A tiny hush fills the lab—at least for the minutes when the monitor lights up and the monkeys begin to “move” without moving at all. Researchers fitted three rhesus macaque monkeys with brain-computer interfaces, and the animals successfully navigated a variety of virtual worlds using only their thoughts.
The experiment, described by Peter Janssen of KU Leuven in Belgium and colleagues, hinges on where the brain signals are recorded. Each monkey received three separate implants, and each implant contained 96 electrodes placed in the primary motor cortex as well as dorsal and ventral premotor cortex. The first region is a familiar target in BCI work tied to physical movement. The other two are believed to contribute to planning movement in a more abstract way—less “do this now,” more “prepare the idea of doing this.”
Those electrical signals were interpreted by an AI model and used to control VR avatars as the monkeys watched a 3D monitor. In initial tasks, the animals controlled a sphere sliding across a virtual landscape from a fixed point of view. Then the challenge shifted. In a more game-like setup, the monkeys controlled animated monkeys from a third-person viewpoint—something like steering while your perspective is no longer locked to the camera angle you’re used to. Researchers say subsequent tests have pushed the interface into virtual buildings, with the monkeys opening doors and moving from room to room.
There’s a reason Janssen sounds especially encouraged. In many earlier BCI trials involving humans, people have had to think about a physical movement—raising or lowering a finger, for example—to move a cursor. Janssen believes the sensor placement in the monkeys may tap into something higher level and more intuitive. “We cannot ask these monkeys, of course, but we just think that it’s a more intuitive way of controlling an a computer, basically,” he said, adding that current BCIs can feel like “trying to move your ears,” a description that captures how unfamiliar and frustrating they can be for people learning to use them.
In principle, the work could matter for patients with paralysis. Janssen argues that, once researchers pinpoint exactly where the equivalent areas are in the human brain, the approach might allow people to explore virtual environments—or more practically, control electric wheelchairs—without needing to map their intentions to awkward, learned finger-like signals. But that next step isn’t immediate. “There’s a bit of work necessary to know exactly where to implant a human because a lot of these areas are not very well known in humans, where they are exactly,” Janssen said. “But once we figure that out, it should be possible. It should actually be easier because you can explain to the human what they are supposed to do.”
Andrew Jackson at Newcastle University in the UK highlighted another feature that’s easy to overlook: the monkeys were able to control movement from different viewpoints and across different contexts in the same overall way. If the system taps brain activity that already encodes movement more abstractly, it may help the interface stay flexible—less like memorizing one fixed mapping, more like adapting to a new game. Jackson compared it to pressing buttons in different contexts. “I’ve
got a bunch of different buttons I can press, and in different games I have to work out the specific mapping between those different buttons and and the particular game,” he said. “But it’s a pretty easy thing to do because there’s only so many combinations I need to try. If the new game actually involved me putting down the controller, going over and opening my fridge or something, then it would be much harder.”
Even with this promising direction, the path from lab VR to everyday control is uneven—BCIs have shown remarkable feats, but also limitations and mechanical fragility. Several simpler BCIs in humans have already been carried out, including a trial in which a man with paralysis flew a virtual drone through an obstacle course by thinking about moving his fingers, and another where a person imagined writing with a pen and brain signals were converted into text.
And in 2024, Neuralink announced it had installed its BCI in a human for the first time, enabling cursor control—though it was later revealed that after just a month, 85 per cent of electrode threads had shifted, sharply reducing the person’s ability to control a computer. Neuralink has faced criticism for alleged animal cruelty in its experiments, which Musk denied, and it had faced US government investigations before it seemingly stalled under President Donald Trump.
The new monkey work is published in Science Advances (DOI: 10.1126/sciadv.adw3876). Whether abstract movement planning in motor-adjacent regions translates cleanly to human experience is still an open question. But the fact that these animals navigated virtual spaces—switching perspectives and performing actions like opening doors—suggests the brain may be more flexible than we usually assume, at least when we’re reading it carefully enough.
To teach in the time of ChatGPT is to know pain
Artemis II astronauts splash down off San Diego after lunar loop
Genetics test could, in principle, identify which identical twin fathered a child