I argue that modal terms have an epistemic interpretation on which concessive knowledge attributions are semantically contradictory. This is compatible with the fallibilist view that the basis on which we know something need not entail it, but not with the view that what is known need not be epistemically certain or necessary. The apparent contradictoriness of concessive knowledge attributions is not due to mere implicature, nor to assertion updating the modal base. And it is contextually invariant. Concessive knowledge attributions contrast markedly with concessive assertions and Moorean conjunctions, whose infelicity is plausibly due to norms of assertion. I briefly explain why the strict fallibilism I recommend is compatible with our ordinary use of ‘know,’ and with our knowing much on the basis of perception. Its contextual shiftiness closely parallels our variably strict use of temporal and other invariant terms with strict application conditions.

中文翻译：

---

认知可能性、让步知识归因和易错论

我认为，情态术语有一种认知解释，在这种解释上，让步的知识归因在语义上是矛盾的。这与易错论者的观点是一致的，即我们知道某事的基础不一定必然包含它，但与已知的东西在认识上不一定是确定的或必然的观点不同。让步知识归因的明显矛盾不仅是由于暗示，也不是由于更新模态基础的断言。而且它是上下文不变的。让步的知识归因与让步的断言和摩尔连词形成鲜明对比，后者的不恰当似乎是由于断言的规范。我简要解释了为什么我所推荐的严格易错论与我们对“知道”的日常使用以及我们基于感知的了解是相容的。它的上下文变化与我们在严格的应用条件下对时间和其他不变术语的不同严格使用密切相关。