The point generalizes. Consider any worthy goal G. What matters is achieving the goal. To improve our chances of achieving it, we should tend to respond rationally to evidence, and act in a way that is "justified" in this conditional sense. But it would be a mistake to think that this justification is what matters. To say, "You ought, given the evidence, to X" does not necessarily mean that you ought (objectively) to X. The evidence may be misleading. So if we interpret it as a kind of wide-scope requirement [as Clayton suggests], it may be that you ought to reject the evidence rather than do X. (That might be irrational. Does this matter? We're supposing the normative goal here is G, not "being rational".)
Even if there's something to be said for being rational (perhaps it is an intrinsically valuable character trait, for example), it at least seems to be "counter-deliberative", as Clayton explains:
Let us say that the rational choice is the choice that sides with what the agent not unreasonably takes the balance of reasons to require. Whenever the agent is in a choice situation, she will not be able to distinguish in thought the choice that is on balance best supported by the reasons from the one that is best supported by the reasons as she takes them to be. But which one is the BASIS of her choice from her perspective? The reasons and not the reasons as she takes them to be.
This ties in with Kolodny's transparency account: the rational choice is that which it seems we ought to do. From the first-personal perspective, we cannot distinguish appearance from reality. (To believe that P is to consider P true.) In deliberation, we hope to identify the best choice. We do this by settling on what seems best. But we wouldn't treat the appearances as having any independent force, over and above the objective facts. I can treat the fact of P as a reason, but not (typically) the fact of my believing P. Note the strangeness of the following reasoning: "I can't decide whether it would be best to X or to Y. I suppose X-ing would be more rational. Oh, so that's another thing in its favour! I shall do X!" Also, Clayton points out that we would advise people to believe what's true, not what they have evidence for.
That's all to suggest that we shouldn't confuse the normativity of ends and their rational means. I'm sure there was some further reason why I wanted to write about this topic, but I can't for the life of me remember what it was. So I'll finish with a quote from Hare:
The winner of a game of backgammon is the player who first bears off all his pieces in accordance with the rules of the game, not the one who follows the best strategies. Similarly in morals, the principles which we have to follow if we are to give ourselves the best chance of acting rightly are not definitive of 'the right act'; but if we wish to act rightly we shall do well, all the same, to follow them.