Biotech Breakthrough: Lab-Grown Human Brain Cells Learn to Play Classic Game Doom
In a striking demonstration of biological computing, Australian startup **Cortical Labs** has successfully trained approximately **200,000 living human brain cells**—grown on a specialized microchip—to play the iconic 1993 first-person shooter *Doom*.
The achievement, announced in late February 2026 via a company video, builds on the team’s earlier 2022 experiment where a larger cluster of neurons (around 800,000 to 1 million) learned the simple paddle game *Pong*. That “DishBrain” system captured global attention for showing how lab-cultured neurons could adapt and “learn” through electrical feedback.
The new setup uses the company’s **CL1** biological computer, a multi-electrode array chip that interfaces directly with the neurons. Game data from a *Doom* variant (often based on the open-source Freedoom for compatibility) is translated into electrical signals sent to the cells. In response, the neurons fire patterns that the system interprets as controls—moving the character, turning, or shooting enemies.
Independent developer Sean Cole reportedly used Python scripting and the Cortical Labs API to implement the *Doom* integration in about a week, highlighting how accessible the platform has become for experimentation.
The cells don’t “see” the screen or play like a human; instead, they respond to sensory-like electrical inputs and receive feedback (rewards or penalties) to refine their behavior over time. Performance remains beginner-level—”like someone who’s never touched a computer,” per the company—but the jump from *Pong*’s 2D simplicity to *Doom*’s 3D navigation, threat detection, and action demonstrates significant progress in neuronal adaptability.
Cortical Labs positions this as proof-of-concept for “synthetic biological intelligence,” a hybrid of living neurons and silicon that could offer energy-efficient alternatives to traditional AI for certain tasks. The CL1 units, commercially available since 2025, cost around $35,000 each and are already shipping to researchers.
While far from conscious gaming or sci-fi sentience, the experiment raises fascinating questions about learning, computation, and the boundaries between biology and technology. As one researcher noted in the announcement video: “They are learning.”
This “Can it run Doom?” milestone—long a tongue-in-cheek test for unconventional hardware—has now officially entered the realm of wetware.