Holding something soft during a frightening movie might seem like a reliable route to comfort. But what happens when the object in someone’s arms starts breathing faster, as though it is frightened too? An experiment with a furry robot suggests that this small physical change can intensify the holder’s bodily response.
The findings add a qualification to research on comforting touch. Participants who held an accelerating robot showed a larger heart-rate increase than those holding a steadily moving or motionless version. The clearest evidence concerned physiological arousal, while later reports of fear showed no significant difference between groups.
Psychologists and robotics researchers from the University of Amsterdam and the University of British Columbia describe the work in Emotion. Led by Zachary Witkower, the team examined a form of communication that social robots can reproduce without words or facial expressions: the movement of a breathing body.
A pet-sized robot with a moving ribcage
The researchers built a robot roughly the size of a small house cat, covered in soft fur. Beneath that surface, a motor moved a plastic ribcage and flexible belly. The mechanism let them control the apparent breathing pattern while keeping the object’s appearance similar across conditions. The pet-like design made close holding more feasible without trying to reproduce the appearance and social expectations of a human partner.
Participants hugged it against their chests, with one hand underneath and the other on top. This created close contact with the moving body, delivering haptic feedback through pressure and motion. A finger sensor recorded their heartbeats while they watched a computer screen. Participants were asked to keep watching the screen and avoid extra movements that could interfere with the finger sensor’s readings.
The final analysis included 103 UBC undergraduates, averaging about 20.6 years old. Researchers recruited 107 students but excluded four because a technical error prevented matching their heart-rate and questionnaire records. Each participant experienced one randomly assigned robot condition, rather than comparing all three.
One group held a robot that remained still; another felt steady movement at about 14 breaths per minute. In the third, the robot began steadily, then accelerated toward 30 breaths per minute during the frightening clip. This touch-based interface allowed the team to vary the rhythm without introducing a frightened human partner.
The heartbeat difference appeared around the scare
Everyone watched the same sequence, beginning with an uneventful video of a snail crossing a plank. Later came an 84-second excerpt from The Shining, chosen because previous research had validated its ability to elicit fear. Headphones supplied the soundtrack and reduced disruption from incidental mechanical sounds.
The main comparison examined heart activity in the 20 seconds immediately before and after the scary excerpt. In the accelerating-breath group, average heart rate rose from 82.0 to 88.6 beats per minute. The robot’s movement remained fast around that post-film measurement period before gradually slowing. Follow-up analyses excluding students who recognized the movie excerpt left 79 participants and produced a similar pattern, strengthening confidence that film familiarity did not explain the result.
The motionless-robot group showed a smaller increase, from 80.7 to 82.7 beats per minute. The steady-breathing group changed from 80.3 to 81.8, an increase that was not statistically significant. The accelerating group’s change was significantly larger than either comparison group’s.
That pattern suggests a particular movement rhythm affected the stress response, beyond the general experience of hugging a moving object. However, the calm and motionless groups did not differ significantly from each other. The experiment therefore does not establish that steady robotic breathing offers extra protection over holding the same plush object still.
Recognizing fear differs from feeling it
Participants also evaluated the robot afterward, judging the fast-breathing version as more afraid than either alternative. That supports the idea that the movement conveyed an emotional signal. It does not mean the machine experienced fear, only that people interpreted its behavior that way.
The distinction matters when discussing emotional contagion, the spread of emotion between individuals. This study connects perceived fearful behavior with an increased bodily response in a fear-inducing setting. It does not show that the robot alone frightened otherwise relaxed participants.
Nor did every measure tell the same story. Questionnaires completed up to about five minutes after the interaction found no significant group differences in current fear, negative emotion or reported activation. A person’s emotional response includes subjective experience as well as measurable bodily changes, and those outcomes need separate consideration.
Heart rates had returned toward baseline by the interaction’s final moments. The delayed questionnaires cannot establish whether subjective fear differed at the precise moment of peak arousal. They do show that the experiment did not detect a lasting group difference in reported fear at that later assessment.
What a robotic hug can reveal about people
“If the person you’re holding is scared, their body might give you signals that increase your own fear,” Witkower said. The robot offered a controlled way to investigate one possible signal. Its programmed movements removed the many changing expressions and behaviors that would accompany a frightened human partner.
Real physical contact involves more than a moving chest, though. Grip, warmth, familiarity and expectations can shape an encounter. Those factors limit how directly the laboratory result can predict what happens when someone holds a partner, child or pet.
The research also leaves the immediate biological mechanism unresolved. Participants’ own breathing was not recorded, so the team could not determine whether their respiration followed the robot’s rhythm. Future work could measure respiration alongside heart rate and examine how expectations or perceived threat contribute.
The sample consisted of university students, and data collection took place in 2017 and 2018. The study was not preregistered, although the authors report setting their hypotheses before collecting data. Extending the work to children, older adults and people with anxiety disorders would help establish its relevance to broader social support experiences.
Designing machines that shape how interaction feels
For technology designers, the result suggests that simulated bodily rhythms deserve attention alongside screens, sounds and facial expressions. A robot’s physical behavior can affect someone holding it, even when the device has little resemblance to a person. Softness alone does not determine the interaction’s effect. A movement felt against the chest can provide another channel while someone’s eyes and ears remain directed toward a film.
One possible application is more evocative virtual reality, interactive films or games. Expanding and contracting devices could add a bodily cue to an emotionally charged scene. The study raises that possibility without testing whether it improves entertainment or learning.
Researchers also envision wearable devices that convey rhythmic pressure, and future calming tools might use related principles. Those therapeutic possibilities require their own trials. This experiment did not demonstrate treatment for anxiety or establish an effective breathing prescription.
Its contribution is more specific: during a frightening scene, an apparently frightened object in someone’s arms amplified their heart-rate response. That finding helps explain why physical reassurance depends on what the contact communicates. A hug can carry information as well as comfort.
Dig deeper into touch and emotional responses
These studies examine how touch communicates emotion, regulates threat responses and informs the design of calming devices.
- A systematic review and multivariate meta-analysis of the physical and mental health benefits of touch interventions: Synthesizes evidence on touch interventions and the conditions associated with benefits, providing broader context for the comforting effects that the robot study qualifies. (Nature Human Behaviour, April 2024)
- A calming hug: Design and validation of a tactile aid to ease anxiety: Develops a breathing cushion and tests its effects on students anticipating a mathematics test, offering a separate investigation of calming tactile technology. (PLOS ONE, March 2022)
- Affect contagion between mothers and infants: Examining valence and touch: Examines how mothers’ affective states relate to infants’ physiological responses and how physical contact contributes to stress transmission. (Journal of Experimental Psychology: General, May 2017)
- The communication of emotion via touch: Tests whether people can convey distinct emotions to unfamiliar partners through touch, establishing context for studying nonvisual emotional communication. (Emotion, August 2009)
- Lending a Hand: Social Regulation of the Neural Response to Threat: Investigates how holding a partner’s hand changes neural responses to anticipated threat, illustrating the protective side of interpersonal contact. (Psychological Science, December 2006)
Research findings are available online in the journal Emotion.
