I'm quite unconfident about this whole line of argument, and concerned that we're heading for some moral conclusions based on appeals to this argument. If you have to get into odd discussions about the truth and meaning of mathematical entities to make a metaethical l argument, I doubt you have a good metaethical argument.

I don't know about that. People continually make metaethical mistakes by assuming that "morality is defined by your brain" is the same as "morality is about your brain", and draw all kinds of faulty conclusions, like that unless there's a stone tablet somewhere out in space with the Thousand Commandments of Morality written on it, it must be okay for people -- in different societies, with different beliefs -- to torture children if they want to (because hey, if morality isn't objective it must be relative, right?). That's exactly the error being talked about here, collapsing levels, and I think it's kind of an important one, metaethically.