Formalizing contextuality in sequential scenarios

Abstract: This paper provides a framework for characterizing sequential scenarios, allowing for the identification of contextuality given empirical data, and then provides precise operational interpretations in terms of the possible hidden variable model explanations of that data. Sequential scenarios are different in essence from non-local scenarios as each measurement instrument is allowed to change the state as it enters subsequent measurement instruments. Thus, it is necessary to formulate the possible state update in any hidden variable model description. Here we develop such hidden variable models for sequential scenarios, and we propose the notion of no-disturbance: an instrument $A$ does not disturb another instrument $B$ if the statistics of $B$ are independent of whether $A$ was measured or not. We define non-contextuality inequalities for the sequential scenario, and show that violation implies that the data cannot be explained by a hidden variable model that is both deterministic and not disturbing in this sense. We further provide a translation from standard contextuality frameworks to ours, providing sequential versions which carry over the same inequalities and measures of contextuality, but now with the sequential interpretations stated.