[Eeglablist] componet and reference

Arnaud Delorme arno at salk.edu
Fri Jul 30 12:22:31 PDT 2004


Kevin Guise wrote:

>I have a question about the component finder in eeglab.  How does this work, is this finding positive and negative peaks and plotting where they origionate from? 
>
What component finder are you referring to? Is it the "envtopo()" 
function, plotting the data envelope and the contribution of component 
activities to this envelope? If yes, components are selected based on 
how much of the variance of the average data  components account for, 
over the selected time window and assuming a mean of 0. You may also 
select components based on the percentage of variance accounted for 
(variance of data minus component compared to variance of data) in the 
graphic interface.
If there is a large ERP peak, components contributing most to the peak 
will be selected (since the mean is assumed to be 0, the "variance" is 
largest at the ERP peak).

Best

Arno





More information about the Eeglablist mailing list