Invasive brain implant

Our brain works with weak electrical signals. Every movement, every thought generates such signals between nerve cells. An invasive brain implant is a small electronic component that picks up exactly these signals. “Invasive” means that surgery is required for this: the skull is opened and the component then sits directly on or in the brain tissue. The counterpart consists of methods where only a cap with sensors rests on the head. These are more comfortable to wear, but receive a much fuzzier signal because the skull and skin lie in between.

What patients regain with it

For some people, such an implant is the only remaining way to communicate. After an accident with a severed spinal cord, the brain keeps functioning, but the commands no longer reach the muscles. An implant can intercept these commands and forward them to a computer. Study participants have used this to move mouse cursors, type messages, and control robotic arms.

A second application works in the opposite direction. Instead of reading signals, the implant itself emits weak electrical impulses. In Parkinson’s disease, this often significantly calms the typical trembling movements. This form of deep brain stimulation is not a vision of the future but has been approved since the 1990s and has been used in more than a hundred thousand people worldwide.

At the same time, the topic is sensitive. Anyone reading out brain signals is working with the most intimate data a person has. On top of that come the risks of any brain surgery: infections, bleeding, scar tissue. That is why doctors, lawyers, and ethicists argue over when the benefit even justifies such a risk.

From electrode to letter on the screen

At the center are electrodes, that is, fine conductive wires or plates. They lie so close to the nerve cells that they register their tiny voltage fluctuations. Some systems use a few dozen electrodes, modern prototypes more than a thousand. The more measurement points there are, the more precisely one can distinguish what the person actually intends.

The measured signals are initially just noise made of numbers. This is where software comes in that learns from examples. During a training phase, the patient imagines certain movements while the system records along. This way, the program learns which signal pattern belongs to which intention. Afterward, it can predict the desired movement from new signals.

A comparison helps: the implant is the microphone, the software the translator. A good microphone alone is not enough if no one understands the language. This is precisely why the field has moved so quickly in recent years. The electrodes got better, but the real leaps came from more powerful learning methods.

Neuralink, competitors, and the headlines about them

The best-known company is Neuralink, founded by Elon Musk. In 2024, it implanted a device in a human being for the first time; the patient used it to control a computer and play chess. Such reports have long since reached the financial news, because a market is emerging around the field. Less well known are competitors such as Synchron, Blackrock Neurotech, or Precision Neuroscience, some of which have been working with patients for years.

Anyone reading such reports should keep two things separate. Approved medical devices such as Parkinson’s stimulators are part of everyday clinical practice. Implants for mind-reading aimed at paralyzed patients, on the other hand, are still in studies with a handful of participants. It will likely be years before broad approval is reached.

A common misconception is also that such implants could read thoughts in the literal sense. They recognize movement intentions and planned speech quite well, because clearly delineable brain regions become active for these. No system as yet reads out arbitrary inner thoughts. Likewise, the vision of using them to make healthy people smarter is currently marketing, not medical reality.