When two neuroscientists announced an online contest in September to decipher the workings of a simulated set of neurons, neither thought the event would attract much attention beyond a small group of their colleagues. But 25,000 visitors to the site later, the researchers think they may have found a new method for stimulating the curiosity of researchers from a whole host of fields.
"The idea of a puzzle really tickled people," says Carlos Brody, a postdoctoral researcher at New York University. He and his former adviser, Princeton University neuroscientist John Hopfield, challenged the community to figure out the principles underlying a neural network they'd created that responds to sounds. Contestants could feed the program their own sound files and analyze the neural net's simulated bursts of activity, or they could look at archived responses to sound files that Brody and Hopfield had presented to the neural net. Researchers were then challenged to use the principles derived from Brody and Hopfield's program to build a neural network that could recognize the spoken word "one."
And the winner is? Twenty groups submitted answers to both contests, but the first to get it right was a team led by David MacKay, another former student of Hopfield's now at Cambridge University in the United Kingdom. (MacKay says he used no insider information.) The team noticed that it didn't matter how fast a test word was spoken. As long as the right sound elements occurred in the right order, the artificial neurons gave the right response. This could only occur, they reasoned, when neurons associated with different elements of the test word fire synchronously. "Whether [this mechanism] is actually being used in the brain, I don't know," MacKay says, "but it's a great idea." MacKay's first-place finish garnered $500 and a handheld computer.
Hopfield says such contests sharpen neuroscientists' ability to analyze experimental data. "We thought it would be instructive for the neurobiology community, especially the young community," Hopfield says. Solving the puzzle was possible if a researcher just stepped through it logically, says David Tank, a neuroscientist at Lucent Technologies' Bell Labs in Murray Hill, New Jersey. "I think it turned out to be a valuable thing to try."
Other researchers are not sure about the role of brainteasers. "I think it's kind of fun," says Larry Abbott, a physicist at Brandeis University in Waltham, Massachusetts, "but I don't think it's a sensible way to disseminate scientific results in general." Abbott thinks the future of this kind of contest--if there is one--lies in posing unsolved problems and coordinating researchers' activities via the Web. "The key is to come up with the right questions."