Lab-Grown Human Brain Cells Learn to Play Doom in Groundbreaking Biocomputing Experiment

In a development that blurs the line between biology and computing, Cortical Labs has announced that its CL-1 chip—composed of 200,000 lab-grown human neurons—has learned to play the iconic 1993 video game Doom. This breakthrough represents a dramatic evolution from the company's 2022 demonstration, when it first showed brain cells playing the simpler game Pong.

From Pong to Doom: A Quantum Leap in Biological Computing

The achievement marks a significant step forward in the field of biocomputing, where living neurons are integrated with electronic systems to create hybrid processing units. Interestingly, while the 2022 Pong experiment used 800,000 neurons, the current Doom system operates with just 200,000 neurons—demonstrating remarkable efficiency improvements in how the neural networks are trained and utilized, despite Doom's greater complexity requiring more sophisticated decision-making and spatial awareness.

According to reports, the system works by translating visual data from the game screen into electrical stimulation patterns that the neurons can process. The cultured brain cells then respond with their own electrical signals, which are interpreted as control inputs for the game's protagonist, known as Doomguy. This bidirectional communication between biological and digital systems represents a remarkable feat of neuroengineering.

How It Works: The Science Behind Biological Gaming

The CL-1 chip, often referred to as "DishBrain" in earlier research publications, consists of approximately 200,000 human neurons grown in a laboratory dish. These neurons form natural connections with each other, creating a rudimentary neural network similar to—but far simpler than—those found in human brains.

The neurons are cultured on a multi-electrode array that can both stimulate the cells with electrical impulses and record their responses. When Doomguy encounters an enemy or obstacle on the screen, that visual information is converted into specific patterns of electrical activity delivered to the neurons. The cells process this information and generate their own electrical responses, which the system interprets as movement commands.

What makes this particularly remarkable is that the neurons appear to learn from feedback. When Doomguy takes damage or fails to progress, the neural culture receives feedback signals that help it adjust its responses over time—a process analogous to how biological brains learn through trial and error.

Implications for AI and Neuroscience

This experiment has profound implications for multiple fields. For neuroscientists, it provides a unique window into how neural networks process information and learn from experience. Unlike traditional artificial neural networks in computers, these are actual biological neurons forming real synaptic connections.

For the AI industry, biological computing presents an intriguing alternative to silicon-based processors. Neurons are incredibly energy-efficient compared to traditional computer chips, and they excel at certain types of pattern recognition and adaptive learning that still challenge conventional AI systems.

Dr. Brett Kagan, chief scientific officer at Cortical Labs, has previously stated that these experiments help researchers understand the fundamental principles of intelligence and learning. "We're not trying to create consciousness," Kagan clarified in past interviews, "but rather to understand how networks of neurons process information and adapt to their environment."

Ethical Considerations and Future Directions

The advancement also raises important ethical questions about the nature of biological computing. While these neuronal cultures are far simpler than even the most primitive animal brains and lack anything resembling consciousness, the use of human-derived neurons in computing systems will require careful ethical oversight as the technology advances.

Looking forward, Cortical Labs envisions applications beyond gaming demonstrations. Potential uses include:

• Drug discovery: Testing pharmaceutical compounds on human neurons to predict effectiveness and side effects
• Biosensors: Creating highly sensitive detection systems for environmental monitoring
• Adaptive controllers: Developing biological processors that can learn and adapt in ways that traditional computers struggle with
• Neural interface research: Advancing brain-computer interface technology by better understanding how neurons communicate with electronic systems

The Broader Context of Biological Computing

Cortical Labs is not alone in exploring the frontier of biological computing. Research institutions worldwide are investigating ways to harness biological systems for computation. Some teams are working with bacterial colonies, others with engineered proteins, and still others with synthetic biological circuits.

However, Cortical Labs' use of actual human neurons represents one of the most direct approaches to understanding how biological neural networks function. The choice of Doom as a testing platform—while attention-grabbing—is scientifically sound. The game's 3D environment, enemy AI, and resource management create a complex problem space that tests the neurons' ability to learn, remember, and make decisions.

What's Next?

While watching neurons play Doom makes for compelling headlines, the real significance lies in what this technology could enable. As the field of biological computing matures, we may see hybrid systems that combine the efficiency and adaptability of biological neurons with the precision and scalability of silicon processors.

For now, Cortical Labs continues to refine its technology, exploring more complex tasks and working to understand the fundamental principles that govern how these neuronal networks learn and adapt. The company has indicated that future experiments will involve more sophisticated challenges that push the boundaries of what biological computing can achieve.

The question is no longer whether biological systems can be integrated with digital technology, but rather how far we can take this integration—and what new capabilities will emerge from the marriage of wetware and software.

As this field evolves, one thing is certain: the future of computing may be more biological than we ever imagined.