A brain-computer interface has enabled two people with paralysis to type at speeds approaching what able-bodied people achieve on a standard keyboard. One participant with a spinal cord injury reached 110 characters per minute — roughly 22 words per minute — with a word error rate of just 1.6 percent.
The results, published in Nature Neuroscience on March 16, represent a significant step forward for communication technology in people who have lost the use of their hands and voice.
How It Works
The BrainGate system uses microelectrode arrays — tiny sensors smaller than a grain of rice — implanted in the motor cortex, the region of the brain that controls movement. Rather than typing by imagining handwriting or using eye movements to select letters one by one, participants imagined making finger movements as if they were typing on a standard QWERTY keyboard.
Each of the ten fingers maps to three positions: extending upward, pressing down, or curling into the palm. These 30 distinct finger positions encode all 26 letters, a space key, and punctuation marks. When participants attempted these finger movements without any physical motion, the electrodes detected the corresponding brain activity.
A neural network interprets the signals in real time, while a language model refines the output into coherent text. The system works continuously at whatever pace the user chooses.
The QWERTY layout wasn’t arbitrary. Most paralyzed individuals already have mental maps of the keyboard from years of typing on phones and computers. And QWERTY errors typically occur between neighboring keys — substitutions unlikely to form valid words — making it easier for the language model to catch mistakes compared to handwriting-based systems, where similar-shaped letters create ambiguity.
Results From Two Participants
Two participants tested the system as part of the BrainGate2 clinical trial. Both used the device from their homes, not a research lab.
Participant T18, who has a cervical spinal cord injury, received 384 electrodes across both brain hemispheres. They achieved 110 characters per minute with a 1.6 percent word error rate — performance the researchers describe as comparable to able-bodied typing accuracy. Calibration required as few as 30 sentences.
Participant T17, who has ALS, received 128 electrodes on one side of the brain. They reached 47 characters per minute — still significantly faster than eye-gaze alternatives. The difference in electrode count likely explains the performance gap between the two.
The Problem It Solves
Eye-gaze tracking is currently the primary communication method for people with severe paralysis. Users look at letters one by one to spell words — a slow process that causes eye strain and fatigue. Many people with advanced ALS cannot reliably operate eye-gaze systems at all. Abandonment rates are high.
“For many people with paralysis, when losing use of both the hands and the muscles of speech, communication can become difficult or impossible,” said Daniel Rubin, a senior author on the study from Mass General Brigham. “BCIs are on track to become an important new alternative to what’s currently offered.”
Limitations
The study only tested two participants with different conditions, electrode counts, and placements. Future research needs to establish broader applicability.
Neural signals also drift over time. Decoders trained on one day’s data degraded after several days without recalibration. The results remained usable, but future versions will need to incorporate automatic decoder updates during normal use.
The system currently requires surgical implantation of electrodes — an invasive procedure that limits who might benefit and raises questions about long-term reliability.
What This Means
The immediate application is communication. The research team envisions future versions integrating directly with email, messaging applications, and text-to-speech output. They’re also working on wrist gestures for accessing numbers, special keys, and password entry.
Longer term, the same approach could enable more complex movements. “Decoding these finger movements is also a big step toward being able to restore complex reach and grasp movements for people with upper extremity paralysis,” said Justin Jude, the study’s first author.
The BrainGate consortium, led by Brown University’s Carney Institute for Brain Science, has been developing implantable BCIs since 2004. This latest result suggests the technology is maturing from laboratory demonstrations to something that might actually help people in their daily lives — though commercial availability remains years away.