Brain-Computer Interface

When you move a hand, nerve cells in your brain send out weak electrical signals. These signals can be measured — either with electrodes on the scalp or with fine wires directly in the tissue. A brain-computer interface picks up such readings and translates them into commands for a machine. The detour through muscles and nerves in the arm is eliminated. So anyone who merely thinks about a movement can use it to move a mouse cursor or control a robotic arm. The German name is 'Gehirn-Computer-Schnittstelle,' but the English abbreviation BCI is the common term.

What it means for people with paralyzed hands

The most important use case is medical. After a spinal cord injury or with diseases like ALS, the brain often continues to work completely normally. Only the pathway from the brain to the muscles is interrupted. A BCI builds a new connection — not through the body, but through technology.

In studies, paralyzed patients have used it to type sentences on a screen, guide robotic arms, and control video games. In recent speech studies in the US, transmission rates reached around 60 to 80 words per minute. That is close to normal speaking speed and many times faster than eye-tracking control systems.

A second reason for the great interest is economic. Companies like Neuralink, Synchron, and Precision Neuroscience have raised billions in investor money. Their bets extend far beyond medicine, all the way to controlling everyday devices. Whether this market will actually emerge remains open — the medical application is the proven part.

From electrode to mouse click

It always begins with measurement. In non-invasive systems, a cap fitted with electrodes is placed on the head and measures voltages at the scalp, known as an EEG. This is safe, but imprecise: the skull blurs the signals like a frosted glass pane. Invasive systems place electrodes surgically in or on the brain. They deliver much clearer data, but require a surgical procedure.

Then comes the translation. A program, usually a trained AI model, searches the measurement curves for recurring patterns. To do this, the system is first calibrated: the user repeatedly imagines the same movement while the model learns the associated signals. In the end, it assigns a command to a pattern, such as 'move cursor left.'

It’s important to understand what a BCI cannot do. It does not read thoughts in the sense of sentences or memories. It only recognizes the signal patterns it has been trained on — usually imagined movements or attempted speech. Everything else remains noise to the device.

Between the clinic, stock market news, and gaming accessories

In the news, the term usually appears in connection with study results. Since 2024, Neuralink has implanted devices in several people, while the company Synchron guides its device through a blood vessel to the brain, thereby avoiding open surgery. Such reports move the stock prices of medical technology companies and regularly spark debates about data privacy.

There are also simple EEG headbands sold commercially for meditation or concentration training. They measure rough activity patterns and often cost under 300 euros. Apart from the measurement principle, they have little in common with medical implants. Anyone buying such products should read the marketing claims with skepticism.

A related term is deep brain stimulation, which has been used against Parkinson’s disease for years. It delivers electrical current into the brain instead of reading out signals — that is, the reverse direction. The question of who owns the recorded brain data remains largely unresolved legally. Individual US states such as Colorado have already passed their own laws on this.