Modeling Human Reasoning on Fuzzy Implication Based on Rule Evaluation Task

Abstract

Although human reasoning on implication is an essential part of both natural fuzzy reasoning and of fuzzy systems, it is still not fully elucidated. Logical puzzles such as Wason Selection Task suggest that it varies greatly depending on external parameters such as social context. In this paper, we introduce fuzzy implication to the study in the hopes of developing methods to elucidate how people reason about implication and what those results imply about the so-called errors of implication shown in classical studies of human reasoning. A survey is conducted containing questions with varying degrees of fuzziness. The subject is asked to evaluate the degree to which the examples follow a given rule. The results are analyzed by applying various methods of fuzzy implication to each response, and suggest that common methods of implication rules closely match most human reasoning.