Do children really have a trust bias? Preschoolers reject labels from previously inaccurate robots but not inaccurate humans

AbstractPast research suggests that young children have a bias to believe what they are told so that they often trust an informant regardless of the informant’s previous accuracy. With the ubiquity of new technology, children regularly come in contact with non-human agents such as robots, yet little is known how children are trusting and thus willing to learn from these artificial beings. In our study, 3.5- to 5.5-year-old children (N=120) watched a single informant (either a robot NAO or a human adult) name familiar objects either accurately or inaccurately. The same informant subsequently tested children on their willingness to accept novel labels for novel objects provided. While children trusted the accurate robot and the accurate human to the same extent, they were less likely to accept information from the inaccurate robot than the inaccurate human. This suggests that preschoolers may not readily extend their “trust bias” to robots as informants.

Return to previous page