Project Summary
This project will examine how children’s beliefs about social categories develop using experimental studies and
model-based analysis. During development, children learn to think of people as members of categories: Teacher,
woman, Republican, Muslim, New Englander, and White are but a few examples. Children also form beliefs
about why members of these categories are the way they are: For example, children might believe boys and
girls play with different toys because they were born that way (what we call natural determination) or children
might believe that boys and girls were mostly taught those preferences (what we call social construction). The
purpose of this project is to understand what leads children to infer natural determination or social construction.
To advance understanding of how these beliefs develop, we need to answer three questions: (1) What are
children’s initial impressions of social categories? 2) How do children interpret what adults say about social
categories? (3) How do children revise their initial impressions in the presence of adult speech? To answer these
questions, we have three aims: Aim 1 will test novel social categories marked by extrinsic cues (e.g., clothing
and accessories) or intrinsic cues (e.g., skin color and face shape). We will measure children’s baseline beliefs
about categories marked by these two cues. We then need to know what children infer based on what adults
say about social categories. Aim 2 will explore generic language, which is language that makes generalizations
about category members: e.g., “boys like blue.” We will vary whether adults use generic language about
biological properties (“they are lactose intolerant”), cultural properties (“they believe the sun is their god”), or
some combination of both. Finally, we need to know how children revise their beliefs in response to adult speech.
Aim 3 will use model-based analysis to quantitatively describe how children update their baseline beliefs about
categories using their interpretation of evidence from language. The proposed research seeks to reconcile
evidence that children often infer natural determination and evidence that children flexibly consider multiple
hypothesis about social categories during development. Specifically, the project pursues an account in which
children’s probabilistic world knowledge leads them to infer natural determination more often than social
construction even though the underlying cognitive processes responsible for their inferences are formally
unbiased. Independent of this specific account, a computational model of how children’s beliefs about social
categories develop would substantially advance theory. It would also aid interventions aimed at reducing harmful
beliefs children can develop about social categories. To this end, we designed experiments that will generate
the data necessary to develop and compare computational models. The use of the online lab to reach families
from across the United States will maximize the generalizability of this project’s findings. By making this data
publicly accessible on Open Science Framework, other researchers can use the data for model-based analysis
too, which will maximize the impact of the proposed experiments.