Facebook Twitter

Will lie detector work? Maybe not at first blush

SHARE Will lie detector work? Maybe not at first blush

Researchers in Minnesota say they have developed a new lie detector technology that could allow swift screenings at airports. But a University of Utah expert doubts it would work for mass screenings.

A report in Thursday's edition of the scientific journal Nature says a new thermal imaging system can tell with 85 percent accuracy if a person is lying. The study was led by Dr. James A. Levine, an investigator from the Mayo Clinic in Rochester, Minn.

Levine's co-authors are Norman L. Eberhardt of the Mayo Clinic and Ioannis Pavlidis of Honeywell Laboratories in Minneapolis, Minn.

The technique relies on recording heat patterns in subjects' faces.

"This technique has an accuracy comparable to that of polygraph examination by experts and has potential for application in remote and rapid screening, without the need for skilled staff or physical contact," says a Nature abstract.

The technique probably would not be valuable for mass screening, contends John C. Kircher, a professor in the U.'s department of educational psychology. He has carried out extensive experiments in detecting lies, with the help of both students and members of the general public.

According to a press release from the Mayo Clinic, the new technology showed whether a person was lying in more than 80 percent of cases studied. Patterns of heat distribution on a face "change dramatically with lying," the release adds.

Levine, an endocrinologist, and the others on the team found that people who are about to perform a deceptive act give off physiological signals, including excessive blood flow to certain areas of the face. When these signals are noted by cameras that take photos of the heat distribution of the face, a technician may conclude that the person is lying.

"The thermal imaging technology detects the subtle changes in metabolism in parts of the body. When an individual is exposed to the thermal imaging camera and is being deceptive, the computer detects the warming around the eyes," Levine said, according to the press release.

In clinical trials of mock crimes, the system correctly categorized 83 percent of subjects as guilty or innocent. Levine conceded that further testing and development would be needed before the device could be used to screen at airports.

A polygraph machine typically measures stress caused by the subject's need to lie. Respiration, heartbeat, blood pressure or sweating may change, showing that he is attempting to cover up what some experts call "guilty knowledge."

The polygraph machine will show a change in heartbeat, breathing, even the skin's electrical conductivity that varies with sweating.

Over the decades, a huge literature has grown up on the science, says Kircher.

"Our research is done mostly in the laboratory, where we have students or people from the community commit a mock crime," he said.

The researchers take a "Mission Impossible" approach. A subject may receive an envelope that instructs him to listen to a tape recording. The recording may either tell him details of some "crime" he knows about, or tell him to wait until a certain secretary has left her desk, then "steal" money from her purse.

When a "theft" happens, he said, "that's the guilty condition" that the subject will want to hide.

Then the subject undergoes a lie-detector test.

In the laboratory, using standard equipment, "our accuracy rates are around 90 percent" in detecting if the subject is lying about the "crime," he said. In field studies, accuracy varies from 88 percent to 95 percent.

The figures can be checked in cases where a person confesses after a polygraph examination.

"We believe that the accuracy of a properly conducted polygraph test, under optimal conditions, is probably on the order of 90 percent," Kircher said. That is, when a well-trained examiner is looking for evidence of some particular event.

"You're testing them about a typical behavioral action," he said.

A reason that the standard test works is that the guilty party is afraid of detection. Without wishing to, his body will signal that he is asked a question about which he knows some secret he does not want to divulge. He may sweat and his heartbeat may change.

To weed out spurious reactions, an investigator may ask the same question repeatedly. During half an hour of testing, that lets him more accurately decide which is a real signal and which is just noise in a person's response.

"Just asking a few questions at an airport, that's not going to happen," Kircher said.

If the accuracy in lab tests is the 80 percent claimed by Levine and associates, it will drop in a mass setting like an airport. It may dip until it is little more than a coin flip, or no use at all.

Actually, it could be far worse than useless.

If the test is 80 percent accurate, that still means 20 percent are false positives, where innocent people seem guilty and require further questioning.

"Thousands of people you're screening are going to appear deceptive," when they really are not.

That's not to say that Kircher thinks the technique has no potential.

"It doesn't surprise me that these investigators are finding thermal changes in the face," since lying triggers other physiological changes, he said.

"My concerns about the report . . . are in how they suggest it might be used in airports."

The Mayo Clinic study had a sample size that was "really small," he noted. "They need to replicate those results."

E-MAIL: bau@desnews.com