Hurkyl,
Please see my answer to some old post of you in this tread.
If I have a list of properties that something called a "set" obeys, I don't need to know anything about the existence of a "set" in order to reason about those properties.
(This is a familiar idea even in "everyday" logic; we often call it a "hypothetical scenario" in such a context)
Any x in some theory cannot be but a model(X), therefore we always have to be aware to the combination of model() and X, where model() is the global state and X is some local state.
Model() is the container (global state).
X is the content. (some local state).
x is the product of model(X) (the combination or relations between global and local states).
Shortly speaking any theory is first of all model() or if you like an empty container (a global state) "waiting" to some X (some local state) to be its examined concept.
From this point of view, any theory must be aware to the relations between the global and the local, otherwise it cannot use its full potential.
If Peano Axioms is a theory then first of all is a model()(a container) that "needs" some X(a content) to deal with.
For example:
1 is in N (also can be understood as: 1 is a natural number)
x <-- model(X) where x cannot be but a combination of model()(= theory of numbers) and X(= the concept of a number).
Concept_of _a_number is the intuitive ability to count things without any supported theory.
Concept_of_a_number is not a Theory_of_numbers, and theory_of_numbers without Concept_of_a_number is an empty model(=model() ).
It means that if we want to understand x we have to "put on the table" its combination (container-content) property.
Shortly speaking, 1 cannot be but 1 in N, whether you say it or not.
Therefore There is a container concept in Peano arithmetic.
Again, there can be a big problem for us to understand and develop deeper connections between so called different areas of research, if we don't take in account the global-local or container-content relations.
What I wrote here also can explain why ZF axiom of infinity and Peano first axiom are the same axiom.
And this axiom can be called "The forced-induction axiom".
My limited comprehension of your ideas may be at fault here, but this sounds awfully like confusing "x-content" with an "x-model".
All products of some theory are nothing but an x-model, an axiom is a product therefore an x-model.
I am talking about the hierarchy of dependency among these products.
The basic level is the axioms level, and on top of it there is the hierarchy of products, which can exist iff they do not contradict the axioms level.
But when we need an axiom that directly determines the existence of some element, it means that there is no hierarchy here but "the same lady with a different dress".
By the way x <-- model(X) is not under Russell's Paradox, because the global level(=container) transcendent any local level(=content).