Semantical considerations on nonmonotonic logic (Q1060866)

This paper concerns the problem of reconstruction, which may be formulated as follows. Given languages \({\mathcal L}\) and \({\mathcal M}\), and given a set A of conditions expressed in \({\mathcal M}\), find a theory Th in \({\mathcal L}\) which satisfies A. The relation R of the solution of the problem has not, in general, the so called monotonicity property, that is, R(A,Th)\&A\(\subseteq A'\) not necessarily implies R(',Th). In the paper the language of propositional calculus has been taken as \({\mathcal L}\), and \({\mathcal L}\) with a unary operator \(L: {\mathcal L}\to {\mathcal L}\) as \({\mathcal M}\). The following sense of conditions expressible in \({\mathcal M}\) has been adopted. It is assumed that \(A\cap {\mathcal L}\subseteq Th\), and \({\mathcal L}\) has the meaning of the operator \(\widetilde{\mathfrak B}\) of \textit{D. Hilbert} and \textit{P. Bernays} [Grundlagen der Mathematik. Vol. 2 (1939; Zbl 0020.19301)] of provability in Th, e.g. \(A=\{Lp\vee L\neg p| \quad p\in {\mathcal L}\}\) expresses the completeness of Th. Some related facts have been proved in the paper. A large part of the reviewed paper has a philosophical and critical character. Certain disadvantages of non-monotonic logic of \textit{D. McDermott} and \textit{J. Doyle} [ibid. 13, 41-72 (1980; Zbl 0435.68074)] and of \textit{D. McDermott} [J. Assoc. Comput. Mach. 29, 33-57 (1982; Zbl 0477.68099)] have been mentioned there. It concerns mainly the semantics of the operator L. Although one can hardly disagree with the author's opinion that the condition \(L^{-1}(Th)\subseteq Th\) is not adequate if Lp has to be interpreted as ''it is believed that p'', the proposed strengthening (*) \(L^{-1}(Th)=Th\) seems to go too far. In fact, the condition \(Th\subseteq L^{-1}(Th)\) reflects so defined meaning of the operator L, whereas (*) is the result of introducing an ''ideally rational agent'', who believes exactly in what can be proved.