Machine Learning Can Predict Your Personality From Your Eye Movement

Machine learning can now identify five personality traits from the blink of your eye… well, not just one blink. Make that many blinks.

Move over San Francisco, Beijing, and other AI hubs. The University of South Australia and the University of Stuttgart in Germany have developed an AI system capable of predicting personality traits by tracking and analyzing eye movements.

Seeing Eye to AI

Eye movements and personality have always been closely linked. Enhancing our understanding of this could someday help health professionals like psychiatrists and serve as a foundation for robots to “understand” humans better from non-verbal communication

This AI research utilized machine learning to track and monitor eye movements of 42 people. The results were compared to the results of personality tests that the subjects had to take. The algorithms worked well in identifying four of five traits: neuroticism, extraversion, agreeableness, and conscientiousness. The trait that needs some work is openness.

“Thanks to the machine learning approach, we could automatically analyze a large set of eye movement characteristics and rank them by their importance for personality trait prediction,” the group of researchers involved wrote.

They also explained another important aspect of this study: “Going beyond characteristics investigated in earlier works, this approach also allowed us to identify new links between previously under-investigated eye movement characteristics and personality traits.”

Analysis Is in the Eye of the Beholder

The eye tracker didn’t collect data over the course of days or weeks. It only measured eye movement for 10 minutes; the subjects were instructed to make a purchase with a $10 bill at the university shop. After wearing the eye tracker, the subjects took the personality and curiosity tests.

The researchers found that the low amount of subjects didn’t yield enough data to apply to practical applications, but they uncovered a lot of interesting information during their analysis. For example, pupil diameter was found to be important for predicting the trait of neuroticism.

Making Robots More Human

The group hopes to develop machine learning applications to recognize and interpret human social cues and signals. They also have more ideas for how to improve the algorithms, apply them in other contexts, and use them in daily tasks:

“Such knowledge of human non-verbal behavior might also be transferred to socially interactive robots, designed to exhibit human-like behavior. These systems might ultimately interact with humans in a more natural and socially acceptable way, thereby becoming more efficient and flexible.”

We can’t wait to see where this research leads! What applications can you see for this AI system?