Towards root cause analysis in hybrid dynamic domains

Abstract

Reasoning about actual causes of observed effects is fundamental to the study of rationality. As such, this important problem has been studied since the time of Aristotle, with formal mathematical accounts emerging recently. We live in a world where change due to actions can be both discrete and continuous, i.e., hybrid. Yet, while there has been extensive research on actual primary and indirect causes in discrete dynamic domains, only few recent studies address causation in such hybrid domains. Building on recent progress, in this thesis I propose a first definition of primary cause in a hybrid temporal action-theoretic framework. My proposal is limited to primitive observations/effects. I also show how a variant of my definition can be interpreted from a counterfactual perspective and hint how the account can be modified to work with conjunctive/disjunctive effects. My proposal is set within a hybrid variant of the situation calculus. I show that my formalization has some basic intuitive properties.