Humean's Fake: Joycean Error Theory

I've been reading The Myth of Morality, by Richard Joyce. Joyce is a proponent of moral error theory: He thinks that when we use moral language we are simply in error, because there are no such things in the world.

His argument centers on the idea of the categorical imperative. A hypothetical imperative has the form, "If you want to achieve A, you ought to do X." This sort of statement is uncontroversial: there is no doubt that such statements are sometimes true. Kant thought, and Joyce agrees, that categorical imperatives are central to moral thought. Categorical imperatives make the claim, "You ought to do X," without any "if..." clause. Such claims are absolute.

As I understand it, Joyce's argument runs as follows. (He lays out his argument very nicely, but this is my formulation of it, not his.)

1. Moral language requires categorical imperatives.
2. Categorical imperatives cannot be legitimately questioned.
3. Practical rationality is the only source of statements that cannot be legitimately questioned.
4. But practical rationality cannot provide a basis for (moral) categorical imperatives. 
5. Therefore, moral language is in error.

(2.) is simply a consequence of the definition of a categorical imperative. Joyce gives each of the other points careful consideration. In the next few posts, I will look at how he deals with attempts to deny each of them.