Reading focus
Syllabus: How Do We Learn From Other People?
Compare familiar-object judgments with novel-label choices, then check the whole-sample result against the subgroups defined by explicit reliability judgments.
Why you’re reading it. Treat source evaluation as something to measure separately from accepting new information.
A question to consider. Why does the association between recognizing the accurate speaker and choosing that speaker’s labels fail to show that explicit recognition causes selective trust?
Why this reading
Recognizing that a statement is wrong differs from using a speaker’s history when learning something new. This short experiment separates those achievements and provides an entry into the study of selective trust.
Before you read
Testimony is information communicated by another person. The children first encounter familiar objects whose names they know; they then hear competing names for unfamiliar objects. The first phase permits checking a claim against existing knowledge. The second requires evaluating a source when the answer cannot be checked directly.
Read for
- Trace how the speakers acquire histories of accurate and inaccurate labeling, and how speaker identity and position are controlled.
- Separate explicit judgments about reliability from subsequent choices of novel labels. Inspect both the whole sample and the subgroups defined by reliability judgments.
- Examine the discussion of memory, language conventions, and whether a speaker’s apparent accuracy should generalize beyond naming objects.
Consider afterwards
- A child identifies the previously accurate speaker but chooses the other speaker’s novel label. Which distinction does this reveal, and what follow-up would help explain the mismatch?
Editorial context
The study involved 53 children from Harvard child-care centers. Overall novel-label performance did not differ significantly from chance; selective trust appeared in the subgroup that correctly identified the reliable or unreliable speaker on both explicit judgments. Those groups were defined by measured performance, not random assignment, so their association does not establish that explicit reliability judgments cause selective learning. Koenig and Harris (2005) extend the investigation, while Tong and colleagues (2020) synthesize epistemic and social cues across studies. Neither the narrow sample nor this short task warrants a universal age threshold for trust.