ss_blog_claim=a290fbfb2dabf576491bbfbeda3c15bc

Sunday, December 02, 2007

From the smartest artificial brain to the first artificial life

The Blue Brain Project

Scientists rely on computer models to understand the toughest concepts in science: the origin of the universe, the behavior of atoms, and the future climate of the planet. Now a computer model is being designed to take on the human brain. Neuroscientist Henry Markram of the Brain Mind Institute at the École Polytechnique Fédérale in Lausanne, Switzerland, has spent the last 15 years painstakingly mapping cells from the living brains of rats so that he can create a neuron-by-neuron simulation of the brain. With assistance from IBM (whose nickname, Big Blue, helped name the project), Markram hopes to have a virtual human brain, with all its 100 billion neurons, functioning by 2015.

Scientists still don’t understand many of the most essential functions of the brain, like memory or the fundamentals of brain disease and treatment. Markram’s model will electronically mirror the real brain’s biological behaviors, imitating mathematically the interactions among individual neurons and the effects of neurotransmitters on those cells.
The model will also be adjustable so that it can explore unusual physiology (a higher-functioning left hemisphere, say, or a weakened hippocampus) and environmental changes (like the effects of taking a pharmaceutical).
The data can then be interpreted via computer images. “We are building a generic template,” Markram says, “which will allow us to reconstruct a brain according to any specifications.”

To grind through the immense amount of data, IBM has custom-tailored one of its most powerful supercomputers, capable of processing more than 22 trillion operations per second. With this computer, Markram has created a preliminary model of a neocortical column—a set of about 10,000 cells that work together—equivalent to one in the brain of a 2-week-old rat. “We have achieved the ability to build a brain microcircuit, an elementary unit, and now it’s just a matter of scaling up,” he says.

Many of Markram’s colleagues think that he is too ambitious, that a model of billions of neurons, no matter how intricate, cannot tell much about the functions of a real brain. “People think that it is impossible,” he admits. “They believe that we don’t understand enough about the brain to build it.” He counters that mysteries of brain circuitry will be resolved as the project moves forward over the years.

If the Blue Brain team succeeds, scientists will for the first time have a meaningful physical model of the human brain. So could a fully functioning virtual brain have the ability to create thoughts of its own? Markram isn’t counting on it, but he will be watching closely if Blue Brain begins to make its own decisions, providing unique outputs to identical inputs in a way that is beyond chance or chaos theory and achieving something that has never been observed in a computer: consciousness. “Once we build the whole brain,” he says, “if consciousness emerges, we will be able to study it systematically and understand exactly how it emerges.”


Susan Kruglinski

No comments:

 
ss_blog_claim=a290fbfb2dabf576491bbfbeda3c15bc