[Eeglablist] continuous or epoched data for fourier transform

[Arnaud Delorme]

Dear Andrea,

There is nothing wrong with analysing power in continuous data :-).
In any case, if you use the pwelch method (i.e. the standard EEGLAB  
plotting function), FFT is applied on "data epochs" extracted from the  
data, and the power value on each of the window is averaged. This is a  
standard way of reducing noise when computing FFT (and power) on  
continuous (or epoched) data.

Arno

> I would like to investigate power of different frequency band of
> continuous eeg data. However, I have thought of adding events and
> epoching data (because during artefact rejection, data needs to be
> compared with video coding results).
>
> Although reading earlier discussions on fourier transforms, I have not
> understood if it is a problem to use the transform on epoched data.
> Should I rather use continuous data?
>
> I would be really glad if you could share your experience (and maybe
> even some scripts) on the matter!
>
> Thanks a lot!
> Andrea