MIT algorithm can tell a fake smile better than humans

You would think that when you’re frustrated, you don’t smile, but MIT researchers have found that you actually do. According to MIT News, the computers that found this phenomenon to be true can also differentiate between a fake smile and a real smile, better than human beings can. Ehsan Hoque, a graduate student in the Affective Computing Group of MIT’s Media Lab is the lead author on a paper written about the project and published in the IEEE Transactions on Affective Computing. He says, “The goal is to help people with face-to-face communication.” The algorithm can be used in the future to help computers determine the moods of their human users and respond accordingly. The computers could also help individuals who have difficulty making meaning of expressions of the people around them.

When the research was in progress, Hoque and his group asked their sample audience to give them an expression of delight or an expression of frustration. These expressions were recorded on webcams. Then, the sample group was either given an online form to fill out, which was designed to elicit frustration or made to watch a video that would result in delight. Hoque says that when the subjects were asked to make a frustrated expression on their own, they would not smile. However, when they filled out the form, which would delete all the information that the subject added in after hitting ‘submit’, 90 percent of the subjects did smile. The smiles that were a result of the ‘delightful’ video and the frustrating form, did not appear too differently when seen by the human eye, but the computer program they developed could tell the difference.

More from News & Analysis

What is the US HIRE Bill and why is India’s $250-billion IT sector worried? Is the internet dead? What's this theory that OpenAI's Sam Altman says might be true?

When users smiled out of frustration, their smiles set on much more rapidly and faded just as fast. However, the smiles that were a result of delight, faded in gradually. The point is that just because an individual smiles, it doesn’t mean they’re necessarily happy. The timing of the smile is key and several public figures, like former British Prime Minister Gordon Brown have been criticized with phony smiles because they didn’t time their smiles correctly. The idea behind this project is to develop the computer algorithm to a point where a computer containing the algorithm becomes more understanding, intelligent and respectful.

The algorithm essentially tracks the movement of facial muscles, which can then be quantified with the Facial Action Coding System, a tool developed in the 1970s. Fake smiles tend to be made with the zygomatic muscles, which are a voluntary set of muscles designed to lift the corners of your mouth. A genuine smile is controlled by an involuntary set of muscles, which raise your cheeks and result in the wrinkling under your eyes (smiling with your eyes!).