Quest to Model the Human Brain Nets a Billion Euros

65 21 Loading

Is a billion euros enough to understand the human brain? The Human Brain Project thinks it’s a good start, and evidently the European Commission agrees. On January 28, the Human Brain Project was one of two projects to be awarded a billion in backing from the European Commission’s Future and Emerging Technologies (FET) Initiative.

Henry Markram, the project’s founder and co-director, hopes that over the next decade the project’s consortium of 80+ institutions will use up to an annual $100 million in funding to build a complete digital model of the human brain.

The better we know the brain, the better we can diagnose and treat neurological disease, and maybe—in the greatest feat of natural reverse engineering to date—the better we can build computers and software as flexible, powerful, and efficient as the brain itself. At least, that’s the goal.

Markram says, “It’s an infrastructure to be able to build and simulate the human brain, objectively classify brain diseases, and build radically new computing devices.” See the following HBP video for more:

The Human Brain Project is a collaboration of over 80 institutions slated to run from 2013 to 2023 and receive up to $100 million in funding per year. But the original undertaking dates back to mid-2005 when IBM and Swiss University École Polytechnique Fédérale de Lausanne (EPFL) launched the Blue Brain project under the direction of Henry Markram founder of the EPFL’s Brain Mind Institute.

The Blue Brain project went on to use supercomputing resources provided by IBM (a BlueGene/L followed by a BlueGene/P supercomputer) to create a working model of a single cortical column in a rat brain—about 10,000 neurons.

The model is still being refined using new data, and researchers run it every two weeks. To guage the model’s accuracy, the team attempts to recreate experiments already conducted on real neurons and claim it is approaching high levels of realism.

However, the Blue Brain project’s loftier goal of simulating the entire human brain—a feat Markram has claimed can happen inside a decade—required a larger organization and more resources. The Blue Brain project was the proof of concept and will be the command center for the larger Human Brain Project, conceived in 2010.

Now they’ve got a billion euros in the pipeline—what can we expect from the Human Brain Project? For one thing, like other ambitious and data-heavy operations, the project will require exascale computing. That’s 100 times more powerful than any supercomputer currently in existence. Another well-funded program gunning for exascale bodes well for the world of supercomputing, if nothing else. And the project could advance our understanding of neurological disorders.

Henry Markram will direct the Human Brain Project.

But what about AI? Markram is careful to note that the goal is not to recreate human intelligence, but to map the connections that give rise to human intelligence.

As Markram told National Geographic, “People think I want to build this magical model that will eventually speak or do something interesting. I know I’m partially to blame for it—in a TED lecture, you have to speak in a very general way. But what it will do is secondary. We’re not trying to make a machine behave like a human. We’re trying to organise the data.”

Though it is undoubtedly now the most well-funded undertaking to understand the brain, the Human Brain Project is far from alone.

IBM and DARPA’s SyNAPSE recently completed a 100 trillion synapse simulation based on the connections in a macaque brain. Spaun is a working (albeit very simple) cognitive computer. Meanwhile, Ray Kurzweil’s latest book, How to Create a Mind, outlines his ideas on how to reverse engineer the brain—his new post at Google could result in some interesting brain-related projects. And not to be outdone by the EU, the Obama administration is expected to announce the details of a ten year, multi-billion dollar project to map the human brain in the coming weeks.

Governments and academic institutions seem comitted to throwing everything but the kitchen sink at the problem of the brain. It’s reminiscent of Apollo or the Human Genome Project. And arguably, giant lump sum public investments (and the inevitable international competition that goes with them) are needed to jumpstart such massive scientific endeavors.

But politics and public funding are fickle beasts. Equally important are the new neural networks—80+ institutional collaborators in the Human Brain Project alone—forming in the global brain. Whether or not we reach the lofty goal of fully modeling the brain in a decade, we’ll certainly have learned and begun to apply much from the process.

Discussion — 21 Responses

  • Andrew Atkin February 26, 2013 on 3:18 pm

    A great weakness of neuroscience today is the inability to see relationship between raw information, and its neurological correlate. Because of this we sometimes assume that a mental disorder is primarily a function of faulty neurology (what we can see) rather than informational structures driving the “faulty” neurology (which is what we can’t yet see). This in part may be because drug companies funding research are only interested in understanding mental disorders on the level of neurobiology, so they can sell their chemical solutions (drugs – not psychotherapy).

    Regardless, these simulators could help drive our understanding of this informtation/neurology relationship. Though in saying that, there is the consciousnesses itself, and how that thing relates to the brain may be another story again.

    Trauma-imprint theory can revolutionise our understanding of (most) mental conditions, as it relates to information distorting neurological function. If the reader is interested:

    http://andrewatkin.blogspot.co.nz/2009/06/understanding-mental-sickness.html

    • DigitalGalaxy Andrew Atkin March 1, 2013 on 2:12 am

      I just read your blog, quite interesting stuff!

  • anthrobotic February 26, 2013 on 5:15 pm

    Ohhhh, this brings up human/machine morality! Because if you build a brain, do you have to be nice to it? – http://goo.gl/ySZcK

    • DigitalGalaxy anthrobotic February 26, 2013 on 6:29 pm

      Good question! Depends, does the brain feel? Or is there “nobody home”?

      More importantly, how do you tell if a brain IS feeling anything?

      • anthrobotic DigitalGalaxy February 26, 2013 on 7:44 pm

        Exactly!

        And really, how do you tell if YOU or I are feeling anything? We can easily argue that we are, and others accept that. But we’re supposed to say that – we know that those are the “rules.” If a machine can do the same… then… it gets sticky.

        Is human feeling, awareness, consciousness – is it anything more than programmed stimulus-response? What are macro/evolution and micro/socialization if not extremely slow coding processes?

        Pouring on even more shameful self promotion, there’s also this:

        “Unless we invoke religion or genetic manipulation by aliens and stuff, we have to assume that intelligence, on whatever level, is an emergent characteristic of life itself. It’s therefore only rational to assume that, given enough processing power and the right sets of instructions, an AI or NBI might one day decide to observe itself, then tell us about it, then tell us it doesn’t want to be turned off.”

        Excerpted from “Can a computer be as intelligent as a human? Or, Asking the Wrong Dumb Question. Get it?” – http://goo.gl/NboN1

        • DigitalGalaxy anthrobotic March 1, 2013 on 2:05 am

          I read the blog, not bad! You just might start a trend in self-quotations! :)

          I think the question is much less about “intelligence”, as it is about “cognizance”. The two are very different things.

          One involves problem solving capabilites.
          One involves qualia.

          They don’t both neccesarily go together.

          We might well indeed make a computer that is as intelligent or moreso than ourselves, meaning it has equal or greaer capacity to autonomously solve problems.

          But qualia are different. We know that our consciousness is more than stimulus-response because we have qualia.
          The question then becomes, “is there anybody home?” Can the computer feel? Can the computer expereince thought? Can the computer make true “free will” decisions independantly of its algorithms? A person or animal can be very unintelligent, and still expereince emotion, thought, self-awareness, and free will. Can the computer, regardless of its level of intelligence?

          At this point we can probably say a firm “no”, because we can track every electron inside that computer if we felt the need to. If the computer is feeling anything, there must be some electrons that account for that somewhere. If the computer is making an autonomous “free will” descision independant of its programming, there must be some electrons somewhere that account for such an action. Right? Or is there more to a mind than electron flow?

          It is at this point that we begin to leave the realm of things that have a nice, tidy material explanation. Qualia and free will do not fit into an entirely material account of the mind. There isn’t much room for cognizance when all there is to the mind is electrons flowing, whether you are talking about a biological brain or an artificial one. So, where do they come from?

          It isn’t about how intelligent we can make a computer. We could make a computer that could build a Dyson sphere and it still might not feel a thing or be aware that it exists at all. It’s about what gives humans the ability of sensing qualia, what gives us and other animals cognizance, and what gives us true free will. In my view (just my opinion!), that is something spiritual. And the question then becomes, can a computer be a spiritual entity?

          If the computer says, “I feel pain”, and there are no electrons we can track in its chips that account for that, is the computer lying? Or, does it have the same qualia we do, qualia that are not based on electron flow in chips? Can it make a true free will desicion? At that point, we have crossed out of the realm of physical science.

          What do you think? What translates a neural pain signal into actual pain you literally feel? Is there a possible physical mechanism of action for that? And if there is not, then is a spiritual explanation appropriate?

          These conversations are important to have, because soon, these questions will leave the realm of philosophical discussion and become quite real.

          • anthrobotic DigitalGalaxy March 5, 2013 on 3:23 am

            Self-quoting is where it’s at!

            Agreed, the spiritual issue is where it gets muddy. For now, we can trace the electrons through a machine and physically and electrically qualify a machine’s behavior. But realistically, we’ll probably see that capability in human brain scanning within 10-30 years.

            Or maybe not – really, whether we find ourselves able to do that, or unable – either way it’s profound.

            Anyway, this is a well-stated response. I’d love to debate it further, but that would result in more shameless self-quotation and sub-references to previous work and so on.

            Any work published anywhere other than comments? I’d be interested.

            -Reno at Anthrobotic.com

          • DigitalGalaxy DigitalGalaxy March 5, 2013 on 8:02 pm

            Thanks! I have been considering starting a blog/writing a book, but I feel like I just don’t have enough substance to go on yet. Our research is still in such embryonic stages! I feel like posing questions is the stage we need to be in now; what happens when artificial brains get to X stage in their development? How do we account for chemical reactions? Is a virtual chemical reaction for dopamine the same as a physical one? Can chemical reactions be simulated on a probabilistic basis, or does the simulation need a million little dopamine “molecules” spinning around? What is the role of naturally-occurring DMT (a so-called “LSD drug”) in our brains? Ect, ect.

            I’ll start following your blog, I read some articles and they were pretty dead-on as to what we as a species need to be considering (as well as containing amusing prose)! I’d rather have at least some sort of cognitive framework at the ready for when these proceedings produce real live simulations, as opposed to just test runs.

  • DigitalGalaxy February 26, 2013 on 6:30 pm

    Interesting, but until nanobot level probes can be placed inside the brain at a neuron level, it will only be guesswork as to which brainwaves are actually going where.

  • Greendogo February 26, 2013 on 8:14 pm

    A waste of public money. Let companies like IBM take care of commercializing this.

    • DigitalGalaxy Greendogo March 1, 2013 on 2:19 am

      There are pros and cons to government research and corporate research. Neither one is perfect. Corporate research is always biased by the bottom line, government reasearch is often hampered by red tape. Six of one half dozen of the other.

  • Robert Schreib February 28, 2013 on 3:30 pm

    If SETI and other groups use an Internet connected array of personal computers around the world volunteered by the public, to use the downtime on their PCs for aims that will benefit all humanity, couldn’t they do the same thing to amass a enormous PC computing and memory capacity via the Internet, to fully simulate a human brain cyber model?

    • DigitalGalaxy Robert Schreib March 1, 2013 on 11:10 am

      Unfortunately no. The main issue would be lag between nodes. In the brain, or in a supercomputer, there is no lag. Over the Internet, there is abundant lag. SETI or folding@home can use distributed supercomputer networks because all the equations can be solved intependantly of each other, in more or less any order. A brain would have to operate in real time, and lag would pose a major problem there.

      Good thinking though!

  • Tober March 5, 2013 on 12:12 pm

    I think that project is going to wrong way. The first step is to know what we are looking for. Some assumptions, ideas must be the starting point. It must be an extensive theory-very understandable and like all big staff – simple. Of course, if viewed from the right angle.

    • DigitalGalaxy Tober March 5, 2013 on 7:07 pm

      Well, true, but it would be helpful to have a brain model ready when we finally do find what is is we are looking for. We need to be able to simulate every neuron in order to simulate brain waves…it’s like having to have a model airplane before you can make a real one.

      • Tober DigitalGalaxy March 6, 2013 on 3:19 am

        When I see some EEG- I lose hope in any precision approach “based” on it.
        1 neuron – 1000 stimulating and suppressive connections. billions of neurons.
        So… “model” will be the solution. Every theory is better than nothing.
        Parallel solving… Programming software which simulate human consciousness and revelation….brain anatomy. Human versus evolution. At some point similarity must must be shown. If there is finances for an independent project-initially-I will be glad to make the model.

        • DigitalGalaxy Tober March 11, 2013 on 1:16 am

          Oh, I think we can simulate billions of neurons; we just need a bigger computer! If each neuron has 1,000 “actions” it can take, and there are 86 billion neurons in the brain, then we need a computer capable of 86 trillion “actions” per second. That’s not that big a deal when old processors from ten years ago can do 10 billion actions per second. Supercomputers can do 4 petaflops, which is a 4,000 trillion actions per second. So, I think we will be able to do complete calculations on a supercomputer!

          • Tober DigitalGalaxy March 19, 2013 on 11:10 am

            For years technically part is not the problem. We have powerful computers and nearly a million programmers. When someone says that with 80 teams is going to solve the problem for 10 years, it sounds like … boring. The key is to admit that there is a problem – in the sense that no one knows how things will develop in the next 10 years, or where to go (in fact – it is not the task of programming) And then just look for the pattern. When you see it you’ll know it’s correct.

          • DigitalGalaxy DigitalGalaxy March 19, 2013 on 10:53 pm

            Interesting way of looking at it! You might have a point!

  • andy_vela May 1, 2013 on 2:24 pm

    Very interesting, especially the computing power needed to simulate the brain, this project has a great future in the medicine too.

  • Josh Koma February 4, 2014 on 5:11 pm

    If these robots achieve intelligence to Human Level, then they will realize their superior abilities over human beings, and Eu-genesis will have to occur. Given the resource, Terminator or the Matrix or iRobot will definately occur. Such projects should never be funded because they leave humanities future hanging in an irregular balace!