More than 100 years ago, The New York Times imagined a future where a machine would be able to tell if people were lying or telling the truth.
Today, that machine exists: It’s called a lie detector — it’s a $2B-per-year industry, and it’s based almost entirely on shoddy pseudoscience.
In a recent feature, Wired went deep on a new device, Converus EyeDetect, that’s trying to change this by enlisting eye-scanning technology. But is tech really any better at detecting lies?
Why do lie detectors suck?
A traditional polygraph test rests on the premise that when a person lies, she will produce a unique physiological response. Basic “indicators” like blood pressure, pulse, and breathing rate are measured during a round of questions; if there’s a spike, it’s interpreted as deception.
These tests have been discredited by numerous studies, even Supreme Court cases — yet 2.5m of them are still conducted in the US every year.
One company thinks it can do better…
The Converus EyeDetect promises to boost the accuracy of lie detection from 65% to 86% by “capturing imperceptible changes in a participant’s eyes.” It already boasts 500 paying customers in 40 countries.
But according to Wired reporter Mark Harris, who tested the device, the EyeDetect falls back on the same assumption as the polygraph, that deception can be physically measured in a scientifically sound way.
What’s really being measured here is fear and anxiety. And as we all know, sh*tting your pants is not an indication of guilt.