Changes in Beliefs About Category Homogeneity and Variability Across Childhood

Abstract

Effective category-based induction requires understanding that categories include both fundamental similarities between members and important variation. This article explores 4- to 11-year-olds' (n = 207) and adults' (n = 49) intuitions about this balance between within-category homogeneity and variability using a novel induction task in which participants predict the distribution of a property among members of a novel category. Across childhood, children learned to recognize variability within categories-showing increasing sensitivity to the role of property type and domain in constraining inferences. Children below the age of 6 showed evidence for a domain-general assumption that categories are homogeneous-generalizing properties broadly to 100% of category members. These studies support important developmental changes in category representations that may influence category-based induction.

Generalizing About Striking Properties: Do Glippets Love to Play With Fire?

“…Other directions for future research would be to establish whether property extension was influenced not only by what evidence was presented to the children, but also by how evidence was presented, as a recent study (Lawson, 2017) suggests that presenting evidence exemplars at the same time (i.e., simultaneous presentation) or one by one (i.e., sequential presentation) influenced property projections. Moreover, it would be worthwhile to explore how our items relate to a finding in the literature that relates to children’s and adults’ intuitions about the balance between within-category homogeneity and variability in induction (Brandone, 2017). For example, individual cats may differ considerably in color but also on other transient properties including being asleep or sick.…”

“…For example, individual cats may differ considerably in color but also on other transient properties including being asleep or sick. Brandone (2017) suggests that children begin to differentiate their inference making based on conceptual knowledge about the nature of the property they are asked to generalize and the category domain by 6–8 years age. Thus, further experiments with older children could reveal how children’s inference making about novel kinds is influenced by such factors.…”

FN400 amplitudes reveal the differentiation of semantic inferences within natural vs. artificial domains

“…The categories in the N domain include naturally occurring objects ( e.g ., animal), and the categories in the A domain include produced or manufactured objects ( e.g ., tool) 4 , 5 . Researchers suggest that the N and A domains are distinct in the nature of their ontology 6 – 11 . The categories in the N domain are generally tightly structured, coherent, and share many similarities ( e.g ., animals have similar internal parts, external structure, and behaviors).…”

“…Many previous studies have examined category-based inferences in the N domain; relatively fewer studies include stimuli in both N and A domains during category-based inference tasks. Several studies have shown that knowledge about the nature of a category, including its ontological domain, such as N and A categories, influences inferences 6 – 8 , 11 . For example, in a study by Coley et al .…”

“… 8 , only category coherence expectations were found to predict inductive inferences for the N domain, but both knowledge and category coherence expectations predicted inductive inferences for the A domain. Moreover, several studies have found that adults drew more inferences within the N than within the A domain 6 – 10 . Researchers suggested that the varied patterns of inferences between the N and A domains can be explained by the differences in the categorical coherence and homogeneity between N and A domains (ontological distinction account) 6 – 8 , 10 , 12 .…”