An Eyewear-Based EEG Interface for Temporal-Region Monitoring and Emotion Recognition
- 1Guangdong-Hong Kong-Macau Institute of CNS Regeneration, Jinan University, Guangzhou; and BIAI Laboratory, BIAI INC, Shenzhen, China.
Abstract
Conventional electroencephalography (EEG) devices are often impractical for ambulatory settings due to their reliance on specialized hardware and expert operation. To address this, we developed a user-friendly eyewear-based EEG system featuring dry titanium electrodes positioned approximately 5 mm inferoposterior to the T7/T8 sites (near the ears), with metal nose pads serving as references. We evaluated the system across 103 recordings spanning five paradigms: alpha reactivity, blink-related transients, fatigue, attention, and video-elicited emotion. Following quality control, 93.2-100.0% of epochs were retained. The system demonstrated high sensitivity to ocular transients, showing significant amplitude increases during the blinking task. The emotion task yielded the most robust state-related features, with statistically significant differences in differential entropy across negative, neutral, and positive video categories, achieving balanced accuracies of 0.882 for two-class and 0.667 for three-class prediction. Despite challenges at the multiple testing correction stage, the alpha reactivity and fatigue contrast tasks were still noteworthy to some extent. The attention paradigm revealed a consistent increase in differential entropy during focused states, and the eyewear system outperformed the Muse headband in group-held-out classification. These findings provide strong evidence for the feasibility of temporal-region eyewear EEG acquisition, highlighting its potential applications in the real-world, particularly for emotion recognition.
Related
- VAD-EmotionNet: a spatial-spectral attention network for EEG emotion recognition in the 3D affective space
- Disrupting the Right Inferior Frontal Gyrus Increases Social Brain Activity During Emotion Recognition
- ASD-FER: An Annotated Multi-Class Benchmark and Comparative Study for Facial Emotion Recognition in Children with Autism Spectrum Disorder
- EEG emotion recognition via multilevel entropy analysis: The power of entropy of entropy on phase dynamics
- Probing EEG Foundation Model Attention Maps for Lightweight Interpretable Emotion Recognition
BibTeX
@misc{li2026eyewearbased,
title = {{An Eyewear-Based EEG Interface for Temporal-Region Monitoring and Emotion Recognition}},
author = {X. Li and L. Chai and S. Xu and J. Ni and K.-F. So},
howpublished = {bioRxiv},
year = {2026},
month = nov,
doi = {10.64898/2026.11.02.763849},
url = {https://www.biorxiv.org/content/10.64898/2026.11.02.763849v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - An Eyewear-Based EEG Interface for Temporal-Region Monitoring and Emotion Recognition
AU - X. Li
AU - L. Chai
AU - S. Xu
AU - J. Ni
AU - K.-F. So
PY - 2026
DA - 2026/11/02
PB - bioRxiv
DO - 10.64898/2026.11.02.763849
UR - https://www.biorxiv.org/content/10.64898/2026.11.02.763849v1
AB - Conventional electroencephalography (EEG) devices are often impractical for ambulatory settings due to their reliance on specialized hardware and expert operation. To address this, we developed a user-friendly eyewear-based EEG system featuring dry titanium electrodes positioned approximately 5 mm inferoposterior to the T7/T8 sites (near the ears), with metal nose pads serving as references. We evaluated the system across 103 recordings spanning five paradigms: alpha reactivity, blink-related transients, fatigue, attention, and video-elicited emotion. Following quality control, 93.2-100.0% of epochs were retained. The system demonstrated high sensitivity to ocular transients, showing significant amplitude increases during the blinking task. The emotion task yielded the most robust state-related features, with statistically significant differences in differential entropy across negative, neutral, and positive video categories, achieving balanced accuracies of 0.882 for two-class and 0.667 for three-class prediction. Despite challenges at the multiple testing correction stage, the alpha reactivity and fatigue contrast tasks were still noteworthy to some extent. The attention paradigm revealed a consistent increase in differential entropy during focused states, and the eyewear system outperformed the Muse headband in group-held-out classification. These findings provide strong evidence for the feasibility of temporal-region eyewear EEG acquisition, highlighting its potential applications in the real-world, particularly for emotion recognition.
ER -