Neuroadaptive Interface 2D: EEG Band-Power Attention Classifier

This is the 2D counterpart to the 3D neuroadaptive-interface scene, computing the identical DSP: 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, then combined into an engagement index (beta over alpha+theta) and a fatigue index ((theta+delta) over alpha+beta). Instead of a mock adaptive panel, this version exposes an explicit threshold-based classifier that turns those two continuous indices into a discrete FOCUSED / FATIGUED / RELAXED state — the decision layer a real neuroadaptive interface runs before it changes anything on screen. Watch the scrolling raw waveform, the live band-power bars and the state banner respond as you change the mental-state preset, signal noise, eye-blink artifacts and the classifier's own thresholds.