Neuroadaptive Interface: EEG Band-Power Attention Classifier

A synthetic scalp EEG signal — built from delta, theta, alpha, beta and gamma oscillators — is decomposed back into band power with five real second-order bandpass filters running at 256 Hz, exactly the front-end used before any real neurofeedback or brain-computer-interface classifier. The resulting engagement and fatigue indices drive a closed-loop mock UI panel that changes its own font size, contrast and complexity live, the same feedback loop that lets adaptive interfaces enlarge text or simplify a layout when EEG indicates visual fatigue or cognitive overload. A 3D scalp model shows eight 10-20-system electrodes color- and size-coded by local band power, with adjustable mental-state presets, signal noise and eye-blink artifacts to see how the classifier degrades under realistic conditions.