Research Explained

Can a Paralysed Person Type as Fast as Someone Who Isn't?

A March 2026 study found two paralysed BrainGate participants typing at 110 characters a minute, calibrated from home in as few as 30 sentences.

A BrainGate participant typing by imagined movement, March 2026. Photo: Brown University.

Losing the ability to speak or move does not mean losing the intention to communicate. That gap, between what a paralysed brain still wants to say and what a paralysed body can no longer do, is what the BrainGate consortium has spent more than two decades trying to close with a device the size of a baby aspirin, implanted directly into the brain's motor cortex. On 16 March 2026, in the journal Nature Neuroscience, the consortium reported its fastest result yet: two participants typing continuous text nearly as quickly and accurately as someone using a keyboard with their hands.

How does a chip in someone's brain actually turn a thought into a typed letter?

The implant is a microelectrode array, originally known as the Utah Array, roughly 100 hair-thin electrodes pushed a few millimetres into the motor cortex, the part of the brain that plans hand and arm movement. When a participant imagines moving their hand to type a letter, the array picks up the resulting pattern of neural activity and sends it to an external decoder, a piece of software trained to translate that pattern into a command, in this case a keystroke on a QWERTY layout mapped to imagined finger movements. Neither participant could actually move their fingers. The system reads the intention to move them.

Is 110 characters a minute actually fast?

For context, it is roughly 22 words a minute, a meaningful fraction of typical two-handed typing speed and well ahead of earlier BrainGate results built around moving a cursor letter by letter across an on-screen keyboard. The two participants in the March 2026 study had lost the ability to type conventionally for different reasons: one had advanced amyotrophic lateral sclerosis, a progressive disease that had taken away voluntary muscle control, and the other had a cervical spinal cord injury causing paralysis below the neck. For both, the study reports a 1.6% word error rate, a figure the authors describe as comparable to able-bodied typing accuracy, achieved after calibrating the decoder on as few as 30 sentences of attempted typing, and done at home rather than inside a research lab.

How long has this technology actually been in development?

Considerably longer than the March 2026 headline suggests. BrainGate was originally developed out of Brown University's Department of Neuroscience, with the first human implant in 2002 in a single participant. An initial clinical trial ran from 2004 to 2006 with four participants who had tetraplegia. By 2006, a participant was using the array to control a computer cursor and open email. In 2012, in a milestone published in Nature, participant Cathy Hutchinson used a robotic arm guided by her own implant to drink from a cup of coffee independently for the first time in fifteen years. The ongoing BrainGate2 trial, running since 2009 across multiple sites including Brown University and the Mass General Brigham Neuroscience Institute, is the platform the March 2026 typing result was built on: the same basic electrode design, considerably better decoding software.

Does the implant keep working as well years after it is put in?

Not uniformly, according to a separate study published in 2025 that followed 14 BrainGate participants and 20 implanted arrays between 2004 and 2019, an average of 2.8 years of follow-up and as long as 7.6 years for some participants. Across more than 2,300 recording sessions, the average array picked up spiking activity from 35.6% of its electrodes, a figure that declined by about 7% between the first and last three months of use, with more than half of arrays still retaining at least a fifth of their electrode yield after three years. Eleven of the 14 arrays kept providing usable movement signals throughout the study. Three did not, and the researchers could not fully explain why, pointing to a combination of possible factors: how the disease underlying each participant's paralysis had progressed, differences in surgical placement, the body's own scar-tissue response to a foreign implant, and manufacturing variation across arrays built over 15 years.

What would it take for this to reach more people?

Justin Jude and Daniel Rubin at Mass General Brigham led the March 2026 study, with Leigh Hochberg of Brown University, who called the result proof of "the strength of academic and university-based researchers working together" on restorative neurotechnology. That collaboration is also the limit: BrainGate remains a research-only implant, available through clinical trials at a handful of academic hospitals, requiring brain surgery, and dependent on hardware whose long-term reliability the 2025 study shows is still not fully understood. Faster typing solves one problem. Whether the electrode itself lasts as long as the person using it needs it to is the one BrainGate's own data has not yet answered.

Institutions in this article: Brown University; Mass General Brigham Neuroscience Institute.

Frequently Asked Questions