Attitudes, Conditionals and Margins for Error

Abstract

This paper is about The Qualitative Thesis, the thesis that if you are not sure that ϕ is false, then you are sure of the indicative conditional ϕ > ψ just in case you are sure of the material conditional ϕ > ψ. Following contextualists about indicative conditionals like Bacon [2015], we will understand this thesis in a local way—roughly as saying that if you are not sure that ϕ is false, then you are sure of the proposition expressed by ϕ > ψ in your context just in case you are sure of the material conditional ϕ > ψ. To state this precisely, let Sc,w(JϕKc) mean that the speakers in c are sure of JϕKc in w. Then: The Local Qualitative Thesis. For any world w and context c, if ¬Sc,w(J¬ϕKc), then: Sc,w(Jif ϕ, then ψKc) if and only if Sc,wJϕ ⊃ ψKc. We investigate the epistemological consequences of The Qualitative Thesis. We characterize The Qualitative Thesis in standard formal frameworks for studying the logic of attitudes and conditionals. With these characterization results in hand, we develop a connection first observed by Ben Holgu´ın (p.c.) between The Qualitative Thesis and a plausible margin-for-error requirement on rational sureness. We show that The Qualitative Thesis is inconsistent with the margin-for-error principle. We propose a new shifty semantics for indicative conditionals. We say that the meaning of an indicative conditional is partly determined by the conditional’s local informational environment—the conditional’s local context—which, in turn, is systematically shifted by attitude operators. Our account validates The Qualitative Thesis, but dispenses with its undesirable epistemological consequences.