3 amplitudes and amplified urge ratings. Social exclusion also modulated N2 amplitudes, suggesting attenuated conflict detection when social support was present. To enhance real-time applicability, the study outlines a future-ready pipeline integrating edge computing for low-latency EEG preprocessing with large language models in a multi-agent framework, enabling multi-modal inference across neural, behavioral, and contextual data streams. This architecture offers a scalable approach for adaptive, closed-loop interventions, providing both mechanistic insights and a translational pathway for precision tobacco control strategies.
