4 ms·
In the context of set theory, by "self-reference" I would assume you mean either 1) impredicative definitions, i.e., definitions where the bound variables used
by math-anon- 11y ago
In the context of set theory, by "self-reference" I would assume you mean either 1) impredicative definitions, i.e., definitions where the bound variables used in the definition are permitted to range over the whole universe to which the very thing being defined belongs, or 2) non-well-founded sets, sets that can have themselves as members or be defined in terms of infinitely descending membership chains like ...ϵ n4 ϵ n3 ϵ n2 ϵ n.
ZFC admits the first by not limiting the range of the bound variables of the axiom schemas of Separation or Replacement (unlike the Separation axiom schema of Bounded Zermelo Set Theory), but it disallows the second (due, for example, to Foundation).
So at least one form of self-reference, impredicative definitions, are permitted by ZFC.
- ganzuul 11y agoBy self-reference I mean recursion. ED: Thanks for the reference. It appears to be a much more accurate account of the rift which I tried to talk about. I have previously been informed that my conclusion that ZFC does not permit recursion, as in a cyclic graph, was correct, and this certainly is further evidence regarding this issue.