Google Open-Sourced a Fly Brain

English
BBetter Stack
Computing/SoftwareVideo & Computer GamesInternet Technology

Transcript

00:00:00So, Google and Janelia just open-sourced a fly's brain, the complete worrying diagram of a real
00:00:04male fruit fly's brain and nerve cord, every neuron and every connection, for anyone to use.
00:00:09And naturally, the internet has done what the internet does, the fly's been beating Minecraft,
00:00:14playing Beat Saber, trading crypto, learning Python, and of course, playing Doom. Now you
00:00:19might be wondering what the hell am I even talking about, open-sourcing a fly,
00:00:22but it's probably easier to just fly in and show you what I mean.
00:00:30Now I want to start out by showing you what the internet did with this fly, and then I'll explain
00:00:34later. Just know that most of these demos do have an asterisk to them. It's obviously not as simple
00:00:38as downloading a sentient fly brain and just connecting them to things. With that in mind
00:00:43though, my favourite demo was posted by Evan Sinclair-Smith, where he showed the full connectome
00:00:47running inside Minecraft, with simulated neural activity driving a fly around the Minecraft world,
00:00:51and even having interactions like player approach, attack attempts, and searching for food and light.
00:00:56I just think it's absolutely crazy that we have a fly in Minecraft that essentially is using the
00:01:00mapping of a real fly's brain. My next favourite demo was seeing the fly play Beat Saber,
00:01:05absolutely nailing the track Among Us Trap remix, which is a sentence I never thought I would say
00:01:09on this channel, but here we are. And naturally after this, when you get into video games,
00:01:16you also have to test it out on Doom. As Alex says here, each Doom frame stimulates sensory neurons,
00:01:22neural activity is mapped to the game controls, damage triggers stimulus to two dopamine cells,
00:01:26reinforcement, and this demo is actually being streamed on a site at the moment. It's had over
00:01:308,000 attempts, and I don't think it's actually come close to beating the game yet, but I'll
00:01:35definitely be keeping an eye on it. This same guy seems to be having a lot of fun creating what I'm
00:01:38going to dub as Fly Slop. Like, he gave the fly $100 and let it trade Bitcoin, with the dopamine
00:01:44neurons stimulated on profit, and it actually beat a lot of day traders. Boom, roasted. He also built a
00:01:50fly language model, and connected three of them on a social media platform. Perhaps my favourite
00:01:54summary of all of this is Google. Here's a complete map of a fruit fly brain. Internet. Can it play
00:01:59Minecraft? Internet two hours later. Can it launch a meme coin? People even started feeling bad for
00:02:04this fly, so they built it its own heaven, where it just has an open world full of grass and fruit,
00:02:09where it can just fly around and eat whatever it wants. Now, naturally, I had to give this a go
00:02:13myself. So this is a circuit of 80 real cells from that map, and we'll get to what that map actually
00:02:17is in a second. But I've essentially pointed its eyes at this channel page, and every time the cursor
00:02:22lands on the subscribe button, we give it a reward signal, which everyone on X is calling dopamine,
00:02:27and after about 200,000 attempts, the fly now has 100% success rate of subscribing to this channel,
00:02:32but I do hope it only takes you one try. So all of those examples aside, what the hell is actually
00:02:37going on here? Well, what we're actually talking about is called Male CNS Version 1, and it's a
00:02:42kinetocomb, a map of which neuron connects to which, and how many synapses sit between them.
00:02:47Every dot you're looking at is a real cell body, over 166,000 of them. Blue is the optic lobes,
00:02:53grey is the central brain, gold is the descending neurons heading down through the body, and between
00:02:57them, around 125 million synaptic connections. Now, the way you make one of these is a little bit
00:03:02crazy. You take one fly and you slice it 8 nanometers thick. This is one real section through
00:03:07the brain, and there are 134,000 of these, and you image everyone with an electron microscope,
00:03:12and if you zoom in 100 times, that dark bar is a single synapse. On one side is a looming
00:03:18detector, so a cell in the eye that fires when something gets bigger fast, like a hand coming
00:03:23at the fly, and on the other side is a giant fiber, which is the neuron that actually makes
00:03:27the fly jump. That's the escape reflex. After this, Google's team then used something called
00:03:32a flood-filling network, which starts at a single pixel and grows outwards to find every
00:03:36other pixel that belongs to the same neuron through every one of those sections. These are 80 of
00:03:41the real trace cells from that escape circuit, and the gold one is the giant fiber. That
00:03:45one neuron alone has nearly 25,000 synapses landing on it from over 3,000 other cells. The
00:03:50red and orange ones are the looming detectors, with nearly 6,500 synapses between them. All
00:03:55of this work then ends up as a giant table detailing who connects to who, how many synapses there
00:03:59are, and a predicted transmitter, which altogether is 125 million rows. All of that data is now
00:04:06public. There's a web viewer called NeuroGlancer, a query interface called Neuoprint with Python
00:04:10and R packages, and the raw image volumes and connectivity tables are all downloadable. But
00:04:15as I mentioned with the demos earlier, there is a bit of an asterisk here. You obviously don't
00:04:18download a sentient fly brain. The map just tells you which neurons are wired to which. It doesn't
00:04:23tell you what happens when a signal goes down those wires. Think of it like having the wiring
00:04:27diagram for a house, but no idea what the switches actually do. In my subscribing fly example,
00:04:32I turn the screen into a few numbers based on where the button is. Those numbers then go into the fly's
00:04:37wiring and are sent through it, but because the map doesn't say how strong any connection is,
00:04:41I just made every single one of them the same. Then what comes out the other end just goes into a
00:04:46small translator that I made, and that essentially has 163 dials, and this translator then decides to
00:04:52move left, move right, or click. The training essentially just tries a load of random settings,
00:04:56sees if the cursor gets closer to the button, keeps the best ones, and then tries this over 200,000
00:05:00times until it's 100% accurate. So technically my fly wasn't a fly, and the same goes for every demo
00:05:06that you saw earlier. They all work the same way. Someone decides what the pixels mean, someone
00:05:10guesses how the cells behave, and a trained network on top does the actual playing. I mean, you probably
00:05:15already guessed that it wasn't a sentient fly. If we were actually at the point where you could
00:05:19download a sentient brain off the internet, I'd probably move to a farm and disconnect from the
00:05:23grid entirely. This doesn't mean that the map is useless though. Nico Christie switched off
00:05:27one type of neuron in the model, watched what happened to the courtship neurons, and got the
00:05:30same result from a real experiment on real flies from 2015. So proper science is the real reason
00:05:36this data actually exists, not the fly slop that we have on Twitter, and we now actually have two
00:05:40complete fruit fly connectomes, one male and one female, and around 5% of that central brain is
00:05:45sex-specific, and that's where courtship and aggression live. So where the two differ, you can
00:05:49actually study that, and where they're the same, you can finally see how much wiring varies between
00:05:54two individuals of the same species, which no one has been able to do on this scale before.
00:05:58Janelia and Google are already working on a zebrafish brain, and then a mouse, and apparently the
00:06:03human brain with 86 billion neurons is still out of reach for now. What do you think of all of these
00:06:08crazy experiments, and have you tried any fly slop yourself? Let me know in the comments down
00:06:12below, while you're there subscribe, and as always see you in the next one.

Key Takeaway

Google and Janelia released a complete connectome map containing 166,000 cells and 125 million synaptic connections of a male fruit fly, enabling simulations in software environments like Minecraft and Doom.

Highlights

  • Google and Janelia open-sourced the complete connectome of a male fruit fly containing over 166,000 cells and 125 million synaptic connections.

  • Researchers sliced a single male fruit fly brain 8 nanometers thick, yielding 134,000 individual sections imaged via electron microscope.

  • A giant fiber neuron in the escape circuit features nearly 25,000 synaptic connections originating from more than 3,000 separate cells.

  • Simulated neural activity mapped from the open-source data drives interactive behaviors inside Minecraft, Beat Saber, and Doom.

  • Researchers connected a circuit of 80 real cells from the map to a cursor, achieving a 100% success rate for clicking a subscribe button after 200,000 training attempts.

  • Data access is provided through NeuroGlancer web viewer, Neuoprint query interfaces, and downloadable connectivity tables comprising 125 million rows.

Timeline

Internet Demonstrations of the Fruit Fly Connectome

  • The open-sourced neural wiring map of a male fruit fly allows the internet to run simulated neural activity inside video games.
  • Evan Sinclair-Smith implemented the full connectome inside Minecraft to drive fly movement and trigger interactions like food searching.
  • Alex mapped sensory neurons and game controls to run a playable version of Doom driven by the fly data.
  • A circuit of 80 real cells trained on a cursor interface achieved a 100% success rate for clicking a subscribe button after 200,000 attempts.

The release of the fruit fly connectome prompted various internet demonstrations integrating simulated neural activity into software platforms. Users mapped sensory inputs and rewards, such as dopamine triggers on profit or cursor position, to game mechanics. These setups demonstrate how raw wiring diagrams can be coupled with external translators and reward loops to produce interactive behaviors.

Anatomy and Mapping Methodology of Male CNS Version 1

  • Male CNS Version 1 maps 166,000 cell bodies and 125 million synaptic connections across optic lobes, central brain, and descending neurons.
  • The physical mapping process requires slicing a single fly brain 8 nanometers thick into 134,000 individual sections.
  • Flood-filling networks trace individual neurons across every section starting from single pixels.
  • The resulting dataset includes 125 million rows detailing cell connections, synapse counts, and predicted neurotransmitters.

Constructing the connectome required physical sectioning of a fly brain into 134,000 slices followed by electron microscope imaging. Flood-filling algorithms traced individual pixels across sections to reconstruct full neurons. The giant fiber in the escape circuit alone receives nearly 25,000 synapses from over 3,000 cells, while looming detectors contribute thousands of additional connections.

Scientific Utility and Future Connectome Projects

  • The raw wiring map lacks active signal strength data, requiring custom translation layers and trained networks for behavioral simulations.
  • Disabling specific neuron types in the model reproduces results identical to real biological experiments from 2015.
  • Comparative analysis between male and female fruit fly connectomes reveals sex-specific wiring differences accounting for roughly 5% of the central brain.
  • Janelia and Google are actively developing connectome maps for zebrafish and mice.

While the map details physical connections rather than live signal dynamics, it enables genuine neurobiological research. Disabling specific neurons yields outcomes matching physical experiments. Researchers can study structural variations between male and female brains, particularly in regions governing courtship and aggression, while development continues on larger vertebrate organisms.

Community Posts

No posts yet. Be the first to write about this video!

Write about this video