Signal complexity and coordinated activity of neuromagnetic dynamics: a few examples, with theoretical and empirical considerations
A study published in Neuroimage, one of the leading scientific journals focusing on method development in neuroimaging and neurophysiology, explored keywords and abstracts from about 8,500 articles published in NeuroImage between 2008 and 2017. In the category of electro- and magneto-encephalography (EEG & MEG), ‘oscillations’ led the list of keywords, with an upward trend. EEG & MEG oscillations are rhythmic patterns of neural activity generated by large groups of neurons firing together. There exist a wide variety of methods to characterize these brain rhythms. One group of methods employs a combination of information theory and non-linear dynamics. In the literature, we can recognize these methods under various names, for example, entropy, signal complexity, or ‘dynamic repertoire’ of neural activity. Such an approach has several attractive properties, offering ‘model-free’ tools with high sensitivity. At this seminar, we will discuss several complexity-based non-linear measures for characterizing individual neural oscillations and their coordinated activity, such as sample entropy, cross-sample entropy, and transfer entropy. We will demonstrate their performance by reviewing a series of MEG studies exploring the somatosensory steady-state response evoked in a reaction to trains of periodically repeated tactile stimuli.