Eon Systems, a San Francisco-based neurotechnology startup, announced this week what they describe as the world’s first embodied whole-brain emulation: a complete digital copy of a fruit fly brain, wired neuron by neuron from biological data, connected to a physics-simulated body that produced multiple naturalistic behaviors on its own. No reinforcement learning. No AI training. Just a biological connectome running in a computer, making a digital body move.
The demonstration builds directly on research published in October 2024 in a special issue of Nature, where Eon senior scientist Philip Shiu and collaborators presented a computational model of the entire adult fruit fly brain, more than 125,000 neurons and 50 million synaptic connections, built from the FlyWire connectome. That model predicted motor behavior with 95% accuracy but had a critical limitation: it was a brain without a body. Sensory inputs came in, neural activity propagated, motor commands fired, and went nowhere. Eon has now closed that loop.
Using the NeuroMechFly v2 embodied simulation framework and the MuJoCo physics engine, the team connected their connectome-based brain emulation to a physics-simulated fly body. Sensory input flows in, activates the connectome, and motor commands flow out to move the simulated body, completing, for the first time, a full perception-to-action cycle driven entirely by a real biological brain map. The work has not yet been formally published and is currently under revision at Nature.
The connectome that made it possible
The FlyWire connectome, the wiring diagram that made all of this possible, is itself one of the most ambitious scientific achievements of the last decade. Led by Princeton University, with major contributions from the MRC Laboratory of Molecular Biology in Cambridge, the University of Vermont, and the University of Cambridge, the FlyWire Consortium mapped every one of the fruit fly’s 139,255 neurons and 50 million synaptic connections. The effort involved more than 146 labs across 122 institutions, cutting-edge AI, professional proofreaders, and even online citizen scientists. It was published in October 2024 in Nature, the culmination of work that started in 2019.
Previous connectome projects existed, but at a fraction of this scale. The C. elegans worm had been mapped at 302 neurons. The larval fruit fly had 3,016. The adult fly brain, at nearly 140,000 neurons, was an entirely different order of magnitude, and the first complete connectome ever produced for an animal capable of walking, flying, navigating, and forming complex memories.
What makes the fruit fly particularly valuable for this kind of research goes beyond size. About 75% of disease-related genes in humans have counterparts in the Drosophila genome, a fact highlighted by Princeton professor Sebastian Seung at Neuroscience 2024. The fly brain uses many of the same neurotransmitters found in humans, including dopamine, glutamate, and acetylcholine. Understanding how its circuits function, and how they break down, has direct implications for research into Alzheimer’s, Parkinson’s, and other neurological diseases that currently have no cure.
From a fly to a mouse to a human
What Eon demonstrated this week is significant not just as a technical milestone, but as a proof of concept for something much larger. The longstanding question in connectomics wasn’t only whether a brain could be mapped, it was whether the map could actually predict how the brain works. As Shiu put it after the 2024 Nature publication: “It’s been unclear how much the connectome would actually allow us to predict neural activity. Now, we and others have found that the connectome really does critically allow us to predict and understand how the brain works.”
Prior work in embodied brain simulation never crossed this specific threshold. DeepMind and Janelia’s MuJoCo fly model, for example, used reinforcement learning to control a simulated body, not connectome-derived neural dynamics. OpenWorm attempted embodiment with C. elegans but with only 302 neurons and limited behavioral range. Eon’s demonstration is the first time a complete emulated brain, derived directly from a biological connectome, has driven a physically simulated body through multiple naturalistic behaviors.
The roadmap from here is clear. Eon’s next target is the mouse brain, approximately 70 million neurons, 560 times the complexity of the fly, and the team is actively building the data infrastructure to attempt it, combining expansion microscopy to capture every neural connection with tens of thousands of hours of calcium and voltage imaging to record how those networks activate in living tissue. After the mouse, the stated goal is human-scale emulation.
Shiu has been direct about what that trajectory means: “This really suggests that getting a mouse connectome, and eventually a human connectome, will be incredibly valuable. We can imagine a world where we can simulate a mouse brain, or eventually a human brain, and really get fundamental insights into the causes of various mental health disorders.”
What happened this week with a fruit fly is the first concrete step toward that world. A biological brain, copied from electron microscopy data, placed into a simulated body, producing behavior. It is not science fiction. It is a preprint waiting for peer review.
So here’s the question we’re genuinely dying to see you answer in the comments: does the idea of copying a human brain into a computer excite you, terrify you, or both at the same time? Would you upload your own consciousness if the technology existed? Drop your take below, there are no wrong answers here, only interesting ones.