When children ignore evidence in category-based induction. Irrational inferences?

The process of induction--generalizing information obtained from limited samples to inform broader understandings--plays a critical role in learning across the life span. Previous research on the development of induction has found important developmental changes in one critical component of induction--how children and adults evaluate whether a sample of evidence is informative about a broader category. In particular, when acquiring knowledge about biological kinds, adults view samples that provide diverse representation of a category (e.g. an eagle, a penguin, and a robin, for the category birds) as more informative than a less diverse sample (e.g. three robins) for drawing inferences about the kind. In contrast, children younger than 8 years often neglect this feature of sample composition, viewing both types of samples as equivalently informative. Is this a case of children making irrational inferences? This chapter examines how these findings can be reconciled with rational constructivist approaches to cognitive development, focusing on (1) the role of the sampling context in determining how learners incorporate information about sample composition into inductive inferences and (2) how developmental differences in learners' intuitive theories influence how they make sense of new evidence. This chapter highlights how strong tests of rational approaches come from incidences where children's performance appears to be quite nonnormative.