Scientists have connected 200,000 human neurons to a chip. And he made them play ‘Doom’

If they tell us that human neurons are playing ‘Doom’, the first thing we would think of is science fiction. However, that is exactly what the Australian company Cortical Labs has shown with your CL1 system: about 200,000 live neurons grown on an array of electrodes on a chip, capable of receiving information from the game and responding through electrical patterns. We are not talking about conventional artificial intelligence, but rather biological tissue interacting with software through an interface designed for that purpose.

Human neurons and ‘Doom’. The demo isn’t just launching the game and letting something random happen. In the material shared by Cortical Labs, those responsible explain that the system receives signals from the video game environment and generates electrical patterns that translate into the character’s actions. This is a form of learning in which the system modifies its response depending on the result obtained. The key here is not skill, but the ability to adapt, which, according to the company, they are managing to train and mold in real time.

How the interaction loop is established. For the experiment to work, it is not enough to display images on a screen. According to CTO David Hogan, an independent developer managed to convert the game’s visual signal into “electrical stimulation patterns” that are applied directly to the cell culture. These stimuli provoke electrical responses in neurons, and certain firing patterns translate into specific actions within ‘Doom’. In this way, the system creates a closed loop in real time in which each decision has an immediate effect on the virtual environment.

look back. In 2021, the same company managed to make a system based on more than 800,000 neurons play ‘Pong’an experiment that required years of scientific work and specific training. That precedent laid the foundations for what would later become the CL1, the equipment presented at the Mobile World Congress in 2025 as the world’s first commercial biological computer. As we explained at the time, the system combines neurons grown on silicon with software called biOS, responsible for exchanging electrical information with living tissue.

It is advisable to adjust expectations. The system, it should be noted, falls far short of advanced human performance. Brett Kagan of Cortical Labs emphasizes that the experiment is not intended to replicate a miniature brain, and rejects the direct comparison: “Yes, it is alive, and yes, it is biological, but it is actually used as a material that can process information in very special ways that we cannot recreate in silicon.” The emphasis, therefore, is not on skill, but on the type of processing that this biological substrate allows.

Starting point. In the video, the team encourages researchers and developers to interact with the CL1 open API. Cortical Labs hopes to address progressively more demanding tasks than a classic video game, although the video itself also recognizes that there is room to fine-tune the feedback of successes and errors. For now, what we have is a proof of concept that shows potential, but whose path will depend on what others manage to build on this platform.

Images | Cortical Labs

In Xataka | Sam Altman has spent his entire life saying one thing and doing exactly the opposite. And this time it didn’t even take 48 hours.