The Inferential Constraint and If φ ought φ Problem
The standard semantics for modality, together with the influential restrictor analysis of conditionals (Kratzer 1986; 2012) renders conditional 'ought' claims like (1) trivially true:
If John's stealing, he ought to be stealing.
While this might seem like a problem specifically for the restrictor analysis, the issue is far more general. For any account must predict that modals in the consequent sometimes receive obligatorily unrestricted interpretation, as in (1), but sometimes appear restricted, as in (2):
2. If John's speeding, he ought to pay the fine.
And the problem runs deeper, for there are non-conditional variants of the data. Thus, the solution cannot lie in adopting a particular analysis of conditionals, nor a specific account of the interaction between conditionals and modals. Indeed, with minimal assumptions, the standard account of modality will render a massive number of claims about what one ought to, must, or may, do trivially true. Worse, the problem extends to a wide range of non-deontic modalities, including metaphysical modality. But the disaster has a remedy. I argue that the source of the problem lies in the standard account’s failure to capture an inferential evidence constraint encoded in the meaning of a wide range of modal constructions. I offer an account that captures this constraint, and show it provides a general and independently motivated solution to the problem.