Revealing the Temporal Dynamics in Non-invasive Electrophysiological recordings with Topography-based Analyses

The fine temporal resolution of electroencephalography (EEG) makes it one of the most widely used non-invasive electrophysiological recording methods in cognitive neuroscience research. One of the common ways to explore the neural dynamics is to create event-related potentials (ERPs) by averaging trials, followed by the examination of the response magnitude at peak latencies. However, a complete profile of neural dynamics, including temporal indices of onset time, offset time, duration, and processing speed, is needed to investigate cognitive neural mechanisms. Based on the multivariate topographic analysis, we developed an analytical framework that included two methods to explore neural dynamics in ERPs. The first method separates continuous ERP waveforms into distinct components based on their topographic patterns. Crucial temporal indices such as the peak latency, onset and offset times can be automatically identified and indices about processing speed such as duration, rise, and fall speed can be derived. The second method scrutinizes the temporal dynamics of identified components by reducing the temporal variance among trials. The response peaks of signal trials are identified based on a target topographic template, and temporal-variance-free ERPs are obtained after aligning individual trials. This method quantifies the temporal variance as a new measure of cognitive noise, as well as increases both the accuracy of temporal dynamics estimation and the signal-to-noise ratio (SNR) of the ERP responses. The validity and reliability of these methods were tested with simulation as well as empirical datasets from an attention study and a semantic priming (N400) study. Together, we offer an analytical framework in a data-driven, bias-free manner to investigate neural dynamics in non-invasive scalp recordings. These methods are implemented in the Python-based open-source package TTT (Topography-based Temporal-analysis Toolbox).