A Computational Approach to Conceptual Blending within Counterfactuals

An important topic in philosophy and the semantics of natural language has been the repre tion and verification of counterfactuals. The majority of approaches have taken a classical vi language where statements can be reasoned about solely in terms of their truth functions. T led to several well known problems in the verification of counterfactuals. We take another v inspired by cognitive linguistics, that language relies on the mappings between the concep structures (or domains) so that phenomena such as metaphor, blending and analogy are no tional uses of language but rather central aspects. To this end, we present an analysis of c factuals based on mapping between different reasoning spaces and the drawing of analogi between these spaces. This analysis is implemented in an artificial intelligence reasoning s The system, called ATT-Meta, was originally designed for uncertain metaphor-based reas (and belief reasoning), and was applied without modification to the task of verification of cou factuals. This was possible because the system approaches metaphor by a pretence me akin to counterfactual reasoning, and because it already contained powerful mechanism resolving conflicts between competing arguments.