Non-genuine MacIntosh logics (Q1316411)

This paper continues the exploration, begun by Chellas and Segerberg (see the review above), of normal modal logics in which the Macintosh rule: If \(\vdash \diamondsuit A\to A\), then \(\vdash A\to \square A\), is admissible for all \(A\), and in particular it examines the relations between this rule and the disjunction rule (DR): If \(\vdash \square A_ 1 \vee\dots\vee \square A_ n\) then \(\vdash A_ i\) for some \(1\leq i\leq n\), and the rule of margins (RM): If, for some \(A\), \(\vdash A\to \square A\), then either \(\vdash A\) or \(\vdash \sim A\). We learn that if a normal logic has DR admissible, then it has RM admissible, but not every normal system with RM has DR. If a normal logic has RM, then it is a non-genuine MacIntosh logic, where an \(M\)-logic is genuine when, for some \(A\), it has \(\vdash A\to \square A\), but neither \(\sim \diamondsuit A\) nor \(\square A\). \(KD\) \((= K+\vdash \diamondsuit \top)\) is the smallest normal logic with RM. The smallest \(M\)-logic is \(K+\vdash \square \diamondsuit \top\). If \(L\) is normal, then \(L+ \vdash \diamondsuit \top\) has RM if \(L+ \vdash \square \diamondsuit \top\) is a non-genuine \(M\)-logic, and a normal extension of \(KD\) has RM iff it is a non-genuine \(M\)-logic. \(K+ \vdash \square \diamondsuit \top\) is non-genuine. This paper goes on to answer two questions of Chellas and Segerberg; both answers are negative. First, the systems \(KB^ n\) \((=K+ \vdash \square \diamondsuit^ n A)\) for \(n\geq 1\), though \(M\)-logics, are non-genuine. Second, if \(L\) and \(L'\) are both \(M\)-logics, it does not follow that \(L+L'\) is an \(M\)-logic, where \(L+L'\) is the intersection of all normal logics containing both \(L\) and \(L'\).