People become more suspicious of a humanoid robot that commits errors — especially when the robot presents itself as an expressive conversation partner, according to a new study published in the journal Science Robotics.

Researchers recruited 50 participants to hold conversations and make joint decisions with Pepper, a commercial humanoid robot designed to be expressive and capable of recognizing emotions. In some sessions, Pepper offered sound advice. In others, it made conversational mistakes — interrupting participants or pushing illogical suggestions.

For some participants, the robot was fully animated, using gestures, eye contact, and nods. For others, it remained motionless. The research team measured four variables: brain activity, levels of the hormone oxytocin, self-reported trust, and the robot's observed influence on participants' decisions.

The researchers found that when people interacted with an expressive robot that violated interaction norms, their oxytocin levels increased. Oxytocin is popularly known as the "love hormone" for its role in social bonding, so the straightforward prediction would be that it declines when a partner disappoints you. Instead, the opposite occurred: the higher a person's oxytocin rose during an expressive robot's errors, the less they trusted the robot and the less frequently they took its advice. The hormone was tracking suspicion, not affection. This finding aligns with a growing body of work showing that oxytocin's effects depend on context, uncertainty, and perceived threat rather than uniformly promoting bonding.

Errors damaged trust and diminished the robot's influence regardless of whether it was expressive. What expressiveness changed was how participants' brains processed the moment.

Reading someone's brain during a real conversation is difficult because the conventional method requires lying motionless inside an MRI scanner. Instead, the team used functional near-infrared spectroscopy (fNIRS), a portable sensor worn on the forehead that tracks oxygen levels in the brain while people move and talk normally. Other teams are now using similar wearable brain imaging systems to study social cognition in natural encounters between people — research that is not possible when subjects are confined to MRI scanners.

The researchers closely monitored two brain regions: the dorsolateral prefrontal cortex, which monitors uncertainty and flags when expectations or norms are broken, and the medial prefrontal cortex, which supports "mentalizing" — the everyday cognitive work of inferring what another party intends.

When an animated robot erred, participants appeared caught off guard and had to work harder to make sense of an awkward social situation. Activity rose in both brain regions, and the two began working together more closely. That coordinated activity predicted the rise in oxytocin levels, which in turn predicted falling trust and reduced influence on participants' behavior. In contrast, this coordinated brain activity was absent in participants who interacted with expressionless robots.

The findings carry weight as robots move into homes, hospitals, and workplaces, where trust determines whether people use them at all. A common design assumption has been that lifelike, socially expressive robots earn more trust, which protects a robot's reputation even when it makes mistakes. However, research is beginning to show that this assumption is faulty. The study demonstrates that expressive cues appear to shift how people perceive a mistake — out of the category of technical malfunction and into the category of social violation, similar to those that occur between people. A motionless robot's error looks mechanical, while the same error from an animated robot engages the social judgment machinery people use to evaluate one another.

This framing reflects how researchers increasingly treat trust as a multilevel phenomenon — spanning individuals, relationships, networks of people, and societies — rather than a single attitude. Much of the existing research on oxytocin involves humans interacting with humans, where the hormone is tied to bonding, though context and perceived threat can flip those effects.

All participants in this study were young men, and the researchers used a single robot design. A key next step is testing whether the same oxytocin-linked vigilance appears in women, mixed groups, other cultures, and other robot designs. The brain sensor also reached only the front of the brain, leaving deeper regions involved in social processing unmeasured. The team also plans to examine whether robots can repair trust after a mistake by acknowledging the error, apologizing, or signaling good intent — the way people do after awkward or uncomfortable interactions.

The study authors are Hasan Ayaz, Professor of Biomedical Engineering, Science and Health Systems, Drexel University; Ewart J. de Visser, Technical Director, Warfighter Effectiveness Research Center, United States Air Force Academy; Frank Krueger, Professor of Systems Social Neuroscience, George Mason University; and Yigit Topoglu, Research Scientist, Warfighter Effectiveness Research Center, United States Air Force Academy. This article is republished from The Conversation under a Creative Commons license.