This talk explores the fundamental distinctions between bipartite quantum correlations across spacelike and timelike separated measurements. While spacelike separated measurements yield local probabilities that remain independent of one another, sequential measurements are governed by the Lüders-von Neumann distribution, which introduces measurement back-action. This measurement dependence creates a temporal asymmetry that provides a clear arrow of time not present in spatial correlations. To mathematically encode these temporal expectation values, the pseudo-density matrix (PDM) formalism is introduced. Unlike standard density matrices that are strictly positive semi-definite, PDMs can admit negative eigenvalues, which act as a witness to quantum correlations that imply causation.