A beautiful and surprising (to me at least) result around the axiom of choice is that, while full $\mathsf{AC}$ is preserved by forcing, a model of $\mathsf{ZF}$ + "There are no amorphous sets" may have a (set-)generic extension in which there do exist amorphous sets. This was proved by Monro, On generic extensions without the axiom of choice; see also Asaf Karagila's summary here.

I'm interested in not-too-strong sufficient conditions on a c.t.m. $\mathcal{M}\models\mathsf{ZF}$ to have no generic extensions in which amorphous sets exist. Specifically, I'm curious if the following model-theoretic condition does the job:

Say that a c.t.m. $\mathcal{M}\models\mathsf{ZF}$ is expansive iff for every first-order theory $T\in \mathcal{M}$ in a finite language with infinite models, every infinite set in $\mathcal{M}$ is the underlying set of a model of $T$ in $\mathcal{M}$.

(Note that such an $\mathcal{M}$ does correctly compute whether such a $T$ has infinite models.) For instance, expansiveness prevents the existence of infinite Dedekind-finite sets, since we can take $T$ to be the theory of a discrete linear order. My question, then, is how expansiveness, amorphousness in particular, and forcing interact:

  • Is every set-generic extension of an expansive c.t.m. also expansive?

  • If not, is there a set-generic extension of an expansive c.t.m. which has amorphous sets?

(Incidentally, it's not immediately clear to me that expansiveness is any stronger than "Infinite = Dedekind-infinite"!)

  • 1
    $\begingroup$ If I'm not mistaken, an infinite set $X$ is a universe for any such $T$ iff $X$ is orderable and there is a bijection between $X$ and $X^2.$ The forward direction is by considering $T$ to be (a finite fragment of) PA and the backward direction by taking the E.M. model generated with $X$ as a set of order-indiscernibles. This would mean expansive models are precisely those satisfying choice by Tarski's characterization of choice. $\endgroup$ Oct 24, 2022 at 2:05
  • $\begingroup$ Incidentally, it would be interesting to see if just "Infinite = Dedekind-infinite" is sufficient to get "generically no amorphous sets." $\endgroup$ Oct 24, 2022 at 3:26
  • 1
    $\begingroup$ Related? mathoverflow.net/a/412400/7206 $\endgroup$
    – Asaf Karagila
    Oct 24, 2022 at 5:56
  • 1
    $\begingroup$ Why ask about CTM instead of just asking about the theory ZF + "it is not forceable that there is an amorphous set"? The CTM aspect seems irrelevant. $\endgroup$ Oct 26, 2022 at 12:29
  • 1
    $\begingroup$ @ElliotGlazer: No, it does not. Indeed, not even DC is enough. Feferman's model, which is $L(\Bbb R)$ of the Cohen model satisfies DC, and by abstract nonsense results, the Cohen model is a generic extension of it (we add a set of reals, each real is in the model, and the set is in a generic extension). So even if DC holds, there can be a generic extension with an amorphous set. Alternatively, my paper with Jonathan Schilhan shows even higher DC won't do in a direct proof. $\endgroup$
    – Asaf Karagila
    Oct 28, 2022 at 1:00

1 Answer 1


Turning my comment into an answer, an $X$ which is the universe of any finitely axiomatized theory with an infinite model must be orderable, and there must be a bijection between between $X$ and $X^2.$ Both of these follow from $X$ satisfying a large finite fragment of PA. In fact this is a characterization of such $X$ by the E.M. model construction, treating $X$ equipped with an arbitrary order as the generating set of indiscernibles. The resulting model is equinumerous with $X$ since $|X|=|X|^2$ implies $|X|=|X|^{<\omega}.$

Just from $|X|=|X|^2,$ we can deduce that a model is expansive precisely if it models choice, by Tarski's equivalence of AC with every infinite set being in bijection with its square. Trivially, such models satisfy generic non-existence of amorphous sets.

There are still a lot of open questions regarding generic non-existence of amorphous sets. Asaf's posts summarize the state of knowledge regarding this principle: it's strictly stronger than nonexistence of amorphous sets, since it fails in the Cohen model, and it's implied by "all sets are almost well-orderable." I don't even know how to show the latter is strictly stronger.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.