No abstract

No abstract

Abstract. A logic of diagnosis proceeds in terms of a set of data and one or more (prioritized) sets of expectancies. In this paper we generalize the logics of diagnosis from [27] and present some alternatives. The former operate on the premises and expectancies themselves, the latter on their consequences.

Abstract. This paper contains different approaches to solve the problem how to construct the ultimate position out of one's interventions in a discussion after possibly one or more position changes. In all approaches it is the aim to come as close as possible to human reasoning. Therefore all logics are adaptive logics. The first logic is an extension of an adaptive translation into S5 of the RescherManor mechanisms. The second one is a dynamic proof theory based on a technique using indices. In the end a satisfactory solution is given by a dynamic proof theory expressing the idea of prioritized compatibility, i.e. compatibility step by step. IntroductionApparently the position of someone in a discussion equals the consequence set of all statements he made during that discussion. Is this consequence set really the best representation for all there is to conclude from the sequence of his interventions? Next to the contents of the interventions, the sequence also contains an ordering in time. This could be more important than it seems at first sight, for there could have been an evolution in the participant's position.The interesting point of a discussion is the confrontation with other positions and the possibility to become convinced by arguments in favour of the latter. So the temporal aspect of a discussion should not be ignored for it is * The research for this paper was supported by the Research Fund of Ghent University and indirectly by the Flemish Minister responsible for Science and Technology (contract BIL98/37).† I am greatly indebted to Diderik Batens for several ideas in this paper. 278Liza Verhoeven crucial information necessary to interpret the dynamics of a participant's position and therefore to interpret his ultimate position. This means we have to consider a person's interventions as forming a prioritized inconsistent base. 1 A technique introduced and elaborated by N. Rescher and R. Manor appropriate to this context is reasoning from consistent parts of the inconsistent whole. I shall call these mechanisms Rescher-Manor mechanisms. Different approaches relevant to this situation are presented in [3]. The aims of this paper are first to make the Rescher-Manor mechanisms adaptive so that they link up better with human reasoning, second to extend these mechanisms in the hope that it brings a significant enrichment and third to look for other adaptive strategies that could be more efficient for the case of rational discussions. I shall restrict my attention to three types of RM consequences (to be defined in section 3). I shall translate them to the semantics of the modal logic S5 in section 4, so they are made adaptive. The first idea is to incorporate the RM consequences in the context of the discussive logic D2 r Joke Meheus introduced in [4]. It comes down to making a selection on the models, this technique is already used in [1], here it will be in section 5. Section 6 is an introduction to dynamic proof theories 2 . Two such proof theories are presented here for the case of rational discussio...