In communicating about certainty, speakers make choices among available formulations and hearers will aim to recover speaker intentions. In two studies, we assess speakers' production choices and hearers' interpretations to test (a) how maximal certainty is formulated, (b) whether those formulations adjust depending on context, and (c) whether speakers' context-driven adjustments are apparent to hearers. We compare the lower-certainty formulation `I believe that the deadline is tomorrow' [`believe'] with two high-certainty formulations, `I know that the deadline is tomorrow' [`know'] and `The deadline is tomorrow' [bare assertion]. Following Williamson (2000) and De Rose (2002), it is unclear which one of the latter two conveys higher epistemic standards. Given the unclear picture, we investigate when (if ever) `know' should be felicitous to utter over the bare assertion. One reason could be that `know' may be uttered felicitously for a wider range of contexts than the bare assertion (De Rose, 1992).Furthermore, `know' might be a useful linguistic tool for speakers to structure the subsequent dialogue to their liking. By presupposing content speakers assume or act as if the conveyed information was already shared knowledge and not up for debate. Thus, hearers might be more inclined to accept and accommodate e.g. Lewis (1979) presupposed content than asserted content.We investigated whether interlocutors align in the way they convey and recover meaning from statements about degrees of belief, comparing their behaviour across cooperative and uncooperative scenarios. Our results suggest (a) that speakers use know>bare assertion>believe for content with successively lower evidentiality scores and that hearers likewise infer know>bare assertion>believe in the same relative ordering. Regarding (b), speakers used `know' strategically in the uncooperative scenario to overstate their knowledge indicating that the usage of `know' is context-dependent. Regarding (c), hearers seemingly fail to recover these production strategies. This may be due to our experimental design where we investigated comprehension from a bystander point of view, or might similarly suggest that speakers succeed with their strategic approach.