In this thesis I ask what the role of imagination is in our representation of the world and interactions with it. A standard answer to this question is that imagination has no direct role: imagination’s proper function is rather to allow us to disengage from reality; its motivating power, if it has any, is basically limited to children’s pretence. I argue that this standard answer is mistaken: imagination’s role is much larger than that. I consider a number of cases – including cases of superstitious and religious actions, or so-called ‘expressive behaviours’ – where we are moved to act by representational states that are not sensitive to real-world evidence, nor integrated in our whole system of beliefs. I argue that at least some degree of sensitivity to evidence and inferential integration are necessary for a state to count as belief; hence the representational states that play the relevant motivating role in the cases I consider cannot be beliefs. I suggest that imagination is the best alternative candidate. This supports an account of imagination according to which its functional profile is the same of belief with respect to emotional and behavioural outputs, while it critically differs from belief with respect to cognitive inputs (and related normative constraints). Imaginings dispose us to act and react in the same ways in which belief do; but they are not (nor ought be) formed and maintained in response to real-world evidence as beliefs are (and ought to be). On this view, many cognitions that are standardly classified as beliefs – like superstitious, ideological and religious ‘beliefs’ – turn out to be better understood as imaginings. Imagination does not just allow us to ‘escape’ from reality into fictional worlds, but plays a key, direct role in our representation of (and practical engagement with) reality.