Because there are ways of engaging with mathematics that don't involve the axioms of ZFC.
One simple enough for me to know and regale only involves two axioms:
- There is a successor function S(x), which maps one unique input to one unique output.
- The number 0 is never the output of S(x).
You can add another axiom (the positive integers is the smallest such collection of these), to get the positive integers, as we conventionally understand them. And off of these, addition and multiplication can be built. But without that additional axiom, there can be other closed loops of any finite length which operate in a circle, such that the successor of yota is gamma, and the successor of gamma is yota, with no proper mathematical relation to the rest of the integers.
In such a system distinct from ZFC (iirc, predating it), and as such, lacking R entirely, does it make sense to talk about how much a yota amount of dimensions would be, by trying to shove it into R^n?
And more broadly, I'm certain that "any possible mathematical axiom" would extend to things far weirder than this.