Our goal is to understand the cognitive processes involved in tool-innovation. We look at a group whose cognitive processes are still in development and who find it remarkably difficult to innovate tools: human children. In our pioneering work on children's tool-innovation [http://dx.doi.org/10.1016/j.cognition.2011.01.003] we gave children a simple physical problem: a bucket containing a sticker needed to be retrieved from the bottom of a tall, transparent, vertical tube. Crows have solved variants of this task by bending wire in to a hook to retrieve the bucket. We expected children to easily fashion a hook from a straight pipe cleaner to solve the task, given their expertise using pre-made tools in everyday life. To our surprise 3- to 5-year-olds rarely innovated a hook, and it was only at 8 years that the majority of children succeeded. We confirmed the generalisability of our finding and now ask: Why is tool-innovation so difficult for human children?