Completeness and conservative extension results for some Boolean relevant logics (Q908897)

This paper is a study of the Boolean extensions of some interesting relevant logics, specifically of the system T of ticket entailment, its contraction-free counterpart TW, and RW, the contraction-free counterpart of the system R of relevant implication. The systems are axiomatised by adding: 1. \(A\to.B\to C\vee \neg C\), 2. \(A\to B\to.A>B\), 3. A\&\(\neg A\to B\) to the non-Boolean systems, with `A\(>B'\) defined as `\(\neg A\vee B'\). (See \textit{A. R. Anderson} and \textit{N. D. Belnap jun.} [Entailment, Vol. 1 (1975; Zbl 0323.02030)] for the axiomatisations of the original systems.) The unexpected strength of these axioms is needed to accomodate T and TW. [Cf. \textit{R. Routley} and \textit{R. K. Meyer}, Stud. Logica 33, 183-194 (1974; Zbl 0316.02030).] The systems are then fitted with semantics which are just as one would expect from Routley and Meyer [loc. cit.] and \textit{R. Routley}, \textit{V. Plumwood}, \textit{R. K. Meyer} and \textit{R. T. Brady} [Relevant logics and their rivals, Vol. 1 (1982; Zbl 0579.03011)]. In particular, the usual postulates \(p_ 1)\) \(R_ Oaa\) and \(p_ 2)\) ROabc\(\Rightarrow Rabc\) are replaced by the stronger \(p_ 1')\) \(R_ Oab\) iff \(a=b\). Consistency and completeness are shown with an argument adapted from \textit{R. Routley} and \textit{R. K. Meyer} [Stud. Logica 32, 51- 68 (1973; Zbl 0316.02029)]. The arguments for conservative extension become more interesting. First, it is shown that the Boolean extension of T with or without intensional conjunction (fusion) is a conservative extension of the original system. The argument relies upon a proof, given in the paper, of a (carefully chosen) version of semantic normality, as would be expected from \textit{R. Routley} and \textit{R. K. Meyer} (1974; loc. cit.). As there, it is shown that a normal T model that refutes some given formula can be ``changed into'' a Boolean T model that does the same. However, the recipe for constructing the corresponding Boolean model is an adaptation of the reconstruction used by \textit{A. Q. Abraham} [The semantics of classical relevant logics, Thesis, Univ. Canterbury, New Zealand (1977)]. Semantic normality is not available for the contractionless systems, so the usual argument for conservative extensions must be modified for Boolean TW and RW. In particular, when moving from a TW or RW model to its corresponding Boolean model, one is forced to introduce a new \(0^*\) as well as a new 0. The use of some results of \textit{J. Slaney} [Stud. Logica 43, 159-168 (1984; Zbl 0576.03014)] to show that TW and RW each have a single canonical model which will refute all of their respective non-theorems makes the task somewhat easier. However, Slaney's results do not include fusion in TW, so that connective is abandoned. Further, it is shown that the Boolean extensions of TW, RW, T and R with the sentential constant t are not conservative extensions of their originals (with t), since \(t\vee (\sim t\&t\to A)\) is not a theorem of R (with t), but is valid even in the Boolean TW semantics. The authors report that it was this odd behavior of t that led to the results of \textit{R. K. Meyer}, \textit{S. Giambrone} and \textit{R. T. Brady} [Stud. Logica 43, 247-256 (1984; Zbl 0586.03016)].