Incorporating insights from category learning and from real-world structure into racial/ethnic bias reduction approaches in early childhood

We expand on the broadly accepted view that children’s racial biases emerge from being exposed to racial inequalities in their environment by documenting one aspect of real-word structure that is likely to contribute to the development of racial biases – children’s books. In Study 1 we employ meta-analytic methodology to document the frequency with which different racial groups are depicted in children’s books. In Study 2, we document the contexts surrounding the depiction of racial groups in children’s books. These studies showed that characters from minoritized racial backgrounds are not only numerically underrepresented in U.S. children’s books, but there are also meaningful differences in the themes that are more likely to be associated with different racial groups. From a category learning perspective, such input would result in decreases in representational similarity across racial groups and thus is likely to contribute to the emergence of racial biases in childhood. In Study 3, we provide the first analysis of the materials used in existing media-based racial bias reduction studies with children, documenting the frequency and themes in those materials. This analysis showed that existing studies often failed to represent characters from different racial/ethnic backgrounds in ways that would counteract the skew of the input described in Studies 1-2, suggesting that existing studies may not have been well-suited to support bias reduction at scale. We discuss the implications of these findings for theories of bias development and present a new mechanistic framework for increasing intergroup similarity in young children.