SIGNAL-RESPONSIVE INSTRUMENT — SYNTHETIC PERFORMANCE SCORE

A living score,
woven from signals.

Shape an evolving visual field with a synthetic EEG-like feature; let a separate synthetic ECG beat interval set the pulse. The mappings are authored, adjustable rules—not readings about a performer.

Display metaphor, not a hologram. The primary labels are “synthetic EEG-like feature” and “synthetic ECG-like beat stream”; “mind projection” and “heart projection” are metaphorical display labels only: this does not read thoughts, depict any real person’s mind or heart, or diagnose, monitor, or provide medical guidance. The ECG-like stream represents a synthetic electrical trace; no PPG is included.

PERFORMANCE VIEW — 2D LIVING SCORE

Field / pulse / thread

SYNTHETIC PERFORMANCE MODE

FIELD SPREAD —FRAME DRAW —

ECG R EVENT

EEG INJECTION + τLINK INJECTED · SIMULATION ONLY

FUTURE AUTHORED INPUTSVOICEMOVEMENTMUSICPlaceholders only · no mic, camera, or fabricated measurements

01 / SYNTHETIC EEG-LIKE FEATURE “MIND PROJECTION” · METAPHOR ONLY

Synthetic EEG-like feature

Oscillatory EEG-like signal · arbitrary units

SIGNAL AMPLITUDELOCAL SIMULATION

EEG-like stream10.0 Hz · synthetic rhythm

02 / SYNTHETIC ECG-LIKE BEAT STREAM “HEART PROJECTION” · METAPHOR ONLY

Synthetic ECG-like beat stream

ECG-like electrical trace · arbitrary units

SIGNAL AMPLITUDELOCAL SIMULATION

ECG · electrical, not PPG72 BPM · simulated

SYNTHETIC COUPLING · NO LIVE ANALOG ACQUISITION

ψEEG-like
synthetic stream

INJECTED PARAMETER

β 0.36

SIMULATION ONLY

Coupling strength ββ = 0 → no injected link

NO LINK · 0STRONGER INJECTION · 0.80

Synthetic lag τ180 ms

β scales an EEG component patterned after synthetic beat times, delayed by τ. This is an experimental software injection—not physiology, causal evidence, or a person-specific relationship.

ECG
synthetic electrical trace⌁

Shared-clock / common-noise confound

Turn up a synthetic disturbance added independently to both streams. Similar-looking signals can arise without injected EEG↔beat linkage.

Common disturbance0%

LIVE MODEL SETTINGS

Tune the score

EEG-like rhythm10.0 HzDominant synthetic oscillatorECG beat rate72 BPMSynthetic R–R intervalEEG feature → field0.78 gainSlow envelope controls visual spreadPulse depth gain0.72 gainR–R interval sets pulse spacing; pulse gain sets depth

Analog inputFUTURE STUBNo electrode, device, microphone, or hardware connector is implemented.

KNOWN-GROUND-TRUTH CHECK

Robustness bench

Held-out seeds test recovery of injected β and τ, plus ECG beat alignment, after synthetic timing and sample damage. Generation and fitting use the same synthetic beat-template family, and the estimator receives known clock-offset corrections. The β/τ results are a matched-model implementation check—not a blind estimator or evidence of EEG/ECG generalization.

Relative time offset120 msTimestamp jitter12 msIndependent noise0.12Dropped samples10%

Ready · held-out seeds 23, 41, 67, 89

MEAN |β̂ − β|—recovery error · unitless

MEAN |τ̂ − τ|—recovery error · ms

ECG BEAT ALIGNMENT—matched-beat MAE · ms

Ground truth: β = 0.36, τ = 180 ms, 72 BPM. Matched-model implementation check only: generation and fitting use the same synthetic beat-template family and known clock-offset corrections; not a blind estimator or evidence of EEG/ECG generalization.

What the bench tests (and what it does not)

For each held-out seed, the simulator creates separate EEG-like and ECG-like arrays, applies known relative clock skew, independent timestamp jitter, added noise, and independent sample drops, then estimates β/τ from the EEG stream against ECG-derived synthetic beat times. The fixed model is not tuned on those held-out seeds, but generation and fitting share the same synthetic beat-template family and fitting receives the known clock offsets. This is a matched-model implementation check, not a blind estimator or evidence of EEG/ECG generalization. Beat-alignment error is calculated only for detected beats matched to a ground-truth beat; the detected/true count is reported separately. Common noise is a separate confound example, not an injected β link.

MAPPING CONTROL CHECK

Shuffled-code mapping-control check

Compares mapped field output under signal-driven, time-shuffled, and zero-input conditions. Condition labels receive randomized software codes, but the code/scorer retains their mapping throughout; this is not a genuinely blinded or human-perception study.

Ready · 240 frames · randomized software condition labels

CODE — mapping error —CODE — mapping error —CODE — mapping error —

Mapping error is mean absolute field-spread error versus the signal-driven target. The signal-driven arm is zero by construction (its target is generated by the same mapper from the same frames), so this is a matched-model software check, not independent validation. Frame-draw latency is measured locally by the browser. Shuffled and zero-input arms are explicit controls.