University of Michigan

Fall 2012 Workshop in Philosophy and Linguistics


 
 
Dynamic Thoughts on Ifs and Oughts
Malte Willer

Abstract

Iffy oughts figure prominently in a variety of paradoxes of deontic logic: Chisholm's (1963) Contrary-to-Duty Paradox, Forrester's (1984) Gentle Murder Paradox, as well as the recent Miners Paradox from Kolodny and MacFarlane (2010). A satisfying response to these paradoxes is a non-negotiable component of any adequate semantic story about conditionals and their interaction with deontic modals. I demonstrate that such a story can be told but doing so requires that we supplement a semantics that pays proper attention to the sensitivity of ifs and oughts to contextual information with a dynamic conception of logical consequence.

I begin by outlining a simple semantic analysis of conditionals and deontic modals that is in the spirit (but not the letter) of Kratzer's semantic analysis. The analysis is not unfamiliar—it is plus or minus a bit, the one that is prominent in Kolodny's and MacFarlane's solution to the miners paradox—but it needs to be supplemented with an adequate notion of logical consequence. Thinking of validity in a classical fashion as necessary preservation of truth at a point of evaluation is good enough to block the miners paradox but, as I shall observe, does not generalize to cover the remaining puzzles about deontic conditionals. A dynamic approach to validity, in contrast, does not only resolve the miners paradox but also the more classical paradoxes about conditional obligations from Forrester and Chisholm, and this is more than a sheer coincidence since it correctly predicts that the right logic for deontic conditionals is nonmonotonic.

The resulting framework for deontic conditionals is attractive since it avoids the classical deontic paradoxes while preserving factual as well deontic detachment for iffy oughts. Some further refinements of the basic analysis are possible and will be discussed in the concluding sections of the paper. Of special interest here is the issue of order-sensitivity: in standard dynamic frameworks, consistency is sensitive to the order of premises. I show how this result can be avoided while preserving the key dynamic insights into the semantic analysis of deon tic conditionals.