14 hours ago
A long face is not the only sign that someone is down in the dumps. How people walk is revealing too, particularly the swing of the arms and legs, researchers say.
Scientists asked volunteers to guess people’s emotions from video clips of them walking and found that bigger swings portrayed more aggression while smaller swings implied fear and sadness.
Tweaking the videos to make the swings longer or shorter made the emotions easier to infer, according to the study, suggesting that the coordinated swing of the arms and legs was a key feature people picked up on.
The work expands the list of cues that humans draw on to make speedy assessments of people’s emotions, and highlights the specific movements that convey most about the range of feelings people have, the scientists say.
“Walking is one of the most familiar and well-practiced whole-body movements for humans,” said Mina Wakabayashi, a researcher at the Advanced Telecommunications Research Institute International in Kyoto, Japan, and the lead author of the study. “Because of this, changes in emotional state may naturally appear in the way we walk.
“In our results, movements with larger arm and leg swings were more likely to be perceived as angry, whereas movements with smaller swings were more likely to be perceived as sad or fearful.”
For the study, the scientists asked actors to recall life events that provoked anger, happiness, fear or sadness and then walk a short distance while dwelling on each memory. The participants wore tight clothing and reflective markers, allowing the researchers to create point-light videos from the side and front which captured their gait without their facial expressions and other bodily cues.
Volunteers then watched the videos and declared which emotion each gait evoked. They recognised all of the portrayed emotions at better than the level of chance, the study found. “To some degree, the walkers’ intended emotions were indeed perceived by the observers,” the researchers write in Royal Society Open Science.
A second experiment delved into the particular movements that betrayed the walkers’ emotions. To do this, the researchers took gait videos from people expressing neutral emotions and manipulated the clips to exaggerate or dampen down the arm and leg swinging. Again, the observers saw more pronounced swinging as aggressive, with less seen as sad or fearful.
Emotions are likely to influence all manner of other ways in which people move their bodies, and the Kyoto team hope to explore these in future work.
“Being able to infer emotions from body movement may help us understand others quickly during social interactions, even without words,” Wakabayashi said. That could mean spotting people’s emotions from afar and changing how we approach them, depending on whether they seem angry or sad, she added.
There are also potential applications of the work. If scientists can reliably predict people’s emotions from their movements, it could help identify vulnerable or threatening people in CCTV footage, or lead to wearable devices that monitor people’s mental states.
Researchers in Texas showed last month that a machine-learning algorithm could predict anger, sadness, joy and fear from a person’s gait, though with limited accuracy. One potential advantage, they say, is that it may be harder to fake a gait than speech or facial expressions.
Dr Gu Eon Kang, a bioengineer at the University of Texas at Dallas and a co-author on machine-learning study, said another potential application was an “AI-based virtual aid” that was able to interpret a person’s emotions from their gait and respond accordingly.
