This chapter clarifies the nature and semantic properties of metacognition in non-humans. As is the case for every philosophical inquiry concerning animal minds, studying animal metacognition should provide new perspectives on the structure of mental content and on mental activity in general. This exploration proceeds in four steps. First, the experimental evidence for animal metacognition is presented and the difficulties of a metarepresentational view on metacognition summarized. Second, the possibility of alternative, non-propositional semantic structures is discussed. Third, a specific representational format that might be sufficient for animal metacognition to develop is examined. Finally, some objections are addressed, and further developments of the proposal are considered. The question of whether human metacognition uses a separate feature-placing format, or is absorbed within the propositional mode of thinking, is still open, and current research may soon come up with interesting new constraints.