Development of innovative tools for the application of argumentation in behavioral economics

Consumer purchasing decisions can be considered as a form of human reasoning based on preferences. There are two main schools of thought about preferences. While mentalism claims that preferences reflect a person’s true mental state, behaviorism is the view where preferences are considered a mathematical construct.

According to behaviorists, it is people’s actions, not words, that determine their preferences. Economists support these consumer behavioral preferences with the Revealed Preferences Theory (RPT), also known as consumer theory.

Since reasoning involves preferences, it is instructive to generalize RPT to artificial intelligence (AI), now dictated by mentalism.

With this in mind, Professor Van-Nam Huynh of the Japan Advanced Institute of Science and Technology (JAIST) and Assistant Professor Nguyen Duy Hung of Sirindhorn International Institute of Technology, Thammasat University in Thailand recently generalized consumer theory to AI reasoning with arguments — a type of reasoning that draws its inspiration from the process of exchanging arguments between people to draw conclusions in everyday life.

In an article published in International Journal of Approximate Reasoningresearchers present theoretical foundations and analytical tools for practical applications of argumentation in the analysis of mentalism and consumer behavior.

Prof. Huynh highlights the novelty of their work: “This paper combines two lines of research: AI-based reasoning and behavioral economics. In particular, it explores the relationships between economic rationality and argumentation semantics, between consumer preferences and AI agent preferences, and between consumer purchasing behavior and AI agent reasoning.”

In this work, the contributions of the scientists are threefold. First, they developed the Revealed Preference Argumentation (RPA) framework. Researchers have argued that the existing framework is governed by the opposite mentalistic interpretation of preferences. In doing so, they reconstructed and unified the two main approaches to RPT in terms of argumentation, showing that consumer analyzes based on RPT, including various rationality checks of consumer behavior and extrapolations of such behavior, can be interpreted as computational tasks in RPA.

Researchers then successfully integrated mentalism and behaviorism to provide an integrated preference argumentation (IPA) framework. They determined that South Africa is just a special case of IPA with only “revealed” preferences. This finding is particularly important because the existing preference-based argumentation framework is presented as an IPA framework with only “stated” preferences.

Finally, the researchers developed comprehensive IPA algorithms, rigorously setting their accuracy and termination to a general class IPA framework. Researchers successfully implemented the algorithms in Prolog, a logical programming language related to artificial intelligence and computational linguistics, and obtained an IPA reasoning engine. Then they tested the developed tool for effective analysis of consumer behavior based on RPT.

Taken together, this work represents a remarkable advance in the largely unexplored area of ​​consumer behavioral economics. “This paper provides not only theoretical and algorithmic foundations, but also programming tools for applications of argumentation in behavioral economics in analyzes of consumer behavior, such as rationality checks, consumer preference retrieval and behavior extrapolations,” notes Prof. Huynh.