Math 557 Sep 19

Henkin Theories

Key Concepts

  • An \(\mathcal{L}\)-theory \(T\) is a Henkin theory if for every sentence of the form \(\exists x \psi\) there exists a constant \(\mathcal{L}\)-term \(t\) such that \[ T \vdash_{\mathcal{L}} \exists x \psi \; \to \; \psi_{t/x} \]

  • Questions:

    • What are Henkin theories needed for?
    • Do they exist?
    • Can we extend a given consistent theory to a Henkin theory that is still consistent?

Problems

Exercise 1
As before, let \(\mathcal{A}_T\) be the term model of \(T\). Pinpoint exactly where we might run into difficulties when we are trying to prove \[ \mathcal{A}_T \models \exists x \psi \; \iff \; T \vdash_{\mathcal{L}} \exists x \psi \]

Exercise 2 (Interlude)
What if in \(\exists x \psi\) the variable \(x\) is not free? Does that make a difference regarding being a Henkin theory?

Exercise 3
Can you find a couple of examples of Henkin theories?

The basic construction step for extending an \(\mathcal{L}\)-theory \(T\) to a Henkin theory is for every sentence of the form \(\sigma \equiv \exists x \psi\),

  • add a new constant symbol \(c_\sigma\) to \(\mathcal{L}\)
  • add the formula \(\exists x \psi \; \to \; \psi_{c_\sigma/x}\) to \(T\).

Exercise 4
Why could the addition of a constant theoretically lead to \(T\) being inconsistent, even though \(T\) itself remains unchanged?

And why does \(T\) actually remain consistent?

The Henkin extension \(T_H\) of \(T\) is obtained by an iterative process:

Single iteration step:

  • \(\mathcal{L}' = \mathcal{L} \cup \{c_\sigma \colon \sigma \: \mathcal{L}\text{-sentence of the form } \exists x \psi \}\)
  • \(\Gamma = \{ \exists x \psi(x) \to \psi_{c_\sigma/x} \colon \sigma \: \mathcal{L}\text{-sentence of the form } \exists x \psi \}\)
  • \(T' = T \cup \Gamma\) (an \(\mathcal{L}'\)-theory)

Exercise 5
Why is \(T'\) above not necessarily a Henkin theory?

Iteration process:

  • \(\mathcal{L}_0 = \mathcal{L}, \; T_0 = T\)
  • \(\mathcal{L}_{n+1} = \mathcal{L}'_n, \; T_{n+1} = T_n'\)
  • \(\textstyle \mathcal{L}_H = \bigcup_{n \in \mathbb{N}} \mathcal{L}_n, \; T_H = \bigcup_{n \in \mathbb{N}} T_n\)

Exercise 6
Why is \(T_H\) a Henkin theory?

\(T_H\) is no longer a purely symbolic extension of \(T\) (in the sense that we simply extend the language), since we add new sentences to \(T\) (the sentences in \(\Gamma\) for each iteration step).

Take-home Problem


Show that for every \(\mathcal{L}\)-formula \(\varphi\), \[ T_H \vdash_{\mathcal{L}_H} \varphi \; \iff \; T \vdash_{\mathcal{L}} \varphi \]