How can you build a brain? The $1.6 billion Human Brain Project in Europe, being billed as the most ambitious neuroscience effort ever, intends to find out.
The effort is being led by the Ecole Polytechnique Fédérale de Lausanne (EPFL) in Switzerland. EPFL Professor Henry Markram, who is heading up the Human Brain Project, told the BBC that the researchers will attempt to understand "what makes the human brain unique, the basic mechanisms behind cognition and behavior," how to better diagnose brain diseases, and how to build new technologies inspired by the brain's computational processes.
The effort expects to take 30 months to set up the model platforms, and it held its first meetings with scientists from more than 130 partner institutions on Monday.
On its Web site, the project says that "one of the major obstacles to understanding the human brain is the fragmentation of brain research and the data it produces." The most urgent need, the project said, is a "concerted international effort" that uses emerging technologies to integrate data "in a unified picture of the brain as a single multi-level system." The goal of the 10-year-long effort is nothing less than building "a completely new information computing technology infrastructure " for use in neuroscience and brain-related research.
Co-funded by the European Union and designated as an EU Flagship initiative, the project will model the brain with supercomputers and then use the model to test simulated drugs and treatments for brain-related diseases and other medical problems.
Six research platforms are planned, covering neuroinformatics, brain simulation, high-performance computing, medical informatics, neuromorphic computing and neurorobotics. In 2016, the platforms will become available to the project's scientists as well as other researchers, and the resources will made available on a competitive basis, just as large telescopes or supercomputers are made available to researchers.
The key resources to be available are simulations, high-performance computing, neuromorphic hardware, and databases. In addition to data obtained from simulations and research using the platforms, information from the thousands of neuroscience papers published annually will be collected.
Neural Network Simulations
As might be expected, the management of these resources will require a huge effort. The neuroinformatics platform, for instance, will generate data that will be integrated into a brain cartography that ranges from an individual brain cell to an entire brain. The medical informatics platform will be utilized to detect and identify pathologies, using anonymous clinical data gleaned from hundreds of patients via participating hospitals and pharmaceutical companies.
Medical informatics will identify biological signatures of brain disease so as to assist early diagnosis. Pharmaceuticals are involved, in part, because better diagnosis, coupled with better simulations of disease and drugs, could lower the cost of drug discovery.
The neuromorphic computing platform will utilize microchips that emulate the function of neurons, in order to better understand learning and neural resiliency. The researchers will integrate these neural network simulations into virtual robots, and eventually may employ them in physical robots.