Brain on Fire

Nonetheless, the University of Pennsylvania has licensed the technology developed there -- on which it has applied for a patent -- to the California firm No Lie MRI Inc. The firm planned to scan for lies the brain of its first customer yesterday. But at the last minute, with NBC and CBS camera crews standing by to record the event, she decided she didn't want to put to the test her assertion that she had not cheated on her husband while he was in alcohol rehab, according to Joel T. Huizenga, the company's founder.

No Lie MRI's Web site has proclaimed that the company hopes to revolutionize truth telling in America, offering "objective, scientific, mental evidence, similar to the role in which DNA biological identification is used," to everyone from the FBI, CIA and NSA to the Department of Homeland Security.

No Lie is not alone. Its Massachusetts competitor, Cephos Corp., has licensed competing fMRI lie detection technology from the Medical University of South Carolina.

The boundless desire for a way to dig through deception is why political consultant John Zogby, president of Zogby International, expects the new brain scanning devices to be in widespread use in the 2008 presidential election. He can clearly see a demand to discover what voters really think of candidates -- and their commercials.

Brainy Game

At the Treatment Research Center at Penn, Daniel D. Langleben -- who was the first to show that lies and truth look different in the brain -- is proud of more modest accomplishments. He can detect whether you're lying about playing cards.

Here's the setup: He gives you two playing cards. He doesn't know which they are. You memorize them -- for example, the seven of spades and the five of clubs. Then he asks you to lie about whether you have one, and tell the truth about whether you have the other. Into the fMRI you slide, and cards flash randomly on a screen above your head, one every three seconds. From the pattern of your brain's reaction to them, his team can tell you which two are in your mind, and which one you were lying about, with what he characterizes as 86 percent accuracy. As the gear is tweaked, he expects the accuracy rate to rise well north of 95 percent.

The "Guilty Knowledge Test," as it is known, is considered hot stuff in the research community.

"No question it is important," says Antonio Damasio, the neurologist who is director of the Brain and Creativity Institute at the University of Southern California. For 15 years, Damasio, the author of "Descartes' Error: Emotion, Reason, and the Human Brain," has pioneered the understanding of how the brain processes feelings and decision making.

"It is a very deep problem," Damasio says. "I don't do any work on lie detection. But you are in essence having to detect a discrepancy between an overt behavior and an internal representation. It is complicated enough to find out what is going on when the idea and the behavior are consistent."

Card recognition, unfortunately, is not a terrific representation of the real world. Human lies are rarely so neat and clean, yes and no, with zero shades of gray.

So Langleben's next round of research is more ambitious. He is trying to determine whether his machine can handle a much more common set of lies -- for instance, those told by somebody applying for a job who has cooked his résumé.

He invites a reporter to be only the second subject in this round of tests.