Empirical Evaluation of Simple Reinstatement in Formal Models of Argumentation

Abstract

This paper presents an empirical study of simple reinstatement, which purports to replicate existing findings on imperfect reinstatement and to investigate potential explanations, such as the disruption of the suspension of disbelief and a variation of it. The motivation behind it derives from the importance of replication studies and from the significance of the interpretation of this empirical finding for argumentation, as imperfect reinstatement is often used as supporting evidence in today’s work on graded acceptability. To this end, an experiment was designed where people evaluated the conclusion of an argument. The ratings from receiving arguments sequentially were compared against ratings given under two alternative manners of argument presentation: receiving all arguments at once and receiving them sequentially after being introduced to the relevant theory. A significant effect of the latter was found on ratings, but not of the former. The results successfully replicated the existing findings but did not validate the two aforementioned explanations. In order to decide whether said findings can be interpreted as supporting evidence for graded semantics, alternative explanations need to be investigated first (such as directionality of attacks and order of arguments), along with whether these findings can be generalized to fit the wider spectrum of the theoretical features they seem to instantiate.