By Jess Goulart
Photo courtesy of Joseph Elsbernd.
Stone Age, Bronze Age, Golden Age, will the 21st century come to be known as the Robotics Age?
The Ecole Polytechnique Federale De Lausanne (EPFL), located in Lausanne, Switzerland, bets yes. For the past five years, they’ve been working to engineer the world’s first fully synthetic brain capable of emulating lightning-fast human thought processes.
The project is called the Blue Brain Project (BBP) and was founded in 2009 by Dr. Henry Markram. The eventual culmination of their efforts will be a supercomputer that’s wiring and software successfully parallels our own, so to speak, a lofty but worthwhile goal.
In the simplest terms possible, the human brain works by way of tiny cells called neurons, each acting as a little battery. When a neuron becomes activated, or charged, it sends out an electrical pulse which a neighboring neuron can read, in turn transmitting a pulse of its own. A human adult has around 86 billion neurons, all firing in an intricate, almost instantaneous pattern.
For a synthetic brain each “neuron” needs the equivalent of a laptop backing it, the complete brain having billions of laptops. In an interview with Seed Magazine, Dr. Markram said “this is the first model of the brain that has been built from the bottom-up. There are lots of models out there, but this is the only one that is totally biologically accurate. We began with the most basic facts about the brain and just worked from there.”
Neuroscientist Dr. Markram earned his PhD from the Weizmann Institute of Science in Israel, discovering during his tenure there a link between acetylcholine, a neurotransmitter, and memory mechanisms. Later, as a professor there, he developed a theory called the “liquid state machine,” a computational construct that mirrors neural networks. Thus began the BBP.
To power the project, IBM built the supercomputer Blue Gene. It’s comprised of four gigantic machines that process data at 56 teraflops (a single teraflop the equivalent of one trillion floating-point operations per second). The machines can currently simulate up to 1,000 neurons with their 16,000 processors, and most recently successfully simulated a single cortical column in a rat’s brain. The size of a pinhead, that column alone houses 10,000 neurons, and a rat has nearly 100,000 of them. By comparison, a human has up to two million.
Painstaking research goes into analyzing every detail in animal brain circuitry to better build out BBP. Their observations are translated into mathematics, and the program is run through Blue Gene every two weeks or so, with current models reporting a “high level of realism.”
To continue the project, the next generation of supercomputers is needed. The current classification of fastest computers is called the peta-scale computer, named for it’s ability to compute at speeds faster than a “petaflop.” There are a handful of peta-scale computers like Blue Gene or the K Computer in Japan, which are used for research or to analyze and help address pressing global needs like climate change, energy, space, and economics.
It’s also quite possible that we will one day ask them the meaning of life, and be told 42.
Exascale computers are the next generation and will allegedly be able to compute at speeds rivaling the human brain. At an IT conference in 2008, it was predicted that exascale technology will be available by 2018. The BBP hopes to use the first exascale computer, called DEEP, to run a human brain simulation by the 2020s. DEEP is currently being constructed by the Julich Research Center in Germany.
Is this uncanny vision, or idealism? In 2013 the EU cast its vote when it awarded the project $1.3 billion. On the other hand, the scientific community remains largely critical due to the enormous expense. It’s been called “an expensive pipe-dream,” and many leading computational neuroscientists refused to comment for this article. Dr. Eero Simoncelli, at the Center for Neuroscience at NYU tells BTR “the bottom line [which I’m sure you’ll hear from many others] is that massive detailed simulations seem unlikely to lead to a real understanding of brain function. And at best, they represent only one of many potentially useful research paradigms.”
For his part, Dr. Markram continues to rally support, causing waves with a recent TED talk in which he claims BBP is only ten years off of completion, which you can check out below.
It’s easy to see why this idea is so appealing. Apart from the implications it could have for the development of AI, the scientific community could gain invaluable insight into the human brain. Success means drug testing with no risk of side effects on humans, better surgical practices once brain circuitry is fully understood, as well as better treatments for disorders like manic depression and schizophrenia once their effect on brain waves is analyzed. Even complex emotions like love and hate could be better understood in terms of the neural pathways involved.
So genius or insanity? The world will have to wait and see.