Abstract
Could someone who wants a gin and tonic have a normative reason to drink petrol and tonic? Bernard Williams and Michael Smith both say, 'No'. They argue that what an agent has normative reason to do is determined by rational deliberation that involves correcting the agent's beliefs and current motivations. On such an account of normative reasons, an agent who is motivated to act in some way due to a false belief does not have reason to act in that way. I argue that the agent could have reason to drink the petrol, because an agent's epistemic circumstances, what that agent can come to know, can be as relevant to what the agent has reason to do as other aspects of his circumstances. Moreover, if an agent's epistemic circumstances are taken into account when determining what the agent has reason to do, this can still give an account of reasons that is normative, ensures that the agent and onlookers agree on what the agent has reason to do, is appropriately connected to rationality and fairly represents the agent's beliefs about the knowledge they need to have to know how they have reason to act.