Social priming enhances young children’s trust in robots

This study examined preschoolers’ selective trust in a robot versus a human informant, and whether social priming may increase trust by making the robot’s social characteristics more salient, using the word-learning paradigm. Children also completed the Theory of Mind (ToM) scale, along with an adapted naïve biology task and an Attribution of Mental States questionnaire assessing their beliefs about robots. In Experiment 1, eighty-six Italian children (aged 3 and 5) observed a competent robot (NAO) correctly labeling familiar objects and an incompetent human mislabeling them. The two informants then provided novel labels for unfamiliar objects. Three-year-olds endorsed labels from both informants equally, while five-year-olds preferentially endorsed the labels provided by the robot. ToM related to trust only for judgment scores (which informant was correct), an association largely explained by age. Experiment 2 investigated whether exposure to social priming would increase children’s trust. Eighty-five Canadian children (aged 3 and 5) completed the same selective trust procedure, except that they first viewed a social priming video designed to increase the robot’s perceived social agency. Both age groups preferred learning new labels from the robot, suggesting that social characteristics support children’s trust in robot informants, particularly among younger preschoolers. ToM predicted children’s endorsement of the robot’s labels, although this effect was largely explained by age. Together, the findings suggest that, by age 5, preschoolers exhibit robust sensitivity to the epistemic characteristics of informants, regardless of their agency. Younger children, in contrast, appear more likely to trust a robot displaying salient social characteristics.