The Physics of crypto-nonlocality

In 2003, Leggett introduced his model of crypto-nonlocality based on considerations on the reality of the photon polarization. In this note, we prove that, contrary to hints in subsequent literature, crypto-nonlocality does not follow naturally from the postulate that polarization is a realistic variable. More explicitly, consider physical theories where: a) faster-than-light communication is impossible; b) all physical photon states have a definite polarization; and c) given two separate photons, if we measure one of them and post-select on the result, the measurement statistics of the remaining system correspond to a photon state. We show that the outcomes of any two-photon polarization experiment in these theories must follow the statistics generated by measuring a separable two-qubit quantum state. Consequently, in such experiments any instance of entanglement detection -and not necessarily a Leggett inequality violation- can be regarded as a refutation of this class of theories.