Facial expression analysis systems are unreliable, experts report
An analysis of over 1,000 studies shows that systems that recognize emotions from facial expressions are unreliable — frowning indicates anger only 25% of the time.
An analysis of over 1,000 studies shows that systems that recognize emotions from facial expressions are unreliable — frowning indicates anger only 25% of the time.
It is widely believed that a person's emotions can easily be determined from their facial expression. However, recent studies indicate that the movement of facial muscles cannot predict a person's behavior or feelings. Meanwhile, various fields are actively using software that evaluates people based on their faces.
Some scientists argue that there are unique facial expressions that reliably point to six categories of emotion: anger, sadness, happiness, disgust, fear, and surprise. But after analyzing more than 1,000 published findings on facial movements and emotions, independent experts found that the standard approach does not reflect the real differences in how people convey and interpret emotions on their faces.
For example, a gloomy expression or a smile can express several emotions depending on the situation, personality, or cultural context. When people frown, they are angry only 25% of the time. In the remaining cases, a furrowed expression may be a sign of intense concentration or pain. The same applies to smiling — most people smile without feeling joy.
Book a visit — we will pick a convenient time.