Inferring children's categorizations from sequential touching behaviors: an analytical model

A child is assumed to belong to 1 of 2 classes: categorizer or noncategorizer. To determine which, 4 toy animals and 4 toy vehicles were randomly arrayed for touching for 2 min. The task was to infer whether the child was a categorizer or a noncategorizer for vehicles and similarly for animals. A model is proposed that assumes a child's sequence of touches follows one probability distribution if the child is a categorizer and another distribution if the child is a noncategorizer. The proportion of children in each category and the probability of a child being a categorizer for, say, vehicles are among the quantities that can be estimated. Data from 18-month-old children are illustrative. The model appears efficient and robust.