Face Masks Disrupt Facial Perception Amid Coronavirus Crisis, Says Study

Identifying faces of people in masks has created a unique challenge amid the coronavirus crisis, and a recent study published in the journal Scientific Reports shows the impact of this obstacle — and potentially substantial consequences.

RELATED: HOW URBAN SURVEILLANCE AND FACE RECOGNITION ALGORITHMS TRACK COVID-19

Face masks disrupting person-to-person recognition

The new study hails from researchers at Ben-Gurion University of the Negev (BGU) in Israel, in addition to York University in Canada — and investigates the impact face masks have had on person-to-person identification.

"For those of you who don't always recognize a friend or acquaintance wearing a mask, you are not alone," said Professor Erez Freud — a faculty member at York University, Toronto, who earned his doctoral degree from BGU, and Professor Tzvi Ganel, who is the BGU department of psychology's head of the laboratory for visual perception and action.

"Faces are among the most informative and significant visual stimuli in human perception and play a unique role in communicative, social daily interactions," noted the researchers, SciTech Daily reports. "The unprecedented effort to minimize COVID-19 transmission has created a new dimension in facial recognition to mask wearing."

People with low face recognition ability more affected

To survey the effects of mask-wearing, Professors Freud and Ganel used an augmented version of the Cambridge Face Memory Test — standard for assessing human perception of faces, which includes both masked and unmasked faces. The study was executed online with a substantial group of nearly 500 people.

The researchers discovered how the rate of successfully identifying someone veiled in a mask was reduced by 15%. "This could lead to many errors in correctly recognizing people we know, or alternatively, accidentally recognizing unfamiliar faces as people we know," said Professor Galia Avidan — a member of the BGU's department of psychology, in addition to the department of cognitive and brain sciences, with expertise in facial perception and recognition.

"Face masks could be even more challenging to people whose face recognition skills are not ideal to begin with and cause greater impairment," added Avidan.

Face masks force us to evaluate facial features individually

The research team's work also showed how masks interfere specifically with extracting a holistic impression of faces — forcing facial recognition to operate on a feature-by-feature process, which reduces the accuracy of human face perception, in addition to taking more time.

"Instead of looking at the entire face, we're now forced to look at eyes, nose, cheeks, and other visible elements separately to construct an entire facial face percept — which we used to do instantly," said the researchers.

Only face recognition AI can easily overcome face masks

These altered modes of performance — compounded by the modification of how humans process faces to a piecemeal style — might have significant effects on the daily tasks of life as we know it, like social interactions and many other events where personal interaction is required, including education.

"Given that mask wearing has rapidly become an important norm in countries around the globe, future research should explore the social and psychological implications of wearing masks on human behavior," said Ganel. "The magnitude of the effect of masks that we report in the current study is probably an underestimation of the actual degree in performance dropdown for masked faces."

It's hard to say when the coronavirus crisis will end. Likewise, no one can say when the masks come off. As of writing, most public and commercial spaces require face masks to enter, which means no one knows anybody at first sight from the face alone. We can't say for sure what this will mean for society, going forward. But beyond the existential dread of walking in a sea of even more alien faces — the only entities unaffected by obscured faces are facial recognition programs running on AI algorithms. So while we feel more like strangers to one another, the machines are meanwhile becoming more familiar.