There are oodles of different traits that people claim can predict your personality — your star sign, your handwriting, your favorite ice cream flavor — but only a few are based in science. The gold standard for personality tests, scientifically speaking, is known as the “Big Five” model. Researchers recently got artificial intelligence one step closer to predicting at least four of those five personality traits in individual people without them ever taking a test. It’s all from how their eyes move.
The Big Five personality traits include openness, conscientiousness, extraversion, agreeableness, and neuroticism. (If you forget, just remember the acronym OCEAN.) In a study published in the journal Frontiers in Human Neuroscience, Tobias Loetscher of the University of South Australia and his colleagues showed how their eye-tracking method successfully predicted four of those five (all but openness), plus something called “perceptual curiosity,” or a person’s tendency to investigate new sights, sounds, or other sensory stimulation.
It may seem unbelievable, but using eye movement to judge personality isn’t all that far-fetched. After all, you’ve probably figured out that a stranger with darting eyes was a bit neurotic, or concluded that an acquaintance that loves prolonged eye contact is extraverted. Past research bears this out: one study showed that optimists will spend less time looking at gross medical images than pessimists, and another showed that people high in openness spend more time looking at locations in abstract animations.
The problem, Loetscher and his team write, is that these studies simply showed that certain personality traits were correlated with certain eye movements. But what if the movement of your eyes could actually predict your personality traits? That could open up a lot of possibilities, not least of all in the realm of robotics. If robots can watch your eyes to know what kind of person you are, they might be able to adjust their behavior on the fly and create a better experience for you, the user.
For the study, the researchers asked 42 students to fill out a personality test to assess their “Big Five” personality traits, then wear eye-tracking smart glasses while they performed a task designed to mimic a real-life scenario: walking around campus and visiting a shop. Then the team fed that data into a machine-learning algorithm, which crunched the numbers and made correlations between the participants’ personality traits and the way their eyes moved.
The glasses tracked a wide variety of eye movements, including where the participants looked, the size of their pupils, and the number and type of saccades — those quick movements of the eyes between fixation points, like the way your gaze might flit back and forth between someone’s eyes when you make eye contact. The algorithm found that certain types of personalities could be characterized by certain types of eye movements. For neuroticism, it was all about blink rate; for conscientiousness, it was the speed of the saccades. Agreeableness could be predicted by how much a person gazed at the bottom right of a scene, while extraversion centered on the size of the pupils and the number of small saccades. Perceptual curiosity was linked to the variation in the size of a person’s pupils.
These are all simplifications, of course — the algorithm worked with a complex mix of factors to make its predictions. The power of those predictions aren’t exactly mind-blowing: just seven to 15 percent better than chance. But it’s a small study, and Loetscher has hope for the future. “Machine learning usually requires thousands or millions of datasets to make highly accurate predictions,” he told New Scientist. “We expect it to get much better.” When it does, there are ethics to consider — seeing as disgraced data analytics firm Cambridge Analytica was targeting people with political ads based on their personality, more advanced personality prediction could open an even bigger can of worms. Be careful where you look.
Check out my related post: Do dogs mirror their owners’ personalities?