EEG-based approach for recognizing human social emotion perception

Abstract

Social emotion perception plays an important role in our daily social interactions and is involved in the treatments for mental disorders. Hyper-scanning technique enables to measure brain activities simultaneously from two or more persons, which was employed in this study to explore social emotion perception. We analyzed the recorded electroencephalogram (EEG) to explore emotion perception in terms of event related potential (ERP) and phase synchronization, and classified emotion categories based on convolutional neural network (CNN). The results showed that (1) ERP was significantly different among four emotion categories (i.e., anger, disgust, neutral, and happy), but there was no significant difference for ERP in the comparison of rating orders (the order of rating actions of the paired participants); (2) the intra-brain phase lag index (PLI) was higher than the inter-brain PLI but its number of connections exhibiting significant difference was less in all typical frequency bands (from delta to gamma); (3) the emotion classification accuracy of inter-PLI-Conv outperformed that of intra-PLI-Conv for all cases of using each frequency band (five frequency bands totally). In particular, the classification accuracies averaged across all participants in the alpha band were 65.55% and 50.77% (much higher than the chance level) for the inter-PLI-Conv and intra-PLI-Conv, respectively. According to our results, the emotion category of happiness can be classified with a higher performance compared to the other categories.

Deep Learning in EEG: Advance of the Last Ten-Year Critical Period

“…In their experiment, two person need to rate their emotions induced by the same piciture one by one. They extracted the intra-brain and inter-brain phase synchronization features from emotional EEG signals and applied a CNN model to evaluate [102]. As we know, deep learning needs parameter tuning and it is time-consuming.…”

Multimodal Emotion Classification Method and Analysis of Brain Functional Connectivity Networks

“…The evaluation of brain functional connectivity networks and graph theory have become powerful tools to help study emotion classification. Zhu et al analysed recorded EEG and utilized the PLI to capture the emotional perception of phase synchronization and classified emotions based on a convolutional neural network (CNN) [19]. Wang et al constructed PLV-based brain functional connectivity networks, extracted two features of functional integration and functional separation, and analysed the differences in brain connectivity of discrete emotions [28].…”

Extracting Information on Affective Computing Research from Data Analysis of Known Digital Platforms: Research into Emotional Artificial Intelligence

“…The authors of [11] investigated everyday social interaction between two or more people. The focus of the research was on the perception of emotion in terms of event potential and phase synchronisation.…”