PROJECTS
CORTEX was a three-year project into non-invasive control for amputees and paraplegics, reaching up to 94% intent accuracy. MuseScope is the macOS app for recording EEG, replaying sessions and training the classifiers.
VENTA / WINCHESTERMUSE 2 / EEG / PPG / IMU / REPLAY
Research programme
CORTEX combines EEG capture, signal processing and personal classifier training. Guided recordings provide the training examples. Recorded sessions can be replayed through the processing and prediction workflow to test the classifiers against the same data.
Native macOS application
MuseScope is the native macOS app built for the EEG work. It connects directly to Muse 2 over Bluetooth and displays raw and cleaned EEG, motion, telemetry and available PPG. Sessions are saved as replayable .musesession packages with SQLite, CSV and JSON data. The app also runs guided capture, classifier training and live prediction.
Capture and prediction
MuseScope checks signal quality, packet loss, motion, camera reliability and model readiness before making a prediction. It returns Unknown when those checks fail. Classifiers learn specific personal labels from guided recordings; they do not read arbitrary thoughts.
EEG CAPTURE AND TRAINING
MuseScope handles the Bluetooth connection, guided recordings, signal inspection and classifier training in one application. Recorded sessions can be replayed through custom Rust FIR bandpass filters to compare raw and cleaned signals.
Read how VENTA builds specialist software, or see the government and defence focus.
PUBLIC DATA / ENTITY RESOLUTION
ATLAS collects information on targets from large public data sources. Corporate disclosures, gazette filings and sanctions registers feed an entity-resolution graph that connects organisations, directors and ownership across jurisdictions. Source citation hashes keep each connection traceable, and the system supports offline, air-gapped deployment.
STOCK PREDICTION / QUANTITATIVE MODELLING
AEVIUM is a stock prediction model built on the foundations of CORTEX. It uses Merton jump-diffusion equations and Monte Carlo simulations to model asset trajectories, including sudden market shocks. The model produces 68% and 95% confidence intervals and adapts its volatility estimates to changes in interest rates and liquidity.
ONGOING RESEARCH / THEORETICAL PHYSICS
Σ₀ investigates universe formation inside black holes and the nature of time. The numerical work traces photon paths around rotating black holes using fourth-order Runge–Kutta integration over the Kerr metric. It models frame-dragging and tests non-singular cosmological interior bounce hypotheses, including an interior expansion formulation intended to avoid gravitational singularities.
2024 TO 2025 / EXTERNAL EVIDENCE
A peer-reviewed evaluation compared four consumer wireless EEG systems with research-grade hardware. All four captured alpha activity. The consumer systems had limited bandwidth, and the overall event-related potential timing was distorted.
“Assessing their signal quality is essential.”Lee, Kim and Ahn, Biomedical Engineering Letters
A systematic review covered 58 studies. Only two identified specific artefact categories. The authors also found that auxiliary sensors such as IMUs remain underused, despite their value for detecting movement.
“The relaxed constraints of the acquisition setup often compromise signal quality.”Arpaia and colleagues, Sensors
CORTEX FAQ
CORTEX was Connor Colyer's three-year project into non-invasive control for amputees and paraplegics. Its algorithm reached up to 94% intent accuracy. MuseScope supports the EEG capture and classifier training work.
MuseScope is the native macOS app built for the EEG work. It connects to Muse 2 over Bluetooth, displays EEG and other available sensor streams, records and replays sessions, and runs classifier training and live prediction.
No. The classifiers learn specific personal labels from guided recordings. MuseScope returns Unknown when signal quality or model readiness is too low.
MuseScope records raw packets, four EEG channels, motion, telemetry, available PPG, quality snapshots, physiology values and markers in a replayable session package with SQLite, CSV and JSON data.
MuseScope blocks a prediction when signal quality, packet delivery, motion, required sensors or model readiness fail the label's checks.
No. CORTEX is a research project, and MuseScope is research software. It is not a medical device or diagnostic tool.