Confounding effects of indirect connections on causality estimation

摘要: Addressing the issue of effective connectivity, this study focuses on effects indirect connections inferring stable causal relations: partial transfer entropy. We introduce a Granger causality measure based multivariate version The statistic takes into account influence rest network (environment) observed coupling between two given nodes. This formalism allows us to quantify, for specific pathway, total amount mediated by environment. show that entropy is more sensitive technique identify robust relations than its bivariate equivalent. In addition, we demonstrate confounding variation in detectability links. Finally, consider problem model misspecification and effect robustness connectivity patterns, showing misspecifying may be an even model-free information-theoretic approach.