Experiments for Assessing Floating Reinstatement in Argument-based Reasoning

Various Artificial Intelligence semantics have been developed to predict when an argument can be accepted, depending on the abstract structure of its defeaters and defenders. These semantics can make conflicting predictions, as in the situation known as floating reinstatement. We argue that the debate about which semantics makes the correct prediction can be informed by the collection of experimental data about the way human reasoners handle these critical cases. The data we report show that floating reinstatement yields comparable effects to that of simple reinstatement, thus supporting preferred semantics over grounded semantics. Besides their theoretical value for validating and inspiring argumentation semantics, these results have applied value for developing artificial agents meant to argue with people.