AI-Based Classification Framework for EEG Pattern Analysis Across Brain Lobes in Resting and Cognitive States

This study employs EEG to explore neural mechanisms underlying critical thinking by examining brainwave activity in both resting and critical thinking states. It aims to identify EEG patterns, particularly in the alpha, beta, and gamma frequency bands, associated with critical thinking skills. Ten healthy participants, aged 20 to 27 and of both genders, took part in the study. They completed four activities: a resting task with eyes closed and minimal movement, and three tasks (Sets A, B, and C) involving ten brain teaser questions each. Data were collected using a 14-channel mobile EEG device, the EMOTIV EPOC+. During preprocessing, a Butterworth bandpass filter was applied to the EEG data. In the feature extraction stage, Power Spectral Density (PSD) methods, including Welch and Burg for linear analysis, and Convolution Operation for nonlinear analysis, were used to derive statistical features such as maximum, minimum, mean, median, mode, standard deviation, and variance. For classification, Decision Tree (DT), K-Nearest Neighbor (KNN), and Multi-Layer Perceptron (MLP) algorithms were applied. Results indicate that the highest classification accuracy across all brain lobes, particularly in beta and gamma signals, was achieved with Welch-based feature extraction, three statistical features, and KNN classification. The gamma signals from the frontal lobe provided the highest accuracy at 89.7%, suggesting that the frontal lobe is the most effective region for critical thinking tasks.