71 Comments
User's avatar
+ and -'s avatar

Are lobsters next?

Himanshu Agarwal's avatar

If brain emulation becomes real, the definition of “being alive” will completely change.

Boxo McFoxo's avatar

This an impressive simulation. But it's just not honest to call this 'brain emulation', a 'brain upload' or to say that this is doing anything like 'sensorimotor loop in simulation'. Aside from the fact that a connectome is not a brain, and so we have no idea whether the parts that have been filled in by ML actually function like a brain, the motor control in this framework is not even driven by the brain simulation. The output from the 'brain' is not a sequence of motor commands. It is a steering mechanism, a 2-dimensional descending signal (essentially, turn left or right, speed up or slow down). That is then fed into a series of CPG oscillators, outside of the brain emulation, that model fly movement in response to that 2-dimensional descending signal. Since outputting a 2-dimensional descending signal is not what a fly brain does, the simulated brain is not operating as a fly's brain does. It's machine learning, clamped into the shape of a fly connectome, that has a resting state of 0Hz, being zapped with simple inputs, not virtual sensory data.

"What you are seeing is not an animation. It is not a reinforcement learning policy mimicking biology. It is a copy of a biological brain, wired neuron-to-neuron from electron microscopy data, running in simulation, making a body move." - I'd go so far as to say this is an actual lie about how NeuroMechFly v2 actually works, and not a lie that the authors themselves make in their paper. The actual movements that the fly makes come from ML.

Arbituram's avatar

Yes, thank you. A connectome is very far from a full brain simulation. Also, the 95% accuracy claim for the earlier Shiu et all study (which I don't want to bad mouth, it's a very cool study) is misleading; it could predict neural flows from input to output for a couple of behaviour types, but on a one input one output basis, and didn't come close to modelling e.g. learning.

Parsa's avatar

Is the video showing NeuroMechFly v2? I thought it's building on "previous work including NeuroMechFly v2"

Boxo McFoxo's avatar

It is NeuroMechFly v2, using some signals that were not used in the original NeuroMechFly v2 paper.

U steve's avatar

AWG...you are so far ahead of the cutting edge...amazing.

I imagine Peter Diamandis would be eager to sign up his agents for the first human trials.

John Conklin's avatar

A monumental milestone!

John Conklin's avatar

You're going to wreck Ray's 86% Accuracy! I think the required scaling will happen before '45.

Dr. Alex Wissner-Gross's avatar

I think that falls under the category of “good problem to have.”

John Conklin's avatar

Indeed. This is definitely a big step in the continuation? start? of a world where the line between "biological" and "digital" starts to disappear. It shall be interesting! ✌️🖖

Francisco De La Torre Cevallos's avatar

Alex. You are amazing !!

Bill Jacobs's avatar

And so, it begins!

Tony Paez's avatar

Amazing.

Tony Paez's avatar

Amazing. This totally blowes me away!

Brad Rothrock's avatar

Fascinating. Your next Doctor Alex

Peter Michael Echols's avatar

Awesome! Proud of you. Why am I not surprised by your involvement? If you decide to download yourself, please make a meat backup so we don't lose you.

matt wayrynen's avatar

Wow, Alex, congratulations! To say this is astonishing is the understatement of the century. I was trying to wrap my head around it and couldn’t quite find the words—until I read your post on the website: “We’re opening doors to possibilities we can barely imagine.” That really says it all.

Vishal's avatar

Sooner or later, we will be living in San Junipero.

Suguna Jayarajan's avatar

if behavior can emerge from a connectome + simulation loop, then the line between “biological mind” and “computational system” starts looking a lot thinner than we usually imagine.

John Smart's avatar

Brilliant work Alex! This was definitely a key next step for Shiu and Eon. I do wonder how if the interface to the body is deeply bio inspired. Also if the fly has a working memory of the type Dhruv Grover at UCSD has found in his fly experiments. What Id love to see is Dhruv set up a version of his apparatus in your fly emulator and see if you get the same patterns and behaviors he calls fly emotions and fly consciousness. Emulating those next. If you can, would be epic! Flies Possess More Sophisticated Cognitive Abilities than Previously Known https://share.google/fCJyjUmt7LP9utuVf

Clancy mcquigg's avatar

Tell eon systems they need a foreword link in their website so we can share with others who may be interested in their work