The Fake Smile Algorithm

How many times have you just said, “I’m fine” and smiled when all you wanted to do was to crawl under the covers with a glass of wine? Well, researchers at MIT are figuring out how to tell when your smile is real and when you’re just faking it.

In a paper entitled, “Exploring Temporal Patterns in Classifying Frustrated and Delighted Smiles,” the authors found that their computer program had a 92 percent success rate in determining when test subjects were smiling because they were delighted or smiling in frustration. Smilers were recorded in two experiments. First, they were told to recall times when they were either delighted or frustrated. Second, they were subjected to either delightful (babies) or frustrating (computer glitch) stimuli. While the MIT testers did not smile when recalling frustrating incidents, they did smile when experiencing actual frustration 90 percent of the time.

Here’s the rub: Humans can only differentiate the two smiles at a rate below random chance. So, it appears that computers can teach us a thing or two about improved face-to-face communication (which is the goal of this study). 

It’s a matter of time before the Matrix can tell when you’re lying. Watch this space for the Smile Differentiation App on your smartphone.