History of RIDE


RIDE is a method for decomposing and reconstructing ERP and analyzing the trial-to-trial variability information of ERP. It was initiated and developed by Guang Ouyang, Changsong Zhou and Werner Sommer from 2011.


The main utility of RIDE is to overcome the limitation of conventional stimulus-locked average ERP that the waveform is blurred by trial-to-trial latency variability. This limitation leads to many problems such as distortion of ERP waveform and attenuation of ERP effects, and so on.

In general, RIDE separates the ERP (with RT) to three component clusters: the stimulus-locked component cluster S, the response-locked component cluster R and the central, ‘neither-nor’ component cluster C. The separation paradigm can be extended to more components (e.g., more than one C components, or only S and R, or only S and C in the data without RT). But the attempt to separate larger number of components is not suggested because it always complicates the problems and the reliability of separating large number of components will be compromised.

The basic idea of RIDE was firstly published in 2011 (Ouyang, et al.) which presented the basic algorithms of RIDE and the first application of RIDE. After two years of constantly upgrading, the algorithms of RIDE converged to a robust version, all of the new algorithms are summarized in the toolbox paper Ouyang et al., 2015. Briefly, the new RIDE employs L1-norm minimization to prevent serious distortion encountered by least-square-based algorithm. Besides, several algorithms for refinements of the waveform were also depicted in Ouyang et al., 2015.





<- Home