If I replace the below axiom in ZFC (or NBG) by the below inference rule, there are any consequence in what can be demonstrated?
Axiom: If two sets (or classes) have the same elements then their are equal. $$ \forall{ }A\forall{ }B(\forall{ }x((x\in A) \Leftrightarrow (x\in B))\Rightarrow A = B) $$
Inference Rule: From $\varphi(A)$ and $\forall{ }x((x\in A) \Leftrightarrow (x\in B))$ infers $\varphi(B)$. $$ \{\varphi(A), \forall{ }x((x\in A) \Leftrightarrow (x\in B))\} \vdash \varphi(B) $$
Comment
This substitution is based on Leibniz Law (but different of this Math.SE question do not quantify about predicates). This substitution have three motives:
- Play with logic.
- Remove the equality symbol.
- Remove the axiom (I like natural deduction, when possible): I imagine that this "formalism" have special significance for classes, since a class is defined by the logical property of their elements then one can imagine that the extension (see this Math.SE answer) is "not a truth" but only a "program" to perform a proof (see this Math.SE answer).