Investigating Patterns for Self-Induced Emotion Recognition from EEG Signals

Area of Science:

• Neuroscience
• Affective Computing
• Biomedical Engineering

Background:

• Current emotion recognition relies on external stimuli (images, sounds, videos).
• Neural patterns for self-induced emotions have not been previously investigated.
• Understanding endogenous emotions is crucial for comprehensive emotion recognition.

Purpose of the Study:

• To infer neural patterns and signatures of self-induced emotions using electroencephalogram (EEG) signals.
• To analyze features, electrode distribution, and neural patterns for discriminating discrete emotions.
• To explore the potential for developing endogenous emotion recognition methods.

Main Methods:

• Recorded EEG signals from 30 participants while they watched movie clips eliciting six discrete emotions (joy, neutrality, sadness, disgust, anger, fear).
• Participants self-induced emotions by recalling movie scenes.
• Analyzed high-frequency EEG rhythms, electrode distribution (temporal, prefrontal, occipital lobes), and neural patterns.

Main Results:

• High-frequency EEG rhythms from bilateral temporal, prefrontal, and occipital lobes showed high performance in emotion discrimination.
• Specific neural patterns and brain topography distributions were identified for each of the six discrete emotions.
• Achieved 87.36% accuracy for positive vs. negative emotion discrimination and 54.52% for six-category classification.

Conclusions:

• EEG signals contain distinct patterns for self-induced emotions.
• High-frequency rhythms in specific brain regions are key discriminators of endogenous emotions.
• This research contributes to the development of comprehensive endogenous emotion recognition systems.