A certain number of cases suggest that our willingness to ascribe “knowledge” can be influenced by practical factors. For revisionary proposals, they indicate that the truth‐values of “knowledge” ascriptions vary with practical factors. For conservative proposals, on the contrary, nothing surprising is happening. Standard pragmatic approaches appeal to pragmatic implicatures and psychological approaches to the idea that belief formation is influenced by practical factors. Conservative proposals have not yet offered a fully satisfactory explanation, though. In this article, I introduce and (...) defend a third conservative proposal which I call “Refined Invariantism”. The two main claims of this proposal are that (1) we should distinguish between high stakes cases in which the subject does not believe (that he knows) the target proposition and those in which he believes (that he knows) the target proposition and that (2) we should adopt a psychological treatment for the first kind of case and a pragmatic treatment based on the epistemic standards for appropriate assertion and action for the second kind of case. I argue that this new combined approach avoids the main pitfalls of its two conservative rivals and that it gives new life to the generality objection levelled against revisionary views. (shrink)