It is well-documented that children do not begin school as blank slates but that they bring with them extensive knowledge about how the world around them works. This conceptual knowledge, embedded within rich theoretical structures, is not always accurate and requires change through learning and instruction. Yet some ideas – such as object motion – appear to be particularly resistant to such change. So how can conceptual change be achieved or facilitated? Collaboration, for one, has long been recognised as a (...) beneficial learning and teaching approach, including early science education. However, for deep-rooted ideas collaborating with others may not always have the desired impact. Instead, the notion of self-collaboration is considered in this review. The current state of research in the field of predictive and underlying knowledge in childhood is outlined and different models of how the knowledge systems relate to each other are discussed. While further work is still needed to establish a clearer picture of how self-collaboration might effect conceptual change, research to-date clearly identifies significant differences between predictive and underlying knowledge structures throughout childhood, how these structures can be related to traditional conceptual change theories, and how they may play a role in future learning and instructional approaches. (shrink)