Back to list

Detail of contribution

Auteur: Alex SILK

Deontic Conditionals: Weak and Strong

Abstract/Résumé: The literature on the Miners Puzzle and information-sensitivity in deontic modals and conditionals has focused on data with weak necessity modals (“WNMs”) like ‘should’. This is problematic. What makes the Miners Puzzle a puzzle, I argue, are features specific to WNMs. Given an independently motivated semantics for WNMs we can explain the data without complicating our semantics for modals and conditionals more generally. In the Miners Case many find (1)–(2) compelling. (1) We should block neither shaft. (2) If the miners are in Shaft A (/Shaft B), we should block Shaft A (/Shaft B). Intuitively (2) says what’s best on a condition; it doesn’t impose an obligation on us conditional on how the world is unbeknownst to us. But (3) does seem to impose such an obligation, as (4) seems to grant such a permission. (3) If the miners are in Shaft A (/Shaft B), we must block Shaft A (/Shaft B). (4) If the miners are in Shaft A (/Shaft B), we may block Shaft A (/Shaft B). This is part of why many disprefer (3)–(4). But if (3) is accepted—as if we stress the gravity of the miners’ plight—(5) seems false; similarly with (4) and (6). (5) We may block neither shaft. (6) We must block neither shaft. (3)/(5) and (4)/(6), unlike (1)–(2), aren't jointly acceptable. The Miners Puzzle is no puzzle at all when expressed with ‘must’ or ‘may’. I treat ‘must’ and ‘may’ as expressing ordinary necessity and possibility, respectively: Given a premise set P, ‘Must p’ says that p follows from P and ‘May p’ says that p is compatible with P. This correctly predicts that if (3)–(4) are accepted, (5) is rejected. But suppose context provides a conversational background like G that reflects how our obligations depend on our knowledge (AK, AK̅, BK, BK̅ are worlds characterized wrt the miners' location and whether we know it). (6) G(AK) = {blA} G(BK) = {blB} G(AK̅) = G(BK̅) = {blN} If it isn’t definitively settled that we won’t learn the miners' location, the context set includes AK̅/BK̅. This correctly predicts that (3)–(4) aren’t accepted. On my view, what makes WNMs weak is that they express conditional necessity: Roughly, ‘Should p’ says that ‘Must p’ is true at all closest q-worlds, for some contextually supplied live condition q. This correctly predicts that (1)–(2) are jointly acceptable. For (1), general pragmatic principles suggest that if it's more likely that we won’t learn which shaft the miners are in, q will be this proposition ¬L. But principles of local interpretation suggest that there will be a reading of (2) for which q = L. On these readings, (1)–(2) are accepted wrt G. Using ‘should’ in (1)–(2) thus allows us to plan for blocking neither shaft while also remaining open to the possibility that we’ll learn the miners' location and planning for this contingency.