On the status of reflection and conservativity in replacement theories of truth (Q1782246)

The Liar and related paradoxes seem to show an inconsistency in our ordinary concept of truth. \textit{K. Scharp}, in [Replacing truth. Oxford: Oxford University Press (2013)] and subsequent works, has proposed replacing the ordinary concept of truth with two distinct concepts, ascending truth and descending truth (ADT). The present paper argues two points against Scharp's strategy. (i) A reflection principle for descending truth (If a sentence is proven, then it is D-true) is inconsistent with ADT, but a reflection principle for ascending truth (If a sentence is proven, then it is A-true), though consistent with ADT, is insufficient to validate the sentence, and hence cannot act as a reflection principle. (ii) \textit{W. Dean} has argued [Philos. Math. (3) 23, No. 1, 31--64 (2015; Zbl 1380.03068)] that reflection principles are linked to mathematical induction within the relevant theory from this and the failure of its reflection principles, it follows that ADT must reject some ordinary mathematical uses of ``true''. \textit{H. Field} [Saving truth from paradox. Oxford: Oxford University Press (2008; Zbl 1225.03006)] argues on Gödelian grounds that consistency statements and reflection principles need not be derivable; this may exempt ADT from proving its own reflection principle, but not from its inability to prove reflection principles for Peano arithmetic or Zermelo Fraenkel. ADT thus fails to provide for all ordinary uses of a truth predicate in mathematics. In appendices, there is a proof that ADT is consistent (more constructive than those previously published), and that ADT is conservative for ZFC.