One possibility is that the error-theorist accepts so-called norming but not normative propositions, e.g. the distinction between:
- It is permitted that A.
- I permit you to A.
Now, there doesn't seem to be much trouble saying, "If I permit you to A, then it is permitted that A," or we might qualify by saying, "If most people permit you to A, then it is permitted that A" (or something else, some other bridge-sentence) so the error-theorist would be moved to deny such things too. Or, if not outright deny them, weaken them by equating the "permitted-that" clause so strongly with the "permitted-to" clause that there is no strong reason to use the that-term instead of the to-term.
More acutely, a Mackiean error theorist might say that permitted-that clauses are acceptable but not in the "objectively prescriptive, and weird, fact" way (that Mackie inveighed against).
Remarks on "objectively prescriptive facts would be too weird" (added)
The statement, "Deontic properties would be objectively prescriptive," or, "There are objectively prescriptive facts (OPFs)," has played out in a few different ways, historically. The way that Mackie focuses on is:
- Identifying a deontic fact, or knowing a deontic property to be true of some object, would instill some willingness, even if not overriding, to comply with the prescription attendant upon the fact/objective property relation.
There are also:
Instilling a desire or willingness to comply makes for a deontic fact after the fact: this is sort of metaethical constructivism. So a deontic fact isn't knowable before we are motivated to act accordingly, but such knowledge is an effect of something in our pure willing.
There are facts such that, if they exist, their existence is accompanied by prescriptions, and somehow the facts themselves can be seen as issuing these prescriptions. And somehow, not complying with those prescriptions amounts to not accepting the facts. Perhaps the facts still count as known in some "clinical" manner, perhaps not, but the key is in rejecting-a-proposition being separable from believing-a-proposition in such a way that we can reject a proposition on some partly factual level while believing it on some other strictly factual level.
So: OPFs are such that either (A) the provided motivation is an effect of our knowledge of the facts, (B) the motivation is the cause of this knowledge, or (C) motivation and knowledge are simultaneous. The error theorist might say, then, that (A) is not logically given (the syntax of a fact, so to speak, is not a formal cause of anything with the syntax of a prescription) and so would have to be metaphysically grounded instead. But this is "weird" and unrealistic, perhaps (though see Korsgaard's argument that human beings and their actions themselves count as OPFs in this sense).
It's less obvious to me, though, that (B) or (C) is weirder than is compatible with a (naive) scientific worldview. At any rate, the error theorist can make a further distinction between normative propositions and deontic propositions sharply: a normative proposition can be true when there has been the required deed of norming (of giving a permission, say), but e.g. if A says, "I obligate B to do x," B can fully accept the fact that A has said such a thing while disregarding, in their actions, the obligation imposed on them by A: the normative fact is not an OPF of any of the above types (A), (B), or (C).
Divine-command theory, then, might be seen as equivocating (in historical arguments) between normative and deontic import. "God permits you to x," ends up with the motivational force of (A/B/C), but not on a normal truth-functional basis, but due to God's ominous powers. But so there would be a possible version of divine-command theory that was also, perchance, an error theory, bizarre as that might seem.
POSTSCRIPT: now, for all that, I don't want to pigeonhole error theorists as actively committed to depraved statements. Yes, an error theorist will be able to sincerely say, "Torturing children for years is morally permissible," and if we tried to maneuver them into only putting it that way, we would be doing them a disservice. For they would be minded to add, "... is morally permissible in the abstract sense that positive obligations and substantive prohibitions do not exist."
This doesn't mean that error theorists secretly desire that atrocities be committed. And having a world where non-error theory dominates the presuppositions of the discourse doesn't seem to have meant having a world where people less often committed atrocities. So we should be careful bordering on tolerant in arguing that error theorists are "dangerous" or offensive (i.e., we should not argue that they are specifically dangerous/offensive). Naive? Shallow? Factually mistaken? Yes. And maybe without objective moral intentions, they cannot have Kantian moral worth attach to their actual behavior. But so even if their behavior might not be from duty, it can still be according to duty, and I myself would prefer to leave them room to see the error of their ways.