Successful modal reasoning depends on WHAT went WHERE.

Modal reasoning requires quantifying over possibilities (e.g., categorizing events as necessary, possible, and impossible). This suite of abilities matures during the preschool years, but toddlers perform better in some modal reasoning tasks than others. We propose that these mixed findings reflect a cognitive distinction: toddlers perform worse when reasoning about ambiguous locations of objects (WHERE) than when reasoning about ambiguous kinds of objects (WHAT). In Experiments 1 and 2, we directly compared these forms of ambiguity using matched tasks. Consistent with the hypothesis, toddlers performed substantially better when reasoning about ambiguous kinds than ambiguous locations. Experiment 3 revealed a similar asymmetry in language comprehension, with preschool children showing better comprehension of modal terms such as "can" and "have to" in a What-based task than in a Where-based task. Thus, modal reasoning emerges earlier than previously suggested, and diversity in results reflects differences in how kind and location information are represented and processed.