Don't try to fake pain, computers beat humans at spotting it: Study
Researchers have discovered a computer-based vision system that is able to tell a faked expression of pain from a real one better than humans can.
In addition to exposing pain malingering, where a person might exaggerate or fake pain symptoms in drug-seeking behavior or for other motives, computers could spot deceptive behavior in job screening or in the fields of law, medicine and security, they said.
In a study by U.S. and Canadian researchers, 205 human participants tried to determine the truthfulness of the pain expressions of people in short videos.
Some of the people in the clips were undergoing a cold presser test, with their hands held in icy water to measure their tolerance to pain, with a resulting genuine facial expression of pain.
Others in the clips were faking similar expressions of pain.
"Human subjects could not discriminate real from faked expressions of pain more frequently than would be expected by chance," University at Buffalo communications Professor Mark G. Frank says.
The researchers used a computer-vision system designed to recognize facial expressions in real time.
They developed the system to determine how accurate machine vision is versus human vision and found it could recognize deceptive human facial expressions by extracting spatiotemporal signals humans are unable to identify.
"In highly social species such as humans," says researcher Kang Lee, a child studies expert at the University of Toronto, "faces have evolved to convey rich information, including expressions of emotion and pain. And, because of the way our brains are built, people can simulate emotions they're not actually experiencing so successfully that they fool other people."
"The computer is much better at spotting the subtle differences between involuntary and voluntary facial movements," Lee says.
The best clue to a falsified expression, the study found, is how and when the mouth opens and closes, with the mouths of people faking an expression opening too regularly and showing less variation.
Subtle clues picked up by the computer system allowed it to "detect distinctive, dynamic features of facial expressions that people missed," Bartlett says. "Human observers just aren't very good at telling real from faked expressions of pain."
The study, by researchers from the University at Buffalo, the University of California, San Diego, and the University of Toronto has been published in the journal Current Biology.