The Worm in the Machine
November 17, 2014 10:15 AM   Subscribe

A Worm's Mind In A Lego Body: Timothy Bubisce of the OpenWorm project (previously) has uploaded a neural mapping, or connectome, of the C. elegans worm as software into a Lego robot. The result? It kinda sorta behaves like a worm would. So, not quite the Kurzweillian dream of uploading one's consciousness into a machine, but still fascinating.
posted by Cash4Lead (23 comments total) 32 users marked this as a favorite
 
That's amazing. Not being a worm biologist I can't say how worm-like the behavior is, but even the obstacle avoidance coming from a neural network seems like a big win.
posted by DU at 10:27 AM on November 17, 2014 [1 favorite]


Huh, how curious. Wouldn't this connectome need to be interfacing with a physical body that has some isomorphism with that of C. elegans in order for it to, I guess, "work right"?
posted by clockzero at 10:28 AM on November 17, 2014


This fucker could probably replace me at work.
posted by Foci for Analysis at 10:32 AM on November 17, 2014 [9 favorites]


The connectome has been translated into software, so it doesn't need the C. elegans body. But then this raises the question of whether this software is identical to a C. elegans' neural network or just a very, very good imitation.
posted by Cash4Lead at 10:38 AM on November 17, 2014 [1 favorite]


We've known for a long time exactly how the nervous system of C. elegans is wired. The real advance here is putting an abstracted piece of that into software. Neat post, cheers.
posted by a lungful of dragon at 10:41 AM on November 17, 2014 [3 favorites]


Who's in charge of the cstross-signal this week?
posted by signal at 10:43 AM on November 17, 2014 [1 favorite]


I anxiously await the new generation of Roombas that come out of storage when it rains and dance all over the kitchen floor.
posted by overeducated_alligator at 10:45 AM on November 17, 2014 [14 favorites]


Foci for Analysis: This fucker could probably replace me at work.

How do you know this hasn't already been done? Maybe you are the worm, but you don't even know it.
posted by surazal at 10:48 AM on November 17, 2014 [6 favorites]


I work on connectome mapping in flies. This is cute, but from a science perspective it's just not interesting. Backing up when detecting an obstacle is a bit of a cheat in particular, since the main "there's an obstacle in front of me" neuron is directly connected to the "move backwards" neuron. You don't need 300 neurons for that, you need 2 (or maybe a couple more because the "move backwards" neuron talks to motor neurons).

It's a weirdly un-biological implementation, as well. C. elegans neurons are graded, not spiking, and so they pass on an analog signal from one neuron to another, and we don't even know the sign — i.e. excitatory or inhibitory— of a huge number of synaptic contacts in it. Also, a great deal of the connectome is not voltage-gated synapses, which only activate when the presynaptic neuron is sufficiently depolarized. Instead, they use gap junctions, which are like direct connections that pass on the state of the neuron no matter how activated it is, not only when its membrane potential gets above a certain threshold. This is to say nothing of neuropeptides, which are pervasively used by the animal to change the excitability properties of both synaptically connected AND distant neurons. It's a tremendously difficult wiring diagram to interpret. To add additional problems, a non-moving non-stimulated worm will still have an active nervous system providing some sort of internal state. This internal state modulates the way the animal responds in a number of ways, some profoundly. The simple neuronal model presented here will basically become utterly quiet.

I don't like to be totally negative, but I just came from a conference full of great worm talks that really demonstrated how complex the way the animal uses its small nervous system is, even when you can actually record the activity of every neuron at once (which you can do now, more or less). The scientists doing the impressive work of figuring out what really happens need to make cuter YouTube videos apparently.
posted by Schismatic at 10:57 AM on November 17, 2014 [51 favorites]


Amazing how this subtle reproduction of sophisticated worm behaviour resembles incoherence.
posted by Segundus at 10:58 AM on November 17, 2014 [1 favorite]


MetaFilter: a conference full of great worm talks
posted by Foosnark at 11:00 AM on November 17, 2014 [6 favorites]


not quite the Kurzweillian dream of uploading

Just a few more bits, baby, just a few more mega-mega-mega bits.
posted by sammyo at 11:03 AM on November 17, 2014 [1 favorite]


Schismatic: The scientists doing the impressive work of figuring out what really happens need to make cuter YouTube videos apparently.

Let's not get too carried away with the cute youtube video idea, lest we forget what happened to the "Shrimp on a Treadmill" guy. ;^)
posted by surazal at 11:05 AM on November 17, 2014 [3 favorites]


THE ROBOT ISN'T EVEN WORM-SHAPED

WHAT THE HELL SCIENCE
posted by prize bull octorok at 11:10 AM on November 17, 2014 [3 favorites]


I just came from a conference full of great worm talks

I choose this to parse this as talks, on any subject, being given by Shai-Hulud.
posted by ROU_Xenophobe at 11:15 AM on November 17, 2014 [15 favorites]


I'm with Schismatic on this (ie, what is the real scientific purpose of this little robot?).I will say that it initially captured my imagination with description, but then you see a video of it clunking into a wall.

Look at real Caenerhabditis elegans crawling around. Beautiful, no?

One of the things that made C elegans a powerful research model are the genetic studies that you can do with them. While it might take yrs of work (if not a decade) to make a rat model for whatever genes or disease you want to study, nematodes - with careful breeding/selection - weeks to a few months. Within very short time periods, you can study the effects of a particular gene on aging or apoptosis, etc.

Although I would be happy if someone could build a robot or robots to replace me, so I can nap/relax while my robot works away.
posted by Wolfster at 11:20 AM on November 17, 2014 [3 favorites]


Is this something I would have to own a prostomium to understand?
posted by CynicalKnight at 11:20 AM on November 17, 2014 [4 favorites]


The Internet: "The [blah, blah, blah] need to make cuter YouTube videos."
posted by meinvt at 11:44 AM on November 17, 2014 [1 favorite]


THE SPICE MUST FLOW
posted by Damienmce at 12:56 PM on November 17, 2014 [2 favorites]


surazal: Foci for Analysis: This fucker could probably replace me at work.

How do you know this hasn't already been done? Maybe you are the worm, but you don't even know it.
That's simple. When I wiggle my hands, the shadow on the opposite side of my cubicle.... oooooh..... shit.
posted by IAmBroom at 12:58 PM on November 17, 2014 [1 favorite]


(I hope this is not too big of a derail, as I think others might have this question too. Mods, I trust your judgment)

I'm really really interested in the growing overlap/intersection between animals and robotics. Cyborgs, digital modelling of animal minds, etc. Knowledgeable people of this thread, where do I go to keep up to date on this? Journals, Websites, Reporters?
posted by DGStieber at 2:20 PM on November 17, 2014


I choose to parse this as talks, on any subject, being given by Shai-Hulud.

A large, toothy Sandworm nervously fidgets behind a podium, shifting from side to side, glancing down too frequently at his papers, talking about semiotics in a web 2.0 world.
posted by Ray Walston, Luck Dragon at 3:19 PM on November 17, 2014 [3 favorites]


Thanks Schismatic for providing some background to this story and the complications around C. elegans work. This is fascinating and inspiring!

what is the real scientific purpose of this little robot?

Fun?
posted by sneebler at 6:36 AM on November 19, 2014


« Older “Remember me as a revolutionary communist.”   |   Is Texas getting ready to kill an innocent man? Newer »


This thread has been archived and is closed to new comments