[Eeglablist] using calibration pulses and establishing 2-channel references

arno arno at salk.edu
Thu May 11 03:10:54 PDT 2006


Dear Ben,
> 1. I have a set of calibration pulses in my data.  They are marked by
> a "condition code" of 0.  First, how can I extract just this data (I
> can't find a way to select data of a specific condition code). 
> Second, how do I actually use the calibration pulses to normalize the
> EEG data?
>   
The menu Edit > Select epoch/events allow you to select epochs 
containing a specific code (or to select events). You may also use the 
command line function pop_selectevent(). Then to normalize the EEG data 
(which is, I have to admit, a strange concept for me), you should do it 
from the command line.

For instance, to normalize all data trials (each trial individually but 
without using a loop)

EEG.data = (EEG.data - repmat(mean(EEG.data, 3), [1 1 
EEG.trials]))./repmat(std(EEG.data, [], 3), [1 1 EEG.trials]);
> 2. Our EEG is recorded as being referenced to M1.  After we record, we
> usually re-reference to (M1+M2)/2 by subtracting channel M2/2.  In
> EEGLAB, when I re-reference my data to M2, is that doing the same
> thing?  Or is it subtracting M2 (not M2/2)?
>   
If you rereference your data to M2, then it will be referenced to M2. 
You cannot reference directly to an existing reference. Using the GUI, 
you should enter M1 as the actual reference and compute average 
reference, then enter M1 and M2 for your new reference. Note that using 
the average reference does not affect the rereferencing process.

Alternatively you can do it from the command line

EEG.data = EEG.data - repmat(EEG.data(index_m2, [EEG.nbchan 1 1]));

Arno





More information about the eeglablist mailing list