Headline: Living human neurons learn to play Doom — and the experiment has crypto culture written all over it
Melbourne startup Cortical Labs has taken the old “Can it run Doom?” meme into biological territory. In a video posted last week, researchers showed clusters of living human neurons controlling the 1993 shooter Doom: gameplay is converted into electrical stimulation delivered to cultured neurons, and the neurons’ activity is decoded back into in-game controls so the cells can move, aim and fire.
How it works
- The work runs on Cortical Labs’ CL1 device, where roughly 200,000 living human neurons are grown on a multi-electrode array. That array lets researchers both stimulate the tissue electrically and read neural responses in real time.
- Game state is encoded as electrical inputs to the cells; spikes and patterns in the culture are translated into joystick and firing commands. The set-up uses a custom platform that exposes high-level Python controls so researchers can interact with the neural tissue without low-level plumbing.
- Cortical Labs first demonstrated Pong on neurons in 2021 as an early test. After building the higher-level platform, a collaborator named Sean wrote the Doom interface in days rather than the 18 months it would have taken before.
Learning, AI and limits
- The neurons “learn” through reward signals: small rewards for aiming at enemies and larger rewards for actually eliminating targets. Over time those feedback signals reinforce the behaviors that produce game success.
- Cortical Labs layers artificial intelligence on top of this loop to optimize how game information is encoded into electrical stimulation. “The cells are actually learning the input,” an application scientist, Alon Loeffler, explained. “But then the AI is trying to improve that input to try and get the cells to do what we want them to do.”
- Importantly, the team stresses this is not the neurons “understanding” Doom. Loeffler: “The system doesn’t actually know it’s playing Doom. It’s getting electrical signals and then spitting out responses.” There are no pain receptors or brainlike structures for higher-order cognition — the experiment shows neural adaptability, not human-like thought.
Why this matters (and why crypto people might care)
- Doom has long been the benchmark for novelty ports: it’s been forced to run on everything from gut bacteria and pregnancy tests to PDFs, robot lawn mowers and even blockchain networks. That cultural history helps explain why researchers chose the game — it’s a playful, public-facing demo that signals technical reach.
- The experiment sits at a crossroads of neurology, AI and unconventional computing — an area that naturally intrigues communities interested in pushing compute to new substrates, including blockchain and decentralized systems. It’s a vivid demonstration that living tissue can be part of a closed-loop information-processing system, even if today’s behavior is simple and stimulus-driven.
Takeaway
Cortical Labs’ Doom demo is a striking showcase of neural adaptability and an example of how biological systems can be interfaced with software and AI. It doesn’t create human-level intelligence, but it does underline neurons’ inherent capacity to respond to and learn from patterned input — and it adds a memorable entry to Doom’s long history of being used as a proof-of-concept for ambitious tech experiments.
Read more AI-generated news on: undefined/news
