A neuroadaptive interface closes the loop between an operator's brain state and the automation itself: instead of a human moving a lever, spectral EEG power drives the controller directly. This simulator synthesizes the three classic scalp bands (θ 4–7 Hz, α 8–12 Hz, β 13–30 Hz) as a function of the current cognitive workload, and folds them into the engagement index used in NASA's original adaptive-automation research:
EI = β / (α + θ) (Pope, Bogart & Bartolome, 1995)
Higher task load → α suppression (event-related desync) + β & θ rise → EI ↑
Under-load/boredom → α rebounds, β & θ fall → EI ↓
EI is smoothed with an exponential moving average (time constant τ ≈ 1.5 s, matching a sliding-window spectral estimate) before it reaches the controller — raw single-epoch EEG is far too noisy to switch automation on directly.
EMA(t) = EMA(t-dt) + (EI(t) − EMA(t-dt)) · (1 − e^(−dt/τ))
Controller (3 discrete levels, hysteresis + cooldown):
EMA > upperTh → automation level += 1 (offload the overloaded operator)
EMA < lowerTh → automation level -= 1 (return control, re-engage a bored operator)
upperTh = 0.55 + sensitivity, lowerTh = 0.30 − sensitivity
Each automation step removes roughly a third of the effective task difficulty the operator experiences next — a genuine feedback loop, not a one-shot readout: automation state changes workload, workload changes EEG, EEG changes automation state. Predicted performance follows the classic Yerkes–Dodson inverted-U, peaking at a moderate engagement level and falling off at both boredom and overload:
performance(EI) = 100 · exp( −(EI − EI_opt)² / (2σ²) )
- Task difficulty — the demand placed on the operator before any automation offload.
- Controller sensitivity — widens or narrows the hysteresis band, trading switching frequency for lag.
- EEG noise — amplitude of trial-to-trial spectral variability layered onto each band.
- Loop toggle — switching it off freezes automation at "Assisted" so you can compare a closed loop against fixed automation.
This is the same architecture used in NASA Langley's engagement-index adaptive-automation studies and in modern neuroadaptive cockpit/vehicle-automation research — not a mind-reading device, but a feedback controller reacting to a genuine, well-studied EEG correlate of cognitive state.