Pain-Scanning Computer Can Tell if You're Really Hurting on the Inside

By Gary Cutlack on at

A team at the University of Toronto has developed a computer system that's able to differentiate between real and fake pained facial expressions, with the code apparently able to do a better job than humans at telling if we're really in pain or just lying about it for attention and biscuits.

The team believes humans can only differentiate between real and pretend pain by guesswork and random chance, getting a 55 per cent accuracy rate at best if they've been trained to look out for the signs of genuine distress. But the university researchers' system can accurately catch genuine pain 85 per cent of the time, with the idea being to apply what they've learned about natural facial expressions to eventually distinguish genuine "health, physiology, emotion, or thought" processes from fakes.

If all goes well, expect to see it testing emotional reactions on the Jeremy Kyle show from 2019. [Current Biology via The Register]

Image credit: Pain from Shutterstock