Stanford: Decoding Inner Speech from Motor Cortex

• Source: https://med.stanford.edu/news/all-news/2025/08/brain-computer-interface.html
• Type: technical-report
• Institution: Stanford
• Date Ingested: 2026-04-05T20:00:00Z
• Tags: inner-speech, neural-decoding, stanford, motor-cortex, paralysis

Key Contribution

Decoded private inner monologue from motor cortex microelectrode arrays. 4 patients with severe paralysis. Key finding: inner speech patterns are similar but smaller versions of attempted speech patterns in motor cortex.

Summary

Stanford researchers have demonstrated the ability to decode inner speech — the private internal monologue that occurs without any physical movement or vocalization — using microelectrode arrays implanted in the motor cortex of paralyzed patients.

Study Details

• Patients: 4 individuals with severe paralysis (ALS, spinal cord injury)
• Recording method: Intracortical microelectrode arrays (Utah arrays) in motor cortex
• Task: Patients asked to think words silently (inner speech) and also to attempt to speak them
• Analysis: Neural patterns during inner speech compared to attempted speech

Key Finding

• Inner speech = attenuated attempted speech: Neural patterns during silent inner monologue are structurally similar to patterns during attempted speech, but with reduced amplitude
• Same neural substrate: Motor cortex encodes both inner and attempted speech — they share neural circuitry
• Decodable signal: Despite reduced amplitude, inner speech patterns are distinct enough to be decoded
• Implications: BCIs designed for attempted speech may be adaptable for inner speech with sensitivity improvements

Methodology

• Microelectrode arrays capture single-neuron and multi-unit activity from hand/arm motor cortex (speech-related areas)
• Machine learning classifiers trained on attempted speech patterns
• Same classifiers applied to inner speech data with recalibration
• Statistical analysis confirms structural similarity between modalities

Clinical Implications

• Opens path to "thought-to-text" BCIs — communication without any physical effort
• Could benefit patients with locked-in syndrome who cannot even attempt to speak
• Inner speech decoding is inherently private — patients could think messages silently
• Reduces cognitive load compared to current BCIs that require attempted movement

Significance

This is one of the first demonstrations of decoding inner speech — the most intimate form of human communication — from implanted brain electrodes. The finding that inner speech patterns mirror attempted speech patterns (just smaller) is a fundamental neuroscience insight that has immediate engineering implications: existing BCI architectures designed for attempted speech can be adapted for inner speech by improving signal sensitivity. For patients with complete paralysis, this could eventually enable direct thought-to-text communication. The 4-patient study provides compelling evidence, though larger studies are needed to confirm generalizability.