There is no paradox of logical validity (Q2254573)

The paper deals with the notion of logical validity expressed by a ``predicate that applies to a premise and a conclusion if and only if the argument from premise to conclusion is logically valid''. Some authors (Field, Shapiro, Beall and Murzi among them) argue to the effect that ``Peano Arithmetic (PA) supplied with a logical validity predicate is inconsistent in much the same manner as is PA supplemented with an unrestricted truth predicate''. Let \(\mathrm{Val}(x,y)\) be the validity predicate ``that holds of the Gödel code \(\ulcorner\Phi\urcorner\) of \(\Phi\) and the Gödel code \(\ulcorner\Psi\urcorner\) of \(\Psi\) (in that order) if and only if the argument whose sole premise is \(\Phi\) and whose conclusion is \(\Psi\) is logically valid''. Let the validity predicate be subjected to the following introduction rule (\(\mathrm{VS}_1\)): If \(\Phi\vdash\Psi \), then \(\varnothing\vdash \mathrm{Val}(\ulcorner\Phi\urcorner, \ulcorner\Psi\urcorner)\), and elimination rule (\(\mathrm{VS}_2\)): \(\varnothing\vdash \mathrm{Val}(\ulcorner\Phi\urcorner, \ulcorner\Psi\urcorner) \rightarrow (\Phi\rightarrow\Psi)\). Consider the Gödelian diagonalization lemma applied to the predicate \(\mathrm{Val}(x,\ulcorner\perp\urcorner)\) to obtain a sentence \(\Pi\) such that \(\Pi \leftrightarrow \mathrm{Val}(\ulcorner\Pi\urcorner, \ulcorner\perp\urcorner)\) is a theorem. Then one can construct a formal derivation of \(\perp\) by using diagonalization, \(\mathrm{VS}_1\), \(\mathrm{VS}_2\) together with the standard first-order logic. Exactly this derivation is sometimes considered to constitute the so-called ``paradox of logical validity''. However, as the author claims, ``there is no paradox of logical validity since the construction of the paradox requires a context into which we can diagonalize. Such a context requires, in turn, a logical validity predicate (either primitive or defined in terms of a logical validity connective and the truth predicate). Such a predicate, however, requires Gödel coding or some other naming device. The presence of such coding function, however, brings with it violations of the substitutivity requirement, thus preventing \(\mathrm{VS}_1\) and \(\mathrm{VS}_2\) from being logically valid.''