Communicating with others is not as simple as broadcasting a message; that message must be calibrated to one’s audience. How exactly we are able to formulate an understanding of others’ minds is a quintessential question in philosophy, psychology, and cognitive science, among other fields, but it is a well-established fact that we are especially adept at doing so. Close observation of conversational interaction offers a window into at least some of the tools that enable us to navigate the knowledge states of others and tailor messages to suit particular recipients. This dissertation explores social and moral norms that govern management of relative authority over knowledge in conversational interaction.This dissertation draws on several sources of data, including a corpus of recorded unstructured conversation between close friends, recordings of the collaborative completion of an experimental task by strangers, and survey data from online participants making social judgments about written transcripts. This dissertation presents evidence, across four chapters, that participants in conversation orient to a normative expectation that what one actually knows (epistemic status) will be accurately reflected in how one talks about that knowledge in interaction (epistemic stance), both within an individual (don’t misrepresent your own knowledge) and between individuals (don’t treat others like they know more or less than they actually do). Violations of this norm are sanctionable offences and participants take measures to avoid making violations (Chapter 1), make off-record challenges when violations are suspected (Chapter 2), and admonish and harshly judge parties who have committed obvious violations (Chapters 3 and 4), behavior which both reflects and reinforces the norm.