Single-Channel Ear-EEG for Emotion Monitoring: A Feasibility Study

Existing electroencephalographic (EEG) devices require dedicated, intrusive accessories aesthetically unsuitable for daily use, as their bulky and conspicuous design makes them visually obtrusive and impractical for daily or social use. This feasibility study evaluates EEG signal measurability using a single-channel near the ear, a step toward developing a wearable adornment with an integrated EEG sensor. EEG signals were recorded from six participants exposed to specific audiovisual stimuli inducing different valence and arousal levels. Signals were labeled based on the participants' self-reports. The EEG data were segmented into 1-s epochs, and both absolute and relative spectral power features were computed via Fast Fourier Transform. Two main analyses were conducted: the first assessed EEG signal quality, while the second involved statistical evaluation of the extracted features using either one-way ANOVA or the Mann-Whitney U test, as well as machine learning-based emotion recognition employing a Sequential Feature Selector in conjunction with a Support Vector Classifier. The acquired signals were consistent with the characteristics of conventional EEG signals. Arousal (or valence) levels were analyzed while the other emotional dimension was held constant at either a low or high level. Arousal level (high vs. low) classification achieved an average accuracy of 76.4% ± 15.1% under low valence and 71.9% ± 7.9% under high valence conditions. Valence level detection (positive vs. negative) exhibited an average accuracy of 76.4% ± 3.4% and 75.3% ± 10.5% under concurrent low and high arousal conditions, respectively. The preliminary results confirmed the feasibility of the proposed approach for EEG-based emotion monitoring in daily life.