[Eeglablist] Power Spectra - Window Type

[Arnaud Delorme]

Hi Marco,

if you are using epoched data, there is no overlap by default. FFT is performed on each epoch unless you use the 'continuous', 'on' option. If you are using the 'continuous' option or if you have continuous data, then yes, the default data segment overlap is 50%.

Best,

Arno

On 9 Feb 2013, at 06:24, Marco Montalto wrote:

> Thanks so much Arno for answering all of my questions! I am truly grateful for the that!
> 
> One very last question regarding (3): if I am using epoched data in a STUDY should I therefore assume that the overlap is the default 50% then?
> 
> Thanks once again!
> 
> Regards,
> Marco
> 
> On 09 Feb 2013, at 06:14, Arnaud Delorme <arno at ucsd.edu> wrote:
> 
>> Dear Marco,
>> 
>> yes, you are right. If the 'continuous' option is left to 'off', then the FFT decomposition will be perform on whole data epochs. If 'continuous' is set to 'on' then each data epochs will be considered as continuous EEG data and smaller data segments will be extracts.
>> 
>> 
>>> So really my question is: if 'continuous' is left 'off' (which is the default value), (1) is the entire (in my case) 2-second epoch being used as window size for the FFT
>> 
>> 
>> Yes
>> 
>> 
>>> (2) what window type, if any, is being used for the FFT
>> 
>> 
>> Your full 2-second window, tapered by a hamming function to minimize overlap.
>> 
>> 
>>> (3) can I somehow change the amount of overlap?
>> 
>> 
>> No. To change overlap, you would have to build a STUDY with the continuous data.
>> Best,
>> 
>> Arno
>> 
>> 
>