Single-Channel EEG-Based Cognitive Load Assessment in Online Learning: A Hybrid Deep Learning Approach

This paper evaluates whether a single-channel consumer EEG device, the NeuroSky MindWave Mobile 2, can distinguish easy from difficult educational-video content. Using a public dataset with nine usable learners, it combines raw EEG waveforms and band-power features in a CNN+LSTM+Attention model. Within-subject accuracy reaches up to 78.5%, compared with 55% for conventional feature-based classifiers, while regularization keeps validation accuracy around 68–73%. The authors characterize the work as a feasibility study, caution against optimistic within-subject evaluation, and release a reproducible pipeline plus a notebook tool that maps estimated cognitive load onto video timelines.