[Eeglablist] ICA on multi-epoch data

arno arno at salk.edu
Tue Feb 13 01:07:22 PST 2007


Dear Peng,

> I'm using EEGLab to perform ica decomposition. I have an EEG data
> of 120 epochs (trials).
> When doing ica, would the data from the same channel (in different
> epoch) be considered together? If so, how were they combined

Yes, they will be considered together. They are not combined (except if 
you use the SOBI
ICA algorithm). If you use the standard runica() Infomax algorithm, each 
time point is
considered independently. You can shuffle them within and across epoch 
and this would
not change the ICA results (except from minor changes due to starting 
conditions and
random shuffling of time points at each iteration of the algorithm).

Best,

Arno




More information about the eeglablist mailing list