[Eeglablist] ERSP baseline

[Scott Makeig]

arno wrote:

>> After reading through the timef code, it seems that the baseline 
>> window include all the center points for the FFT window, so a -1 to 0 
>> (seconds) baseline would have its last FFT window centered on either 
>> -1 or 0 (ms) and data from 0-63 or 64ms would contribute to the mean 
>> baseline power.  If anyone disagrees please let me know.
>
> Yes, this is correct and this is a good point. We should use baseline 
> that ends before time 0 (using the 'baseline' command line option). 
> This is especially important for low frequencies.
> Best, Arno 

I have updated the timef() help message to make this fact explicit. We 
should also add to the tutorial the caustion that to insure no post-0 
data are used the baseline period must end 1/2-window before latency 0. 
Getting an adequate amount of baseline data may then require longer 
epochs, though (e.g., > -1 s).

Scott