Following the initial mapping of an insect's brain, Google Research has released MaleCNS v1.0, a comprehensive 3D model of the entire brain and central nervous system of an adult male fruit fly. While previous efforts had identified roughly 139,000 neurons, this updated connectome leverages AI to reconstruct over 166,000 neurons by combining millions of 2D images into complex neural shapes.
Beyond the scientific milestone, software engineers have immediately pivoted to testing the model's functional capabilities in virtual environments. Developers Alex Wormuth and Jessica Paquette have demonstrated that the MaleCNS connectome can be used as a biological engine to drive gameplay in classic titles.
From Neuroscience to Gaming
In the case of Doom, Wormuth has implemented a system where each frame of the game stimulates specific sensory neurons. The resulting neural activity is then mapped directly to game controls. To facilitate learning, the system uses reinforcement: when the fly takes damage in-game, a stimulus is triggered in two PPL101 dopamine cells, simulating a biological reward/punishment mechanism. This project is fully open source, allowing the community to monitor the training of the 166,700 neurons in real-time.
Similarly, Paquette has applied the connectome to Super Mario 64, utilizing GPT Astra for "vibe coding" the implementation. Early results show Mario interacting with the environment, albeit clumsily, mirroring the erratic movements of a real fly.
Expanding Virtual Biological Simulations
The trend of integrating biological connectomes into software extends further. Recent reports indicate that the MaleCNS v1.0 model has also been successfully run inside Minecraft, where simulated neural activity drives the movement of a virtual fly.
This progression marks a shift from static mapping to active simulation. While earlier reports on AlexTech.ai focused on the structural discovery of the fly's brain, the current phase explores how these biological blueprints can be operationalized as controllers for digital agents.
No comments yet. Be the first!