It has recently been argued that testimony’s epistemic role exceeds the exchange of propositional knowledge (Hazlett 2025; Hills 2020; Fraser 2021; Sliwa 2023; Malfatti 2024): speakers can change not only what but how we think about some topic. For example, by influencing what we attend to, what we take as significant, the associations we make, and the inquiries we embark on.

In the literature, it’s implicitly assumed these cases involve some kind of epistemic dependence, beyond dependence for knowledge (Ibid). What’s unclear, however, is how to make sense of this: what is the nature of this dependence? More specifically, what precisely does the hearer depends on the speaker for? What is the relationship between the speaker’s utterance and the hearer’s resulting epistemic states or conduct? In what sense, if any, is the hearer taking the speaker’s word for something?

This article has two aims: 1) to elucidate the problem and lay out some desiderata for an adequate answer, and 2) to offer a tentative solution: at least in some cases, speakers issue indirect directives – to compare two things, to explore the implications of applying a certain concept to a situation – and recommend following these as epistemically fruitful.