Mothers optimally sample information for their child and children learn best from maternal sampling

According to recent computational approaches, when children are presented with information by knowledgeable others, children can make the pedagogical inference that their partner has chosen the best possible data in order for them to learn from (Bonawitz &Shafto, 2016, 2017). But do caregivers –the child’s first teachers –really choose the best possible data for their child to learn? The current study examined the extent to which caregivers (and their children)sample information to fill gaps in children’s knowledge of object-label associations. Furthermore, we examined the repercussions of such pedagogical sampling in terms of children’s retention of labels for objects they elicited as opposed to labels elicited by their mothers. The results suggest that mothers are worthy of the pedagogical assumption in that they not only choose information tailored to fill their child’s knowledge gaps, but that children also appear to learn best when information is specifically elicited by their caregivers. In contrast, children did not sample information that fills gaps in their knowledge of the object-label associations presented. Our findings speak to the pedagogical role of caregivers in interactions with their childrenand the power of social learning in early childhood.