A new causal-network approach removes the freedom-of-choice assumption in Bell tests

Edited by: Svitlana Velhush

In standard Bell inequalities, the measurements are assumed to be independent of the source of correlations. A new work shows that this assumption can be replaced by rigorous, experimentally testable bounds.

A group of researchers from the International Institute of Physics in Brazil, Griffith University in Australia, Sapienza University of Rome and other centers presented a preprint on 10 November 2026 on arXiv. They embedded the standard Bell scenario into a larger causal network with several independent sources. The resulting linear-programming MDC inequalities give lower and upper bounds on the degree of dependence of the measurements directly from the data.

If these bounds are incompatible, nonlocality is certified without an a priori assumption of freedom of choice. The method works even under noise and does not require ideal correlations, as in the original Fritz construction.

Most unexpectedly, genuine quantum nonlocality can be confirmed for correlations exceeding the Tsirelson bound — up to a CHSH value of about 3,03. This overturns the familiar notion that exceeding the Tsirelson bound automatically indicates "fake" nonlocality.

The approach is applicable to multi-source and multipartite Bell tests and is significantly simpler than existing methods in computational complexity and noise resilience.

The results are set out in the preprint arXiv:2611.14791 of 10 November 2026.

Experimenters now have a tool that allows them to certify quantum nonlocality under real, non-ideal conditions without relying on unprovable assumptions about freedom of choice.

3 Views

Sources

  • Bounding Freedom of Choice via Causal Network Embeddings

  • ar5iv.labs.arxiv.org

Comments

Read more articles on this topic:

Did you find an error or inaccuracy?We will consider your comments as soon as possible.