Inferentializing semantics (Q975776): Difference between revisions

This paper is a contribution to ``inferentialism, a doctrine that the meaning of an expression is generally in its inferential role''. In order to sort out problems in inferentialism, the author offers a ``very general framework for measuring inferentializability of semantic systems''. He starts from a simple, naive stage and goes into more elaborate stages. To summarize the contents, this review goes the other way round. A language is, by definition, an ordered pair \(\langle S,F\rangle\), where \(S\) is a set of sentences and \(F\) is a collection of subsets of \(S\). A member of \(F\) is called a form. (Intuitively, it is the set of substitution instances of a sentential form.) A generalized semantic system is a pair \(\langle\langle S,F\rangle,V\rangle\) of a language \(\langle S,F\rangle\) and a set of valuations \(V\subseteq\{0,1\}^S\). When one ignores \(F\), `generalized' is dropped. In either case, the notion of consequeence, \(\models\), results. On the inference side, the basic ingredient is a pair \(\langle\Gamma, \Delta\rangle\), where \(\Gamma,\Delta\subseteq S^V F\), called a parametric Semi-Quasi-Inferon, shortened to pSQI-on. A pPSQI-structure is a pair \(\langle\langle S,F\rangle,\vdash\rangle\), where \(\vdash\) is a set of pSQI-ons (and where `P' stands for `Proto'). These prefixes are removed in the following manner: `p' when \(F\) is not considered, `P' when \(\vdash\) is a finite set, `S' when every \(\Gamma\) is finite, and `Q' when every \(\Delta\) is a singletons. Obviously, a class of structures is more inclusive when it carries more prefixes. The author shows that inclusions are genuine, and thus these classes form a hierarchy. On semantic systems, he deduces interesting collections of valuations, like saturated and compact systems. The main topic is the relation between \(\models\) and \(\vdash\), namely what kind of semantic systems are determined by what kind of inferential systems. Through the analysis of consequence and actual situations he concludes that only pSI- and pI-systems are of interest. They are exemplified by familiar propositional systems and true arithmetic provided by the \(\omega\)-rule, respectively. The author states that the definition of languages is overly simplified, axiomatic systems are not considered, and the framework here is not adequate for the analysis of quantifiers, hence for predicate calculi. Also, the framework of inferential structures is static: just sets of inferons are considered, and nothing about the process of inferring. The \(\omega\)-rule has ramification, like the recursive \(\omega\)-rule.