I've been reading Zimmerman's, Living with Uncertainty and some remarks of his concerning the subject 'ought' inspired this.
Consider the view that obligations are rationally identifiable as such. If there's an obligation to act against one's own reasonable verdictive moral judgments (say, but doing what one rationally judges is worse than what one reasonably judges is best), it's not rationally identifiable as such. Consider a case where a subject reviews what she takes to be her options: A, B, and C and comes to judge correctly that:
(1) A is uniquely best and B is better than C. D (i.e., do nothing) is worst of all.
As a matter of fact, however, she cannot bring it about that A obtains. [In a vending machine, there's (A) a small child, (B) a small puppy, and (C), a small kitten. Agent has sufficient change to release any of these trapped critters, but the mechanism that would release the small child is broken.] So, if we assume that OIC, our subject is obliged to either bring it about that B or C. She knows B is better and that she could bring it about that either B obtains or C obtains. She reasonably but mistakenly believes that she could bring about A, B, or C.
Facts about what can be done are, well, facts. Facts like that don't directly affect what's reasonable to believe about what can be done, will be done, should be done, etc... So, suppose we identify reasonable judgment with justified judgment with permissible judgment.
(2) Agent judges reasonably/justifiably/permissibly that she should bring A about.
From (i) the principle that obligations are rationally identifiable as such and (ii) the observation that she cannot rationally identify any (alleged) obligation to bring it about that B obtains and refrain from trying to bring it about that A obtains, it seems to follow from (2) that:
(3) Agent's obligation cannot be to do other than A.
It follows from OIC that:
(4) Agent's obligation cannot be to do A.
What's Agent to do? You can say that A ought to act against A's judgment about what ought to be done, but this is just to give up the principle that obligations are rationally identifiable as such.
Now, consider (LINK)
Link: If you judge that you ought to A (and oughtn't refrain from so judging), you ought to A.
In other words, you gotta do what you (judge permissibly) you gotta do. It seems that the argument above suggests that the justification for normative judgment is sensitive to objective facts (e.g., facts about what can be done) or LINK is incompatible with OIC.