Doom on a Brain-on-a-Chip: Living Neurons Learn to Play Doom in Cortical Labs’ Bio-Computing Breakthrough

In this Nature Briefing episode hosted by Benjamin Thompson, a Doom-playing brain-on-a-chip story reveals how biology and computing collide. An Australian team, Cortical Labs, has taught human neurons cultured on a silicon chip to play the classic game through an AI translator that converts game events into neural signals and back again. Building on Pong experiments, the project aims to explore brain-inspired computing and new drug-testing models, while highlighting the enduring appeal of Doom as a research meme. The segment emphasizes energy efficiency, determinism, and the playful yet serious edge of science where games become a laboratory for the brain's potential.

Overview and context

Benjamin Thompson hosts this Nature Briefing segment, which uses a historically meme-worthy idea about running Doom to illuminate groundbreaking work at the intersection of biology and computation. A team in Australia, Cortical Labs, has placed living human neurons derived from stem cells on a silicon chip and connected them to an AI system that translates the game's screen events into electrical signals for the neurons, and then translates the neurons' responses back into in-game actions. The project extends a prior line of experiments that started with Pong and gradually increased in complexity, with the aim of understanding how neural tissue can learn to adapt to changing inputs and environments. Doom provides a compact, dynamic testbed because it is small, deterministic, and culturally resonant, making it an effective conduit for science communication about brain-inspired computing.

"it's a workhorse of science" - Rachel Fieldhouse, Nature

How the experiment works

The Doom-on-a-chip setup encompasses roughly 200,000 neurons differentiated from human stem cells and grown on a silicon chip. An intermediate AI translates the on-screen actions in Doom (move, shoot, pickups) into electrical impulses that stimulate the neurons. The neurons, in turn, emit signals that are interpreted by the AI to drive the character’s movements in the game. A researcher from Cortical Labs described the AI as the eyes and the hands of the computer, effectively seeing through the neural network and acting upon it. This integration represents a continuum of learning—from the researchers' Pong experiments in 2022 to the Doom challenge—where the brain-like substrate adapts to a more complex, dynamic task. The researchers emphasize they are not trying to create a mind inside a dish, but rather to study learning processes in a simplified neural substrate.

"the AI was the eyes and the hands of the computer controlling what was going on" - Rachel Fieldhouse, Nature

Why it matters: applications and implications

Beyond novelty, the work is touted as a potential route to energy-efficient computing that draws inspiration from biology. The team argues that brain-like tissue might perform certain tasks with far less energy than conventional AI hardware, presenting a possible complement or alternative to traditional neural networks in industrial AI applications. There are also envisioned uses in drug testing and development, where tissue-like models could offer more human-relevant readouts than conventional cell cultures or computer simulations. The project sits at the intersection of neuroscience, bioengineering, and computer science, with ethical considerations around how living neuronal tissue is used and interpreted in research settings.

"an energy-efficient alternative to AI computing" - Rachel Fieldhouse, Nature

Context, history and broader landscape

The Doom-on-a-chip story sits alongside other explorations of living systems interfacing with machines, from Pong played by rat neurons to bacterial cell simulations and even experiments aiming to run Doom in space. The broader takeaway is that Doom endures as a powerful vehicle for illustrating principles of neural computation, learning, and interdisciplinarity. The combination of an open-source game, a clear experimental setup, and a compelling narrative makes this a particularly effective vehicle for communicating cutting-edge STEM ideas to non-specialist audiences while inviting researchers to rethink how computation and biology can inform each other.

Quotes in context

"it's a workhorse of science" - Rachel Fieldhouse, Nature