Phase-Amplitude Coupling of EEG Applied in Music-Induced Emotional Recognition Tasks

• Published in: 2024 20th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD) pp. 1 - 6
• Main Authors: Zhao, Mohan, Fortunelli, Gianmarco, He, Zhenghao, Gong, Xiaoliang, Cohn, Anthony
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.07.2024
• Subjects: Affective computing; Couplings; Electroencephalography; Emotion recognition; Feature extraction; Knowledge discovery; Music; Picture archiving and communication systems; Support vector machines; Telematics
• DOI: 10.1109/ICNC-FSKD64080.2024.10702312

Phase-Amplitude Coupling (PAC), as an important electroencephalogram (EEG) feature, which contains information about the brain's telematic mechanisms and cross-frequency relations, has gained more attention recently compared to conventional spectral features. Music has been recognized as a powerful way to induce emotions, and among the various emotional responses to music, musical chills are a unique and notable phenomenon, researching which can improve our knowledge of both musicology and cognitive neuroscience. Our experiment induced emotions, quantified into levels of arousal, valence, and preference, in 19 subjects using specific music extracts. During the experiment, the subjects experienced over 50 instances of musical chills with their EEG being recorded. This work extracts PAC, 15 conventional spectral statistical features, and 7 brain network features, and utilises a support vector machine (SVM) for emotion classification and recognition of musical chills. The results of emotion classification show that PAC performs the best in all emotional levels - arousal with an accuracy of 87%, valence 88% and preference 86%. In the musical chills recognition task, even though the result of PAC is lower than in the emotion classification task, it is the highest compared with the other two features. Furthermore, PAC with gamma band always shows the best results. These results suggest that PAC can be an essential factor for emotional recognition tasks, thus giving more insights into the field of affective computing.