Development of a Measurement Procedure for Emotional States Detection Based on Single-Channel Ear-EEG: A...

Area of Science:

• Neuroscience
• Affective Computing
• Biomedical Engineering

Background:

• Real-time emotion monitoring is crucial for adaptive systems in healthcare, automotive, and workplace settings.
• The Circumplex Model of Affect (valence-arousal) provides a framework for understanding emotional states.
• Electroencephalography (EEG) offers a non-invasive method for capturing brain activity related to emotions.

Purpose of the Study:

• To investigate the feasibility of classifying emotions using EEG signals from a single mastoid channel.
• To identify specific EEG features predictive of valence and arousal dimensions.
• To assess the potential of minimal EEG configurations for real-world emotion monitoring.

Main Methods:

• Acquired EEG data from 24 participants using a single mastoid channel (OpenBCI Cyton board) while viewing emotion-eliciting videos.
• Extracted spectral features (power in frequency bands) and temporal features (entropy-based complexity measures).
• Employed classical statistical analyses and artificial neural networks with explainable AI (Gradient × Input, Integrated Gradients) for feature identification.

Main Results:

• Confirmed the physiological validity of single-channel ear-EEG for emotion classification.
• Identified absolute beta (β) and gamma (γ)-band power, spectral ratios, and entropy-based metrics as significant contributors.
• Demonstrated reliable and interpretable affective information extraction from minimal EEG setups.

Conclusions:

• Minimal EEG configurations, specifically single-channel ear-EEG, can effectively capture and classify affective states.
• Findings support the potential for wearable, real-world emotion monitoring systems.
• Further research is needed to address practical challenges like long-term comfort and stability for sustained use.