Human Brain Cells on a Chip Learn to Play Doom in One Week
Key Takeaways
- ▸Cortical Labs successfully programmed human brain cells on a chip to play Doom in one week using a new Python interface, dramatically reducing the time and expertise required compared to their 2021 Pong demonstration
- ▸The neuronal chip learned faster than traditional silicon-based machine learning systems, though performance remains below human levels
- ▸The breakthrough brings biological computers closer to practical applications like controlling robotic arms, with Doom's complexity serving as a better proxy for real-world tasks
Summary
Australian biotech company Cortical Labs has achieved a significant milestone in biological computing by programming human brain cells grown on microelectrode chips to play the classic first-person shooter game Doom. Independent developer Sean Cole accomplished this feat in approximately one week using a newly developed Python interface, marking a dramatic improvement over the company's 2021 Pong demonstration that required years of painstaking effort. The neuronal computer chip, utilizing roughly 800,000 living brain cells, learned to play Doom faster than traditional silicon-based machine learning systems, though its performance remains well below human players.
The breakthrough represents a major leap in accessibility and programmability of biological computing systems. Unlike the earlier Pong project that demanded specialized biological expertise, the new Python interface allows developers with minimal biology experience to program these living neural networks. The chip performed better than random chance but has significant room for improvement with more advanced learning algorithms, according to Brett Kagan of Cortical Labs.
Experts view this development as a critical step toward practical real-world applications of biological computers. The complexity of Doom—requiring real-time decision-making, pattern recognition, and navigation through uncertain environments—more closely resembles practical tasks like controlling robotic arms. Researchers at the University of Reading are already exploring similar applications using hydrogel-based biological computers. While fundamental questions remain about how neurons process visual information and understand game objectives without sensory organs, the demonstration showcases the unique information-processing capabilities of biological materials that cannot be replicated in silicon.
- Experts highlight the system's ability to handle complexity, uncertainty, and real-time decision-making as key advantages for future hybrid computing applications
Editorial Opinion
This development represents a genuine inflection point for biological computing—not because brain cells can play video games, but because the technology has crossed the threshold from laboratory curiosity to programmable platform. The one-week timeline using standard programming tools signals that wetware computing might finally escape the confines of specialized research labs. However, the field still faces fundamental questions about how these systems process information and whether they can scale beyond proof-of-concept demonstrations to deliver on the promise of energy-efficient, adaptable computing for robotics and other real-world applications.
