Researchers at Cortical Labs have successfully trained living human brain cells to play the video game DOOM [1].
This experiment demonstrates the potential for bio-computing, where biological neurons perform tasks traditionally reserved for silicon chips. The project provides a new framework for studying neurological diseases and the fundamental ways biological systems process information [4, 5].
The project began in 2022 and continued through 2026 [1, 2]. Scientists at the Australian start-up observed the cells in a petri dish, integrating them with a glass chip to facilitate communication between the biological cells and the digital game [1, 3].
In the initial phase of the experiment, the team trained approximately 200,000 brain cells [2]. Later iterations of the study expanded the scale of the biological network, utilizing 800,000 brain cells in a petri dish to interact with the 1990s-era shooter [3].
"Scientists have taught living human brain cells how to play the seminal video game 'Doom'," Cortical Labs said [1]. The process involved creating a feedback loop where the cells received sensory data from the game and sent signals back to control the movement of the character.
This achievement represents a shift in how researchers view the intersection of organic matter and software. By treating neurons as a computational resource, the team has created a hybrid system that can learn and react to environmental stimuli in real time.
"Researchers at Australian lab have taught neurons on a glass chip to play a 90s video game," neuroscientists said [6]. The team continues to explore how increasing the number of cells affects the learning speed, and efficiency, of the biological network.
“Scientists have taught living human brain cells how to play the seminal video game 'Doom'”
The ability to integrate human neurons with digital interfaces suggests a future where biological computing could outperform traditional hardware in specific tasks. While the use of DOOM serves as a proof-of-concept for learning and interaction, the broader implication is the development of 'wetware' that could model human brain function more accurately than current AI, potentially accelerating the discovery of treatments for brain-related disorders.
