|
Abstract:
|
The high temporal resolution of brain imaging and neural recording techniques such as LFP, EEG, and MEG offers the possibility to investigate dynamic interactions between brain areas (also known as dynamic functional connectivity). However, external stimulus presentations and the subject's engagement in a task may both induce nonstationary effects, called evoked responses, which complicate the application of standard methods for estimating dynamic functional connectivity (such as sliding-window Granger causality). Conventionally, evoked responses are estimated by averaging stimulus-aligned signals across repeated trials. However, this practice may confound functional connectivity analyses due to trial-to-trial variability in the timing, amplitude, and shape of evoked responses. We propose dynamic functional connectivity analyses which account for this variability in evoked responses and therefore correctly recover the underlying connectivity structure in cases where conventional evoked response estimation fails. In addition, our method for estimation of single-trial evoked responses may yield interesting insights on trial-to-trial variability and its relation to behavior.
|
