Capturing Argument in Agent-Based Models

Area of Science:

• Computational Social Science
• Artificial Intelligence
• Cognitive Science

Background:

• Agent-based models (ABMs) are prevalent for studying complex systems and emergent properties.
• Belief and opinion dynamics in ABMs are relevant to online social media and scientific discourse.
• Existing ABMs have not fully captured rich argumentation scenarios.

Purpose of the Study:

• To introduce a framework for agent-based models of argument.
• To distinguish and analyze three notions of argument: reasons, syllogisms, and dialectics.
• To provide an organizing scheme for comparing and choosing argument models.

Main Methods:

• Distinguishing arguments as propositional content (reasons), premise-conclusion relationships (syllogisms), and conversational deployment (dialectics).
• Utilizing the NormAN framework, which builds ABMs on Bayesian networks, as a reference model.
• Analyzing the continuum of complexity in modeling each notion of argument.

Main Results:

• A novel organizing scheme for comparing and selecting agent-based argument models.
• Clarification of how the three notions of argument constrain each other.
• Demonstration that the NormAN framework captures familiar argumentation facets and dialectical influences on argument diffusion.

Conclusions:

• Agent-based models can effectively represent complex argumentation, including dialectical aspects.
• The NormAN framework provides a versatile approach for modeling argument exchange.
• Further research can explore how dialectical considerations impact argument diffusion in populations.