The logics of strict-tolerant logic (Q253131)

A predicate \(T\) in a language is a transparent truth predicate if, for every valuation \(v\), for the language and every sentence \(A\), \(v(T\langle A\rangle) = v(A)\), where \(\langle A\rangle\) is a name of \(A\). Because of the semantic paradoxes, like the Liar and especially Curry's, it is well known that a suitably expressive language cannot contain such a predicate and also be completely governed by classical logic. Many accounts of truth and the paradoxes would apply a 3-valued logic, like Kleene's K3 or Priest's LP (based on the same matrices). The present paper investigates another, related approach, due to \textit{D. Ripley} [Rev. Symb. Log. 5, No. 2, 354--378 (2012; Zbl 1248.03012)] and \textit{P. Cobreros} et al. [Stud. Log. 100, No. 4, 855--877 (2012; Zbl 1260.03058); J. Philos. Log. 41, No. 2, 347--385 (2012; Zbl 1243.03038)]. This is based on a ``strict-to-tolerant logic'' ST that contains a non-transitive consequence relation, whereby \(\Gamma \models \Delta\) just in case, for every valuation, \(v\), if \(v(\gamma) = 1\), for every \(\gamma \in \Gamma\), then \(v(\delta) > 0 \) for some \(\delta \in \Delta\). This is supposed to deal with the paradoxes while also having a transparent truth predicate together with full classical logic. The authors here demonstrate, however, that when meta-inferences are taken into account the logic is significantly weaker than classical logic, and indeed is very close to the paraconsistent LP. This result draws on a distinction between the internal and external logics of the language, which for ST do not coincide. While the internal logic may be classical, the external is LP. The authors establish the latter proof-theoretically, applying a three-sided sequent calculus, motivated by the 3-valued semantics for ST and its consequence relation.