Blue Brain: IBM and Swiss university to model human brain

This week saw the launch of a 10-year (or so) project between IBM and a Swiss university team to model the entire human brain. The Blue Brain project will involve building a new computer based on IBM's Blue Gene supercomputer (one of the top 5 fastest). The team will start by modelling the electrical structure of neural circuits repeated throughout the brain and then map and model their behavior. Once complete, they will move onto creating a molecular model of the neurons involved and a complete neocortex (the largest and most complex part of the human brain) before modelling the rest of the brain. The end product will then (hopefully) give us an understanding of certain malfunctions in the brain's "microcircuits" which are thought to cause psychiatric disorders such as autism, schizophrenia, depression and internal doubts that Apple would ever go Intel.

[Via NewScientist]

[matt @ Dec 19th 2005 2:05AM]

What no skynet jokes here?

[alienhead @ Dec 19th 2005 2:05AM]

one more step closer to skynet.

[alienhead @ Dec 19th 2005 2:05AM]

How's that?

[Cullen @ Dec 19th 2005 2:05AM]

what happens when the computer realizes its trapped in 50 8' steel racks in the basement of an office building?

[Andrew @ Dec 19th 2005 2:05AM]

In the order of what scientists know most about "the brain": the peripherial nervous system (this is the part that carries information away from and to the spinal chord, this is also where we get most of our information about exactly how things like neurons work), the anatomy of the "central" nervous system (scientists have a pretty good idea about where the basic structures of the brain are but have less information about exactly what they do), actual neurons and so-called pathways in the brain (everything scientists know about what goes on in the brain is based on knowledge about the periphery because it is extemely difficult (i.e., has never really been accomplished) to isolate part of a living brain and figure out exactly what it does), then the final question is, how do all the things I've just mentioned work together to do the things we do -- this is the hardest part. Making the leap from what happens at the level of a single neuron to what happens at the level of the individual (e.g., basic behavioral relations) is like trying to fill the grand canyon with cotton candy. We're not even close do doing that kind of thing. In other words, while modeling the human brain might be interesting from a computer science point of view and might even result in some interesting AI, it's going to tell us much less than what we might gain from investing that money elsewhere. Even 10 years down the road, this project will not succeed in making any reasonable approximation of a human brain, much less being able to identify the conditions that lead to psychological disorders.

There's another point of evidence to consider. Our current understanding of the brain has resulted in the following number of drugs discovered to treat a mental/behavioral illness/psychological state: zero! To date, every known psychoactive drug has been discovered either by accident, by trial and error, or by taking a drug that is known to cause some effect, changing its chemistry slightly, and testing its new effect. The point I'm trying to make is, we don't know anywhere near what we would need to know to simulate a brain. We could waste money and time and careers trying to simulate it, but we're never really going to know unless we study it directly.

[Whiplash @ Dec 19th 2005 2:05AM]

I agree with Andrew. I think also though that it will probably be impossible to simulate a living, thinking, self aware, organic structure like the brain with a "dead" machine that pushes around 1's and 0's. We need a new type of computer to even start to grasp the complexity of the brain.

[John @ Dec 19th 2005 2:05AM]

Skynet? What the heck happened to HAL? Since when did a classic like 2001 get passed over by schlock (good schlock) like Terminator?

In any case, if HAL is ever going to learn to sing "Daisy" it'll need to operate with optical, quantum or even DNA processors. Not there yet.

[Spencer @ Dec 19th 2005 2:05AM]

I for one will welcome our new computer overlords!
:0

[:0 @ Dec 19th 2005 2:05AM]

What they really should devote all this incredible round-the-clock computing power to is figuring out the reasoning behind why Bill chose to name his company Microsoft instead of Huge-n-Hard.

[hafa @ Dec 19th 2005 2:05AM]

i see judgement day coming, 2 days after they build this monster

[Asz @ Dec 19th 2005 2:05AM]

Funny that Bluegene looks like a giant version of Nintendo's Revolution prototype.

[uclatommy @ Dec 19th 2005 2:05AM]

I imagine that they would not build a model of a brain from scratch, but would probably copy the exact structure and configuration of each of the billions of neurons in someone's brain which, when run, would probably be an exact model for that person's memories, behaviors and thoughts. Perhaps they would use a brain from a body donor who had just died. Imagine if you had died and they built an electronic model of your brain and ran the simulation. It would be as if you regained consiousness, but you would awaken and everything would be dark because you have no sight, and you have no feeling of your body or hearing or a sense of smell-- only your thoughts. Perhaps you would even be able to recollect your death. Also, if you are lucky, the researchers found a way to convert sound into neural signals that the model brain would understand and the researcher is able to communicate with you. You would learn that you have become an experiment and that all the thoughts you are experiencing is only simulation and that you would only be alive so long as the experiment is running.

[uclatommy @ Dec 19th 2005 2:05AM]

please dont pull the plug!!