The normality of error

Human Foreknowledge

“…Should this defense of the unidirectionality of indeterminate knowledge fail, there are other paths to supporting it. In particular, we might sign up for a safety-based or normality-based account of knowledge (in the style of Smith 2016; Goodman and Salow 2018, ms.;Carter and Goldstein 2020). It seems plausible to ground the unidirectionality of knowledge in the unidirectionality of certain important normality relations.…”

“…Moreover, there are some nearby states whose existence we have no reason to doubt, such as historically possible knowledge (what would be expressed by ◊𝐾) and epistemically possible knowledge (corresponding to ⧫𝐾). For discussion of these principles, and proposals to think of epistemically possible knowledge as providing account of justification, see Stalnaker (2006Stalnaker ( , 2019; Bird (2007); Ichikawa (2014); Carter and Goldstein (2020). 12 Williamson (2000) is often associated with this idea, but Williamson adds in the idea that this norm is constitutive of the speech act of assertion.…”

Knowledge from Probability

“…Note that, unlike standard models of belief-revision, there may be no v that is the result of discovering p in w -for example, if p is incompatible with the state of the world in w. And since only truths can be discovered, normality structures allow us to easily model iterated discoveries (in contrast to formally similar models of theories of belief-revision like AGM [1], first developed in [22], which do not handle iterated belief-revision). 6,7 2 Reducing normality to probability Appealing to notions of comparative normality (or comparative plausibility) is, by now, a familiar idea in theorizing about knowledge and belief. The main advance of this paper is to explore the consequences of an analysis of these notions in terms of the result of conditioning a prior probability distribution on the agent's evidence.…”

“…See[44,45,46,17,21,11,19,18,6,33,4,7,34,16] for related ideas about knowledge,[39,40,41,42,43] for related ideas about justified belief, and[27,35] and references therein for related ideas about non-monotonic reasoning.2 See[38,55,25] for precedents in the case of knowledge, and[29,56,24,5] for precedents in the case of belief.3 See[10,2,3,41,42,19,18] for recent discussion of the kind of cases in § §3-5, and[50,51,53,54,8,17,47,45,6,12,7,18] for recent discussion of the kind of cases in §6.…”

“…(Thanks to AybükeÖzgün for drawing our attention to this work.) Note that what "world" the agent is in changes as they get new evidence; those who prefer to reserve the word "world" for something unchanging can substitute "situation", "case", or "centered world" 7. The framework presented here sides with[18] over[19] by treating normality relations as holding between worlds rather than states.…”