Wow, nuerons in a petri dish learn to fly.

[wesmorris]

From Wired's site

Somewhere in Florida, 25,000 disembodied rat neurons are thinking about flying an F-22.

To me, that seems like a very big deal.

What do you think?

Are there ethical problems with it? What if they want to try it with human brain cells?

Regardless, talk about a brave new world. And it's just the tip of the iceburg do you think? I do. What a trip.

I wonder which stock one could buy to get a piece of that action.

 

[Athelwulf]

Why wouldn't it be ethical?

 

[Avatar]

another article : http://www.napa.ufl.edu/2004news/braindish.htm

-----
to me it also seems like a very big and a bit scary deal
what if they become insane?

 

[invert_nexus]

Wes,

Are there ethical problems with it?

I think you misunderstood the point of the study. It's purpose is not to create a brain for an airplane, it's to give these neurons a body so that they can study the network better. This isn't about creating cyborgs, it's about attempting to understanding how it is that neurons do what they do.

And, it's only a start. The interface is primitive at best. Electrodes it says. What is needed is some kind of a synaptic interface with the neurons. A chemical transfer of information rather than electrical.

I'd like to see some glial cells added to the network as well.

Also, I wonder what type of feedback mechanisms they use to enhance the network over time.

"To control the simulated aircraft, the neurons first receive information from the computer about flight conditions: whether the plane is flying straight and level or is tilted to the left or to the right. The neurons then analyze the data and respond by sending signals to the plane’s controls. Those signals alter the flight path and new information is sent to the neurons, creating a feedback system."​

Now that certainly doesn't say much. The feedback mechanism must be exceedingly primitive. I can't imagine it even being at the level of a pseudoeye. To me, this stinks of just putting the neurons on a plate and seeing what they do. There is no real control over the experiment. They are just hoping that the neurons figure out what to do and how to act under the circumstances.

It has a long way to go to get any real use out of this. But, observing the actions of the neurons should be a learning experience and the system can be refined over time.

What if they want to try it with human brain cells?

I'd love to see this done with human brain cells. I wonder how differently human neurons are to rat neurons in a dish. We all know that the brain structure is vastly more complex but what about the actions of neurons in an unfamiliar environment? I've never actually seen any comparisons between the two in this regard.

Avatar,

what if they become insane?

It will need to become far more advanced than it is now for a state of insanity to occur. We're not dealing with a brain. We're dealing with a bunch of neurons.

 

[wesmorris]

Wes,

I think you misunderstood the point of the study.

A smidge presumptuous of you. Maybe you misunderstood the reason for my question.

It's purpose is not to create a brain for an airplane, it's to give these neurons a body so that they can study the network better. This isn't about creating cyborgs, it's about attempting to understanding how it is that neurons do what they do.

Yeah, as is obvious by the text. My point is, if you start using rat brain cells for this and human brain cells for that, and the things are learning, is that then for instance, part of the identity of the person from which the cells came... maybe it isn't, but shouldn't the question be asked?

And, it's only a start. The interface is primitive at best. Electrodes it says. What is needed is some kind of a synaptic interface with the neurons. A chemical transfer of information rather than electrical.

Of course.

I'd like to see some glial cells added to the network as well.

I'm sure they're anxiously awaiting your e-mail.

Also, I wonder what type of feedback mechanisms they use to enhance the network over time.

"To control the simulated aircraft, the neurons first receive information from the computer about flight conditions: whether the plane is flying straight and level or is tilted to the left or to the right. The neurons then analyze the data and respond by sending signals to the plane’s controls. Those signals alter the flight path and new information is sent to the neurons, creating a feedback system."​

Me too actually. How does it know it did something bad? Is it "aware"? (I doubt this thing is, but regardless, it begs the question). Where's the line you're not supposed to cross?

 

[invert_nexus]

Yeah, as is obvious by the text. My point is, if you start using rat brain cells for this and human brain cells for that, and the things are learning, is that then for instance, part of the identity of the person from which the cells came... maybe it isn't, but shouldn't the question be asked?

The old question on where memory lies within the neuronal net. Is it carried within the nerve cells themselves somehow in a chemical manner? Or is it carried in the shape of the network?

Personally, I think that human memory is too complex to be conveyed be a single cell as is the case with the planaria experiments. Once the neurons are removed from the brain they are no longer part of us. Any new net that they will become a part of will be just that, a new net. Perhaps there will be cellular memories of its old environment that would have it attempt to form a new network on similar lines, but the chances of such happening are slim to none.

I'm sure they're anxiously awaiting your e-mail.

Already sent and they've responded. Glial cells to be added next week. Stop.

Asshole.

Me too actually. How does it know it did something bad? Is it "aware"? (I doubt this thing is, but regardless, it begs the question). Where's the line you're not supposed to cross?

Is it aware is a tough question. I doubt if it could be. Not because of the mass of brain cells exactly. Insects are aware. And so are other animals with exceedingly simple brains and ganglia. (Not talking about self-awareness.) But, the sensory inputs must be far to vague for any type of real awareness to take place. It seems to be just some type of electroshock treatment.

As to the line, it would be far, far away. Self-awareness is a thing that requires far more complexity than a few neurons in a petri dish are able to provide. There must be all the automatic functions of the brain that dish up and serve sensory and associative computations to the consciousness. This isn't going to happen by accident. Not in a single lifetime anyway.

The question once more comes down to cellular memory. Is it possible for neurons removed from their native environment to form into a functioning brain? I seriously doubt it. Each would try to form the part of the network it was in before. To get a complete brain you'd need neurons from each part of the brain. Or the whole brain. In which case the network would already be preformed.

 

[Avatar]

do the neurons in different parts of the brain have different structure or the difference comes just from their possition in the whole network system?

 

[vslayer]

not sure how they managed to program neurons, anyone know

 

[one_raven]

It sounds to me like they are making it out to be much more than it really is.
Every claim they made was vague enough to not really "say" anything meaningful.

"It's as if the neurons control the stick in the aircraft, they can move it back and forth and left and right," said Thomas DeMarse, a professor of biomedical engineering at the University of Florida who has been working on the project for more than a year. "The electrodes allow us to record the activity from the neurons and stimulate them so we can listen to the conversation among the neurons and also input information back into the neural network."

Back and forth and left and right resulting from purposeful stimulation with that intention?
Not likely, or they would have said that. More likely simple electrostatic feedback that holds no significant meaning whatsoever.
Input information back into the neural network? Come on! They are making it sound like it is a well-defined processor and memory bank.
I picyire it more like, "Hey, look what happens on the oscilloscope ehan you poke it here."

Maybe I'm just cynical, but if that had anything that even approached significance, there would be at least minor details.

I'll wait until the findings come out.

 

[Dilbert]

i do not think you are cynical; it is vauge to say the least.
But, lets not forget that it has potential. It might be vauge today, but 50 years from now we may look back and see this as one of the most crucial discoveries.

one_raven, in your quote Thomas DeMarse said that it was AS IF the neurons controlled the stick. What does that mean? That they did not really control the stick but they either thought Forward, Backwards, Left or Right and someone did it and then they got the feedback or what ?

 

[wesmorris]

Seems to me though that any successful nueronal/electronic interface is proof of concept of something much much larger.

 

[invert_nexus]

do the neurons in different parts of the brain have different structure or the difference comes just from their possition in the whole network system?

You know. I'm not entirely certain, come to think of it. But, it almost has to. I haven't read anything specific about base neuronal organization in different areas of the brain, but they have to be different.

For instance, in the cortex it is only the surface of the cortex where the neurons dwell and do their thing. Beneath this grey matter is the white matter which is axonal connections.

Beneath this in the limbic system and in the cerebellum and the other areas of the brain, I can't imagine that they are organized in the same manner as the cortex.

We've got a new brain expert in town, maybe she'll poke her head in here and clue us in.

Input information back into the neural network? Come on! They are making it sound like it is a well-defined processor and memory bank.
I picyire it more like, "Hey, look what happens on the oscilloscope ehan you poke it here."

That's how I read it to.

not sure how they managed to program neurons, anyone know

They're not. They're letting the neurons program themselves. Hoping something good will happen. The article mentions a feedback system, but I think that this is another exageration.

Seems to me though that any successful nueronal/electronic interface is proof of concept of something much much larger.

It's a starts. But, I wonder if they wouldn't be better off by giving the neurons a virtual worm to control or something else equally simple.

 

[Avatar]

You know. I'm not entirely certain, come to think of it. But, it almost has to. I haven't read anything specific about base neuronal organization in different areas of the brain, but they have to be different.

For instance, in the cortex it is only the surface of the cortex where the neurons dwell and do their thing. Beneath this grey matter is the white matter which is axonal connections.

Beneath this in the limbic system and in the cerebellum and the other areas of the brain, I can't imagine that they are organized in the same manner as the cortex.

We've got a new brain expert in town, maybe she'll poke her head in here and clue us in.

maybe you can point her/him to this thread...

so .. the white stuff, etc. isn't it just different environment?
anyways I don't know about this and it would be great to hear the/a brain expert

It's a starts. But, I wonder if they wouldn't be better off by giving the neurons a virtual worm to control or something else equally simple.

maybe a virtual tank?
maybe the military need an ( a not so) artificial intelligence to pilot their unmanned craft..

anyways.. I agree that a snake game would be a better option in terms of understanding how it works, but controlling an aircraft certainly shows that this opens far wider possibilities than artificial worm control

I really think that (if true) this is one of the greatest achievements of science, the first baby step of something more grand to emerge
p.s. and this is scary too
now we do not have to design from scratch an artificial intelligence, we can just grow one

 

[c20H25N3o]

Seems to me though that any successful nueronal/electronic interface is proof of concept of something much much larger.

I agree. But this is where ethics does come into play surely!
We start off with something simple, we get excited by our progress, we seek more and more parts of biology to fuse to machines. We gain a greater understanding of how to make biological cells perform greater and greater tasks. You know what, I reckon by the end of the process we will have created a whole human being out of bits of another human being! Then guess what - we can get the created beings hands to actually fly the aeroplane! What a buzz that would be. What shall we call the being? Who's his daddy? Who cares right? Wrong. Big no no - start again!

This is so flawed it makes me laugh. But let the 'intelligent' have their fun - whatever!

 

[wesmorris]

The thing is: How the shit do they set the goal? I mean, whassup? How does it know "don't crash"? How does it know what its supposed to be doing? I don't get it. How do you "learn" when there is no goal? If there IS a goal, how do the nuerons know? Is it just negative feedback? If so, in what form? Shock I guess? Why do random nuerons care if they get shocked?

 

[c20H25N3o]

The thing is: How the shit do they set the goal? I mean, whassup? How does it know "don't crash"? How does it know what its supposed to be doing? I don't get it. How do you "learn" when there is no goal? If there IS a goal, how do the nuerons know? Is it just negative feedback? If so, in what form? Shock I guess? Why do random nuerons care if they get shocked?

It's like an orgasm for them. They love it. You know they do

 

[Avatar]

I'm guessing that in the future citizenship will be given to intelligences not people.
Like the Swiss citizenship to an A.I. in Neuromancer.
Just because there will be no more clear line between a human and a machine.
A citizenship and rights given to whatever form of an intelligence requests it or is given it. (Opens possibilites for dolphins*, dogs, etc )

*

“Humans think they are smarter than dolphins because we build cars and buildings and start wars etc...and all that dolphins do is swim in the water, eat fish and play around. Dolphins believe that they are smarter for exactly the same reasons.”
--Douglas Adams

 

[Avatar]

Why do random nuerons care if they get shocked?

maybe because they are not random any more?
I mean.. if they have created a network by themselves, then there is some drive behind it, and stopping the drive (their network expansion, functionality, whatever) annoys the network

 

[invert_nexus]

maybe you can point her/him to this thread...

Her name is Hypatia. You can ask her if you want. I generally don't like to go bothering strangers.

anyways.. I agree that a snake game would be a better option in terms of understanding how it works, but controlling an aircraft certainly shows that this opens far wider possibilities than artificial worm control

But, that's the thing. This isn't about what possibilities there are for application of this in a technological world. It's about understanding how the neurons function. This is why the article so overdramatized it.

I imagine that this is the exact purpose of connecting it to a flight simulator. The journalistic value of it. Hoping for investment.

That and the difficulties in designing new software rather than some off the shelf simulator. But, surely a worm simulator couldn't be that difficult. But, the way that the ganglia of a worm connect to the worm would still be far more complicated than is possible in this electrode method. I would imagine. That and planaria neurons lack axons and dendrites, being a whole 'nother critter.

But, still. Something simpler would be better.

The thing is: How the shit do they set the goal? I mean, whassup? How does it know "don't crash"? How does it know what its supposed to be doing? I don't get it. How do you "learn" when there is no goal? If there IS a goal, how do the nuerons know? Is it just negative feedback? If so, in what form? Shock I guess? Why do random nuerons care if they get shocked?

Exactly. It's random. There must be an emotional system to make the 'brain' prefer one situation over another. Blue sky above, for instance. It seems that they're just hooking it up and hoping that it works.

And, the thing about shocking neurons is that it halts their behavior. If you shock an area of the brain containing a certain word, for instance, you don't say that word. You forget it.

 

[Avatar]

Well, I have a theory how you can arrange possitive/negative feedback
I thought of this just now when comparing it to some basic single celled organisms
they also are not intelligent, but they get and understand feedback from the outer world (basic photosynthesis for instance => light is good, no light is bad)
so in our braindish case we can simulate it
they must get some kind of energy to sustain, grow themselves, right?
so we hook up an additional (be it visable light) sensor and for instance when the brain crashes the plane, we minimize the food source and also lessen the light on the sensor (double feedback. plane crash = no light = no food // light = food // plane flies = food + light)
so even the most simplistic life-form understands (and we have life here) that in order to keep alive, sustain itself (basic instinct, property of all the life) it needs to fly the plane and not crash it