The brain response displays strong variability from trial to trial in cognitive tasks as shown by the reaction times. This causes problem in the conventional ERP method based on averaging a larger number of single trial EEG. The averaging scheme distorts the waveform and is not able to cope with the problem of brain response variability, which could be an important facet of individual differences. We proposed a theoretical framework and model of ERPs consisting of temporally overlapping components locked to different external events or varying in latency from trial to trial. Based on this model a new ERP decomposition method was developed: Residue Iteration Decomposition (RIDE). RIDE decomposes ERP into different components with different extents of latency variabilities, and retrieves the single trial information of each component in terms of amplitude and latency. The waveform pattern and single trial information of ERP sub-components separated by RIDE were shown to be closely coupled with cognitive sub-processes.
Ort: INB Seminarraum
