A team of researchers at Meta has created a revolutionary new type of computer interface: a wristband that reads the tiny electric signals your brain sends to your muscles. It doesn’t need any surgery or electrodes. It’s just a sleek band you put on your hand.

The device, which communicates with a computer using a Bluetooth receiver, facilitates a new type of low-effort control for a wide range of computer interactions. People can even type using the device, though more slowly than at a keyboard.

Your Wrist Becomes a Computer Terminal

It’s an idea that shows up in a million sci-fi stories: controlling a computer simply by gestures or moving your hand. But now, it’s no longer science fiction. When you move your hand, your brain sends an electrical signal to the muscles. This signal, though invisible to the eye, is detectable just beneath the skin. The wrist, packed with muscles that control hand and finger movements, offers the perfect spot to do that.

The device is a slim wristband packed with 48 gold-plated electrodes that detect these signals — called surface electromyography or sEMG — and sends them via Bluetooth to a computer. There, deep learning models trained on data from thousands of volunteers decode the intent. A swipe, a pinch, a letter being “written” in the air tell the computer what you want to do.

The system works for all users. It doesn’t need to be trained for each user. The models are generic and adapt on the fly.

The study involved up to 6,627 participants, depending on the task. Participants navigated menus by rotating their wrists, made selections through pinches and taps, and wrote letters. In all use cases, they were successful in controlling the computer. When writing, for instance, they typed at a speed of 20.9 words per minute. While this is slower than the average smartphone typing, which is usually over 33 words per minute, it’s not that far away, especially considering the lack of experience people had with the system.

Great Potential for People with Disabilities

One of the most promising aspects of the technology is its potential to support people with disabilities. Because the wristband can detect even subtle muscle signals, without requiring visible motion, it could be tailored to people with paralysis or other motor impairments. In fact, the researchers are already looking ahead to accessibility applications. The wristband might one day let someone with limited mobility control a phone, computer, or even robotic tools using only tiny, effortless muscle signals.

To encourage this next phase, Meta is releasing a public dataset of over 100 hours of sEMG recordings from 300 participants. This is bound to be a valuable resource for the broader scientific and developer community who can keep developing and finessing this technology.

The team’s biggest leap, however, may be conceptual: showing that neuromotor decoding advances like other AI fields. Just like language models improve with more data, these sEMG models got better as they trained on more people. To give an idea of the improvements, with even a small amount of personal data (just 20 minutes) the system’s performance improved by 16%. This suggests that in the near future, these systems could improve substantially and be even more fine-tuned for individual users.

“This is only the start,” write the authors. “We’ve discovered that personalizing handwriting models with minimal fine-tuning for individual participants can lead to a 30% improvement in performance. As we gather more extensive training datasets and deploy models in situations where individuals can customize sEMG models to their unique writing style, sEMG decoding models will continue to advance.”

If the mouse and keyboard brought us into the digital world, devices like this could let us live in it more naturally. As the team puts it, they’re not just building an interface. They’re building a new way for humans to interact with computers.