2150 Distinct Patterns of Neural and Behavioral Representations in Smile Expression and Self-Reported Mood During Naturalistic Intracranial Recordings
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
INTRODUCTION: Reliable neural and audio-visual markers to detect disruptions in mood processing are lacking despite their prevalence and adverse clinical impact. In contrast, smiling is a near-universal, stereotyped expression associated with happiness and with shared behavioral and neural representations across individuals. METHODS: Multimodal, multi-day, visual-facial and intracranial electroencephalography (iEEG) recordings were obtained from 15 epilepsy patients. Naturalistic smiles and neutral periods were labeled, and participants periodically provided their self-reported mood (11-point Likert scale: 0-very angry to 10–very happy). Logistic and linear regression models with L1 regularization and with nested cross-validation were employed to decode smiles and mood scores, respectively. Models were provided with facial action units (FAUs) or iEEG features. Significance (α = 0.05) was performed by permutation of observations (n = 100). RESULTS: 1,396 smiles and 3,746 neutral events were labeled across all participants and validated by observed high accuracy of a decoder using FAUs, specifically smile-related FAU12, in individual and multi-subject, leave-one-patient-out modeling (91%, p < 0.01). Smile decoding using iEEG achieved good individual and leave-one-patient-out (79.8%) model accuracy. Broadband slope in temporal regions was the strongest contributing feature and could independently decode FAU12. In 12 subjects with sufficient mood scores, iEEG activity in 5 subjects demonstrated individual and leave-one-patient-out (Pearson’s r = 0.69 and 0.29, respectively, p < 0.05) decoding with low-gamma power in limbic regions predominating as a salient feature. In these 5, smile counts were correlated to mood scores (r = 0.38, p < 0.001). Despite decoding smile and mood in these subjects, there were no significant overlaps in iEEG and FAU features. Independently, smile-related features inconsistently decoded mood. CONCLUSIONS: These results highlight the challenges in decoding mood-related behavioral and neural dynamics. While smile-related features don’t directly generalize, understanding these distinct patterns is key for future research on reliable markers.
Fields of Study & Subjects
Emotion and Mood Recognition
Face Recognition and Perception
EEG and Brain-Computer Interfaces
Publication details
- Journal / source
- Neurosurgery
- Volume
- 72
- Issue
- Supplement_1
- Pages
- 286–287
- Publication date
- 26 March 2026
- Language
- EN