The agent's failure to be moved by the right reason reveals a fault in him, but it does not count against the permissibility of his action. The best explanation of this distinction seems to me to lie in the close connection between permissibility and the guiding of choice.
The question of permissibility is the question, "May I do X?" which is typically asked from the point of view of an agent who is presented with a number of different ways of acting. The question is, which of these may one choose? The question of permissibility thus applies only to alternatives between which a competent agent can choose.
And, Scanlon suggests, we cannot choose what we see as reasons. (Sure, we "decide whether something is a reason or not" (p.22). It is a judgment of sorts, and one we can be held responsible for. But it is not a willed choice, and so doesn't raise questions of permissibility.) We can choose whether to do X or not, but we cannot choose to do X for certain reasons rather than others.
Is that right? Scanlon continues (p.23):
This explanation may seem at odds with the very appealing idea that moral considerations are not esoteric but are available to any agent. If the reasons that a moral principle identifies as relevant are "available" in this sense, why is acting on them -- doing the right thing for the right reason -- not also available to the agent? The answer is that the availability of moral considerations simply means that a normally competent agent ought to be able to understand them and see that they provide reasons. It does not follow that an agent who, for whatever reason, does not see the force of such a reason, is nonetheless in a position to choose to see its force, or to act on it.
In our seminar, Michael raised an interesting comparison with epistemic obligation. It's usually thought that we ought to see or access all the epistemic reasons. Even though it is not strictly a matter of "choice", nonetheless, in failing to recognize the evidence available to us, we fail to believe as we should. So it seems that what matters for responsibility here is simply our rational capacities, and the fact that we could have accessed the relevant reasons (even if not simply by exercising our will).
So why the difference in the practical case? Do we have more freedom to inquire into facts than to change our internal motivations? Or are the resulting obligations different in nature, such that practical obligations have to meet a stricter form of the 'ought implies can' requirement? Any ideas?