Relational models for the modal syllogistic (Q1362593)

Suppose that for every term \(x\) in Aristotle's syllogistic there are some things that are actually-\(x\), some that are necessarily-\(x\), and some that are possibly-\(x\). Label these extensions \(|x|\), \(|x^\square |\), and \(|x^\lozenge |\). And suppose further that \(|x^\square |\subseteq|x|\subseteq |x^\lozenge |\). From these assumptions Thomason proposes a semantics for Aristotle's logic in which \(|x|\) and \(|x^\lozenge |\) are defined in terms of \(|x^\square|\) using two relations \(R\) and \(S\). \(|x|\) is defined in such a way that for any individual \(a\), \(a\in|x|\) iff \(\exists b(b\in|x^\square|\) and \(bRa)\). \(|x^\lozenge |\) is defined similarly: for any individual \(a\), \(a\in|x^\lozenge |\) iff \(\exists b(b\in |x^\square|\) and \(bSa) \). These can be understood as saying that for any individual \(a\) to be \(x\) is for \(a\) to be \(R\)-related to an individual \(b\) that is necessarily-\(x\). For \(a\) to be possibly-\(x\) is for \(a\) to be \(S\)-related to an individual \(b\) that is necessarily-\(x\). Thomason imposes various restrictions on \(R\) and \(S\) and bases models for the syllogistic on these relations. The apodeictic syllogisms valid in all these relational models are shown to be exactly those affirmed in Storrs McCall's (1963) classification of the syllogistic. Thomason does not attempt to give textual support for the \(R\) and \(S\) relations or for the restrictions he imposes on them. One possibility might be to interpret \(R\) as the relation between a substance and an accident that holds when the substance has the accident in it. Then \(S\) is the relation between a substance and an accident if the accident is the kind of thing the substance could have in it. What is less obvious is that Thomason's requirement that \(R\) and \(S\) be symmetric relations fit Aristotle's reasoning.