To Evaluate Testimony, Children Track Where it Originated, and What Others Believe About its Source

• In Event: Reasoning About Epistemic Processes: Children’s Understanding of how Knowledge is Acquired, Transmitted and Socially Constructed

Abstract

When receiving information from others, we must decide whom and what to believe. Even preschoolers are discerning learners, considering whether informants were previously accurate, or had perceptual access to relevant information. But our informants don’t always acquire their knowledge firsthand, and often, those around us may know more about their track record than we do. To evaluate testimony, we must sometimes look past our direct informants, considering: 1) where their information came from, and 2) what others believe about them.

Investigating the first capacity, three preregistered experiments (N=360) investigate whether four- to seven-year-olds consider informants’ sources, testing when children begin favoring well-sourced claims over those that are simply more widely-repeated. See Fig 1 for the procedure and results. Across all experiments, participants observed individuals disagree; when asked to justify their claims, one side always referenced three independent eyewitness sources, and the other referenced only one. After seeing this, participants were asked to endorse one side’s testimony.

Importantly, each experiment manipulated the number of individuals who initially disagreed. In Experiment 1, two agents initially disagreed; by age five, children endorsed the claim supported by more primary sources. In Experiment 2, each claim was repeated by multiple intermediaries (something even adults can struggle to reason about; Yousif et al., 2019). Six-year-olds endorsed the side with more primary sources, while five-year-olds were at chance, suggesting they may prioritize source information only in the absence of widespread repetition. Finally, Experiment 3 equated the number of overall individuals endorsing each claim, pitting three individuals who heard from one primary source against a single dissenter who actually heard from three primary sources. Again, by age six—but not before—children endorsed the testimony of the individual with multiple primary sources, suggesting that older children did not solve Experiments 1-2 by simply siding with the more numerous group.

Together, this work suggests young children already understand the complex social processes that construct the knowledge of those around them—but may only converge on a stable strategy to evaluate repeated testimony by age six, after considerable social experience. The intermediate developmental trajectory of five-year-olds, however, raises an intriguing possibility: maybe young children expect agents to repeat testimony they think is true. If so, widespread repetition of a claim could signal others believe an agent is particularly knowledgeable or trustworthy.

An ongoing preregistered study (N=95 of planned 108 four- to six-year-olds) begins investigating this capacity, testing if young children decide whom to trust by observing whether others believe a source. See Fig 2 for the procedure and results. A third party accepted one agent’s testimony, and subtly questioned the other’s; by age five (but not before), participants judged that the agent whose testimony was challenged was likely wrong. This work provides some first hints that children may not only evaluate informants on the basis of their own experience—they may also notice and consider what others think about them.

Authors

• Rosie Aboody, Harvard University
  Presenting Author

• Sami Yousif, Yale University

• Sehrang Joo, Yale University

• Mark Sheskin, Minerva University

• Frank C Keil, Yale University