Children remain polite to robots even when the machines speak to them in a commanding manner, Polish researchers have found. The finding challenges the assumption that children automatically imitate the communication style of machines.
A team led by Konrad Maj, PhD, at Warsaw’s SWPS university conducted a study involving 251 children aged 7-12 to examine how young people respond to different communication styles used by robots and whether the way a machine speaks influences their own behaviour toward it.
The researchers published their results in the journal „Computers & Education”.
Participants engaged in brief, controlled interactions with a 120-centimeter humanoid robot resembling a child. The robot operated under four conditions: a polite robot with a female persona, a polite robot with a male persona, a commanding robot with a female persona, and a commanding robot with a male persona.
Equipped with sensors, cameras and microphones, the robot could recognize speech, gestures and some emotional cues.
During the experiment, the robot imitated animals and responded to an experimenter’s attempt to take a photo. The robot either politely or firmly refused, and researchers observed the participants’ reactions. Children were also asked whether they believed the robot could be happy, dream or imagine things.
The study aimed to determine whether children would imitate the robot’s communicative style and whether age or gender influenced their tendency to anthropomorphise the robot, that is, attribute human-like qualities such as emotions, intentions or personality.
The results showed that in most cases, children addressed the robot politely, regardless of its tone. Seventy to 80% of participants used standard polite phrases, suggesting that established social norms had a stronger influence than behavioural mimicry.
The analysis also revealed developmental differences. Younger children, aged 7-9, were more likely to anthropomorphize the robot, treating it as a thinking and emotional being. Gender differences were also observed: girls, like the youngest group, attributed more human-like qualities to the machine, confirming previous findings about girls’ greater sensitivity to social and emotional aspects of interaction.
The communicative style and gender of the robot also influenced anthropomorphisation. The polite robot with a female persona elicited the highest average levels, although the differences were not pronounced.
SWPS University’s Konrad Maj said social robots are increasingly used in education worldwide.
“For example, they can adapt teaching methods to individual students’ needs, boost students’ motivation and engagement through fun, or provide immediate feedback. However, it is important to remember that interactions with robots also raise certain concerns, such as whether prolonged contact with them will negatively impact children’s social behaviour,” he said.
He added that understanding how children interact with robots is essential to determine in what situations the interactions are most beneficial and that robots working with children must have an appropriately adjusted communication style.
“A polite one promotes the machine’s personality and strengthens the sense of being a social partner, which can increase engagement in educational activities. However, excessive anthropomorphisation of robots can lead to undesirable educational consequences,” he said.
Interactions with robots are among the topics at the international HumanTech Summit 2025, organized by the HumanTech Center at SWPS University. The fourth edition of the conference will be held in Warsaw on November 20-22.
Speakers include Professor Emily Cross (ETH Zürich, Switzerland), an expert in the neurocognitive foundations of human-robot interaction; Jessica M. Szczuka, PhD, from the University of Duisburg-Essen, who investigates how digital technologies such as AI, chatbots, robots and VR are transforming intimacy, sexuality and interpersonal relationships; and Peng Liu, PhD, from Zhejiang University, China, whose research focuses on machine psychology and human-AI collaboration in contexts including automated vehicles, decision support systems and generative AI.
Katarzyna Czechowicz
kap/ zan/
tr. RL