Actual Causation in CP-logic

TL;DR

This paper proposes using CP-logic, a probabilistic logic programming framework, as a better alternative to structural models for analyzing actual causation, emphasizing its dynamic semantics and default/deviant distinction.

Contribution

It introduces CP-logic as a novel framework for actual causation, addressing limitations of structural models by incorporating dynamic semantics and default/deviant distinctions.

Findings

CP-logic captures the deviant/default distinction in causation.

It provides a formal, dynamic semantics for causation stories.

Offers a new perspective on modeling actual causation.

Abstract

Given a causal model of some domain and a particular story that has taken place in this domain, the problem of actual causation is deciding which of the possible causes for some effect actually caused it. One of the most influential approaches to this problem has been developed by Halpern and Pearl in the context of structural models. In this paper, I argue that this is actually not the best setting for studying this problem. As an alternative, I offer the probabilistic logic programming language of CP-logic. Unlike structural models, CP-logic incorporates the deviant/default distinction that is generally considered an important aspect of actual causation, and it has an explicitly dynamic semantics, which helps to formalize the stories that serve as input to an actual causation problem.