Scientists Train Human Brain Cells on Microchip to Play Doom
Key Takeaways
- ▸Cortical Labs successfully trained 200,000 living human neurons on a microchip to play Doom, building on earlier work with Pong
- ▸The system converts gameplay into electrical stimulation patterns, with neurons responding via electrical signals interpreted as game controls
- ▸Performance remains at beginner level, but demonstrates biological neural networks can learn and adapt to complex 3D gaming environments
Summary
Australian biotech company Cortical Labs has successfully trained living human neurons grown on a microchip to play the classic video game Doom, marking a significant advancement in biological computing. The CL1 biological computer contains approximately 200,000 human neurons cultured on a multi-electrode array that can both receive electrical signals from a computer and send responses back as game control inputs.
The experiment represents a major leap from the company's previous work, where lab-grown neurons learned to play the simpler game Pong. Doom's 3D environments and enemy encounters present substantially greater complexity, requiring researchers to translate gameplay into electrical stimulation patterns that neurons can process. When enemies appear on screen, corresponding electrodes stimulate specific regions of the neural culture, which respond with electrical signals interpreted as movement or shooting commands.
While the neurons demonstrate basic learning and adaptation capabilities, their gaming performance remains rudimentary. Researchers acknowledge the system plays "a lot like a beginner who's never seen a computer" and is far from esports-level competency. The work nonetheless represents a novel intersection of neuroscience, biotechnology, and computing, potentially opening new avenues for understanding how biological neural networks process information and learn complex tasks.
- The research advances biological computing and provides new methods for studying how neural networks process information
Editorial Opinion
This research sits at a fascinating crossroads of neuroscience, gaming, and bio-computing, pushing boundaries in all three domains. While the practical applications remain unclear and performance is admittedly basic, the ability to interface living neurons with digital systems in increasingly complex ways could eventually inform both artificial intelligence design and our understanding of biological learning. The choice of Doom as a test platform is both culturally resonant and scientifically appropriate—its 3D complexity provides a meaningful challenge beyond simple games while remaining computationally tractable for biological processing.
