Fig. 2.

Knowledge similarity. a Responses of Children L and M are coded in Interviews 1 (L1 and M1) and 2 (L2 and M2). To assess knowledge similarity, coding of each trait in each channel is compared. In the example, Child L’s and Child M’s Interviews 1 (L1M1) are compared. A trait is said to match if it was given the exact same code in both cases (both children or both interviews) in at least one channel. b Matching of traits within dyads. Children L and their peers matched in significantly more traits after (L2M2) than before (L1M1) interaction. Particularly, Response Adoption (i.e., modification of children’s responses to resemble those of their partners) was significant for Children L (L2M1–L1M1), whereas not for Children M (L1M2–L1M1). LxMy: number of traits matching between Child L in Interview x and Child M in Interview y, with x/y: 1 or 2. c Knowledge shifts led by Response Adoption. Adopted responses are defined as traits matching between a child’s Interview 2 and her partner’s Interview 1, whereas not between them before interaction in Interview 1. Here, adopted responses were divided into groups of positive, negative, or no knowledge level shift. Responses adopted from their partners led to significantly more knowledge gains than either losses or no changes in Children L. (*p < 0.05, **p < 0.01, ***p < 0.001, ns: non-significant. Error bars represent SEM, and circles are individual data points)