This chapter summarizes the main conclusions from the discussion. Predictive models are flexible, and can serve many roles in cognition. They can do this without apparently using the structures that cognitive scientists have generally assumed must be needed, and without the kind of grounding that many have thought necessary. Prediction as a framework explains the great flexibility of thought, in which people freely intermix ideas from different domains in their thinking.The central role of imitation in the model suggests an unflattering view of human nature, but one that accords with the actual prevalence of ideas, like conspiracy theories, that seem utterly ungrounded. But key questions of mechanism remain, leaving open the possibility of emergence of structure in systems that have little structure built in.