Probing the human brain directly presents great medical and ethical hurdles. Researchers’ efforts are also often stunted by limitations imposed by imperfect models. That latter restriction, at least, may soon ease: IBM recently announced plans to build a digital model of the brain.

IBM's Blue Gene, regarded as the world's most powerful supercomputer, will be used over the next two years by company researchers, in collaboration with the Brain Mind Institute at the Swiss Federal Institute of Technology in Lausanne, to simulate the electrochemistry of the neocortex—the large part of the brain responsible for cognition. The goal is to understand perception, learning and memory. The Blue Brain Project model could also perhaps improve our knowledge about mental disorders, such as autism and schizophrenia.

The effort is not the first to create a computational model of the human brain, but it appears to be the most extensive. “This type of endeavor at this number of circuit elements, with the capacity to include realistic properties, has never been attempted before,” says Yale University neurobiologist Gordon M. Shepherd, who is not involved in the project.

IBM's human brain model, if successful, will surely prove to be useful. But many neurobiologists agree that Blue Brain will not be a substitute for traditional research. “While mathematical simulations are extremely helpful and predictive for hypothesis testing,” advises Jeffrey T. Potts, a neurobiologist at the University of Missouri not on the Blue Brain team, “they will not replace the physiological elegance and complexity of biological life, be it mice, cats, dogs or humans.”