A Logic of Actual Cause for Nondeterministic Domains

Maryam Rostamigiv, Shakil M. Khan, Yves Lespérance, Mriana Yadkoo

Abstract: Crucial to the process of reasoning is the formalization of the actual causes of observed effects. While there has been a lot of work on root cause analysis, most studies have focused mainly on deterministic domains. In this paper, we build on previous work on actual causation in the situation calculus to deal with causation in nondeterministic domains. We use the nondeterministic situation calculus as our base framework for this. We introduce the notions of “CertainlyCauses” and “PossiblyCauses” that enable the representation of actual cause when the agent does not have any control on and does not know the choices that are made by the environment. We then investigate how regression in the situation calculus can be used to reason about these notions.