[Eeglablist] Questions about ICA Filtering...

[arno]

aparkr83 at siu.edu wrote:
> Hello all,
>
> My name is Ashley Parker and I am a researcher in Dr. David 
> Gilbert's Integrative Neurscience Laboratory at Southern 
> Illinois University, Carbondale.  We plan to use ICA filtering 
> for some EEG data we collected from 128 channels.  I was able to 
> get 128 components for a practice file, but for others I 
> recieved only 25 or even 13 components.  I was wondering what 
> might be causing the low number of components on some files and 
> then 128 on other files.  Would any one have any suggestions on 
> how to fix this problem or tell me what I might be doing 
> incorrectly?
>   
This is most likely because the "order" of your data is lower than your 
number of channels. This means that some channels are linearly related 
to each other (because of the use of interpolation methods). This might 
also be the case if you have very few time frames (for instance try to 
decompose ERP and not single trial data) and have low precision for the 
EEG data (whole number).

I hope this helps,

Arno