In learning about the world, children employ a critical developmental tool: carving up the world into categories that allow for efficient learning and inference. But this process presents an inductive challenge for young children, as they must continually assess whether novel information is idiosyncratic, episodic, or superficial, or whether it represents important, generalizable information about a category. In this talk I discuss a series experiments investigating one way in which children might tackle this inductive challenge: attending to whether novel information is being intentionally communicated for their benefit. Building on theoretical and empirical work showing an early sensitivity to intentional communication in infancy (Csibra, 2010; Csibra & Gergely, 2009), I will present a series of experiments testing when and how this sensitivity plays a role in older children's active inductive inference processes as they construct a conceptual understanding of the world. I will propose that during the preschool years children build on this foundation of early sensitivity to intentional communication, developing a nuanced, social-pragmatic learning mechanism that helps them judiciously evaluate the generalizability and importance of novel information.