[superkuh]

This is about 40x more cells than they used to fly a fighter jet in a simulation back in 2004. https://www.nature.com/articles/nrn1572 .

I suppose the claim to fame in this similar study is the use of the title organoid and there's some legitimacy to that. Form and function are intimately tied in the brain and just a bunch of neurons on a petri dish isn't quite an organ.

[sillysaurusx]

Gosh, that study. It was the first time my faith in academia was severely shaken. As a 19yo, I spent around 6 hours reverse engineering what was going on, and then fell into a spiral of “Is this what academia is all about? Really?” Because I was so fascinated by the concept, then it was such a letdown…

It took some time to appreciate that there are some worthwhile ideas in that paper. But it was my first experience with “Academic lies about accomplishment to secure further funding.”

It’s hard to say whether “lie” or “severely exaggerate” is appropriate.

A method existed to measure a signal from a neuron. A second method existed to modify that signal. Whenever the plane crashed, the signal was modified slightly, until the plane flew level.

It didn’t learn to fly. The researcher modified a neuron (steady signal) until it gave the appropriate signal (e.g. zero) to fly level.

Negative signal, plane banks left. Positive signal, plane banks right. If plane crashes, modify neuron until signal is neither positive nor negative. That was the extent of the study.

In that context, do you feel like the neurons learned to fly? Maybe. It’s certainly similar in spirit to reinforcement learning in modern times. But I wouldn’t say that setting a signal to zero is a nice definition of “flew a plane”.

In other words, there was no active feedback; if you pointed the plane in a slightly different direction, it would immediately crash. It wasn’t doing anything more than setting the signal slowly over time to an answer that was known ahead of time. (Keep the plane level by not moving the controls.)

Suppose the neurons learned to draw a straight line. That was essentially what was being demonstrated here. If you substitute “plane crash” with “line becomes a curve”, it becomes much less exciting, to say the least.

“Isn’t that just learning to set a signal to a constant value?” “Yep” “Will it always become the same constant?” “Yep” “Can’t we already do that?” “Yep”

I was so disillusioned that it took many years to stop believing that academia itself was at fault for misinforming the public so badly. After all, it’s almost two decades later, and people still believe “rat brain flies plane” happened in 2004.

If I could go back in time, I’d tell myself not to worry about it; focus on the academics that are working quietly on the frontier, not the ones trying to raise funding for their lab.

At least the neurons in today’s study actually learned to play something. But if the past is any indication, I’d err on the side of skepticism.

[thrown_22]

Do you know the actual study which did this?

Because the Nature citation on where the work was done is literally "The Discovery Channel".

[sillysaurusx]

It was big news at the time. Headlines everywhere. Rat brain flies plane. You’d think that within a few years, maybe rat brains would be the primary way we’d control planes.

I did manage to track down the original study, which at the time wasn’t too easy; this was before scihub. I wish I’d saved it for posterity. My motivation went from “this is exactly what I want to do with my life” to laying on the couch wondering what the heck could possibly be happening, when the whole world believes rat brains are flying planes, vs what was actually demonstrated in the paper.

If you do find it, please post it. It’d be a nice stroll through memory lane, and an interesting retrospective on how to get funding for one’s own research lab. (A handy skill to have, if you don’t mind… well, exaggerating, to say the least.)

[justin_]

I think it might be this one:

Adaptive flight control with living neuronal networks on microelectrode arrays

The brain is perhaps one of the most robust and fault tolerant computational devices in existence and yet little is known about its mechanisms. Microelectrode arrays have recently been developed in which the computational properties of networks of living neurons can be studied in detail. In this paper we report work investigating the ability of living neurons to act as a set of neuronal weights which were used to control the flight of a simulated aircraft. These weights were manipulated via high frequency stimulation inputs to produce a system in which a living neuronal network would "learn" to control an aircraft for straight and level flight.

https://ieeexplore.ieee.org/document/1556108

https://sci-hub.ru/https://doi.org/10.1109/IJCNN.2005.155610...

[sillysaurusx]

Thank you so much! That’s the exact one. I’ll upload it to imgur, since scihub urls tend to die over time: https://imgur.com/a/vdQP17I

The details are presented at the end of page 2. You can see on page 3 that the whole thing is a glorified “I can make neurons go to zero!” machine.

It kills me because there is real value in this work — the ability to modify neurons is a useful thing to be able to do. Showing that you can set them to specific values is worthwhile.

But that wasn’t what it was presented as. The focus was on the idea that the neurons somehow learned something. Maybe they did, and maybe time will prove me a fool by showing that this is how neurons learn in general. But it was so disheartening to spend hours carefully going over every detail, full of excitement at the possibility of machines that can learn… only to realize I could do the same thing by setting a value to zero in python, and that the complicated language seemed designed to conceal this.

I remember seeing a video on the discovery channel where they were interviewing him. He gave a dazzling account of the implications of this work.

Maybe he was predicting the upcoming ML boom. But now that I’ve worked in ML for a few years, I can safely say that this work didn’t help us get here. Maybe it helped biologists figure out how to control the signals that neurons deliver.

(I still think that it’s pretty darn cool that you can manipulate neurons at all, so hats off to whoever figured out how to do that. I just wish it was presented as what it was: the ability to set a value to zero over time via neurons.)

I really want to make a joke like “You know what else would set neurons to zero? Set them on fire and wait,” but the paper showed they could be set to arbitrary values. Which of course was the interesting and valuable part in the first place, not the plane stuff. But planes make good headlines.

[leonmoonen]

Thomas DeMarse is mentioned in some of the 2004 news articles. His Google Scholar profile has the following short paper published at a conference in 2005: https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=Thom...

[thrown_22]

>(Discovery Channel, USA, 22 October 2004)

That's a bit like saying battle bots solved locomotion of robots.