As Nagel puts it, in The View from Nowhere (1986: 138)
If values are objective, they must be so in their own right, and not through reducibility to some other kind of objective fact. They have to be objective values, not objective anything else.
Of course, naturalists are apt to think it question-begging of us to charge them with changing the subject in this way, since whether they've given a successful naturalistic reduction of values (as opposed to just using the word 'values' to talk about something completely differnet) is precisely the issue under dispute. This leads Schroeder, in Slaves of the Passions (p.82) to conclude that:
[F]iguring out whether any reductive view is true is not the sort of thing that we can do without getting our hands dirty.... [T]he way to see whether normative properties can be analyzed, is to propose your analysis, and see whether it checks out.
Perhaps it's poor form to ever deny the need for hard work, but I really do think that sometimes it's easier than Schroeder makes out. For example (possibly borrowed from Parfit?), suppose that someone were to propose that the property of being a normative reason just is (or is constituted by) the property of being a cabbage. I don't think we need to "get our hands dirty" working out all the implications of this proposed analysis to see whether it "checks out". Instead, this analysis is plainly a non-starter. Anyone with the slightest grasp of normative reasons can know that, whatever they are, they ain't cabbages. Likewise, I think, for any other physical entity. Normativity, if there is any such thing, is (like qualia) in a class of its own, and cannot be reduced to any other (independently specifiable) thing.
Here's one way to spin this underlying thought into an argument (what we might call, with a tip of the hat to Frank Jackson, the Normative Knowledge Argument):
(1) The full nature of normative facts can only be grasped via normative concepts. (Someone who lacked all normative concepts would be importantly ignorant; they could not fully grasp the normative facts.)
(2) Natural facts can be fully grasped using merely natural, non-normative concepts.
Therefore, (3) Normative facts aren't natural facts.