Social Causality and Responsibility: Modeling and Evaluation

Causality is a central issue in many AI applications. Social causality, in contrast to physical causality, seeks to attribute cause and responsibility to social events, and accounts for how an intelligent entity makes sense of the social behavior of others. Modeling the underlying process and inferences of social causality can enrich the cognitive and social functionality of intelligent agents. In this paper, we present a general computational model of social causality and responsibility. Our model incorporates the basic features people use in their judgments, including physical causality, coercion, intention and foreknowledge. We propose commonsense reasoning of these features from plan knowledge and observation, and empirically evaluate and compare the model with several other models.

• Type: Conference
• Year: 2005
• Where: IVA
• Authors: Wenji Mao, Jonathan Gratch
• Keywords: Entity Makes Sense, IVA 2005, Physical Causality, Social Causality