Idea

Intuitively, a stable entity is one that does not involve quantification.

For example, the property of loving someone is not stable. It is defined in terms of the quantifier someone.

I think canonical examples of stable properties are Boolean operations (negation, conjunction, …) and combinators (entities expressed by closed terms without occurrences of constants, like ).

Relationship with domain specifiers

Non-stable entities can be “moved” domain specifiers. Loving someone is not the same as loving someone among those who love someone: if you love someone who doesn’t love anyone, then you love someone, but you don’t love someone among those who love someone. On the other hand, stable entities should be “fixed” by all domain specifiers.

Model-theoretic representation

There is a model-theoretic representation of the notion of stability definable in Quantificational substitution structures. It is essentially the restriction of the concept of purity from @Bacon2019SS to QSSs.

Definition

Let be a QSS. An entity is called

  • -stable in when for every -domain ;
  • Stable in when it is -stable for all nice type sequences .

Stable domains

A stable domain is any domain such that is stable whenever it is defined. A stabilizing substitution is any substitution such that is a stable domain. Likewise, the notions of a -stable domain and of a -stabilizing substitution are defined.

It turns out that given minimal assumptions about the existence of stable domains, stable (resp. -stable) entities can be equivalently characterized as those entities that lie in the range of some stabilizing (resp. -stabilizing) substitution.

Proposition

be a QSS containing at least one stable (resp. -stable) -domain for each nice type sequence , and let . Then is stable (resp. -stable) in iff for all nice type sequences there a stable (resp. -stable) -domain such that .

Proof

The left-to-right direction is obvious given the existence assumptions about stable domains. Conversely, take any stabilizing substitution . When is a -domain, we know there is a stable -domain such that . But by (Composition) and the stability of this implies

So, is stable in . The same argument works for -stability. 

Stabilization

We can formulate comprehension conditions for the class of stable entities in a model.

A second class of stability-theoretic conditions consists of comprehension conditions for the class of stable entities. Some applications of Quantificationalism---most importantly metaphysical predicativity---rely on the existence of enough stable properties. Comprehension principles for stability are a natural way of ensuring just that.

Let be a model and . A stabilizer for is a stable entity  such that . The notion of a -stabilizer is defined the same way, but substituting ‘-stable’ for ‘stable.’ We can lift the notion of a stabilizer to domains. A stabilizer for a -domain is a -domain such that stabilizes whenever is defined. Likewise for the notion of a -stabilizer.

Definition

A model is called stabilized (resp. -stabilized) when every entity has a stabilizer (resp. a -stabilizer)

Stabilization does not make sense for regular, Leibnizian models: a stabilized regular Leibnizian model would be one in which every entity is stable, and there are no such models. But stabilization can be had by irregular, quasi-Leibnizian models.