Cognitive Argumentation for Human Syllogistic Reasoning

Abstract

This paper brings together work from the psychology of reasoning and computational argumentation in AI to propose a cognitive computational model for human reasoning and in particular for human syllogistic reasoning. The model is grounded in the formal framework of argumentation in AI with its dialectic semantics for the quality of arguments. Arguments for logical conclusions are constructed via a set of proposed argument schemes, chosen for their cognitive validity, as supported by studies in cognitive psychology. The proposed model with its cognitive principles of argumentation can encompass together in a uniform way both formal and informal logical reasoning, capturing well the empirical data of human syllogistic reasoning in the recent Syllogism Challenge 2017 on cognitive modeling. The paper also argues that the proposed approach could be applied more generally to other forms of high-level human reasoning.