CONICET | Buscador de Institutos y Recursos Humanos

An argument is reinstated when all its defeaters are in turn ultimatelydefeated. This is a kind of principle governing most argument systems in AI.Nevertheless, some criticisms to this principle have been offered in the literature.Assuming that reinstatement is prima facie acceptable, we analyze some coun-terexamples in order to identify common causes. As a result, we found that theproblem arises when arguments in a chain of attacks are related by speciﬁcity:when non-maximally speciﬁc arguments are reinstated, fallacious justiﬁcationsare originated. Particularly, we show how the problem affects DeLP, a systemthat combines a speciﬁcity-based defeat criterion with a warrant process satisfy-ing reinstatement. Following old intuitions by Carl Hempel about the speciﬁcityof inductive explanations, we propose to rethink the concept of speciﬁcity in defeasible argumentation. Our analysis leads us to propose a general requirement ofmaximal speciﬁcity for defeasible reasoning. We identify two kinds of speciﬁcitydefeaters that make an argument non-maximally speciﬁc: proper defeaters and cautious defeaters. While proper defeaters are well-known, cautious defeaters areformally introduced here. A system combining cautious and proper defeaters is deﬁned as an extension of DeLP, and dialectic warrant games are proposed forﬁltering out non-maximally speciﬁc arguments.