Fast-collapsing theories (Q2350213)

The main result of the paper is stated informally as: ``An r.e.-upgeneric theory remains true upon augmentation by knowledge of its own truthfulness.'' and formally as: ``Let \(T_0\) be an r.e.-upgeneric \({\mathcal L}_{\mathrm{EA}}\)-theory. Let \(T_1\) be the union of \(T_0\) with all universal closures of formulas of the form \(K\phi\rightarrow\phi\). Let \(T\) be the smallest theory that contains \(T_1\) and such that for all \(\phi\in T\), \(K\phi\in T\). Then T is true.'' with the notions involved in this result being defined as follows. {\parindent=6mm \begin{itemize}\item[--] \({\mathcal L}_{\mathrm{EA}}\) is the the language of epistemic arithmetic from [\textit{S. Shapiro}, in: Intensional mathematics. Amsterdam-New York-Oxford: North-Holland. 11--46 (1985; Zbl 0559.03036)], that is, the language \((0, S, +, \times)\) of Peano arithmetic, \({\mathcal L}_{\mathrm{PA}}\), complementary with the knowledge operator \(K\). \item[--] \(T_0\) is r.e.-upgeneric iff \(T_0\) is r.e.\ and \({\mathcal M}_{T^\oplus}\models T_0^\oplus\) for every r.e.\ \({\mathcal L}_{\mathrm{EA}}\)-theory (set of sentences in \({\mathcal L}_{\mathrm{EA}}\)) \(T\) that contains \(T_0\). \item[--] Let \({\mathcal L}_{\omega\cdot\omega}\) denote the language obtained from \({\mathcal L}_{\mathrm{EA}}\) by adding operators \(K_\alpha\) for all \(\alpha\in\omega\cdot\omega\). Given an \({\mathcal L}_{\mathrm{EA}}\)-theory \(T\), \(T^\oplus\) denotes the \({\mathcal L}_{\omega\cdot\omega}\)-theory \(\{\phi^+\mid\phi\in T\}\) where for all \({\mathcal L}_{\mathrm{EA}}\)-formulas \(\phi\), \(\phi^+\) is inductively defined as follows, w.r.t. an infinite subset \(X\) of \(\omega\cdot\omega\): {\parindent=12mm \begin{itemize}\item[--] If \(\phi\) is atomic then \(\phi^+\) is \(\phi\). \item[--] If \(\phi\) is of the form \(\phi_1\rightarrow\phi_2\), \(\neg\phi_1\) or \(\forall x\phi_1\) then \(\phi^+\) is \(\phi_1^+\rightarrow\phi_2^+\), \(\neg\phi_1^+\) or \(\forall x\phi_1^+\), respectively. \item[--] If \(\phi\) is of the form \(K\phi_0\) then \(\phi^+\) is \(K^\alpha\phi_0^+\) where \(\alpha\) is the least ordinal in \(X\) with \(K^\alpha\) not occurring in \(\phi_0^+\). \end{itemize}} For instance, for \(X = \{\omega,\omega\cdot 2,\omega\cdot 3,\dots\}\), \[ \bigl(K(s(0)=0)\rightarrow KK(s(0)=0)\bigr)^+\text{ is }K^\omega(s(0)=0)\rightarrow K^{\omega\cdot 2}K^\omega(s(0)=0). \] \item[--] Given an \({\mathcal L}_{\omega\cdot\omega}\)-theory \(T\), \({\mathcal M}_T\) denotes the \({\mathcal L}_{\omega\cdot\omega}\)-structure which is the intended model of \(T\), which for every \({\mathcal L}_{\omega\cdot\omega}\)-formula \(\phi\), assignment \(s\) into \(\mathbb{N}\) and \(\alpha\in\omega\cdot\omega\), satisfies \[ {\mathcal M}_T\models K^\alpha\phi[s] \text{ iff } T\cap\alpha\models\phi^s. \] where: {\parindent=12mm \begin{itemize}\item[--] \(T\cap\alpha\) is the subset of \(T\) all of whose members contain no occurrence of \(K^\beta\) for \(\beta\geq\alpha\); \item[--] \(\phi^s\) is the sentence obtained from \(\phi\) by replacing all free variables \(x\) in \(\phi\) by the numeral that is the value of \(s(x)\). \end{itemize}} \end{itemize}} With the restricted condition that the background theory of knowledge is r.e.\ and built of r.e.-generic pieces along with truthfulness, Reinhardt's conjecture, proved in [\textit{T. J. Carlson}, Ann. Pure Appl. Logic 105, No. 1--3, 51--82 (2000; Zbl 0973.03019)], is a particular case of this result, whose proof uses induction up to \(\omega\cdot\omega\), whereas in [loc. cit.] induction is up to \(\varepsilon_0\cdot\omega\).