MoNETA: A Mind Made from Memristors
The Cat and the Computer
What is it about cats—or in this case, their brains—that can make otherwise rational people act goofy?
A year ago, IEEE Spectrum reported that IBM had won a major computing award for simulating the neurons and synapses in the brain of a cat. When the news broke, initial coverage of the award morphed, as if by a game of telephone, into reports that IBM had managed to simulate an entire cat brain.
Bedlam ensued. A normally mild-mannered scientist who had simulated a small area of the brain of a rat called for the award-winning IBM researcher to be "strung up by his toes." The forums of Spectrum and other relevant sci-tech magazines soon filled with invective and accusations that the entire field of computational neuroscience was a shameful hoax.
The hysteria was a symptom of a larger problem in a field whose goal is to create computer software that approximates the functions of the mammalian brain. Because no one agrees on exactly how intelligence arises in people, it's impossible to know which approach to building it into machines is correct, which approach is barking (or meowing) up the wrong tree, and which is just barking mad.
Into the fray stepped Massimiliano Versace and Ben Chandler [above, center and right], two Boston University computer scientists. Their incisive, repeated commentary on Spectrum's online forum prompted Associate Editor Sally Adee [left] to contact them. They became the sources for a postmortem article, published on IEEE Spectrum Online two months later, that became the foundation of the Wikipedia entry on cat intelligence.
Versace and Chandler are working with HP Labs (the creators of the first usable memristor) and the U.S. Department of Defense on a project to create truly intelligent machines. But they wisely decline to say whether they expect their artificial intelligence to be in rat or cat form. They describe their pathbreaking work in this month's "Brain of a New Machine."