Inner speech in motor cortex and implications for speech neuroprostheses

The 60-second view

What problem?

Attempting to speak is tiring for people with paralysis, and the prospect of decoding private thought raises a real concern.

What did they do?

Recorded multi-unit activity in four participants, compared inner with attempted speech, and decoded imagined sentences in real time.

What changed?

Found a shared representation plus a distinct 'motor-intent' dimension that separates the two, and showed strategies that stop a speech BCI from picking up private inner speech.

Why does it matter?

It treats privacy as a design requirement with a measurable solution, which is the kind of evidence regulators and users will ask for.

Method

Participants

Four BrainGate2 participants

Interface

Intracortical microelectrode arrays in motor cortex

Signal

Multi-unit spiking activity

Task

Attempted speech, inner speech, sequence recall and counting

Decoder

Real-time sentence decoding with intent-gating strategies

Results

The published abstract does not state headline figures, so none are quoted here. See the primary source for quantitative results.

Strength of evidencePeer reviewed, four participants.

Figures are quoted from the published abstract and were checked on 3 Oct 2026.

Limitations

• Free-form inner speech was only partly decodable.
• Small cohort.

What this could enable

Less effortful speech BCIs, and concrete engineering answers to a mental-privacy question.