Imagine this experiment: You're looking at someone with a painful expression on their face.
They've got their arm submerged in a bucket full of water.
It might be ice water, and the person might be grimacing because they're actually freezing and feeling real pain. Or maybe the bucket is filled with lukewarm water and they're acting, completely faking that pain.
Would you be able to call their bluff?
Well, you probably couldn't beat the computer built by UC San Diego researcher Marian Bartlett and her colleagues.
In a new paper, the researchers describe an expression recognition system that detected faked pain 85 percent of the time. Humans in the same experiment only called out the fakers about half the time.
Bartlett says mouth movements were the giveaway.
"It turns out that people open and close their mouths at intervals that are too regular — they're too consistent — when they're faking pain," she said.
This technology currently is being commercialized for use in market research, retail and healthcare settings.
