Tracking Tiny Facial Movements Can Reveal Subtle Emotions in Autistic Individuals

A Rutgers-led study examines how detecting microscopic facial movements, previously overlooked, are key to enhancing emotional recognition in autistic individuals

A study led by Rutgers University–New Brunswick researchers suggests that tiny facial movements – too slight for the human eye to notice – could help scientists better understand social communication in people with autism.

Published in Frontiers in Psychiatry, the study found that while individuals with autism express emotions like everyone else, their facial expressions may be too subtle for the human eye to detect.

“Autistic individuals use the same basic facial movements to express emotions, but their intensity often falls outside the culturally familiar range that most people recognize,” said Elizabeth Torres, a psychology professor at the Rutgers–New Brunswick School of Arts and Sciences. “This disconnect can lead to missed social cues, causing others to overlook or misinterpret their emotions.”

The researchers said individuals on the autism spectrum – especially those who cannot speak or require significant support for movement – also may have more unpredictable and varied facial expressions, making it more difficult for doctors and caregivers to recognize their emotional cues. As a result, some may mistakenly assume these individuals aren't trying to communicate at all.

“But that’s not the case,” said Torres, a computational neuroscientist with more than 17 years of experience working with individuals with autism. “Their emotions and social signals are there – we just haven’t been able to see them properly. This research could help bridge that gap, fostering a better understanding between autistic and nonautistic individuals.”

This unintentional disconnect can contribute to social isolation and misunderstandings about autistic behavior, she added.

The study, led by Torres and her team at the Rutgers Sensory Motor Integration Lab, used a novel data type she developed called micromovement spikes. This method captures microscopic facial movements using statistical techniques developed by Torres and nonlinear dynamics methods developed by Theodoros Bermperidis, a postdoctoral associate.

By recording short, five-to-six-second videos on smartphones or tablets, researchers tracked facial micromovements that typically go unnoticed.

“We wanted to investigate whether brief microexpressions appeared during common emotional expressions, like smiling or showing surprise,” Torres said. “Our goal was to uncover what was really happening beneath the surface when expressions go unnoticed.”

The research team developed an app to guide participants through four stages: practicing video capture, recording a resting face, smiling and making a surprised face. Data was collected in various settings, including schools, therapy gyms and social events, with some participants submitting videos from home.

The study analyzed data from 126 participants, including 55 nonspeaking individuals who communicate by typing. Researchers found that while there were differences in facial micro-movements between autistic and neurotypical individuals – varying by age and sex – the facial muscles responsible for emotional expression were active in both groups.

Torres noted the key difference was in the intensity of these expressions.

"The challenge isn’t a lack of expression – it’s that their intensity falls outside what neurotypical individuals are accustomed to perceiving,” she said. “This means we are quite literally missing each other’s social cues.”

The implications of this research are far-reaching, said Torres, who also created a mobile application to screen, diagnose and track nervous system disorders. As the chief scientific officer of NeuroInversa LLC, a Rutgers spinoff company co-founded with Chris Dudick, she works on using technology to monitor treatment effectiveness over time.

She said this study challenges common misconceptions about autism and introduces a scalable method for understanding social interactions in autistic individuals.

“This research gives us a powerful tool to expand autism studies beyond simply detecting differences,” Torres said. “Now, we can work toward bridging the gap—helping neurotypical individuals recognize different expressions of emotion and fostering better social understanding.”

The researchers said their findings could lead to improved diagnostic methods and new ways to support communication between autistic and nonautistic individuals.

By using accessible tools such as smartphone cameras powered by AI, this study paves the way for more inclusive and real-world autism research, Torres added.

Study co-authors included Bermperidis; former and current doctoral students Richa Rai and Joe Vero; and Neel Drain, a student at Robert Wood Johnson Medical School.

The research was funded by the Nancy Lurie Marks Family Foundation.

Explore more of the ways Rutgers research is shaping the future.