Abstract

One issue which any natural language processing by the computers must face is the contexts that involve propositional attitudes such as belief. This is made more complicated when an object has more than one name, or the reference of a name does not exist. It is thus desirable to have a formalized, rigorous logic dealing with this issue, which may serve as the theoretical basis for future implementation. The present work is such a formalization. The semantical basis of the logic FMP is the notion of modes of presentation, introduced by Gottlob Frege in "On Sense and Reference". The language of FMP includes two belief operators, representing de re and de dicto beliefs, respectively, one of which is similar to the lambda abstract used by Fitting and Medelsohn in First Order Modal Logic. We show that the axiom system of FMP is both sound and complete with respect to a natural semantics. We then apply it to some problems in the philosophy of languages, such as Kripke's puzzle about belief. This is a step toward building a "rational" robot that is able to reason about belief and naming as humans usually do