Slashdot Mirror
Researchers Simulate Building Block of Rat's Brain
slick_shoes passes on an article in the Guardian about the Blue Brain project in Switzerland that has developed a computer simulation of the neocortical column — the basic building block of the neocortex, the higher functioning part of our brains — of a two-week-old rat. (Here is the project site.) The model, running on an IBM Blue Gene/L supercomputer, simulates 10,000 neurons and all their interconnections. It behaves exactly like its biological counterpart. Thousands of such NCCs make up a rat's neocortex, and millions a human's. "Project director Henry Markram believes that with the state of technology today, it is possible to build an entire rat's neocortex. From there, it's cats, then monkeys and finally, a human brain."
What? Your post is so wrong I don't even know where to begin.
First off, why not just use a human brain if you want an identical machine? Well, for sending probes to mars. Or to the depths of the ocean. Or any other place that is too dangerous to send humans, but that a machine could survive in. Even if the brain was a replica of someone's personality, all they'd have to do is find someone who thinks it would be really cool to go to mars, and replicate their brain. It'd be a hell of a lot more intelligent than a traditional AI system at this point.
Secondly, if we want an AI system that better than the human brain, THIS IS THE WAY TO GO. Figure out exactly how the human brain handles thing that are really hard for computers, like object recognition. Once you've got that, you can replace//add on parts that do things better/faster than humans, like math. In terms of adaptability and general purpose use, NOTHING in AI comes anywhere close to the human brain right now. So trying to make an AI system that is better than the brain, a good first step is to try and make the human brain, then start tweaking that.
The point is to try and understand how biological brains do what they do, and how we can make computers do those things (which computers currently suck at). Sure, you can emulate basic behaviour in a pre-define environment, but try making a system that can differentiate a food source the 'rat' may never have seen before based on sight and smell in an environment that it's never been in.
Open Your Mind. Open Your Source.
I disagree. The human brain would be perfect to use as a model to create something "better" than the human brain.
...).
The reason is that the artificial brain would have the benefits of intelligence, creativity, etc... that we see in people, but none of the limitations.
Imagine the smartest person you know, but with essentially unlimited memory, constantly increasing processing capability (with newer/faster processors) and the ability to live forever without a decline in function. That's better than a real human brain for sure.
But that's all science fiction, for now. If we could replicate even the modest abilities of a normal person, or even a primate, we'd be well on our way to true artificial intelligence (if not already there
Not to be a doubting Thomas but I think that they are underestimating the complexity of a brain. There are many different chemicals and biochemical reactions going on in the body, that science has only a vague idea of their mechanisms. Look at any drug in the market, most of them only give conjecture on why they work. My feeling is that until one day when we can create computer models that reliable predict the effects of drugs in the brain or in the body in general, these models are nowhere near what real brains are. But I would also love to be proven wrong.
"and finally, a human brain."
Why stop there?
It's amazing how some people want the computing resources to simulate a rat's brain but still can't simulate a honeybee's brain and the resultant behavioral complexity. After all, a bee's brain has only about a million neurons. It could probably be done on a desktop machine and yet, a bee's behavior is amazingly sophisticated. Is it me or does it seem that some people have no clue as to what constitutes intelligence and would rather spend the taxpayer's money on what can only be qualified as useless goals?
Would it not be much better to implement a downsized version of the human brain (with all the various cortices) and see if it can learn and adapt to the environment? But then again, that would be too much to ask since Markram et al don't have an overall theory of brain operation. It's better to keep your sights as high as possible and have an excuse as to why your artificial brain or cortical column is no more intelligent than a flea: you always need faster and more expensive computers. And more funding. Yeah.
Yet no doubt when a competent emulation of a bird brain exists and is observed flying around, you will raise the bar again. Not long ago recognizing natural speech was offered as you offer the test of flight. We have since moved the bar because our inexpensive, portable, battery powered cell phones now understand the simple noises we make with accuracy approaching our own. Bipedal walking, land navigation, chess and facial recognition are more examples of tests offered that once solved, for some reason, no longer count.
Consider this; we're having to move the bar with greater frequency all the time. At what point does the realization occur that the problem of thought is finite and solvable? I believe that very soon we will have at least parity between ourselves and our machines. Not because the machines are tremendously powerful, but because we're not.
The count of neurons (100G+) and synapses (up to 10K per neuron) is well known. The switching speed of this finite set of electrical and chemical circuits is measured in (comparatively slow) milliseconds. Our brains run on a couple calories a minute and operate at approximately body temperature. In contrast to the infinite supply of uniform opinions offered here that effectively assert that the brain is too elaborate for it's own comprehension, there simply isn't enough space or energy involved to convince me that the brain is some unapproachably complex enigma forever beyond our capacity to emulate.
Every new milestone passed only reinforces my belief, regardless of how fast you raise the bar.
Lurking at the bottom of the gravity well, getting old
Wow, these scientists really were shooting for the stars. Why not start small, like say the brain of a GOP presidential candidate or that of a Britney Spears fan?
I think your critique is woefully out of date. You are correct if you limit the neural network to the basic neural network models of decades past. From what I've seen at conferences in the HPC world lately, the more recent models do more than just use capacity to increase the size and connectivity of the network, but take into account more realistic physical models such as the electrical properties of the brain and mechanisms by which signals propagate both within neurons and across synapses. You're not looking at just a bigger back propagation network with sigmoid nonlinearities here -- the neural modeling world has moved far beyond that, in part due to increased interest and participation of neuroscientists. Unfortunately, most CS folks fail to learn much about the current state of the art beyond the basics such as the material from Simon Haykin's text (which, mind you, is pretty good).
Your body is destroyed and a copy of you comes out the other end, thinking that it's the original ... which makes it me, the original, for all purposes.
it's in my head
I strongly consider you to perform a modicum of research before you regurgitate knowledge you got at a party while partly intoxicated, and hoping to get that girl-in-the-green-dress' phone number.
Oh wait... do you get invited to those kinds of parties? Perhaps you think digital watches are a pretty cool idea?
I have no problem with your religion until you decide it's reason to deprive others of the truth.