This package contains Matlab code for testing time-reversed Granger causality (TRGC) [1,2] as described in [1].
runSimulation.m - Start this to run a sample simulation and produce a plot similar to Figure 1b in [1].
create2dVAR.m - Function which generates a two-dimensional time series from a random autoregressive process
timeReversedGC.m - Function which performs the following statistical tests:
+ difference-based TRGC (based on bootstrapping)
+ conjunction-based TRGC (based on bootstrapping)
+ net Granger causality (based on bootstrapping)
+ standard Granger causality (F-test)
[1] I. Winkler, D. Panknin, D. Bartz, K.-R. Müller, S. Haufe (2016). Validity of time reversal for testing Granger causality. IEEE Transactions on Signal Processing, 64(11), 2746-2760.
[2] S. Haufe, V.V. Nikulin, K.-R. Müller, K. R., G. Nolte (2013). A critical assessment of connectivity measures for EEG data: a simulation study. Neuroimage, 64, 120-133.