Convergent cross-mapping and pairwise asymmetric inference

Convergent cross-mapping (CCM) is a technique for computing specific kinds of correlations between sets of times series. It was introduced by Sugihara et al. [Science 338, 496 (2012).] and is reported to be a necessary condition for causation capable of distinguishing causality from standard correlation. We show that the relationships between CCM correlations proposed by Sugihara et al. do not, in general, agree with intuitive concepts of driving and as such should not be considered indicative of causality. It is shown that the fact that the CCM algorithm implies causality is a function of system parameters for simple linear and nonlinear systems. For example, in a circuit containing a single resistor and inductor, both voltage and current can be identified as the driver depending on the frequency of the source voltage. It is shown that the CCM algorithm, however, can be modified to identify relationships between pairs of time series that are consistent with intuition for the considered example systems for which CCM causality analysis provided nonintuitive driver identifications. This modification of the CCM algorithm is introduced as pairwise asymmetric inference (PAI) and examples of its use are presented.