Photo: Colin Anderson/Getty Images

It's technology of the future being tested today," said the TV news announcer, "a new lie detector that one day could be used to spot terrorists at airports." And my local radio station has been plugging Temple University, in Philadelphia, as the "creator of a new lie detector," based on brain imaging.

High-tech polygraphs are rapidly emerging from the laboratory, accompanied by hyperbole and hopes that they will be useful in the fight against terrorism. But whether they can make a real contribution to law enforcement is more uncertain than most people realize.

Brain Fingerprinting Laboratories Inc., in Seattle (http://www.brainwavescience.com), is aggressively promoting one technique for criminal and other investigations. Using electroencephalograph sensors, Brain Fingerprinting measures an electrical signal from the scalp that is evoked by the sight of a familiar image. For example, show a photo of a crime scene to a suspect and the response can be compared to that from an image of a familiar person. Could these signals be interpreted to indicate whether that person was familiar with the crime scene, or not? The method was used in India for high-profile terrorist investigations and figured in an unsuccessful attempt to stop the execution of Oklahoma death-row prisoner Jimmie Rae Slaughter earlier this year.

Other research groups, including two in Philadelphia, are developing lie detection methods using other neuroscience techniques. My colleague Daniel Langleben at the University of Pennsylvania and his co-workers are using functional magnetic resonance imaging to measure changes in brain function in subjects doing card tricks. And a group at Drexel University is developing a high-tech polygraph system using near-infrared light scattering from the brain.

Defense and security agencies are also funding research in this area. In 2004, for example, the U.S. Navy awarded Li Creative Technologies, in Florham Park, N.J., a US $100 000 contract to study "thermal imaging of the head for sensing and identification of concealed intent." The goal is to develop a "fast screen test for entry to U.S. borders, military checkpoints, and critical facilities."

Can these methods work? Certainly, scientists can measure different brain responses when a subject is shown pictures with varying degrees of meaning, for example a picture of a loved one compared with that of a stranger. But nobody knows how well these new methods will work in real-world settings. Being able to detect a student's lie in a laboratory card trick is clearly not the same as being able to detect terrorists at the airport or reliably uncover knowledge stored in the brain.

Most of these methods are based on an experimental paradigm known as the guilty knowledge test, which has a false-positive rate of about 20 percent. Can we improve on this sufficiently to detect the rare event of a terrorist passing through airport security without incriminating a far larger number of innocent travelers? Not likely.

And brain image or other "scientific" data from the brain will be intensely prejudicial to a present-day jury, much as polygraph results must have seemed incontrovertible when they were first introduced as courtroom evidence. Would a lay jury view a false-color image of the brain, and be able to understand that it shows small changes in regional blood flow in the brain with only a very indirect relation to telling the truth?

Polygraph testing came to be widely used and accepted in the United States and other countries without the careful evaluation that should be required of any investigational tool on which the future of the subject's life depends. Before being applied in socially important settings such as the courtroom, Brain Fingerprinting testing and other neurotechnological methods should be subjected to a rigorous, independent, public assessment.

Our guest editorialist, Kenneth R. Foster (F), is professor of bioengineering at the University of Pennsylvania, former president of the IEEE Society on Social Implications of Technology, and a member of IEEE Spectrum's Editorial Advisory Board.