by staff writer Laura Davis

transparentBrainOnce again, science is taking a stab at catching up with science fiction. In an ambitious undertaking, the Human Brain Project (HBP), an international … well, brain trust, under the direction of renowned neurology researcher Henry Markram, plans to create a working computer model of the human brain. Markram explained the project at TED in 2009 as, “a Noah’s ark of knowledge” about how the human brain works. He went on to explain that the human brain has about ten million synapses in it, and by learning how neurons collect information and transmit it to those synapses, we can understand the “fabric” those synapses create.

By replicating and studying the “electrical objects” formed in the brain as a result of stimuli, the project aims not only come to a more complete understanding of the brain’s structure and organization, but also to discover new applications for that knowledge. HBP has 80 partner facilities across Europe, which, they hope, will provide a broad enough base of data to achieve this lofty goal. The project has been awarded a billion Euros by the European Union, and the Swiss government is providing an IBM Blue Gene computer. Even the touted Blue Gene, though, doesn’t offer enough computing power for the ultimate goals of HBP. They plan to build and run the brain model on an exascale computer.

A what? Follow me, here.

FLOPS (FLoating Point Operations Per Second) is a measure of a computer’s raw operations execution rate. Your computer probably operates in the gigaFLOPS range: billions of FLOPS. Blue Gene operates in the petaFLOPS range (the current pinnacle of raw computing speed): quadrillions of FLOPS, or about a million times faster than your computer. The computer needed for this working brain model will operate in the exaFLOPS range: quintillions of FLOPS, a thousand times faster than a petacomputer, a billion billion operations per second.

There’s a small problem. The exacomputer doesn’t exist yet. Japan, China, and the United States are in a race to build one, but the obstacles are huge. Energy demand and heat management problems grow exponentially with this kind of massive increase in speed. IBM’s Sequoia, a petacomputer which operates at Lawrence Livermore National Laboratory and is currently the world’s fastest computer (16.32 PFLOPS), is considered extremely energy efficient for its speed. It requires 7.9 megawatts of power to run. Your entire house demands kilowatts (thousandths of a megawatt) to run. Sequoia takes up 3,000 square feet of space, it generates enough waste heat to continuously heat 50 single-family homes, and it requires 3,000 gallons of water a minute to cool. Japan claims that we’ll see an exacomputer by 2020, but there are many experts who believe these estimates to be far too optimistic.

The scientists at HBP aren’t sitting around waiting and hoping, though. Before they can make use of an exacomputer, they’ve got a daunting amount of preparation to accomplish. Their neuroinformatics group in Stockholm has to assemble and annotate all the research data needed for the models, so it’s ready to be fed into the simulation system. They’re creating tools to analyze the data, and more tools to build brain atlases for both mouse and human brains. That may sound straightforward, but it’s not. Although a great deal of research has been done on various aspects of brain function and structure, there is no comprehensive database for that research and that is what they must create.

Additional teams in Munich, Madrid, and Edinburgh are providing research on both human and mouse brain structure and organization, and comparisons between the two. In Munich and Lausanne, teams will be working on robotics experiments to be run using the brain model, and in France mathematical and theoretical teams will be working out how to model such complex phenomena as neuroplasticity: the way in which the human brain reorganizes its neural pathways when we learn new things, or in response to lost function. While all of that is happening, yet another team in Sweden will be building the actual software that will combine the data from all of the teams to create and run the brain model.

HBP’s model will build on the work of the Blue Brain Project, which Henry Markram also directs at the École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland. Blue Brain’s goal was to create a synthetic brain by reverse-engineering the mammalian brain (rat, in this case), starting with a simulation of a single neocortical column (the smallest functional unit of the neocortex, which is responsible for higher brain functions like conscious thought and self-awareness), and progressing to building a cellular mesocircuit, made up of 100 neocortical columns. By 2014, Blue Brain plans to complete a cellular rat brain simulation, using 100 mesocircuits. The hope is that the technology and groundwork established by the Blue Brain Project will serve as a substantial basis for HBP’s more detailed model.

What will all of this painstaking and expensive research accomplish? Huge implications for medicine and psychology are obvious. Markram envisions a diagnostic system for brain diseases and disorders that would be based on objective biological data, instead of depending upon the highly subjective, symptom-based criteria currently in use. “We are going to put all the [brain] diseases on the table and start working out mathematically how they are related to each other.” Computer brain modeling will also allow researchers to accomplish more work with far less animal testing.

Being able to replicate the human brain’s efficient and flexible ways of processing data could also revolutionize the way we build and use computers. Neuromorphic computers – computers based on the structure and function of the human brain – could learn, rather than requiring software upgrades, and could achieve far more complex operations than are currently possible.

Now, before any fans of Asimov or Heinlein get too worried, I’d better mention that HBP is planning ahead, and while all of this new technology is under development, its Ethics & Society team is working out how to apply it. Yesterday’s science fiction is becoming tomorrow’s science, and we’re here in the middle, to watch it all unfold. In the words of Robert Heinlein, “everything is impossible, until it is done.”

– 30 –