Maximality vs. optimality in dyadic deontic logic. Completeness results for systems in Hansson's tradition (Q484199)

In the semantics for dyadic deontic logic, a proposition \(X\) is usually treated as obligatory given proposition \(A\) iff \(X\) is true in the best among the worlds in which \(A\) is true. However, there are two main ways of reading `best' in this context -- as maximal (none other greater) or optimal (greater than all others); these two understandings can, in principle, have different effects. Again, it is common in such a semantics to impose a condition that reflects, in first-order terms, part of the content of the second-order constraint that there are no infinite ascending chains. There are at least two main ways of articulating such a constraint -- as what is often called a `limit' condition (if a formula is satisfied by some state then there is a best among the states that satisfy it) or as the stronger smoothness aka stoppering condition (if a formula is satisfied by a state then there is a best among the states that both satisfy it and are greater than or equal to the given state). This gives rise to a four-fold range of closely related semantic options, and the question arises: how far does choice between them modify the induced deontic logic? The paper under review investigates this problem for certain deontic logics in the tradition of Bengt Hansson and Lennart Åqvist, focusing particularly on Åqvist's F (a system allowing not only iterated deontic modalities but also alethic modalities interacting with them) supplemented by the postulate CM of deontic cautious monotony (if each of \(B,C\) is obligatory given \(A\), then so too is \(C\) given the conjunction of \(A\) with \(B\)). The treatment is rich and thorough, and some of its many results are quite surprising. A general pattern that emerges is that the contrast between maximality and optimality readings of `best' is not as significant for the induced logic as one might expect. In particular, the system F+CM is sound and complete for the class of models in which the relation is both reflexive and smooth, independently of whether we use optimality or maximality in the evaluation of obligation (and also independently of whether the relation is required to be complete). If transitivity and completeness are jointly imposed on the relation then we again get the same induced logic (known as G) independently of the maximality/optimality option. As well as its value for understanding the particular system F+CM and its close neighbours, this paper provides a rich toolbox for researchers who wish to ascertain the outcomes of such options for other deontic logics, indeed for any kinds of logic using a preferential type semantics.