Despite their fixation on the problem of deceit, government agencies aren’t interested in funding anything so abstract as basic research. “They want to buy hardware,” Honts says. But without an understanding of the mechanics of lying, it seems that any attempt to build a lie-detecting device is doomed to fail. “It’s like trying to build an atomic bomb without knowing the theory of the atom,” Honts says.
Take the polygraph. It functions today on the same principles as when it was conceived in 1921: providing a continuous recording of vital signs, including blood pressure, heart rate, and perspiration. But the validity of the polygraph approach has been questioned almost since its inception. It records the signs of arousal, and while these may be indications that a subject is lying—dissembling can be stressful—they might also be signs of anger, fear, even sexual excitement. “It’s not deception, per se,” says Judee Burgoon, Nunamaker’s research partner at the University of Arizona. “But that little caveat gets lost in the shuffle.”
The US Army founded a polygraph school in 1951, and the government later introduced the machine as an employee-screening tool. Indeed, according to some experts, the polygraph can detect deception more than 90 percent of the time—albeit under very strictly defined criteria. “If you’ve got a single issue, and the person knows whether or not they’ve shot John Doe,” Honts says, “the polygraph is pretty good.” Experienced polygraph examiners like Phil Houston, legendary within the CIA for his successful interrogations, are careful to point out that the device relies on the skill of the examiner to produce accurate results—the right kind of questions, the experience to know when to press harder and when the mere presence of the device can intimidate a suspect into telling the truth. Without that, a polygraph machine is no more of a lie-detector than a rubber truncheon or a pair of pliers.
As a result, although some state courts allow them, polygraph examinations have rarely been admitted as evidence in federal court; they’ve been dogged by high false-positive rates, and notorious spies, including CIA mole Aldrich Ames, have beaten the tests. In 2003 the National Academy of Sciences reported that the evidence of polygraph accuracy was “scanty and scientifically weak” and that, while the device might be used effectively in criminal investigations, as a screening tool it was practically useless. By then, other devices and techniques that had been touted as reliable lie detectors—voice stress analysis, pupillometry, brain scanning—had also either been dismissed as junk science or not fully tested.
But spooks and cops remain desperate for technology that could boost their rate of success even a couple of points above chance. That’s why, in 2006, project managers from the Army’s polygraph school—by then renamed the Defense Academy for Credibility Assessment—approached Nunamaker and Burgoon. The government wanted them to build a new machine, a device that could sniff out liars without touching them and that wouldn’t need a trained human examiner: a polygraph for the 21st century.
by Wired Staff, Wired | Read more:
Illustration Joyce P. Chan/The University of Arizona
