Influence of prediction time on TE. Results for quadratically coupled AR(10) processes with a coupling delay δ of 21 samples (X to Y). (A) Upper panel - Detection rates versus prediction time u. Open circles: detection of coupling from X to Y (correct detections); Cross: detection of coupling from Y to X (false positives). Interactions can be detected best, if the prediction time u is set to the coupling delay (u = δ = 21). Lower panel: Shift test detection rates (B) TE values versus prediction time u. Solid line: TE values of the generated data. Dotted line: TE values for trial-shuffled surrogate data. The difference between the TE values of the data and the surrogates is largest when the prediction time u is set to the coupling delay.
Lindner et al. BMC Neuroscience 2011 12:119 doi:10.1186/1471-2202-12-119