The Role of Knowledge and Explanation in Attenuating Children’s Expectations of Category Homogeneity

• In Event: 3-079 - Boundaries for the biases in reasoning about categories

Integrative Statement

The ability to make inferences from categories relies on our assumption of category homogeneity: the belief that individuals of the same category share key properties (Gelman, 2003). However, category members are not homogeneous across all properties and vary in important ways (e.g., all snakes lack legs, but only some are poisonous). Effective induction requires knowing when to expect homogeneity versus variability.

Previous work exploring children’s sensitivity to this balance between homogeneity and variability shows that young children display a reliable bias to think about categories as homogeneous (Brandone, 2017; Rhodes & Brickman, 2010), generalizing properties of an individual (e.g., this dax has wings) broadly to all category members (e.g., all daxes have wings)—but this bias decreases with age. Additionally, children as young as 4 show some sensitivity to factors limiting generalizability, tempering their generalizations by category domain and property type (Brandone, 2017; Gelman, 1988).

The present studies aimed to extend previous work by exploring two additional factors that may attenuate the homogeneity bias: (1) knowledge of real-world variability, and (2) explanatory information. These studies used an induction task in which participants predicted the distribution of a property among category members (e.g., whether “all”, “lots”, or “just some” fossas are brown) after learning about a property true of an individual (e.g., “this fossa is brown”).

In Study 1, we examined whether 4- to 6-year-olds’ (N = 110) and adults’ (N = 26) knowledge of diversity in familiar animal categories corresponds to an expectation for diversity in unfamiliar animal categories. As children include more diverse members in their representations of categories (e.g., penguins as a kind of bird), they may infer that categories generally consist of diverse members, thus, attenuating their expectation of homogeneity in unfamiliar categories. Across 24 trials, participants learned about a property true of one member of a familiar or unfamiliar animal category and then estimated that property’s prevalence among the whole category; real-world property prevalence varied across trials to test knowledge of variability within familiar categories. Results suggest that by age 4, children recognized some variability within familiar categories, but the youngest children acknowledged less variability than older participants. Additionally, regardless of age, children with greater knowledge of variability in familiar categories expected greater diversity in unfamiliar categories (Figure 1).

In Study 2, we examined whether 4- to 6-year-olds (N = 132) shift their expectations of homogeneity based on explanatory information provided by others. To test this, we varied the explanations children heard: some heard a “category membership” explanation (e.g., “this animal is brown because it’s a fossa), some heard a “unique” explanation (e.g., “this fossa is special because it’s brown”), whereas those in baseline heard no explanation (e.g., “this fossa is brown). As expected, compared to baseline, children in the category membership condition made broader generalizations (e.g., were more likely to expect “all” members to have the property) whereas those in the unique condition made narrower ones (Figure 2).

Implications for the role of knowledge and explanation in attenuating children’s homogeneity bias will be discussed.

Authors

• Kelsey Moty, New York University
  Presenting Author

• Amanda C Brandone, Lehigh University
  Non-Presenting Author