Slashdot Mirror
Flying By Brain
Garabito writes "Scientists at the University of Florida made a living 'brain' by extracting 25,000 neurons from a rat's brain and culturing them inside a glass dish. Then, the neurons began to extend lines to each other, creating a living neural network between them. The dish had a grid of 60 electrodes connected to a computer running a flight simulator. The scientists were able to train the 'brain' to control the plane in the simulator and to react to conditions of the plane. Are we getting closer to create an artificially made conscious being, or perhaps, a living computer?" AlphaJoe was one of several readers to add a link to Wired's article on the experiment.
so from the point of view of the brain, it's an aeroplane. and it flies around in it's self contained reality.
How did the clump of neurons know what they were trying to accomplish? More precicely, why didn't they try to crash the plane? What sort of positive/negative feedback did they use? I understand that this works, and vaugely how it works, but i can't wrap my poor little brain around what sort of feedback they used!
md5sum
d41d8cd98f00b204e9800998ecf8427e
Am I the only one disturbed by this stuff? I know it's only a rat, but...imagine a world where your brain (sliced and diced) is worth more outside your body than inside. For some reason this kind of reminds me of Larry Niven's classic "Patchwork Girl".
- The race is not [always] to the swift, nor the battle to the strong. -
Bring a whole new meaning of a computer virus ...
So, how is this thing reacting to good and bad?
Did they create a neural net that falls through a given search space to a local or global minimum, or what?
Is "good" a total lack of input, i.e. the plane is flying straight with no lateral or vertical drift, and is degree of input dependent on the amount of lateral motion, etc.?
As I type this, it makes sense that this might be so, but I wonder why the network created a negative feedback system, and not a positive feedback system.
~ Mike
Michael C. Hollinger
I wonder if human neurons would be more effective? Or are all neurons created equal, and only the structure of a brain makes it more or less intelligent? Could we grow rat neurons into a human brain? Maybe we could customize brains for certain abilities, by growing them along certain structures. I don't have alot of personal knowledge here, so i'm just putting out some questions that this brought up for me.
The scientists were able to train the 'brain' to control the plane in the simulator and to react to conditions of the plane.
I seriously doubt this is true as there is very little incentive for this "brain" to perform. When you "train" an organizm you need some feedback loop like bananas, agar gel, money, etc to encourage the organizm to favor one behavior over another. Frankly a collection of neurons just isn't powerful enough to "learn" how to fly a plane.
I know of similar work with sea slugs in an off-campus lab funded partly by UF (the Whitney Lab). I'm not _too_ familiar with it, so this may not be entirely accurate. Basically, they found that neurons in the brain of the seaslugs are always in the same positions as other animals of the same species. They then started training animals, much like pavlov's dogs, to close their siphon whenever they were electrically shocked on their tail (by touching the siphon whenever they were shocked so the animal would relate the 2 stimuli). They then could isolate the neurons in the brain and train then individually. Two neurons in a petri dish would gradually connect and then share information. At the moment the group is working on identifying which genes control what part of the brain, or something like that..
Yes.. you are the only one.
Oreville: Hey Wilbur, you want to try making an Aeroplane?
Wilbur: Nonsense! We dont know anything about Aeroplanes. We should wait until we know more about Aeroplanes before we go building one.
That's the way you make advances in technology that doesnt exist yet. You experiment and gain knowledge so that others can build on that knowledge. Why is it that every time something potentially revolutionary is in the air people are ready to shoot it down?
Your reasoning behind why we shouldnt we carry out these kinds of research is on based on your own fear of the unknown.
Does the question even mean anything?
Years ago, patients with extreme cases of epilepsy were treated by severing the connection between the left and right halves of the brain. The theory was that this would prevent the "electrical storm" of the seizure from propagating from one side of the brain to the other. This would supposedly reduce the frequency and severity of the seizures.
As a result, these individuals had, in their skulls, two independent brains with no communication link between them (a simplification, but mostly accurate). These patients would report strange experiences, such as getting up out of a chair and walking to another room, without having any idea why they were doing it. Essentially, the two halves of their brains were functioning independently, and sometimes "fought" over what the body was going to do.
It's a very interesting question -- did the "person" go into the left half of the brain, or the right? If it went into the left side, for example, what happened to the right side? Is it now a soulless automaton? How can a single person exist in two conscious modes simultaneously? Yet these people live normal lives, for the most part.
Sadly, you are trolling. But you raise an interesting point.
It's not too difficult to find a source of brains - visit your local abbatoir.
Wouldn't want to use the sheep brains though.... Imagine a "mob" of aircraft playing follow the leader...
Seriously, you would want to use something with a life span of more than a few years - besides, how do you do backups? how do you transfer existing knowledge to the new, untrained brain? (I mean more efficiently than us humans manage to using our existing I/O ports).
...because I think (although I am talking out my ass here) neurons are exempt from the auto-immune response, so rejection of donor cells is a non-issue.
If you know, is this true?
Ok, I RTFA and the brain seems to be able to control a simulated airplane, but does the brain know...
...what bad weather is
...why it should avoid bad weather
...what a horizon is
...even what an airplane is
in other words, does it really know why it's doing what it's doing.
Jonathan B.
I wonder what kind of maintinence goes into keeping a 'living brain' computer. Do you have to feed it? Keep it cool? Will it go crazy if you don't give it enough beer?
Now the terrorists will be growing brains to fly planes into american buildings!
Hmmm. Yes, this is the line of thinking I was on here. In the simplest terms, brains are biological computers comprised of neurons rather than transistors. As this technology progresses, researchers will grow more adept at cultivating neural tissues and configuring them for better performance/lower production cost, just as chip manufacturers do. Though the technology is in it's infancy, I see a new industry beginning here, one that makes chips from living proteins instead of silicon wafers. As to the advantages of using living tissue over silicon, I don't know what that would be.
What gets me are the ethical questions that are raised by this kind of research. Given time for the technology to mature, what happens if we produce a sentient cybernetic organism? Or will there be "safeguards" incorporated into the design to forestall this eventuality, in effect lobotimizing the "devices" before the fact. It's very Asimov-ian. Yeah, I know my neural network is going way out on a limb, but the ethical implications of further commoditizing animal tissue are a bit unsettling.
And yes, I am a vegetarian...
"OH SHIT, THERE'S A HORSE IN THE HOSPITAL!"
I should preface this question with the information that I have abosolutely no medical training, nor have I any real understanding of how brain neurons operate. That being said, please be kind in your responses.
.
How are the organic neurons being kept alive in this petrie dish? Surely they would have to have a blood supply or something similar to exist. If they can just live in saline solution or something like that, how it is that possible? Especially given that they are working neurons, not just sitting there doing nothing.
Can they reproduce? (I think I read somewhere that brain cells do actually reproduce, in contrast to the traditional thinking for many years). The actual article says they are:
growing on top of a multi-electrode array
Are there any implications for brain neuron transplants as a result of this type of research?
How are the neurons hooked together? Are they wired up, using impossibly thin wires, or just connected via the array?
What the hell is a multi-electrode array anyway?
Anyway, I guess they are enough questions, although I could probably sit here all day typing away at the million queries this type of research presents me with.
If the pattern goes 9am, 10am, 11am, why isn't noon 12am?
Someone else has probably already said this...
I don't know if mice are self aware, I'm not even sure if I am for that matter, but if that mass of brain cells were capable of 'learning' to control or react to input, was it conciousness? I don't give a rats about the moral implications - I don't care, I just think this is fascinating - very cool!
Just because a lettuce can't scream that doesn't mean it can't feel. Think about that next time you have a salad. At least some cows want to be eaten.
Well its quite simple really, your brain stores knowledge chemically. thats why when you sleep or pass out or even have a near death experience you still remember stuff (mostly). Rats have some of the best reflexes of all creatures in the world, they've adapted to surviving in essentially the same conditions on the ground (ever had a car go past you REALLY fast? magnifiy that to include the scale of rats vs cars). therefore its a simple step for rat brain cells (which have no concept of self afawk) to control an airplane in conditions similar to what they used to control their rat body in. because it has no concept of self (again afawk) theres no problem of it thinking its still a rat, or thinking at all for that matter. therefore you can program them to do whatever you want. our brains are essentially super-complex computers. we have billions upon billions upon billions of 1/0 switches in our brains that allow us not only to follow "computor logic" (think like a computor does) but also to use inductive and deductive reasoning. A rat cant use inductive and deductive reasoning, or at least not nearly to the level we can. Mostly they go by computor logic, a simple matter of goals and obstacles. therefore what your doing with a rat brain is pretty much the same thing, you just put it into 1/0 properly and the cells do the rest. the program has all the info they need and they handle the computing. add enough cells and eventually you reach the critical point where the mass is able to use deductive and inductive reasoning and sensory input to figure out "hey, i exist. so what am i and why am i here?" as opposed to ">need:foodget:food"
A bullet may have your name on it but splash damage is addressed "To whom it may concern."
So this neural net is creating new synapsis and is essencially learning to fly. Now I myself am wondering the extent of the ability of it's growth...Does it remember patterns? can you train it to fly an obstacle course? Just to what extent can it learn? Will it become aware of what it is doing or is it just a set of neurons that have made the appropriate connections to beable to keep a simulated plane from crashing...I can just here greenpeace shouting that humans are playing god again...
The parent certainly wasn't modded "5" for accuracy. Neurons are terminally differentiated and therefore CANNOT divide (or "reproduce," as the parent called it). In fact, if you stimulate an adult neuron with "divide" signals, you often get an apoptotic neuron. Neural STEM CELLS can divide, and some of them hang out near the ventricles in the adult brain and continue to produce neurons throughout life--newly born neurons have even been observed in damaged areas of Alzheimers' brains.
As far as the Wired article is concerned, this sounds pretty cool, but I never trust the popular press for scientific accuracy. The peer-reviewed paper will be worth reading.
or perhaps, a living computer?
Sure, I'd gladly give my brain to this research, or at least some animal's brain. If I have to give a random animal's life so I can have a cool computer that barks like a dog, then so be it, I'm brave and humble enough to make that sacrifice.
Actually, for purity purposes, let's just kidnap some girl off the streets and use her.
"Oh, no, how dare you say that! These fine people are WAY too moral to do something as disgusting a revolting as that! It's just... Oh, wait. Hold on a sec, my cell phone is meowing."
I am NOT a number! I am a - oh wait, I'm number 761710. Look! 761710!
What scares me quite a bit about creating artificial neural networks, is how consciousness and the experience of pain comes into the equation.
... or even boredom.
Does any complex enough system have a consciousness, just as we do? Is that "equilibrium" the system is trying to accomplish experienced as something similar to a person trying to keep their balance on their feet? As a person trying to keep their body away from a surrounding fire?
What if there is a sharp feeling of discomfort in such an artificial system when its input parameters are not within "specifications" (plane flying level)?
Can the experience of pain / discomfort always be measured from outside? Should we continue creating artificial neural networks if we can't answer that question?
Then again - maybe I am being squeamish for no reason.
Certainly not. I think these questions should be seriously considered, since we may eventually (if we haven't already) be creating a real conscious being, perhaps with no way ever of telling the outside world that it experiences a constant feeling of pain....
After all, if your entire existence was flying imaginary planes, maybe that wouldn't be so bad.
My hunch is that these could lead to new, smarter bombs, cruise missiles etc., thus reducing armed forces recruitment demands while advancing the cause of the Crusade, which should please the Christian conservatives to no end. They can call the the control modules RABBAI SADs (for RAt Brain Biometrically Adapted Intelligent Stealth Aeronautical Devices), a name which would no doubt score points with the Evangelicals at the polls.
media girl
Merriam Webster defines conscious as:
1 : perceiving, apprehending, or noticing with a degree of controlled thought or observation
2 archaic : sharing another's knowledge or awareness of an inward state or outward fact
Also, consciousness is:
1 a : the quality or state of being aware especially of something within oneself b : the state or fact of being conscious of an external object, state, or fact
I think the question we are really worried about is: When does a neural network become a living, sentient being?
True, a biological neural network has been alive from day 1, and a consciousness is what we generally attribute sentience to, but consider this:
One poster commented that we might possibly create a sentient being that could feel pain and was, in fact, feeling constant pain, but unable to communicate this to us. What if we simply created the neural network without the ability to ever feel pain? (How, you ask? As I understand, pain is interpreted as a chemical change in a sensory nerve.. I didn't do too well in HS biology, so don't quote me.)
In short, we could make sentient beings with our installed sense of motives (10 laws of robotics perhaps?) and no ability to feel pain. Sentient, but not human, you might say. This would seem, to me, perfectly ethical (speaking to grandparent here) and practical.
The power of Christ compiles you.
A Random Blog
What does it take before a bundle of cells will start to exhibit conscious behaviour. Are there degrees of consciousness?
First degree conciousness starts on average with 33 billion neurons. Some examples have been noted in simpler cases, but beyond 33 billion, examples of 1st degree conciousness abound. The second and third degrees of conciousness occur at roughly equal spacings: 48 and 63 billion. For reasons not yet fully understood, fourth degree conciousness like our own doesn't occur until 100 billion neurons.
OK, that was complete bullshit :)
Seriously, number of neuirons isn't very important. Blue whales have the largest brain on earth, but they're not the most "conscious" as far as I know.
More interesting is the measure of brain size to body size. Plotting that line from the tinitest organism with a brain all the way up to the blue whale, you see a very constant ratio. There are a couple of notable exceptions, though: humans, and... dolphins. We both have abnormally large brains for our body size. In fact, especially with humans, we have ridiculously large heads for our size and probably look quite silly to the rest of the animal kingdom.
My bet is that conciousness is just a label we attach to certain complex behavior that we don't yet fully understand. We have hundreds of billions of neorons and trillions of connections between them -- all receiving input from visual stimuli, audio, tactile, chemical, etc. sources. AD.com's simulationsi a grand attempt to see what we can get out of an articifial neural network of similar capacity to that of the human brain. But event hough they took care to pre-organize their brain into a human-brain-like structure, iut may be trumped by our crude understanding of our own brain's structure. Perhaps they'll find something, but I don't think it will be conciousness on day 1.
The bigger problem is how you define conciousness. Some people will sit and chat to Eliza-like programs and not realize it isn't a human. Other people will be biased and say a machine 10-times more capable than our own brain and won't be concious by definition that it is a machine.
What happends when the "brain" becomes sentient and starts fucking with the results, crashing randomly so that it won't have to fly a fucking plane 24/7 and instead get thrown into the trash were it can adapt and take over the world?
The other question is slightly more serious; Does anyone else wonder how long it's going to be before we have "bio neural networks" embedded in everything from our cellphones to doing fuzzy logic in our washing machines?
weird.
j.
I wonder if Animal X tastes like chicken...
You know you've lost it when you begin signing physical documents with =^_^=
An interesting post - too bad we dont have more discussions like this on slashdot.
Ethical issues are certainly something to be considered - but this does not neccesarily just apply to biological neural networks. I dont see any reason why we shouldnt apply the same concerns to neural networks in software or silicon. Although instinct suggests to me that a biological network is going to be the most similar to the real thing and therefore more likely to offer closer similarities.
My personal take on conciousness is that it is an emergent behaviour. For example imagine a brain that is kept alive- but has never received any sensory input. Its fairly likely that it couldnt be concious - because conciousness requires processes based on accumulated knowledge. Whether that is learned by cause and effect - as a baby learns quickly what actions to get a feed. The more choices we have , the more knowledge we have and the more we are able use these things to effect the world around us or to enjoy the things in the world around us.
It is also important to consider more lowly lifeforms which exhibit conciousness. One of my favorite examples is the "Bower Bird". The bower bird exhibits true creativity. The male bower bird attracts females by collecting colorful petals, butterfly wings and other items. And by arranging these items in a specific way create a beautiful display. (experiments were performed whereby a scientist rearranged pieces - the birds would put them in the correct spot again)
Female birds then select a prospective mate by selecting the nest it finds most appealing.
What this shows is that these birds can be considered truly creative in that they can both create a work whilst also being able to appreciate the work of others.
To me this example highlights the fact that we should not make the mistake of thinking that it is only the larger - higher level animals that exhibit a complex conciousness.
Anyone interested in these kinds of issues and discussions should look at some of the work by Daniel C Dennet
http://ase.tufts.edu/cogstud/biblio.htm
In particular his book
"Conciousness Explained"
Nick...
Electronic Music Made Using Linux http://soundcloud.com/polyp
What if it was a really large dish with a lot of neurons?
Finally! A year of moderation! Ready for 2019?
The brain is fully functional even when sliced in two, however it does lead to some really fascinating side effects brought about by the differing functions of the two sides.
In effect, we all have two brains, they do different things but by communication we end up with a single whole brain, once you cut the CC you're back to two brains, with different capabilities. Most of the time you won't notice the difference because the brains compensate adequately, but in certain situations you can expose some truely bizarre features.
(http://www.schiffermd.com/dualbrain.html)
Here's another interesting link with details about one case which through having an unusual development of language in both sides of the brain the experimenters were able to discover that the two brains (after separation) were vastly different in thier ideas, rigt down to what job the person would like to lead (race car driver vs draghtsman!).
http://www.macalester.edu/~psych/whathap/UBNRP/
NZ Electronics Enthusiasts: Check out my Trade Me Listings
I wonder what it eats.
They have invented the ornithopter. Frank Herbert imagined a feudal galactic civilization in the wake of an anti-AI jihad, where technology raced forward in the shadow of the religous edict "thou shalt make no machine in the image of the mind of a man". Artificial intelligences were verboten, but vast augmentations of the human mind were fair game.
On present-day Earth we struggle with similar taboos, like stemcell research. This ratty project points to a vast potential for human/machine interface and learning. After they perfect the training of these resynthesized rat brains for controlling an airplane, they seed their tanks with human nervous stemcells. Once the training regime is "humanized", these flying tissues might be grafted into existing human brains with more stemcells: brain plugins. We might grow various motorskills, like flying, driving, or space navigation, simultaneously in tanks, while we train our "default" brains a more oldfashioned way, then plug them all in to "graduate".
All those old pictures of "future humans" showed our descendants with big cranium globes. Lots of us have laughed at those pics, because past evolution trends towards bigger skulls have probably stopped with human siezure of our own reproduction. But maybe those big skulls are just artificial expansion bays...
--
make install -not war
Is while some of the popular paths they take can lead to something academically interesting they often don't lead to significantly better understanding, nor really useful.
You stick a bunch of neurons to a computer and after a period of training it does _mostly_ what you want. But once they get to a certain stage of complexity, they don't really know why it works that way - they can't summarize/simplify things (no E=MC^2). It's more like alchemy of old. Stir in a bunch of stuff, and you while know that A+B gets you C but you don't know much else.
Doh stick neurons together and they can learn. Oh wow... Like we didn't know that already. Poke a needle in a frog and it twitches.
Sure you will still have to experiment with neurons, I'm not saying stop science. But lots of this is not good science, nor necessary either (it's only necessary so the scientist can publish some paper and get grants etc).
Sure Alchemy developed into Chemistry and other sciences. But maybe this time scientists knowing what they do should be a bit more scientific, given the possible far-reaching impact of their work. The path many are taking is just like mixing random brews and hoping it works. Hope we skip the consuming mercury, uranium part etc.
As is, for many of the things being researched, we might as well use existing animals as they are, or augment them accordingly instead. For instance, you could use a bunch of trained dolphins in shift to help control and process sonar for a submarine. Same for using dogs to sniff for explosives. It really isn't that hard. You can already interface brains with computers already. In short there are tons of existing prepackaged neurons + supporting "hardware" that do much of what we want.
The dolphins/dogs will get bored? Sure, but once you start using tons of neurons hooked up in complex interlinks (for more features) how'd you know what will happen either, or what is actually happening? Cruel to the dolphins? Maybe. But how about those neurons?
Many animals are pretty good at what they do. And they have very similar requirements to humans (which often means they are well suited to helping us). We can relate to them and they can relate to us (in our limited ways).
If you wire up an animal, you know it is hurting if you are do something bad to it. Whether that is necessary in the big picture is for us to decide, but at least we know we are doing something bad to it.
Whereas if you just keep chucking together more and more neurons together and create symbiotes with rather different requirements and perceptions, things might not be so good, nor go as well.
Just because something is legal doesn't mean it isn't wrong. Anyone who actually spends some time with animals will soon realize that they do have emotions and personalities and other hallmarks of self. To subject them to unnecessary pain and suffering is morally wrong. If using animals as a resource, they should enjoy decent living conditions and a quick and painless death.