- Moral language requires categorical imperatives.
"Institutional," in Joyce's usage, means something that one may or may not adopt. The rules of chess are an institution. I must adopt the rules of chess if I want to play in a tournament, but if I am playing against my 7 year-old nephew, I might intentionally make an illegal move (move into check, for instance, so that she can win). As we saw last time, Joyce thinks that practical rationality is the only normative system for which we do not have the luxury of being able to "step outside of the system."
Some philosophers (Joyce mentions Philippa Foot, but says she later abandoned this approach) have proposed that morality is really a system of hypothetical, not categorical, imperatives. Joyce's only response here is to refer back to his discussion of practical reason as the only non-institutional normative system.
Still later, Joyce considers Harman's relativistic view of morality. For Harman, different moral systems are like different frames of reference in physics: events can be viewed from any system, and no one system is privileged over another. If practical rationality is indeed agent-relative, and if morality can be founded on practical rationality, then it makes sense that the resulting system would be relative rather than absolute.
Joyce responds with the Nazi objection. When the Nazis were put on trial, no one thought it necessary to consider the facts from the point of view of the Nazi ethical system. The judges behaved as if their morality was the only correct frame of reference. Thus, Joyce argues, we do not in practice act as if morality were relative.
My first thought here is, "Well, yeah, but in other cases we act as if morality were relative." For instance, knowing that my friend Joe is Jewish, I have no difficulty saying "Joe ought not to eat that cheeseburger," even if I do not feel that I ought not eat one.
More broadly, I would say that our use of moral language is partly cultural and partly instinctive, and that it would be very surprising if it formed a logically consistent system. It is not surprising, then, if we find we have to modify something in our morality in order to make it logically consistent. Isn't this just what moral philosophers have been doing for centuries? Joyce would respond that to let go of the absolute quality of morality results in a system which is no longer recognizably a moral system.
Finally, I want to note that Joyce's Nazi response is merely argument-by-example, and so is pretty weak. My intuition is that most moral systems have some sort of inbuilt relativity. For instance, quite often there are different rules for "us" than there are for "them." Joyce's intuition is different, but he acknowledges that it is just an intuition, and would require much more research to establish with certainty.
To summarize, we can evade (1.) by saying that morality is an institution that we can choose to adopt. Equivalently (?), we can say that morality is a system of hypothetical imperatives of the form, "In order to act in accordance with moral system X, you ought to do Y." What results is a relative, rather than an absolute, morality.