What’s Your Tell?

Scientists from the University of Oulu in Finland have built and tested a machine capable of spotting and recognizing micro-expressions. According to the team’s initial tests, the technology is even better than humans at the task.

Micro-expressions are brief and involuntary facial expressions. They can be just one- twenty fifth of a second in length, and they come with varying degrees of intensity. Typically, micro-expressions are indicative of an emotion that people are trying to conceal. Unsurprisingly, the movement of the facial muscles involved in micro-expressions are very subtle and are easy to miss.

For the study, the team gathered 20 individuals to watch a series of videos designed to draw out intense emotions, and they were given an incentive if they successfully managed to control their expressions and show no emotion. The group was being recorded on a high-speed camera at 100 frames per second. To get the technology to work, the team used a single frame that shows the subject’s face, which was then compared to other frames to show whether the expression changed. 

Poker Face

The result showed the machine’s remarkable ability to spot micro-expressions. The machine matched, and even surpassed a human’s ability to recognize the expressions, with the algorithm showing 10 percent more accuracy in terms of translating what the micro-expression meant.

However, distinguishing between regular facial movement and spotting swift micro-expressions remained within a human’s expertise, with the machine regularly interpreting movements such as eye-blinking as micro-expressions.

While still in the early stages of the study, researchers consider it a very promising start with plans to further develop the research. Should the algorithm and technology be perfected, applications could range from law psychology to law enforcement, and it could even be used for human resource management.


Share This Article