The piece I'm not quite understanding is. To what extent is this really simulating the brain accurately?
It appears to behave accurately but it seems from reading the background which is largely above my head that those behaviours are in effect implementations not directly from data.
I.e. a pattern in the neural simulation is interpreted as groom. And then a groom behaviour is in effect played back
Walk pattern recognition is translated into walking (effectively this is a animation).
Does it matter? I'm not sure but I think it's the difference between playing a running animation or running in QWOP
They've basically taken three separate models, one for fly vision, one for the fly brain, and one for the fly body, and bolted them together Frankenstein style.
They've taken the connectome, which is a map of how neurons in the brain are connected to each other, and then created a fly brain using artificial spiking neurons connected together using that same connectome.
So the neurons are not remotely accurate. The interesting point they're making is that even with these simplified neurons they still see plausible behaviours (i.e. simulate the presence of sugar by stimulating a gustatory neuron -> neurons associated with lowering the proboscis for feeding are triggered). So they make the case that a lot of information is encoded simply in the connectome.
The body isn't connected up to the brain in the way we'd expect. "Input" comes from a completely separate neural network which they've trained to simulate appropriate CNS neurons, and output is looking at "descending" (efferent I guess) neurons in a very basic way. It's not completely playing an animation, but the level of connectivity is very low dimensional. It's not clear how much control they have, but for example I imagine they have a spiking threshold for the proboscis below which it's lowered and above which it's raised, which is sort of like you being able to stick your tongue completely out or pull it completely in but nothing else.
So it's not especially bioplausible. The most interesting part is that leaky-integrate-and-fire connectome-based brain model which they're using, even though it is also very limited (for example, it doesn't learn).
Demo looks very cool though. Much credit for being so explicit about what's going on in it in that post. And I was immediately filled with ideas about what they could do next to improve it, which to me is a signal of good research.
> The fly body is not currently driven by the full downstream motor hierarchy of the biological fly. Instead, we use a small number of descending outputs as a practical interface between the connectome model and the biomechanics
> [...] Steering in our model is driven through the neurons DNa01 and DNa02 (Yang et al., 2024), which are implicated in turning. Forward velocity is modeled by activation of oDN1
It seems that currently only WASD control is working. But even that is impressive! This is essentially an NPC driven by a real connectome.
I had exactly the same question. In the linked tweet the CEO claims it's not an animation. But in the article they imply certain animations are played when a specific signal is detected.
I think a full simulation without these precreated animations would be more convincing that they actually fully simulated a fly's behavior. It's very easy to make it look like real behavior using animations, videogames do exactly that.
I saw a short clip on this project the other day and was thinking along the same lines! They animated the flies’ mouth part (proboscis?) lowering and I was wondering if they were truly reading the motor commands for that from the brain.
> and while I am not a neuroscientist, I tried to explain this is just a combination of chunks of existing work
Which most complex technology is these days, even the most impactful. Docker is a fancy wrapper around Linux cgroups, Kubernetes combines that with etcd, and stuff like AWS EKS combines that with hypervisors (which I think were based on Linux KVM for a long time).
Have a look at this comment from Ken Hayworth, a highly respected scientist in the field, which I am copying here from a twitter post by him (https://x.com/KennethHayworth/status/2032604687212392562). I also just came back from the Cosyne 2026 conference, and the work was unfortunately not met with great enthusiasm, despite the media attention: My statement regarding the misleading EON Systems “fly upload” video:
The hundreds of researchers who make up the Drosophila neuroscience community are making good progress toward eventually understanding how the intelligent behaviors of a fruit fly are produced by computations in its neural circuits. Obtaining the structural connectome of the fly brain and ventral nerve cord was a significant milestone in that quest, as was obtaining an estimate of neurotransmitter types for each cell type. What is currently most lacking is a catalog of the precise electrophysiological and molecular dynamics of each neuron and synapse type. Dozens of on-going electrophysiological, genetic, and behavioral experiments are beginning to fill in those details. But completing that task will likely take many years, possibly decades, of more research. At the end of that long road, I have no doubt, there will be a detailed paper, published in a high-quality journal with full details and carefully peer-reviewed, which will at long last make the true statement “we’ve uploaded a fruit fly”. And that future paper will have a supplementary video much like the EON Systems one, showing a fly navigating a virtual environment. But, unlike the misleading EON Systems video, that future video will be real… all 100,000+ neurons displaying dynamics that reflect those that would occur in the real fly engaged in the same sensory-motor behaviors. That paper will represent the crowning achievement of a successful Drosophila neuroscience field.
What EON Systems’ misleading video and claim has done today is to try to steal that future victory and take its valor for their own, all in the hopes of raising some cash from naive investors who think they might get to human uploads soon, and all while riding a tide of hype they generated in the gullible public. The result has been a wave of secondary reporting that grossly mischaracterizes the current state of neuroscience progress, implying that it is much further along than it currently is.
As a member of the Drosophila research community, and as a long-term advocate of brain preservation for eventual mind uploading, I feel it is my responsibility to call out this reprehensible behavior. Neuroscience technology is progressing fast enough that we are now able to obtain structural connectomes of small organisms like the fruit fly. But neuroscience understanding is progressing much more slowly. True uploading, even for a fruit fly, is likely years to decades away. Even obtaining a mouse connectome seems likely to be a decade or more away. Human uploading is simply not on any reasonable research or investment timeline, unless such a timeline includes many decades of methodical basic neuroscience research. Of course, we can preserve human brains today using aldehyde fixatives as is done in all of today’s connectomics studies. But we will not be able to upload a human brain for many decades, perhaps centuries to come.
Please do not let today’s real scientific progress in connectomics and brain preservation be drowned out by misleading hype.
The piece I'm not quite understanding is. To what extent is this really simulating the brain accurately?
It appears to behave accurately but it seems from reading the background which is largely above my head that those behaviours are in effect implementations not directly from data.
I.e. a pattern in the neural simulation is interpreted as groom. And then a groom behaviour is in effect played back
Walk pattern recognition is translated into walking (effectively this is a animation).
Does it matter? I'm not sure but I think it's the difference between playing a running animation or running in QWOP
They've basically taken three separate models, one for fly vision, one for the fly brain, and one for the fly body, and bolted them together Frankenstein style.
They've taken the connectome, which is a map of how neurons in the brain are connected to each other, and then created a fly brain using artificial spiking neurons connected together using that same connectome.
So the neurons are not remotely accurate. The interesting point they're making is that even with these simplified neurons they still see plausible behaviours (i.e. simulate the presence of sugar by stimulating a gustatory neuron -> neurons associated with lowering the proboscis for feeding are triggered). So they make the case that a lot of information is encoded simply in the connectome.
The body isn't connected up to the brain in the way we'd expect. "Input" comes from a completely separate neural network which they've trained to simulate appropriate CNS neurons, and output is looking at "descending" (efferent I guess) neurons in a very basic way. It's not completely playing an animation, but the level of connectivity is very low dimensional. It's not clear how much control they have, but for example I imagine they have a spiking threshold for the proboscis below which it's lowered and above which it's raised, which is sort of like you being able to stick your tongue completely out or pull it completely in but nothing else.
So it's not especially bioplausible. The most interesting part is that leaky-integrate-and-fire connectome-based brain model which they're using, even though it is also very limited (for example, it doesn't learn).
Demo looks very cool though. Much credit for being so explicit about what's going on in it in that post. And I was immediately filled with ideas about what they could do next to improve it, which to me is a signal of good research.
> The fly body is not currently driven by the full downstream motor hierarchy of the biological fly. Instead, we use a small number of descending outputs as a practical interface between the connectome model and the biomechanics
> [...] Steering in our model is driven through the neurons DNa01 and DNa02 (Yang et al., 2024), which are implicated in turning. Forward velocity is modeled by activation of oDN1
It seems that currently only WASD control is working. But even that is impressive! This is essentially an NPC driven by a real connectome.
I had exactly the same question. In the linked tweet the CEO claims it's not an animation. But in the article they imply certain animations are played when a specific signal is detected.
I think a full simulation without these precreated animations would be more convincing that they actually fully simulated a fly's behavior. It's very easy to make it look like real behavior using animations, videogames do exactly that.
I saw a short clip on this project the other day and was thinking along the same lines! They animated the flies’ mouth part (proboscis?) lowering and I was wondering if they were truly reading the motor commands for that from the brain.
I wrote this up for the Register, and while I am not a neuroscientist, I tried to explain this is just a combination of chunks of existing work...
https://www.theregister.com/2026/03/16/digital_fruit_fly_bra...
> and while I am not a neuroscientist, I tried to explain this is just a combination of chunks of existing work
Which most complex technology is these days, even the most impactful. Docker is a fancy wrapper around Linux cgroups, Kubernetes combines that with etcd, and stuff like AWS EKS combines that with hypervisors (which I think were based on Linux KVM for a long time).
Have a look at this comment from Ken Hayworth, a highly respected scientist in the field, which I am copying here from a twitter post by him (https://x.com/KennethHayworth/status/2032604687212392562). I also just came back from the Cosyne 2026 conference, and the work was unfortunately not met with great enthusiasm, despite the media attention: My statement regarding the misleading EON Systems “fly upload” video:
The hundreds of researchers who make up the Drosophila neuroscience community are making good progress toward eventually understanding how the intelligent behaviors of a fruit fly are produced by computations in its neural circuits. Obtaining the structural connectome of the fly brain and ventral nerve cord was a significant milestone in that quest, as was obtaining an estimate of neurotransmitter types for each cell type. What is currently most lacking is a catalog of the precise electrophysiological and molecular dynamics of each neuron and synapse type. Dozens of on-going electrophysiological, genetic, and behavioral experiments are beginning to fill in those details. But completing that task will likely take many years, possibly decades, of more research. At the end of that long road, I have no doubt, there will be a detailed paper, published in a high-quality journal with full details and carefully peer-reviewed, which will at long last make the true statement “we’ve uploaded a fruit fly”. And that future paper will have a supplementary video much like the EON Systems one, showing a fly navigating a virtual environment. But, unlike the misleading EON Systems video, that future video will be real… all 100,000+ neurons displaying dynamics that reflect those that would occur in the real fly engaged in the same sensory-motor behaviors. That paper will represent the crowning achievement of a successful Drosophila neuroscience field.
What EON Systems’ misleading video and claim has done today is to try to steal that future victory and take its valor for their own, all in the hopes of raising some cash from naive investors who think they might get to human uploads soon, and all while riding a tide of hype they generated in the gullible public. The result has been a wave of secondary reporting that grossly mischaracterizes the current state of neuroscience progress, implying that it is much further along than it currently is.
As a member of the Drosophila research community, and as a long-term advocate of brain preservation for eventual mind uploading, I feel it is my responsibility to call out this reprehensible behavior. Neuroscience technology is progressing fast enough that we are now able to obtain structural connectomes of small organisms like the fruit fly. But neuroscience understanding is progressing much more slowly. True uploading, even for a fruit fly, is likely years to decades away. Even obtaining a mouse connectome seems likely to be a decade or more away. Human uploading is simply not on any reasonable research or investment timeline, unless such a timeline includes many decades of methodical basic neuroscience research. Of course, we can preserve human brains today using aldehyde fixatives as is done in all of today’s connectomics studies. But we will not be able to upload a human brain for many decades, perhaps centuries to come.
Please do not let today’s real scientific progress in connectomics and brain preservation be drowned out by misleading hype.
-Kenneth Hayworth