Complete Inference of Causal Relations between Dynamical Systems.

From philosophers of ancient times to modern economists, biologists and other researchers are engaged in revealing causal relations. The most challenging problem is inferring the type of the causal relationship: whether it is uni- or bi-directional or only apparent - implied by a hidden common cause only. Modern technology provides us tools to record data from complex systems such as the ecosystem of our planet or the human brain, but understanding their functioning needs detection and distinction of causal relationships of the system components without interventions. Here we present a new method, which distinguishes and assigns probabilities to the presence of all the possible causal relations between two or more time series from dynamical systems. The new method is validated on synthetic datasets and applied to EEG (electroencephalographic) data recorded in epileptic patients. Given the universality of our method, it may find application in many fields of science.