Creating a virtual model of the human brain is one thing. I do it all the time, doodling little cerebrums while I talk on the phone. But getting your model to behave just like its flesh-and-blood counterpart? That’s a Frankenstein moment right there. Researchers with the Swiss-based Blue Brain Project have just created a virtual pack of neurons that acts just like the real thing, and hope to get an e-brain up and running. I hope somebody yelled “It’s alive!”
Nerve cell model. Photo courtesy of the Blue Brain Project
Launched by the EPFL in 2005, the Blue Brain Project is an attempt to reverse-engineer the brain. As most folks know, the human brain is made up of lots (and lots and lots) of neurons – around 100 billion or so. These neurons connect and communicate with each other through a dizzying network of something like 100 trillion synapses. If these numbers are making your own brain hurt, you now have a sense for how hard it is to make an artificial brain.
But all hope is not lost. Genes don’t really code the body like blueprints do for a building, mapping out every single detail; instead, they give a more general instruction and hit the “repeat” button a few million times (e.g. when they give fractal instructions). This means that amid the great complexity of the whole brain, there are structural units that repeat themselves. One such structure is called a neocortical column (NCC): a group of about 10,000 neurons in the cerebral cortex that are organized in a relatively consistent way across the mammalian brain. Millions of these columns compose the whole of your grey matter.
Rather than trying to create a model of the whole brain at once, the Blue Brain Project is attempting to accurately model a single NCC in a rat’s brain. If they can create an artificial column that responds the same way that biological ones do to electrical impulses, they’ll be on the right track to building a good model. And this is exactly what they’ve done: built a virtual copy of an NCC down to the molecule, and had it replicate in simulation the real-world activity of rat brain tissue. And because the NCC’s structure is so consistent across mammals, they could eventually use the same simulation (scaled up, of course) to model the human brain.
Check out this video that flies you through their virtual brain:
The Blue Brain runs on IBM’s Blue Gene/L supercomputer, one of the top five supercomputers on the planet. Containing over 8,000 processors, the Blue Gene provides the tremendous amount of parallel computation (about 22 trillion operations per second) needed to simulate the complex functioning of a virtual NCC. Still, even with their current hardware, the simulation only runs at about half the speed of a biological NCC. Speeding the model up to real-time will take even stronger computers. Also, the current computing power is stretched thin to simulate a single cortical column; imagine the power needed to replicate a whole brain (millions of NCCs). The ultimate goals of the project (e.g. a virtual human brain) will require advances in computer power that just aren’t around yet.
Phase I of the research project is now complete (creating a virtual model that replicates real-life NCC behavior). The next step is to further fine-tune their model, integrating more molecular and genetic information into their simulations. To add these details, the project is upgrading their hardware to a stronger Blue Gene supercomputer that can handle the additional computations. Further stages of the project will aim to build a whole brain (not just a column): first the rat, then onward and upward to other species.
I’d encourage you to take a quick moment at this point to marvel at the awesome power of the three pounds of meat housed in your skull. The fastest, most efficient computers on the planet are barely sufficient to model even a tiny fraction of your neural tissue. Our own engineering algorithms pale in comparison to what mother nature has cooked up, though (to our credit) she had a little more time to work on her science project. Needless to say, this speaks to how much we have to learn from the efficiency of biological adaptations.
A simulated network in action. Photo courtesy of the Blue Brain Project
Like most labs in this flowering economy, the Blue Brain Project needs further funding to take the next step. “It’s not a question of years, it’s one of dollars,” Henry Markram, leader of the Project, told BBC. “The psychology is there today and the technology is there today. It’s a matter of if society wants this. If they want it in 10 years, they’ll have it in 10 years. If they want it in 1000 years, we can wait.” Funding thus far has come from the Swiss government and private grants; IBM sold their supercomputers at a discount to see what kinds of new applications EPFL would come up with.
Despite the similarities to neural nets chasing after AI, the Blue Brain Project isn’t trying to build HAL9000. Their goal is to explore how the brain functions and to provide a useful tool to the field of neuroscience, both for research and clinical purposes. But while their primary goal may be physiological simulation for doctors, they admit they might unlock a few secrets to consciousness along the way. Well as long as it’s convenient…
If the Project can secure more funding – and this recent breakthrough might help – they’ll be able to afford more of the supercomputing power needed to add details and scale up their model. A virtual brain that mimics the behaviors of a real brain would be an invaluable tool for researchers and doctors alike. And who knows? Maybe “It’s alive!” will take on a whole new meaning.
Tags: blue brain, ncc, neocortical column, neurons, virtual neurons
Share
Hopefully their work of physically modelling the brain will lead to successful nanotechnological computing breakthroughs – maybe after they model one human neuron, they can sell that map to IBM to create a self-assembling nanostructure that can be pumped out by the billions, each designed to link up with one another like the ones in our brain…
Then it’ll be easy to model a brain, you’ll just print one.
I mean, the technology is already basically there – as the man said, it’ just a question of dollars as to how soon we deliver.
haha IBM
Just put all sub Atoms in a Holographic DB
For more information about fractals see:
http://www.fractal.org/Julius-Ruis-set.pdf
Not really news tho, this phase was already completed in november 2007. Good article nonetheless!
[...] Virtual Neurons Acting Like the Real Thing – The Blue Brain Project | Singularity Hub. [...]
[...] The Blue Brain Project By archdave Virtual Neurons Acting Like the Real Thing – The Blue Brain Project | Singularity Hub. [...]
[...] Virtual Neurons Acting Like the Real Thing – The Blue Brain Project | Singularity Hub – ai bluebrain ibm brain [...]
[...] minds will someday be housed on different platforms. If you traded in your own neural tissue for a Blue Brain upgrade, would you still desire sex? Would we flood our computer-brains with artificial hormones to [...]
[...] But wait you say, receiving a radio signal in your skull and shooting a laser out of your brain is cool, but where’s the hard science? Well, while the Braingate team is mucking around in your motor cortex, trying to pass signals on to computers, robotic wheelchairs, and prosthetic limbs, they’ll also be studying neuron behavior. As part of their research, Braingate2 hopes to explore how different diseases, emotions, and awareness levels (i.e. if you’re asleep) affect the neuronal firing patterns. These guys are going into the brain to tinker with motor neurons, but they could learn a lot about the brain functions in general. I hope they pass that information on to the groups who are hoping to create computer simulations of the brain, like FACETS or Blue Brain. [...]
[...] and how we can build a computer that will simulate those functions for us to explore. If our earlier article on the Blue Brain Project left you eager to learn more, check out the new presentation that Project Director Henry Markram [...]
[...] and how we can build a computer that will simulate those functions for us to explore. If our earlier article on the Blue Brain Project left you eager to learn more, check out the new presentation that Project Director Henry Markram [...]
[...] – Excerpt – Markram does a good job of breaking down Blue Brain’s approach, as does our earlier article, so I won’t repeat it all. The focus of the simulation is the neocortex, specifically [...]
Deep Computed BCI: A Short Story
Imagine your motor cortex fully activated while you have full muscle tone but both what your cortex says you are experiencing and what you are actually experiencing are not what you body is actually doing. You were trained to do this on a brain computer interface. Highly Skilled lucid dreamers in intense sessions and brain tomography on the level of seismic tomography make this all possible. Accessing the brain thru non-invasive means is vital in Berlin where Brain Computer Interfacers and the Locked-in are moving things with only their minds; however, one might say that all this research is treading water awaiting advances in Neuro-surgery. I’m pitching the thoroughly developed non-invasive technique as a necessary prelude to the invasive interface. I’m just looking for sympathetic places to post the story I’m telling in the form of a fictitious photo journal.
[...] is neither the first Blue Brain Project story, nor even the first video of Markram that we’ve discussed here at Singularity Hub. What keeps [...]
Can anyone tell me that what are the mathematical forms or equations that are primarily fed to the computers? I mean if i excite the brain by showing a flower then what are the forms of equations or mathematical equations that the computers are fed after which the supercomputer is carrying the program?? please help me with this.
LOGICAL EXTRACTION OF NEOCORTEX STRUCTURE
By Dr. Ronald J. Swallow
rswallow@ptd.net
610 704 0914
I do not understand why the neocortex is a mystery to everyone. Its neuron net circuit is repeated throughout the cortex. It consists of excitatory and inhibitory neurons whose functions, each, have been known for decades. The neuron net circuit is repeated over layers whose axonal outputs feed on as inputs to other layers. The neurons of each layer, each receive axonal inputs from one or more sending layers and all that they can do is correlate the axonal input stimulus pattern with their axonal connection patterns from those inputs and produce an output frequency related to the resultant PSPs. Axonal growth toward a neuron is definitely the mechanism for permanent memory formation and it is just what is needed to implement conditioned reflex learning. This axonal growth must be under the control of the glial cells and must be a function of the signals surrounding the neurons.
The cortex is known to be able to do pattern recognition and the correlation between an axonal input stimulus and an axonal input connection pattern is just what is needed to do pattern recognition. However, pattern recognition needs normalized correlations and a means to compare these correlations so that the largest correlation is recognized by the neurons. Without normalization, the PSPs relative values would not be bounded properly and could not be used to determine the best pattern match. In order to get PSPs to be compared so that the maximum PSP neuron would fire, the inhibitory neuron is needed. By having a group of excitatory neurons feed an inhibitory neuron that feeds back inhibitory axonal signals to those excitatory neurons, one is able to have the PSPs of the excitatory neurons compared, with the neuron with the largest PSP firing before the others do as the inhibitory signal decays after each excitatory stimulus, thus inhibiting the other excitatory neurons with the smaller PSPs. This inhibitory neuron is needed in order to achieve PSP comparisons, no question about it. For a meaningful comparison, the PSPs must be normalized. As unlikely as it may seem possible, it comes out that the inhibitory connections growing by the same rules as excitatory connections grow to a value which accomplishes the normalization. That is, as the excitatory axon pattern grows via conditioned reflex rules, the inhibitory axon to each excitatory neuron grows to a value equal to the square root of the sum of the squares of the excitatory connections. This can be shown by a mathematical analysis of a group of mutually inhibiting neurons under conditioned reflex learning. This normalization does not require the neurons to behave differently from that known for decades, but rather requires that they interact with an inhibitory neuron as described.
Thus, by simply having the inhibitory neurons receive from neighboring excitatory neurons with large connection strengths where if the excitatory neuron fires, the inhibitory neuron fires and by allowing the inhibitory axonal signals be included with the excitatory axonal input signals as the inputs to those excitatory neurons, the neocortex is able to do normalized conditioned reflex pattern recognition as its basic function.
If one thinks about it, layers of mutually inhibiting groups of neurons are all that are needed to explain the neocortex functions. The layers of neurons are able to exhibit conditioned reflex behavior between sub-patterns, generating new learned behaviors as observed by the human. With layer to layer feedback, multi-stable behavior of layers of neurons results, forming short term memory patterns that become part of the stimulus to other neurons. With normalized correlations, there is always an axonal input stimulus pattern that will excite every excitatory neuron.
The only way to prove this cortex model is to build a simulator modeling large nets of neurons and to observe resultant human behaviors. Most certainly we will never be able to measure the neuron nets of the cortex due to their small sizes. This means, that research projects must be formed that do these simulations and do not waste R&D efforts to try to measure properties of the cortex as the main means to understand the cortex. Certainly the area to area connection scheme is needed, but it likely can be varied, still with intelligence being exhibited. Trials will be needed to determine the initial connection strengths when initiating the simulator. These connections will need to be simple such as non-zero between corresponding neurons of the mutually inhibiting groups.
Axon growth toward pulsing neurons is the likely mechanism for memory alteration. Alternatively, having neuron axons back away from neurons has no physical basis and it is well known that the number of axons increases with age in the human. Certainly axon connection strengths never become proportional to axon pulsing frequencies, otherwise the nets of neurons will never exhibit permanent past memories, but rather be a function of recent events only. Glial cells are likely participants to axonal growth control. It is likely that they will inhibit axonal growth physically, unless a chemical falls below a concentration. In particular, this would be when the excitatory stimulus (chemically emitted to a neuron by axons to that neuron) to a cell, falls below a critical level, where the correlation between stimulus and connection pattern falls below a limit. The result of such a rule is that learning would only occur if stimulus patterns are new and don’t sufficiently match the connection patterns to neurons. The psychological effect would be a curiosity behavior, observed in humans. Also, it would result in old age reduction of ability to learn, also observed in humans.
Progress in understanding how the brain works has been basically non-existent over the last 40 years due to limits in measurement. Rather, progress requires simulation to work out the missing details. I predict that simulation will dominate the future efforts of researchers.
Also, I predict that special purpose hardware will dominate the approach. Using conventional computers to simulate nets of neurons in real-time will go out of style very soon due to their high cost and poor performance.
Simulation permits an evolution process to arrive upon a successful understanding of the brain. If a logical conclusion is wrong, simulation will eliminate it. If it is right, simulation will verify it.
LOGICAL EXTRACTION OF NEOCORTEX STRUCTURE
By Dr. Ronald J. Swallow
ronswallow@ptd.net
610 704 0914
I do not understand why the neocortex is a mystery to everyone. Its neuron net circuit is repeated throughout the cortex. It consists of excitatory and inhibitory neurons whose functions, each, have been known for decades. The neuron net circuit is repeated over layers whose axonal outputs feed on as inputs to other layers. The neurons of each layer, each receive axonal inputs from one or more sending layers and all that they can do is correlate the axonal input stimulus pattern with their axonal connection patterns from those inputs and produce an output frequency related to the resultant PSPs. Axonal growth toward a neuron is definitely the mechanism for permanent memory formation and it is just what is needed to implement conditioned reflex learning. This axonal growth must be under the control of the glial cells and must be a function of the signals surrounding the neurons.
The cortex is known to be able to do pattern recognition and the correlation between an axonal input stimulus and an axonal input connection pattern is just what is needed to do pattern recognition. However, pattern recognition needs normalized correlations and a means to compare these correlations so that the largest correlation is recognized by the neurons. Without normalization, the PSPs relative values would not be bounded properly and could not be used to determine the best pattern match. In order to get PSPs to be compared so that the maximum PSP neuron would fire, the inhibitory neuron is needed. By having a group of excitatory neurons feed an inhibitory neuron that feeds back inhibitory axonal signals to those excitatory neurons, one is able to have the PSPs of the excitatory neurons compared, with the neuron with the largest PSP firing before the others do as the inhibitory signal decays after each excitatory stimulus, thus inhibiting the other excitatory neurons with the smaller PSPs. This inhibitory neuron is needed in order to achieve PSP comparisons, no question about it. For a meaningful comparison, the PSPs must be normalized. As unlikely as it may seem possible, it comes out that the inhibitory connections growing by the same rules as excitatory connections grow to a value which accomplishes the normalization. That is, as the excitatory axon pattern grows via conditioned reflex rules, the inhibitory axon to each excitatory neuron grows to a value equal to the square root of the sum of the squares of the excitatory connections. This can be shown by a mathematical analysis of a group of mutually inhibiting neurons under conditioned reflex learning. This normalization does not require the neurons to behave differently from that known for decades, but rather requires that they interact with an inhibitory neuron as described.
Thus, by simply having the inhibitory neurons receive from neighboring excitatory neurons with large connection strengths where if the excitatory neuron fires, the inhibitory neuron fires and by allowing the inhibitory axonal signals be included with the excitatory axonal input signals as the inputs to those excitatory neurons, the neocortex is able to do normalized conditioned reflex pattern recognition as its basic function.
If one thinks about it, layers of mutually inhibiting groups of neurons are all that are needed to explain the neocortex functions. The layers of neurons are able to exhibit conditioned reflex behavior between sub-patterns, generating new learned behaviors as observed by the human. With layer to layer feedback, multi-stable behavior of layers of neurons results, forming short term memory patterns that become part of the stimulus to other neurons. With normalized correlations, there is always an axonal input stimulus pattern that will excite every excitatory neuron.
The only way to prove this cortex model is to build a simulator modeling large nets of neurons and to observe resultant human behaviors. Most certainly we will never be able to measure the neuron nets of the cortex due to their small sizes. This means, that research projects must be formed that do these simulations and do not waste R&D efforts to try to measure properties of the cortex as the main means to understand the cortex. Certainly the area to area connection scheme is needed, but it likely can be varied, still with intelligence being exhibited. Trials will be needed to determine the initial connection strengths when initiating the simulator. These connections will need to be simple such as non-zero between corresponding neurons of the mutually inhibiting groups.
Axon growth toward pulsing neurons is the likely mechanism for memory alteration. Alternatively, having neuron axons back away from neurons has no physical basis and it is well known that the number of axons increases with age in the human. Certainly axon connection strengths never become proportional to axon pulsing frequencies, otherwise the nets of neurons will never exhibit permanent past memories, but rather be a function of recent events only. Glial cells are likely participants to axonal growth control. It is likely that they will inhibit axonal growth physically, unless a chemical falls below a concentration. In particular, this would be when the excitatory stimulus (chemically emitted to a neuron by axons to that neuron) to a cell, falls below a critical level, where the correlation between stimulus and connection pattern falls below a limit. The result of such a rule is that learning would only occur if stimulus patterns are new and don’t sufficiently match the connection patterns to neurons. The psychological effect would be a curiosity behavior, observed in humans. Also, it would result in old age reduction of ability to learn, also observed in humans.
Progress in understanding how the brain works has been basically non-existent over the last 40 years due to limits in measurement. Rather, progress requires simulation to work out the missing details. I predict that simulation will dominate the future efforts of researchers.
Also, I predict that special purpose hardware will dominate the approach. Using conventional computers to simulate nets of neurons in real-time will go out of style very soon due to their high cost and poor performance.
Simulation permits an evolution process to arrive upon a successful understanding of the brain. If a logical conclusion is wrong, simulation will eliminate it. If it is right, simulation will verify it.