The mere thought of being interrogated—by a parent, boss or significant other—is enough to make one's blood pressure rise and pulse and breathing rates race. But contrary to popular belief, these signs of anxiety are not reliable indicators of a person's honesty. Instead, researchers are looking into the brain to separate liars from truth tellers.

The act of lying or suppressing the truth triggers activities in the brain that send blood to the prefrontal cortex (located just above the eye sockets), which controls several psychological processes, including the one that takes place when a person crafts a new rather than a known response to something. "Lying is an example of this type of executive response, because it involves withholding a truthful response," says Sean Spence, a professor of general adult psychiatry at the University of Sheffield in England. "When you know the answer to a question, the answer is automatic; but to avoid telling me the true answer requires something more."

Spence and colleagues use functional Magnetic Resonance Imaging (fMRI) technology to determine whether someone is fibbing by tracing blood flow to certain areas of the brain, which indicates changes in neuronal activity at the synapses (gaps between the neurons). "If you're using fMRI, the scanner is detecting a change in the magnetic properties in the blood," he says. More specifically, hemoglobin molecules in red blood cells exhibit different magnetic properties depending on the amount of oxygen they contain. The most active brain regions use—and thereby contain—the most oxygen.

An fMRI is a full body scanner—"rather like being in a long cigar box," Spence says. The subject is completely immersed by the machine, which contains microphones, speakers and a keyboard that enable him or her to communicate with researchers. Researchers compare images taken of the subjects at rest with those snapped after they had been asked to respond to a series of questions.

Polygraph, or lie detector, tests are the most well-known method of discerning fact from fiction, but researchers say they are not reliable because they measure anxiety based on a subject's pulse or breathing rate, which can easily be misread. "They're not detecting deception but rather the anxiety of being…[accused of deception]," Spence says. "It's known that psychopaths have a reduced level of anxiety," that would allow them to fool a polygraph. The fMRI, he says, images the actual processes involved in deception.

Spence's research faced its greatest test in June during a demonstration of the technology on an episode of a three-part English reality TV show called Lie Lab, which studied the truthfulness of a man accused of being a terrorist, a woman who claimed to have been abused as a child, and a woman convicted of poisoning a child in her care. Spence and his team used an fMRI to study Susan Hamilton of Edinburgh, Scotland, who in 2003 was convicted of poisoning with salt a girl diagnosed with a terminal metabolic condition. Hamilton, who was in charge of feeding the child via a feeding tube that led directly into her stomach, was arrested after the girl was admitted to the hospital with massive blood sodium levels. The police testified that a syringe full of salt was found in Hamilton's kitchen, but she denies any knowledge of it. The woman was released from prison last year and has continued to search for ways to publicly prove her professed innocence.

This provided a stage for Spence to extend his research, which until then had only been conducted on young, healthy university students. In most of Spence's tests he and his team asked subjects questions twice, which they would answer first honestly then dishonestly. In Hamilton's case, there were specific events in dispute and she was answering queries about them under duress.

The researchers scanned Hamilton, 42, four times; during each scan they grilled her about the poisoning. With the fMRI, Spence was able to see that she activated extensive regions of her frontal brain lobes and also took significantly longer to respond when agreeing with the cops' account. The results did not prove her innocent, Spence says, but suggested that her brain was responding as if she were innocent.

Spence acknowledges that the results might have been more accurate if he had first done a baseline study that included asking her more general questions unrelated to the charges. Unfortunately, TV is show biz and his time with her was limited. "Being able to study this lady pointed out problems with the technique," he notes. "There are a number of control studies we want to do."

Spence says the technology is not ready to be widely used in criminal investigations, noting that there is a big difference between determining deception in innocuous instances (such as whether someone had a cup of coffee that morning) and in serious crimes. But it hasn't stopped scores of inmates from contacting him in the hope that his technique might help prove their innocence.

Companies have already begun to market fMRI tests as accurate lie detectors, even though Spence says the results achieved during controlled studies are rarely duplicated during far less predictable police probes. Cephos Corporation in Pepperell, Mass., last year began to offer what it calls "commercially available fMRI-based deception detection services," following a 2005 study of 61 individuals that claimed to be able to determine deception with more than 90 percent accuracy. No Lie MRI, Inc., in Tarzana, Calif., offers similar services.

But Spence warns that more research is needed before fMRIs can be accurately used to determine someone's guilt or innocence in a criminal case. He is currently preparing to apply for grants to study the technology on volunteers from different economic backgrounds as well as those with a history of personality problems and depression who are more likely to find themselves in legal hot water.