Towards a Formalisation of Motivated Reasoning and the Roots of Conflict

In current approaches to computational argumentation, values are presented as the means to adjudicate conflict between arguments: arguments stand in “attack” relations; values are posited to hold of an argument as a whole rather than of any constituent parts; and preference rankings between values help to determine a “winning” argument. We propose a novel, formal, alternative approach where values are constitutive in the construction of arguments and yield the source of conflict between arguments - Value-based Formal Reasoning (VFR). It accounts for motivated reasoning, wherein an Agent selects propositions relative to the Agent’s values, uses those selected propositions in their knowledge base, then constructs (instantiated) arguments from the knowledge base. Thus, the arguments themselves reflect an Agent’s values. It sets the root of conflict between arguments in differences between the values associated with the arguments’ constituent propositions rather than just with the arguments as a whole. This contrasts with current work in two respects. First, conflict between arguments is not necessarily only based on some incompatibility between their constituent propositions or more generically between abstract arguments. Second, values contribute to explaining why the Agent constructed the argument rather than just to adjudicate conflicts amongst arguments.