More importantly, the system now allows rats to shoot. A physical trigger mechanism lets the animals activate Doom's fire input, meaning they're no longer just moving through the game, but interacting with it in a way that directly maps to classic FPS controls . It's still a far cry from tactical demon slaying, but mechanically speaking, the rats are now performing multiple discrete in-game actions.
To be clear, none of this involves invasive neural interfaces. The entire system relies on external sensors, motion tracking, and reward-based learning. The rats' physical movements are translated into standard Doom inputs, and correct behaviors are reinforced through the reward system. From a hardware perspective, this remains a hacker-friendly, open-source setup rather than a sealed, bespoke lab instrument.
That distinction matters because this project has never really been about proving that rats understand Doom in any human sense. The update doesn't suddenly mean rodents grasp level design, enemy behavior, or objectives. What it demonstrates is that the technical platform has matured enough to support richer interactions. The rats can now perform multiple distinct in-game actions, and the system can reliably evaluate and reward those actions, so the limiting factor is no longer the hardware or software. Instead, it's training time and experimental design. Teaching an animal to associate specific physical behaviors with abstract outcomes inside a virtual space is slow, and scaling that training takes patience. The updated system opens the door to more ambitious experiments than the original build could support.
The rats still aren't speedrunning E1M1, but the project has clearly moved beyond a one-off stunt. The update shows real technical progress, and it hints at future experiments that could use game engines as standardized, low-cost virtual testbeds. Doom , once again, refuses to die, and indeed, the Doom engine is doing a lot of quiet heavy lifting here. Its lightweight engine, trivial moddability, and decades-long history of running on almost literally anything make it an ideal virtual environment for this kind of work. What looks like a joke on the surface is really a practical choice: Doom provides a controllable, well-understood 3D world that can be bent to experimental needs without reinventing a game engine from scratch.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/virtual-reality/SPONSORED_LINK_URL
- https://www.tomshardware.com/virtual-reality/rats-are-still-being-taught-to-play-doom-now-with-a-curved-amoled-and-a-shoot-button#main
- https://www.tomshardware.com
- New app injects AMD's AI-driven FSR Redstone framegen into unsupported games — pre-alpha OptiScaler build demo shows nice improvements over legacy FSR frame gen
- Experiment to train rats to play Doom reaches a new level; rats can now shoot enemies — wraparound AMOLED screen provides virtual environment for neuroengineers
- Opt-In NVIDIA Software Enables Data Center Fleet Management
- Oracle reportedly delays several new OpenAI data centers because of shortages — tight material and labor supply frustrate expansion plans, possibly by a year or
- Amazon unveils 192-core Graviton5 CPU with massive 180 MB L3 cache in tow — ambitious server silicon challenges high-end AMD EPYC and Intel Xeon in the cloud
Informational only. No financial advice. Do your own research.