Knowledge and source type influence children’s skepticism of misinformation

ABSTRACT

Three-year-olds often respond to lies as if they were true or with no clear rationale. Individual differences influence children’s processing of misinformation. Here, we explore how two contextual cues (children’s conflicting first-hand knowledge and different information sources) affect their ability to correctly interpret and respond to misinformation. 133 three-year-olds from the northeastern United States searched for a hidden object after it had traveled down one of two criss-crossing tubes. Children played this game by themselves (baseline), with a deceptive human (intentionally incorrect about the location of the object), a broken human (unintentionally incorrect), or a broken machine (incapable of intentions and incorrect). Half of the children could see the object as it traveled down the tubes (knowledgeable group), and the other half could not (no-knowledge group). With this first-hand knowledge, children were more likely to search correctly when playing by themselves. Information source further impacted children’s responses. Responding to a deceptive human was the most challenging: knowledgeable children searched randomly; unknowledgeable children inappropriately deferred to the deceptive experimenter. But when knowledgeable children played with either the broken human or the broken machine, they had as many correct searches as when they played alone. Context may have its limits when it comes to influencing children’s responses to misinformation. Specifically, although first-hand knowledge appears to help 3-year-olds correctly respond to some forms of incorrect information, it may not be enough to help them overcome misinformation from explicitly deceptive sources.

Acknowledgments

We thank the children, caregivers, and preschools who participated in these studies. Thanks to Sarah Buhl and the many undergraduate students who helped create stimuli, collect, and code the data.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

The data that support the findings of this study are available from the corresponding author, upon reasonable request.

Open scholarship

This article has earned the Center for Open Science badge for Open Data. The data are openly accessible at https://doi.org/10.17605/OSF.IO/N3F2U

Notes

¹ These data were collected before and during the worldwide COVID-19 pandemic. As did most scientists around the world, we modified our procedures slightly to accommodate social distancing requirements. In Study 1, that meant that 11 children participated on Zoom using prerecorded videos. Those same videos were used with 9 additional children when in-person testing resumed, but masking requirements were still in place. This minimized the need for talking, and also provided a mask-less experience with the stimuli, which we felt was more important than live interaction. In Study 2, children in the deceptive human and broken machine conditions participated in person, but children in the broken human condition participated on Zoom using prerecorded videos. We did not find differences in performance on the search task when data were categorized by procedural changes. Finally, only children in the deceptive human and broken machine conditions participated in the ToM task. We justified eliminating this task in some conditions because first, it allowed us to shorten the online broken human procedure and second, we had not found any indication of ToM influencing children’s performance in the task in previous conditions. We recognize the limitations of having the procedure modified, however we are confident in the pattern of results we found after statistically accounting for these modifications.

² Revise anonymous OSF link to the following: https://doi.org/10.17605/OSF.IO/N3F2U

³ We chose to have the deceptive human receive a star when E1 searched incorrectly in order to provide a reason for her incorrect information (i.e., she is competing for stars). The broken human and broken machine did not receive stars when E1 searched incorrectly because their “brokenness” served as the reason for their incorrect information.