BBP/EPFL
A representation of a mammalian neocortical column, the basic building block
of the cortex. The representation shows the complexity of this part of the
brain, which has now been modeled using a supercomputer.
The Blue Brain Project
By Duncan Graham-Rowe
MIT Technology Review, November 28, 2007
Edited by Andy Ross
An ambitious project to create an accurate computer model of the brain has
reached an impressive milestone. Scientists in Switzerland working with IBM
researchers have shown that their computer simulation of the neocortical
column, arguably the most complex part of a mammal's brain, appears to
behave like its biological counterpart.
The project began with the
initial goal of modeling the 10,000 neurons and 30 million synaptic
connections that make up a rat's neocortical column, the main building block
of a mammal's cortex. The neocortical column was chosen as a starting point
because it is widely recognized as being particularly complex.
The
model itself is based on 15 years' worth of experimental data on neuronal
morphology, gene expression, ion channels, synaptic connectivity, and
electrophysiological recordings of the neocortical columns of rats. Software
tools were developed to reconstruct accurate 3D models of neurons and their
interconnections.
The neuronal circuits were tested by simulating
specific input stimuli and seeing how the circuits behaved, compared with
those in biological experiments. Where gaps in knowledge appeared about how
parts of the model were supposed to behave, the scientists went back to the
lab and performed experiments.
The level of detail of the model can
be taken further. It is still at a cellular level, but the scientists want
to look at the molecular level. Doing so would enable simulation-based drug
testing to be carried out by showing how specific molecules affect proteins,
receptors, and enzymes.
"What we're doing is reverse-engineering the
brain"
Henry Markram, Codirector, Brain Mind Institute
Ecole
Polytechnique Fédérale de Lausanne, Switzerland
"This is an
evolutionary process rather than a revolutionary one"
Christof Koch,
Professor of Biology and Engineering
California Institute of Technology,
Pasadena, CA
By Jonah Lehrer
Seed, March 3, 2008
Edited by Andy Ross
Four black boxes, each about the size of a refrigerator, form the processing
core of a machine that can handle 22.8 trillion operations per second. When
the computer is turned on, all you can hear is the air conditioner. This is
Blue Brain.
The behavior of the computer replicates the cellular
events unfolding inside a mind. "This is the first model of the brain that
has been built from the bottom up," says Henry Markram, a neuroscientist at
Ecole Polytechnique Fédérale de Lausanne (EPFL) and the director of the Blue
Brain project.
When Markram launched the project in the summer of
2005, as a joint venture with IBM, scientists criticized the project as an
expensive pipedream. Neuroscience didn't need a supercomputer, it needed
more molecular biologists. But Markram's attitude was different: "I wanted
to model the brain because we didn't understand it. The best way to figure
out how something works is to try to build it from scratch."
The Blue
Brain project is now at a crucial juncture. It took less than two years for
the Blue Brain supercomputer to accurately simulate a neocortical column,
which is a tiny slice of brain containing approximately 10,000 neurons, with
about 30 million synaptic connections between them. "The column has been
built and it runs," Markram says. "Now we just have to scale it up." Blue
Brain scientists are confident that they will be able to start simulating an
entire brain.
Before he began developing Blue Brain, Markram was best
known for his painstaking studies of cellular connectivity. His technical
innovation was "patching" multiple neurons at the same time, so that he
could eavesdrop on their interactions. This experimental breakthrough
promised to shed light on how billions of discrete cells weave themselves
into functional networks.
When Markram looked at the electrical
language of neurons, he realized that he was staring at a code he couldn't
break. "I would observe the cells and I would think, 'We are never going to
understand the brain.' Here is the simplest possible circuit — just two
neurons connected to each other — and I still couldn't make sense of it. It
was still too complicated."
Neuroscience is a reductionist science.
It describes the brain in terms of its physical details, dissecting the mind
into the smallest possible parts. Over the last 50 years, scientists have
managed to uncover a seemingly endless list of molecules, enzymes, pathways,
and genes. According to Markram, however, this scientific approach has
exhausted itself. "I think that reductionism peaked five years ago," he
says.
Markram hopes the Blue Brain project represents a whole new
kind of neuroscience: "You need to look at the history of physics. From
Copernicus to Einstein, the big breakthroughs always came from conceptual
models. They are what integrated all the facts so that they made sense. You
can have all the data in the world, but without a model the data will never
be enough."
Neuroscience is a thoroughly empirical discipline, rooted
in the manual labor of molecular biology. The sole exception is
computational neuroscience, a relatively new field. But Markram is
dismissive of most computational neuroscience: "It's not interested enough
in the biology. What they typically do is begin with a brain function they
want to model and then try to see if they can get a computer to replicate
that function. The problem is that if you ask a hundred computational
neuroscientists to build a functional model, you'll get a hundred different
answers."
Before the Blue Brain team could start constructing their
model, the collected works of modern neuroscience had to be painstakingly
programmed into the supercomputer. The problem is that neuroscience is still
woefully incomplete. Even the simple neuron remains a mostly mysterious
entity.
Markram focused on a neocortical column in a two-week-old
rat. A neocortical column is the basic computational unit of the cortex, a
discrete circuit of flesh that's 2 mm long and 0.5 mm in diameter. The
cortex consists of thousands of these columns, each with a very precise
purpose and a basic structure that remains the same, from mice to men. The
virtue of simulating a circuit in a rodent brain is that the output of the
model can be continually tested against the neural reality of the rat, a
gruesome process that involves opening up the skull and plunging a needle
into the brain.
Felix Schürmann, the project manager of Blue Brain,
oversees this daunting process. Schürmann shares a workspace with a diverse
group. The 20 or so scientists working full-time on Blue Brain's software
originate from 14 different countries.
Neurons are electrical
processors that represent information as bursts of voltage. They control the
flow of electricity by opening and closing different ion channels. When the
team began constructing their model, the first thing they did was program
the existing ion channel data into the supercomputer. They wanted their
virtual channels to act just like the real thing. However, they soon ran
into serious problems. Many of the experiments generated contradictory
results. After several frustrating failures, the team realized that they
needed to generate the data themselves.
Schürmann led me down the
hall to Blue Brain's wet lab. The room looks like a generic neuroscience
lab. But tucked in the corner of the room is a small robot, about the size
of a microwave, filled with a variety of test tubes. A delicate metal claw
is constantly moving back and forth across the tray, taking tiny sips from
the different liquids. Schürmann says the robot is recording from a cell.
The Blue Brain team genetically engineers Chinese hamster ovary cells to
express a single type of ion channel, then they subject the cells to a
variety of physiological conditions. The robot can generate hundreds of data
points a day, or about 10 times more than an efficient lab technician.
Markram refers to the robot as "science on an industrial scale."
The
patch clamp robot helped the Blue Brain team redo 30 years of research in
six months. By analyzing the genetic expression of real rat neurons, the
scientists were able to construct a precise map of ion channels. This new
knowledge was then plugged into Blue Brain, allowing the supercomputer to
accurately simulate any neuron anywhere in the neocortical column. "The
simulation is getting to the point," Schürmann says, "where it gives us
better results than an actual experiment. We get the same data, but with
less noise and human error."
After assembling a three-dimensional
model of 10,000 virtual neurons, the scientists began feeding the simulation
electrical impulses. Because the model focused on a neocortical column in
the somatosensory cortex of a two-week-old rat, the scientists could feed
the supercomputer the same sort of electrical stimulation that a newborn rat
would actually experience.
After only a few electrical jolts, the
artificial neural circuit began to act just like a real neural circuit.
Clusters of connected neurons began to fire in close synchrony: the cells
were wiring themselves together. Different cell types obeyed their genetic
instructions. The scientists could see the cellular looms flash and then
fade as the cells wove themselves into meaningful patterns. "This all
happened on its own," Markram says. "It was entirely spontaneous."
By
comparing the behavior of the virtual circuit with experimental studies of
the rat brain, the scientists could test out the verisimilitude of their
simulation. "People complain that Blue Brain must have so many free
parameters," Schürmann says. "They assume that we can just input whatever we
want until the output looks good. But what they don't understand is that we
are very constrained by these experiments."
"We have already shown
that the model can scale up," Markram says. "What is holding us back now are
the computers." Markram estimates that to simulate the human brain, you'd
need to process about 500 petabytes of data. That's about 200 times more
information than is stored on all of Google's servers. But if computing
speeds continue to develop at their current exponential pace, Markram
believes that he'll be able to model a complete human brain on a single
machine in ten years or less.
Neuroscience describes the brain from
the outside. The paradox is that we don't experience our matter. "We've got
all these tools for studying the cortex," Markram says. "But none of these
methods allows us to see what makes the cortex so interesting, which is that
it generates worlds. No matter how much I know about your brain, I still
won't be able to see what you see."
According to Markram, the power
of Blue Brain is that it can transform a metaphysical paradox into a
technological problem. "There's no reason why you can't get inside Blue
Brain," Markram says. "Once we can model a brain, we should be able to model
what every brain makes. We should be able to experience the experiences of
another mind."
Markram's ambitions are grounded in concrete steps.
Once the team is able to model a complete rat brain — that should happen in
the next two years — Markram will download the simulation into a robotic
rat, so that the brain has a body. He's already talking to a Japanese
company about constructing the mechanical animal.
Installing Blue
Brain in a robot will allow it to develop like a real rat. The simulated
cells will be shaped by their own sensations, constantly revising their
connections based upon the rat's experiences. "What you ultimately want,"
Markram says, "is a robot that's a little bit unpredictable, that doesn't
just do what we tell it to do." His goal is to build a virtual animal with a
mind of its own.
Markram wants to simulate what that brain
experiences. If he can really find a way to see the brain from the
inside,then he will have given neuroscience an unprecedented window into the
invisible. He will have taken the self and turned it into something we can
see.
Schürmann leads me into a large room. A lone Silicon Graphics
supercomputer hums loudly in the center of the room. This computer
translates the simulation into visual form. The vast data sets generated by
the IBM supercomputer are rendered as short films. Schürmann starts the
digital projector. The screen begins to fill with thousands of colorful
cells. After a few seconds, the colors start to pulse across the network, as
the virtual ions pass from neuron to neuron. I'm watching the supercomputer
think.
Markram says the first step will be to decipher the connection
between the sensations entering the robotic rat and the flickering voltages
of its brain cells. The supercomputer should be able to take its map of the
cortex and generate a movie of experience. "There is nothing inherently
mysterious about the mind or anything it makes. Consciousness is just a
massive amount of information being exchanged by trillions of brain cells.
If you can precisely model that information, then I don't know why you
wouldn't be able to generate a conscious mind."
By Gautam Naik
Wall Street Journal, July 14, 2009
Edited by Andy Ross
For the last four years, Henry Markram has been building a biologically
accurate artificial brain. Dubbed Blue Brain, the project is based in
Switzerland. "We're building the brain from the bottom up, but in silicon,"
says Markram, the leader of Blue Brain, which is powered by an IBM
supercomputer.
Markram has decided to work out the blueprint of its
wiring and then use that map to rebuild the brain in an artificial form. He
focused on a rat's neocortical column, or NCC, an elementary building block
of the brain's neocortex, which is responsible for higher functions and
thought.
A rat's NCC, comprised of about 10,000 neurons and their 10
million connections, functions much like a computer microprocessor. All
mammals have NCCs, and the ones in humans aren't all that different from the
ones in rats. However, humans have far more NCCs.
Markram began by
collecting detailed information about the rat's NCC, down to the level of
genes, proteins, molecules and the electrical signals that connect one
neuron to another. These complex relationships were then turned into
millions of equations, written in software. He then recorded data directly
from rat brains to test the accuracy of the software.
At the Lausanne
lab one recent afternoon, a pink sliver of rat brain sat in a beaker
containing a colorless liquid. The neurons in the brain slice were still
alive and actively communicating with each other. Nearby, a modified
microscope recorded some of this inner activity in another brain slice.
"We're intercepting the electro-chemical messages," said Markram, then
testing the software against them for accuracy.
It takes the power of
one microprocessor to mimic the behavior of a single neuron. To model the
rat's NCC, the IBM computer can perform 22.8 trillion operations a second.
The simulation is rendered as a 3D object. When zapped by a simulated
electrical current, the neurons start to signal to each other and their
wiring lights up. Tests indicate the same areas light up in the model as do
in a real rat's brain, suggesting that Blue Brain is accurate.
The
NCC's wiring fills up with a cascade of myriad signals. The NCC looks like
an incredibly dense tangle of undergrowth. The neurons have synchronized
their behavior. "The cells start to take on a life of their own," says
Markram.
AR Next step: The Human Brain Project, 2013
Billion-Euro Brain
By Jonathon Keats
Wired, May 2013
Edited by Andy Ross
At a TED conference in Oxford in 2009, Henry Markram announced a plan to
deliver a sentient hologram within a decade. He hoped to wipe out all mental
disorders and create a self-aware AI. And he said he would do all this by
building a complete model of a human brain and running it on a
supercomputer.
His Human Brain Project aims to simulate the functions
of all 86 billion neurons in the human brain, and the 100 trillion
connections that link them. Once he has a plug-and-play brain, anything is
possible. You could take it apart to figure out the causes of brain
diseases. You could rig it to robotics and develop a whole new range of
intelligent technologies. You could strap on VR glasses and experience a
brain other than your own.
Markram and his team have simulated the
behavior of a million-neuron portion of the rat neocortex. They hope their
methods can scale. They don't know if they can build out the rest of the rat
brain, let alone the vastly more complex human brain. In January 2013, the
European Commission awarded him a billion euros to try.
There are
three other big brain initiatives:
1 The
Allen Brain Atlas maps the correlation between specific genes and specific
structures and regions in both human and mouse brains.
2 The Human Connectome Project uses noninvasive
imaging techniques that show where wires are bundled and how those bundles
are connected in human brains.
3 The
Brain Activity Map aims to show circuits firing in real time. At present
this is feasible only for fruit flies. Scaled up to human dimensions, such a
map would chart a web of activity but leave out much of what is known of
brain function at a molecular and functional level.
Markram: "The
Brain Activity Map and other projects are focused on generating more data.
The Human Brain Project is about data integration."
As a student,
Markram worked in a lab in Cape Town where he did hundreds of experiments
recording the effect of a neurotransmitter on neurons in the brain stem. He
got his PhD at the Weizmann Institute of Science in Israel and went on to
consecutive postdocs in Bethesda, Maryland, and in Heidelberg, Germany.
In 1995, back at Weizmann, Markram discovered that the pattern of
synaptic connections in a neural network is determined not only by whether
neurons fire together but also by when they fire relative to one another. He
published his results, became a tenured professor, and saw that he had to
aim higher.
Neuroscientist Patrick Aebischer, the president of the
Swiss Federal Institute of Technology, recruited Markram in 2002, and in
2005 bought him an IBM Blue Gene supercomputer.
Markram uses
predictive reverse engineering to leapfrog data from wet labs. He needed
good experimental data for detailed modeling, but without modeling and
simulation the data would be useless. With a multilevel model of the rat
brain as a template, scientists can make progress.
Caltech
neuroscientist Christof Koch: "I like his vision. The guy has cojones."
Markram thinks he can determine the causes of all 600 known brain
disorders: "It's not about understanding one disease. It's about
understanding a complex system that can go wrong in 600 different ways. It's
about finding the weak points."
He plans to make his first brain
model soon and "open up this new telescope to the scientific community" by
2016.
 |
 |