Ross’s Paradox is a problem arising in Standard Deontic Logic. On the SDL interpretation of “ought”, obligations survive weakening, in the sense that, if P entails Q, and it ought to be that P, then it follows that it ought to be that Q. Ross’s paradox involves an instance of this commitment.
1) Jones ought to mail the letter.
2) Jones ought to mail the letter or burn the letter.
According to SDL, (1) entails (2), but, intuitively, (2) seems false in situations where (1) is true. Intuitively, (2) suggests that it is permissible to either mail the letter or to burn it, while (1) would seem to rule out the permissibility of burning the letter (it being assumed that you cannot both mail it and burn it).
I recently noticed a parallel between (2) and the following sentence:*
3) Jones wants a dog or a cat.
(3) is interesting because it can be extended in either of two ways:
3a. Jones wants a dog or a cat, but I don’t know which.
3b. Jones wants a dog or a cat, but doesn’t care which.
(2) can be similarly extended:
2a. Jones ought to mail the letter or burn the letter, but I don’t know which.
2b, Jones ought to mail the letter or burn the letter, but it doesn’t matter which.
It seems to me that the most natural way to treat (3) is as ambiguous between a disjunction of desire attributions (the 3a reading) and a desire with a disjunctive content (the 3b reading). A different approach would be to suppose that when one has a desire towards a disjunctive content, this can be true in virtue of either, a desire towards one of the individual disjuncts, or in virtue of a desire which is equally well satisfied by either disjunct.
On the former approach, it does not follow from my desiring a cat that I desire a cat or a dog. On the latter approach, the first clause of 3a. and 3b. would express the same proposition, and it would follow from my desiring a cat that I desire [a cat or a dog]. The former approach seems better in the case of desire. Sentence (3) is ambiguous in a way teased out by (3a) and (3b), and that difference is captured precisely by the difference between a disjunction of desire attributions and a desire with disjunctive content.
The parallel between (2) and (3), combined with the superiority of an ambiguity treatment for (3), seems to me to provide good reason to accept an ambiguity treatment for (2).
Getting back to Ross’s paradox, it appears that SDL builds in the latter (non-ambiguity) reading of disjunctive ought claims. After all, the three possibilities that would render (2) true are a) that Jones ought to mail the letter, b) that he ought to burn the letter, or c) that he ought to mail or burn the letter, but it is not the case that he ought to mail it or that he ought to burn it. This is simply the semantic content needed on the non-ambiguity approach.
I think this provides some reason for thinking that SDL, and specifically the fact that oughts survive weakening, are not correct for the treatment of the english “ought”.
*Insofar as I can source the observations I mention about this sentence, they come from Lepore’s “Meaning And Argument”, though I don’t have it handy, and don’t intend to be trying to accurately represent his views here.