Field Journal #014: Not All BCIs Go Both Ways
From the QIF Field Journal
Date: 2026-02-18 Source: Conversation with Claude about BCI directionality Derivation details: Entry 60 — BCI Limits Equation Synthesis (Part 1)
I assumed most BCIs were unidirectional — either read or write. Turns out that’s mostly true, but the exceptions matter a lot for security.
Read-only (recording):
- BrainGate (Utah array) — motor cortex decoding
- Emotiv, OpenBCI, Muse — consumer EEG
- Stentrode (Synchron) — endovascular recording
Write-only (stimulation):
- Cochlear implants — auditory nerve stimulation
- Traditional DBS leads (older Medtronic models) — stimulate only
Bidirectional (read + write):
- Medtronic Percept RC — DBS stimulation + BrainSense LFP recording
- NeuroPace RNS — detects seizure onset, then delivers responsive stimulation (closed-loop)
- Neuralink N1 — designed for both recording and stimulation, current trials focus on recording
The bidirectional ones are the scariest from a security perspective because they have the full attack surface — an adversary could potentially read AND write. NeuroPace is especially interesting because it’s already running a closed-loop algorithm autonomously inside the patient’s skull: detect pattern, inject signal. That’s the exact pipeline QIF’s threat model covers.
This matters for the BCI Explorer (Entry 013) — directionality should be a visible property. A read-only device has a fundamentally different threat profile than a bidirectional one. You can’t inject signals through a device that only records.
Connected to:
- Entry 013 — BCI Explorer needs to show directionality per device
- I0 Depth (QIF-TRUTH.md) — directionality is orthogonal to depth; a shallow read-only device may be less risky than a deep bidirectional one
This entry is part of the QIF Field Journal, a living, append-only research journal documenting first-person observations at the intersection of neurosecurity, BCI engineering, and neurorights. The journal exists because neural privacy is a right, not a feature. Tools like macshield protect digital identity on networks; this research works toward protecting cognitive identity at the neural interface.
Written with AI assistance (Claude). All claims verified by the author.