Why Are Lie Detectors Still a Thing?

The earliest imprint that you, dear reader, have probably gotten from the film Blade Runner is of that weird eye-scanning machine, a prop in an interrogation. It's an eerie and invasive test, one that involves a lot of emotional probing, eye scans, and measures of physiological activity to determine whether or the test-taker is capable of human emotion. It's a souped-up lie detector, a device and concept that exists in this particular future.

But the polygraph (that's the academic name for the device, so-called because it actually records breathing rate, pulse, and skin conductivity, i.e. signs of sweating), invented in 1921, is still being used.

This is despite being wrong or inconclusive about 13 percent of the time, according to the American Polygraph Association. The figure will range depending on the sorts of tests you pull stats from, but you can roughly say about 11 to 13 percent of the time you can't come up with an accurate assessment.

That might be too high a number to conclusively stick a crime on someone based on a polygraph test, but somehow US government employers still find it reliable enough to pre-screen potential employees for sworn positions in law, national security, criminal justice, and intelligence.

"The reason the polygraph works is because we can point to a machine and say you're lying, and some people succumb to that," Joe Navarro, a former FBI Counterintelligence Agent told me.

Navarro had been with the FBI for 25 years and, through it, became acquainted with non-verbal communication inasmuch as it helped him catch deception and spies.

The basis for lie detectors is simple: if the body and mind are connected, lying should induce physical signs of stress.

Here's how it goes. The test-taker is hooked up to a number of physiological sensors that measure heart rate, breathing, blood pressure, sweat, and so on. The test-taker is informed of every question that'll be asked on the test, and according to the American Psychology Association, she will be reassured that the test will certainly know that she's lying. No pressure.

The examiner asks a series of "relevant questions" mixed in with "control" questions that are broad, but are designed to induce a bit of stress. They're the sorts of questions that imply you've committed small wrongdoings, like jaywalking or telling white lies, just to see what spikes your blood pressure even a tiny bit.

The relevant questions will ask directly about any crimes the subject may have explicitly committed. When the test is done, the examiner will compare the physiological reactions between control questions and relevant questions. If your blood pressure spikes, if you start sweating or show signs of panicking, in particularly damning questions, they'd catch that.

Understandably, all these government agencies need ironclad confidence in their employees for security reasons. But here's the crux of why the US is still dependent on using lie detectors for their security positions: lie detectors have been deeply institutionalized, and there hasn't been a better alternative.

"People use it because there's a whole industry around it," Navarro said.

Indeed, there's an entire accreditation process to using a polygraph machine. Tuition for a license costs anywhere from $4495 to $5750, and that feeds into a market, which, according to University of Minnesota professor William Iacono, makes it hard for scientists to simply rule the machine out as inconclusive.

"There's a lot at stake. And the people who use polygraphs think that they work very well and that's why they say they're not going to accept inconclusive results [as an option]. To say it's not accurate is not conducive with their goals," he told me.

The National Research Council published a damning 417-page report in 2002, which found that polygraphs weren't only unreliable to use in actual use, but also benefited from inflated success rates reported by practitioners. Yet this wasn't enough to cause any sea change in its use.

"Polygraph testing now rests on weak scientific underpinnings despite nearly a century of study, and much of the available evidence for judging its validity lacks scientific rigor," the committee said in a statement.

Researchers have looked elsewhere, too.

Navarro told me that the P300 test, a test that assesses memories stored in the brain, could be another potential avenue for testing deception. It's mostly effective on non-interpersonal crimes, i.e. criminals who don't know their victims. If you show a non-interpersonal criminal a crime scene (which should trigger the part of the brain that the P300 test watches), you can catch them in the act of lying. But for interpersonal criminals, that's not as effective, as crimes can be acted out in familiar territories.

Listening to voice assessments or watching non-verbal tics or other bodily behaviors might be another useful tactic. But those methods received even less attention as the scientific community has written them off as mostly unreliable.

"Those don't have any scientific basis. They're probably worse than the polygraph," Iacono told me.

It's possible, Iacono continued, that academia have an easier time ruling voice assessments out as as those haven't been as deeply entrenched as the polygraph's century of use.

And yet, polygraphs are still a thing, and likely will continue to be a thing should the US government continue to view its accuracy as "good enough."

Iacono said that about half of US jurisdiction have "stipulated" admission, meaning that both parties have to agree to polygraph evidence to speed along the trial. Though, it's used infrequently as evidence more so than a supplementary "proof" of trustworthiness.

But that the arms of the law still rely on the truthiness of an invention that's still being contested a century after its invention should raise some eyebrows. At least until someone invents the next best thing. And who knows when that will be?