EEG-based emotion recognition is getting smarter, faster, and more real-world-ready. Gkintoni et al. (2025) review 64 studies and outline what’s next.
Emotion decoding from brainwaves is no longer science fiction. In a new systematic review published in Brain Sciences, Gkintoni et al. (2025) explore the evolution of EEG-based emotion recognition systems — from academic theory to real-world application. The paper reviews 64 studies under PRISMA guidelines, analysing the latest breakthroughs in neural network models, signal processing, and ethical implications.
Most promising were deep learning architectures like CNNs and RNNs, especially when paired with physiological signals (e.g., heart rate, skin conductance). Multimodal setups consistently outperformed EEG-only models — some achieving over 90% classification accuracy. Signal processing advances, including spectral feature extraction, functional connectivity, and frontal asymmetry analysis, further enhanced system precision.
Despite the progress, real-time application remains a challenge. High computational demands, inconsistent emotion labeling, and lack of protocol standardisation hinder scalability. The authors also raise essential ethical questions around data security, emotion profiling, and bias in machine learning models — especially when used in healthcare, education, or marketing contexts.
This review stands as both a benchmark and a call-to-action: without stronger infrastructure, ethics frameworks, and collaborative standards, emotion recognition tech may stagnate in labs rather than thrive in everyday life.
From Neural Networks to Emotional Networks: A Systematic Review of EEG-Based Emotion Recognition.
Published in Brain Sciences, February 2025
