Protect the subject
Use informed consent and random IDs. Keep identifying information outside the experiment data.
MEMBER LAB / BCI × AI
Explore the mind.
Test the signal.
Build your own Remote Viewing Imaging System as an experimental workflow for EEG, voluntary subject reports, and AI-assisted visualization.
One experiment. Five separate records.
Record the signal
Capture non-invasive EEG, timestamps, device settings and quality flags. Preserve raw files.
Freeze the observation
Record notes or sketches before seeing target data. Hash and time-stamp the response.
Label the output
AI visualizes only approved notes. Keep prompts, seeds, model versions and outputs separate.
Score blindly
Use randomized decoys, blinded raters, preregistered scores and held-out sessions.
Choose a practical stack.
01 / OPEN RESEARCH
OpenBCI + Python
OpenBCI → BrainFlow → MNE → PyTorch
- Acquire EEG with a supported board.
- Inspect artifacts and baseline features.
- Test supervised models on held-out subjects.
- Keep inference local when possible.
02 / SYNCHRONIZED LAB
EEG + event streams
LSL → XDF → MNE → scikit-learn
- Use a device with a compatible LSL outlet.
- Synchronize EEG with event markers.
- Audit clock drift and dropped samples.
- Benchmark simple classical models first.
03 / VISUAL INTERPRETATION
Notes to illustrations
Consented notes → PyTorch → Diffusers
- Start with synthetic or deidentified reports.
- Generate images from written descriptions.
- Save model version, prompt and seed.
- Never score a generated image as recovered evidence.
OPTIONAL / QUANTUM RESEARCH
Test another
way to compute.
Explore small hybrid models and optimization experiments. Compare cost and accuracy against classical baselines. Quantum computing does not establish remote viewing or imply an advantage for EEG.
Explore 20 quantum blueprints ↗Qiskit
Build small parameterized circuits and simulator baselines. Use IBM Quantum only after checking current hardware access, queue limits and cost.
Official documentation ↗PennyLane + PyTorch
Try hybrid quantum/classical feature classifiers on deidentified features. Keep the same held-out split as your classical model.
Official documentation ↗D-Wave Ocean
Formulate discrete problems such as sensor selection or scheduling as QUBOs. Test local solvers before paid cloud runs.
Official documentation ↗YOUR NEXT EXPERIMENT
Create a research plan.
Choose a stack, define a measurable question, and export a protocol. This planner runs in your browser. Enter no names, health records, or raw EEG. It does not connect to a headset or run AI models.
Use non-invasive, vendor-supported equipment and follow its instructions. Do not stimulate the brain or modify electrical safety isolation. This is a research planning workspace, not medical diagnosis or treatment. Cloud providers may incur charges; do not upload identifiable neural or health data.