Children struggle to identify the defining properties of shapes through elementary school (Satlow & Newcombe, 1998). Perhaps preschoolers struggle to learn these properties because they have little conception of what constitutes relevant shape knowledge or how to go about acquiring it. We outline empirical support for the idea that children’s questions show us not only what a child knows in a given domain, but how that child is approaching the act of knowledge acquisition itself. In this way, studying children’s questions provides information that cannot be gleaned from studies which simply test children’s domain–specific knowledge. To illustrate this point, we examine existing research into children’s questions about animals, to show how this idea applies in relation to an extensive body of research on children’s questioning behavior. Second, we review our own study of children’s questions about shapes, to see how preschoolers’ approach to knowledge acquisition differs in a domain composed of pure nominal kinds rather than natural kinds. Our results suggest that, in contrast to the animal domain, three–year–olds demonstrate very little awareness of relevant information in the shape domain. We conclude by considering why this might be the case and suggesting some ways it could be rectified.