Nameability Supports Rule-based Category Learning in Children and Adults

Abstract

Learning abstract, rule-based categories is crucial to children’s development, from learning how to maneuver through the world (e.g., knowing to stop at a red light) to abstract reasoning (e.g., learning that a three-sided shape is a triangle). How do children learn rule-based categories? Past work with adults suggests that verbal labels may support novel category induction by providing compact hypotheses about category-relevant features. We asked whether children also make use of accessible labels to guide category learning. We tested children’s (N = 97; 4 - 6 years of age) and adults’ (N = 90) ability to learn a novel category in which the colors that perfectly predicted category membership were either easy to name (e.g., “red” and “brown”) or more difficult to name (e.g., “mauve” and “chartreuse”). Both children and adults showed better learning when the categories were composed of more nameable color features compared to less nameable color features, though the effect of nameability on category learning was stronger for adults than children. Children’s success in learning categories with difficult-to-name features was related to knowledge of difficult-to-name colors. Together, these results provide evidence that rule-based category learning may be supported by the emerging ability to form verbal hypotheses about category membership (e.g., “the red ones belong to category A”).

Nameability supports rule‐based category learning in children and adults

“…However, it was not clear that these norms would be representative of children's perceptual discriminability judgments. To obtain a behavioral measure of the discriminability of the color features among children in the current age range, we conducted a norming study in which we asked a separate sample of 3‐ to 6‐year‐old children, as well as a separate sample of adults, to make judgments about color pairs in a speeded match‐to‐sample task (Zettersten et al, 2020). Forty 3‐ to 6‐year‐old children in the Midwestern United States ( M age = 56.4 months, SD = 6.5; range: 45–69 months; 16 female, 22 male, 2 not reported; 87.5% White, 5% more than one race, 5% Asian, 2.5% did not disclose; tested in 2023) participated in the task.…”