"Schroedinger ... was troubled by the fact that according to quantum mechanics one party Alice (A) could seemingly ‘‘steer’’ a remote system of some other party Bob (B) into an eigenstate of some arbitrary observable. ... At present, essentially three distinct types of nonlocal correlations can be identified in nature: Bell nonlocality, Schroedinger’s steerability, and    entanglement [3,4]. Bell nonlocality, the strongest of the three, concerns the existence or not of a local hidden variable model for a set of correlations. If such a model does not exist, then the correlations violate some Bell-type inequality, and are termed Bell nonlocal. Bell nonlocality can be stated without any reference to quantum theory, and could thus be thought of as the most fundamental of the nonlocality types. ... On the lower end of the hierarchy is quantum entanglement, which was formally defined in 1989 by Werner who distinguished it from Bell nonlocality [5]. Entanglement has received a lot of attention as the underlying resource for quantum computation and certain quantum information tasks. ... The concept of steering has caught the least amount of attention. Schroedinger’s intuition that one party can remotely prepare some ensemble of quantum states by local measurement on the other part of an entangled state has been formalized in the Hughston-Josza-Wootters theorem ... We have introduced a novel inequality for identification of EPR steering correlations. Our theoretical and experimental results agree that an EPR steering criterion based on entropic uncertainty relations is capable of detecting non- local correlations which remain hidden to the criteria based on variances. This should encourage further investigation and characterization of EPR steering and its relation to the other forms of nonlocality, as well as application to quantum information tasks such as quantum cryptography."
http://prl.aps.org/abstract/PRL/v106/i13/e130402