The traditional Bayes' rule lays the foundation for causal reasoning and finding relations between cause (input) and effect (output). This causal reasoning is universally applied to all physical processes to establish causal relations. Here we show that it does not establish correct causal correspondence between quantum causes and effects in general. In fact, there are instances within the framework of quantum mechanics where the use of the traditional Bayes' rule leads to inconsistencies in quantum measurement inferences. We consider two such cases, inspired by Frauchiger-Renner's and Hardy's setups, where the traditional Bayes' rule results in paradoxical situations even after assuming quantum mechanics as a nonlocal theory. As a remedy, we introduce an input-output causal relation using the reasoning based on a quantum Bayes' rule. It applies to general quantum processes even when a cause (or effect) is in coherent superposition with other causes (or effects), involves nonlocal correlations as allowed by quantum mechanics, and applies in the cases where causes belonging to one system induce effects in some other system as happens in quantum measurement processes. This enables us to propose a resolution to the contradictions that appear in the context of Frauchiger-Renner's and Hardy's setups. Our results thereby affirm that quantum mechanics, equipped with a quantum Bayes' rule, can indeed consistently describe the use of itself.

• Received 14 August 2022
• Revised 9 June 2023
• Accepted 13 July 2023

DOI:https://doi.org/10.1103/PhysRevA.108.012224