Abstract: This article explores the applicability of causal Bayesian networks in the causal interpretation of quantum mechanics, with a focus on the compatibility of quantum entanglement with causal principles. To address the quantum entanglement problem within the framework of causal Bayesian networks, it is necessary to abandon assumptions such as causal locality, exogeneity of control variables, or temporal ordering. The non-locality of quantum entanglement and the no-signaling theorem cannot simultaneously satisfy the core principles of causal Bayesian networks—namely, the causal Markov condition and the faithfulness principle. Of these two principles, the faithfulness principle can be relinquished. Among the various causal structures that abandon specific assumptions, retrocausality emerges as the most compatible with the structure of causal Bayesian networks. Further analysis of the feasibility of retrocausal interpretations reveals that, while it aligns with the acyclicity requirement of causal Bayesian networks, existing physical models have already exceeded the descriptive capabilities of traditional causal graphs. The article contends that quantum causal research must develop new tools beyond classical causal Bayesian networks to provide a unified interpretation of the causal essence of both quantum and classical phenomena.

Key Words: Causal Bayesian Network; Quantum entanglement; Retrocausation; Non-Faithfulness; EPR Experiment