Researchers have demonstrated that clusters of living human brain cells can be trained to interact with and play the classic video game Doom, marking another step in the development of biological computing systems. The experiment showed that laboratory grown neural cells could process visual information, respond to feedback, and generate electrical activity that translated into in game actions.
The project was conducted by Australian biotechnology company Cortical Labs, which specializes in integrating living neurons with electronic systems. Scientists used microelectrode arrays capable of both stimulating neural tissue and recording electrical activity, enabling two way communication between the biological material and a computer system. The setup allowed game visuals to be converted into electrical signals while neural responses were interpreted as control inputs.
Researchers translated elements of the game environment into distinct patterns of electrical stimulation delivered to the neural culture. In response, the neurons produced electrical firing patterns that were mapped to simple gameplay actions such as moving, turning, and shooting. The biological system received continuous feedback from the game, allowing it to adjust neural activity in real time.
Although performance remained limited, scientists reported that the neural clusters demonstrated measurable learning behavior. Over repeated sessions, the neurons adjusted their electrical responses in ways that improved their interaction with the digital environment, indicating adaptive processing capabilities.
The experiment builds on earlier research by the same organization involving simpler video game tasks. In 2021, the team demonstrated that a larger neural culture could learn to play the arcade game Pong. That earlier project required more than a year of development to establish stable interaction between biological neurons and computer systems. By comparison, the Doom experiment involved fewer neurons and was completed within a shorter timeframe despite the game’s greater complexity.
Doom, first released in 1993, is considered more computationally demanding than Pong due to its three dimensional perspective, navigation requirements, and combat mechanics. Even so, researchers limited the neural interface to basic control actions to match the processing capabilities of the biological system.
Scientists involved in the project emphasized that the neural cultures do not resemble a fully formed brain and lack consciousness. The cell clusters function as biological processing material rather than sentient tissue. Researchers describe the system as a form of biological hardware that may eventually complement traditional silicon based computing.
Biological neural networks are capable of processing information through parallel electrical signaling pathways that differ from conventional digital architectures. Researchers believe these characteristics could support future hybrid systems combining organic and electronic components for specialized computing tasks.
Potential applications include adaptive control systems, robotics interfaces, and bio digital platforms that require energy efficient pattern recognition. However, significant technical and ethical considerations remain before such systems can move beyond controlled laboratory environments.
The demonstration highlights growing interest in unconventional computing approaches that explore how living tissue can be integrated with electronic systems. While current capabilities remain limited to experimental interfaces, the results suggest biological substrates may eventually play a role in specialized information processing technologies.
