Extracting causal relations on auditory evoked potentials in response to amplitude modulated signals

This paper presents three existing methods, the directed coherence (DCOH), the directed transfer function (DTF) and the partial directed coherence (PDC), based on multivariate autoregressive models to evaluate causal relations between electrophysiological signals. In our study, we apply these estimators to a set of human intracranial auditory evoked potentials (AEP) in response to amplitude modulated signals, recorded on epileptic patients. After showing that these AEPs present a frequency peak at the modulation frequency (MF), we present the results obtained for a set of three time series recorded in the primary auditory cortex of one patient. From these results, we establish graphs representing causal relations at the modulation frequency for each method and show that these graphs are similar.