Two Different Ways to Grow a New Hand

On January 28, 2024, surgeons at Barrow Neurological Institute in Phoenix implanted a coin-sized device called the N1 into the skull of Noland Arbaugh, paralyzed from roughly the shoulders down since a diving accident in 2016. Seven weeks later, in a livestream broadcast to the public, Arbaugh moved a cursor across a screen using nothing his body could still move on its own. Asked what it felt like, he did not describe practice or trial and error. “It was like using the Force on the cursor,” he said, “which was such a wild experience the first time it happened.” He was describing something that arrived complete the first time he tried, not a skill he had learned.

Ten months later and roughly seven thousand miles away, a different man received a different implant for a different kind of paralysis. Dong Hui, paralyzed since a car accident around 2018, was fitted in November 2024 with NEO, a coin-sized device developed by the Shanghai firm Neuracle Technology with Tsinghua University, resting on the surface of the brain rather than piercing into it. Where Arbaugh’s account skips straight to a completed sensation, Dong Hui’s account is a diary of specific numbered attempts. “On the ninth day of my training,” he said, “my right hand successfully grabbed a ball without the glove. That was a miraculous moment.” He did not report using a new sense on first contact. He reported nine days of failing, and then, on the ninth, succeeding.

Both men are describing the same general category of event: a device reading neural activity and turning it into a movement a paralyzed body could no longer produce on its own. Their devices are comparable, implanted within the same year to restore much the same kind of lost motor control. The technology is broadly alike. The stories each man tells about getting control of it are not, and only one of the two matches what philosophers who have studied how tools are taken into the body would have predicted.

In 1945, the philosopher Maurice Merleau-Ponty described a blind man’s cane in Phenomenology of Perception to make a specific, narrow point about how tools become part of a body rather than remaining objects held by one. In the cane’s early use, he observed, a person feels the pressure of the handle in the palm, an object being gripped. Only with sustained use does the felt location of sensation migrate outward, until the user reports feeling the curb and the doorway directly, through the cane, the cane itself dropping out of conscious awareness entirely. His point was that incorporation takes practice, because what has to change is not only behavior but where, in the felt map of the body, sensation is located.

Dong Hui’s nine days fit that model closely. A tool, in his account, remained a tool, an external thing he was operating, right up until a specific practiced attempt succeeded and something in how the device felt apparently shifted. Arbaugh’s account fits it barely at all. He reports the completed sensation, wanting something and having it happen, on his very first attempt, with no described interval in which the cursor felt like a foreign object he was operating from outside. If Merleau-Ponty’s account is right, one of these two reports describes something no tool user had experienced before.

There are two ways to read the mismatch, and neither is comfortable. The first is that Arbaugh is honestly wrong about his own timeline: a livestream seven weeks after surgery is not a report of the first attempt but a compressed retelling, with the trial and error already smoothed out of memory by the time a camera was on him. The second is that he is right, and that this kind of tool really is different. The philosopher Andy Clark, in Natural-Born Cyborgs (2003), argued that technology coupled directly to the nervous system might shorten or even skip the incorporation period other tools require, precisely because it skips the sensory and motor loops, the hands and eyes, that a cane still has to pass through. If Clark is right, Arbaugh’s report might be accurate after all, and Dong Hui’s slower nine-day arc might be the one that reflects an older, more conservative interface still routed partly through effort and practice rather than direct decoding.

What happened to Arbaugh afterward makes the second reading harder to hold. In May 2024, Neuralink disclosed that most of the N1’s threads had retracted from his brain, reportedly cutting the usable signal by around 85 percent. The company recalibrated the decoding software rather than replacing the hardware, and by his own later account the device kept working, though by then the story had stopped being one of instant, complete transfer and started looking more like Dong Hui’s: adjustment, retraining, a system being coaxed back toward function rather than simply obeyed. The instant version of the story belonged to the first seven weeks. The version that has lasted, for both men, looks much more like practice.

Of the two, Dong Hui’s nine days may turn out to be the more accurate report of what a hand has to go through to become new again, and Arbaugh’s Force may turn out to have described less a working device than the particular, forgivable excitement of week seven, mistaking the surprise of a working demonstration for a claim about how it had been acquired.