This makes it sound like internal reason is a complex concept. We start with the basic (external) concept of a normative reason -- a fact that counts in favour, has normative force, or whatnot -- and then add a quirky motivational constraint.
But of course that can't really be what Williams has in mind, because he wants to cast doubt on the very coherence of the idea of external reasons. ("What is it that one comes to believe when he comes to believe that there is reason for him to ϕ, if it is not the proposition, or something that entails the proposition, that if he deliberated rationally, he would be motivated to act appropriately?") And if internal reasons = external reasons + a motivational constraint, then any senseless to the concept of an external reason would automatically be inherited by the complex concept of an internal reason.
The remaining option for Williams would seem to be to banish the whole "reason" part of the concept entirely, and understand so-called "internal reasons" purely in terms of the motivational constraint. In this case, to attribute an internal reason to someone is to do nothing more than to make a certain sort of empirical prediction about how acting in a certain way relates to their desires. Our sentence 'A has a reason to ϕ' doesn't merely imply, but is analytically equivalent to, some descriptive sentence along the lines of 'A has some motive which will be served or furthered by his ϕ-ing'. But then, as Parfit trenchantly notes in On What Matters, this (i) is no longer a distinctively normative claim at all, and moreover (ii) trivializes what should have been a substantive normative claim, namely the subjectivist thesis that we only have reason to do what furthers our desires.
Williams thus seems to face a dilemma. Either the concept of a normative (external) reason makes sense after all, or else internal reasons aren't normative. Am I missing something?