Blue Brain Project

[sphenisc]

According to
http://news.bbc.co.uk/1/hi/sci/tech/4071192.stm
"Neuroscientists are to build the most detailed model of the human brain with the help of an IBM supercomputer. "

It may be possible for someone who believes in the existence of a soul to be comfortable with this experiment. However, I suspect that materialists would regard consciousness, awareness and free-will as emergent properties of the brain. Can they condone the creation of a sufficiently detailed simulation of a brain such that these properties may emerge? If so, what constraints on testing should there be, i.e. what (rights?) should such a simulation have?

Any thoughts?

 

[IllegalArgument]

First question that pops to mind. Will it be able to pass the Turning test?

 

[Phil]

According to
http://news.bbc.co.uk/1/hi/sci/tech/4071192.stm
"Neuroscientists are to build the most detailed model of the human brain with the help of an IBM supercomputer. "

It may be possible for someone who believes in the existence of a soul to be comfortable with this experiment. However, I suspect that materialists would regard consciousness, awareness and free-will as emergent properties of the brain. Can they condone the creation of a sufficiently detailed simulation of a brain such that these properties may emerge? If so, what constraints on testing should there be, i.e. what (rights?) should such a simulation have?

Any thoughts?

I'm a little confused by your post. Are you suggesting that materialists would for some reason be opposed to this type of brain mapping?

First, I believe the project is meant to be the most extensive mapping of the brain ever done, aided by the use of a supercomputer. And that's basically it. Unless I missed something in the article, I don't think they're suggesting that the model will be a "working brain" with all its inherent emergent properties.

Having said that, I personally would put no constraints on testing whatsoever. If in fact science did create a brain and consciousness, awareness and free-will did emerge from it, all the better.

I have no problem allowing scientific experimentation to go as far as it possibly can go, in any dicipline. And if we somehow destroy ourselves in a paranoid science fiction/horror story sort of way . . . well, we had a pretty good run.

 

[sphenisc]

I'm a little confused by your post. Are you suggesting that materialists would for some reason be opposed to this type of brain mapping?

Sorry if I wasn't clear - Yes.

I have no problem allowing scientific experimentation to go as far as it possibly can go, in any dicipline.

Mengele would be proud of you!

Unless I missed something in the article, I don't think they're suggesting that the model will be a "working brain" with all its inherent emergent properties.

The site says "Researchers expect not only to get a better understanding of how the organ is wired up but also to use that "atlas" of neurocircuitry to probe how the brain functions - and malfunctions.

The scientists say the project could lead, for example, to new ideas on how psychiatric disorders develop - illnesses such as autism, schizophrenia, and depression."

This suggests to me that the they expect to produce a 'working model' not merely a mapping of locations.

Cheers

 

[IllegalArgument]

Phil,

I think the question he's getting at, which I missed the first time through. If the project created an AI, what rights would the AI have? Could we or should we wipe it's memory clean as a test? That's of course, if this project were to create an AI.

I'll let mercutio handle the obvious what is consciousness, awareness and free-will questions.

 

[kedo1981]

Ok say it works and a self aware sentient “being” brain emerges.
Would it be immoral to make it or others like it, insane so we can study it ?

 

[SGT]

Phil,

I think the question he's getting at, which I missed the first time through. If the project created an AI, what rights would the AI have? Could we or should we wipe it's memory clean as a test? That's of course, if this project were to create an AI.

I'll let mercutio handle the obvious what is consciousness, awareness and free-will questions.

I think that if the AI showed signs of self-awareness (wich I strongly doubt) we would have no right to wipe its memory clean anymore that we would be allowed to experiment with a sentient alien.

 

[Phil]

Mengele would be proud of you!

As would the many people who would benefit from stem cell research.

Supposing a working model resulted from the project, in my view the AI would have no rights whatsoever. It would be totally at the mercy of the creators. They would be free to do with it what they please.

(An aside: This scenario is simply bringing the god/creation dynamic into reality. Humans in this case would be the gods and the AI would be the creation. Most religious folks have no problem turning their lives over to whichever god they believe in. Only in this case, the gods would actually exist.)

Now, is this moral? Is this just?

These are the questions I think you're asking. And I don't know that these are questions that necessarily need to be asked.

Morality is a function of evolution. It may very well be impossible to imagine a future scenario and then to accurately determine which actions if any will be right and which will be wrong. I think we make the mistake of expecting a final product when it comes to moral evolution, just as many uninformed people think in terms of a final product when considering biological evolution.

Were I a member of the project team, I don't think I would have any second thoughts about treating AI as anything other than an artificial thing. I think such would be necessary to take that new technology to whatever next steps await it.

However, I can imagine a later sequence of events that would give rise to things like Azimov's laws, and other tenets that would allow for a more ideal co-existence.

 

[drkitten]

This suggests to me that the they expect to produce a 'working model' not merely a mapping of locations.

With the current state of understanding of the brain, no one working in the field expects to produce a 'working model' at any point in the foreseeable future. I suspect we're at least three Nobel prizes away from a working model for the human brain.

Given how unrealistic your worries are, I doubt anyone on the project has taken (or would take) the effort to address them. The first astronauts to walk on the moon didn't take firearms with them for the same reason -- no reasonable person involved in the project thought that there was any need. On the other hand, NASA got a lot of questions from the space-opera-reading general public about what the astronauts would do if they encountered hostile aliens.

 

[Ashles]

According to
http://news.bbc.co.uk/1/hi/sci/tech/4071192.stm
"Neuroscientists are to build the most detailed model of the human brain with the help of an IBM supercomputer. "

How are they going to model this?

Are they going to create binary analogues of every single electrochemical path in an entire neocortex? Okay. That sounds like an unbelievably large amount of processing power and programing required, but let's for the moment assume they can do it.

How are they going to subsequently model the general operation of a brain, relative weightings of all individual neurons, information transfer, storage etc.? I was under the impression that we were nowhere near understanding this.

Without that knowledge what use is a neocortex model?

Isn't it a bit like building a computer perfectly, but not having any software?

I don't really get what they are hoping to achieve.

Researchers expect not only to get a better understanding of how the organ is wired up but also to use that "atlas" of neurocircuitry to probe how the brain functions - and malfunctions.

So they are building a structure in order to understand how the structure is built?
Surely if they don't know how it is built or wired, then whatever they build will be different and work differently...

No, I'm not getting this.

And I agree with newdrkitten - we are no closer to having to worry about what the 'rights' of this creation will be than we are to worrying about that other Star Trek plot - what might go wrong with the holodeck.

"With an accurate computer-based model of the brain much of the pre-testing and planning normally required for a major experiment could be done 'in silico' rather than in the laboratory.

"With certain simulations we anticipate that a full day's worth of 'wet lab' research could be done in a matter of seconds on Blue Gene."

These sound like some ridiculously over-optimistic claims. Sounds like someone is exaggerating for funding purposes.

 

[drkitten]

Re: Re: Blue Brain Project

How are they going to subsequently model the general operation of a brain, relative weightings of all individual neurons, information transfer, storage etc.? I was under the impression that we were nowhere near understanding this.

Without that knowledge what use is a neocortex model?

Actually, that sounds to me like a reason to build the neocortex model; we finally will have something to test the theory-of-the-week regarding information transfer and storage in the human neocortex on a realistic scale. Or whatever we think the glial cells might be doing this month. Or whatever.

It's exactly because we are nowhere near understanding this that we need to experiment with the possibilities and see what works (and what doesn't).

 

[sphenisc]

Thanks for the comments guys,

From http://bluebrainproject.epfl.ch/
---------------------
The Blue Brain Project was launched by the Brain Mind Institute, EPFL, Switzerland and IBM, USA to simulate the brain from the molecular to the whole brain level.

This project requires precise quantification of the brain's micro-architecture (BMI expertise), the state-of-art approaches to simulate complex systems (IBM expertise) and powerful supercomputing (available in the new Blue Gene/L Supercomputer).

The first phase of the BBP will be to replicate, in software, the Neocortical Column (NCC) composed of 10'000 morphologically complex neurons and 10x8 synapses for real-time simulations.

The neocortical column constitutes the quantum leap from reptiles to mammals and is considered to be the elementary network of neurons that can act as a unit exhibiting some of the most complex functions of the brain. The NCC sees to have been repeatedly duplicated in evolution to form the neocortex - the beginning of mammalian intelligence (see The Brain: Our Universe Within).

An accurate replica of the NCC is the essential first step to simulating the whole brain and will also provide the link between genetic and molecular levels of brain function and cognitive functions. There are numerous other potential benefits of siimulating the Neocortical Column.

The second phase of the project will be to a) simulate a NCC at molecular level detail and b) to simplify the NCC for repeated duplication, recapitulating the evolutionary process, to simulation progressively large brain regions and eventually the whole brain.
----------------
(I assume 10x8 is supposed to be 10^8), the claim seems to involve modelling a two mm long bit of brain tissue (the NCC) at 'molecular level detail' (every single DNA strand, protein and fat molecule, each neurotransmitter and receptor, and all those water molecules...)
Then develop a 'simplified' model NCC and produce a population of them to simulate brain regions. My suspicion is that brains are very expensive organs to maintain and if there were ways of simplifying their construction without significantly altering their functionality then evolution would have found away ( given the usual caveats about evolution and optimisation, local fitness yada,yada)

I hope that answers some of your questions Ashles, but I have to say I'm beginning to agree with your skepticism about the project.
Perhaps if some clearer claims and a time frame is presented, they may be eligible for the JREF challenge... when does 'ridiculously over-optimistic' become paranormal?

 

[Ashles]

Re: Re: Re: Blue Brain Project

Actually, that sounds to me like a reason to build the neocortex model; we finally will have something to test the theory-of-the-week regarding information transfer and storage in the human neocortex on a realistic scale. Or whatever we think the glial cells might be doing this month. Or whatever.

It's exactly because we are nowhere near understanding this that we need to experiment with the possibilities and see what works (and what doesn't).

Well wouldn't it make more sense to pick a much smaller area to test? Perhaps an area that has limited and specific actions eg. Broca's area or something?

Start with a much simpler model and modify it until you start getting results similar to observed behaviour, then you know you're on the right track, rather than just creating an entire inaccurate neocortex. If they've got it wrong somehow, how will they ever find out where and in what way?

ETA - just read sphenisc's post abnd that appears to be what they are doing.

Looks like the claims really were describing experiments decades ahead of what they are actually doing.

 

[drkitten]

Re: Re: Re: Re: Blue Brain Project

Well wouldn't it make more sense to pick a much smaller area to test? Perhaps an area that has limited and specific actions eg. Broca's area or something?

It would indeed. But IBM has a tendency to throw lots of resources at problems, partly for the PR value (it sounds much better to say you're building a neocortex than to build Broca's area), and partly because many of the people behind these things are geeks at heart and want to play with the biggest toys they can.

We've dealt with IBM from time to time at my school, and it's much easier to get them to donate three-quarters of a million-dollar computer than it is to get them to donate half of a half-million-dollar computer. The cost out of pocket to us is identical, but they would rather give us the bigger machine.

 

[IllegalArgument]

Having no ideal what been done in the brain model field. Why don't they try and model a smaller mammal first?

 

[Ashles]

Re: Re: Re: Re: Re: Blue Brain Project

It would indeed. But IBM has a tendency to throw lots of resources at problems, partly for the PR value (it sounds much better to say you're building a neocortex than to build Broca's area), and partly because many of the people behind these things are geeks at heart and want to play with the biggest toys they can.

We've dealt with IBM from time to time at my school, and it's much easier to get them to donate three-quarters of a million-dollar computer than it is to get them to donate half of a half-million-dollar computer. The cost out of pocket to us is identical, but they would rather give us the bigger machine.

Yeah that sounds about right.

I guess that also answers IllegalArgument's question.

Which sounds sexier?
"We're creating a 3D simulation of the human neocortex"
"We're modelling a hamster brain"

Which is probably going to yield useful results more quickly?
"Who cares. It's all PR and PR is delightful!"

 

[sphenisc]

Kedo asked

Ok say it works and a self aware sentient “being” brain emerges.
Would it be immoral to make it or others like it, insane so we can study it ?

My guess it would be insane anyway, given that it would have no sensory input, i.e. complete sensory deprivation apart from the occasional 'test stimulus'. Humans tend to start hallucinating quite quickly under similar conditions, though I'm not sure whether this has been tested to the point of inducing madness - anyone?(perhaps Phil's got around to this?)

 

[Phil]

Kedo asked


My guess it would be insane anyway, given that it would have no sensory input, i.e. complete sensory deprivation apart from the occasional 'test stimulus'. Humans tend to start hallucinating quite quickly under similar conditions, though I'm not sure whether this has been tested to the point of inducing madness - anyone?(perhaps Phil's got around to this?)

Nope. I've only induced madness in people not sensory deprived.

 

[IllegalArgument]

Re: Re: Blue Brain Project

Nope. I've only induced madness in people not sensory deprived.

Great, yet another agent of Cthulhu on this board.

 

[drkitten]

Re: Re: Re: Re: Re: Re: Blue Brain Project

Which is probably going to yield useful results more quickly?
"Who cares. It's all PR and PR is delightful!"

Well, to the marketing department (which is one of the major decision-makers regarding what sort of projects any major corporation will support, fund, or donate equipment to), PR is a useful result. In fact, it's arguably the only useful result.

After all, if there were a possibility of direct financial benefit from this research, then it shouldn't be done publically in the newspapers; it should be done privately so that IBM can get the appropriate patents and trade secrets and maximize the return. If there's no possiblity of financial benefit from this research, then it's a waste of corporate resources (and a drain on shareholder value) to support the research in the first place.

So to be able to fund the research, there needs to be a demonstrated indirect financial benefit. This means, basically, PR.

Now, if you will excuse me, I have to go wash my hands, my keyboard, and my mind. Preferably with benzene.