[Eeglablist] When to select Additive Baseline in ERSP Study Function?

[Arnaud Delorme]

In one case, if you calculate ERSP using (post-stimulus uV^2/Hz)/mean_across_time(pre-stimulus uV^2/Hz) this is a divisive baseline (gain model). In the other one you do  ERSP(post-stimulus uV^2/Hz) - mean_across_time(pre-stimulus uV^2/Hz). This is an additive model.

In case of the checkbox for plotting ERSP, the additive baseline means that the different conditions' baseline is averaged and then subtracted from each of the ERSP condition (independently for each individual subject). So this is using the second additive model. This limitation will be removed in EEGLAB 15 where the type of common baseline used is the same as the one you have used to compute each condition ERSP (may be gain model or additive model).

Best wishes,

Arno

> On Jul 18, 2018, at 11:53 AM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu> wrote:
> 
> Dear Kathleen,
> 
> > Can someone explain when one would want to select  the "assume additive baseline" option when in the Study Function: ERSP. 
> 
> This is my guess--the default ERSP baseline is (post-stimulus uV^2)/mean_across_time(pre-stimulus uV^2). This is to compute change ratio. This is why when we see 6dB increase in ERSP power, we say '200% power compared with baseline'. This approach has a merit in avoiding across-frequency normalization problem.
> 
> However, you can also take an average across conditions to compute baseline value, assuming the baseline calculation is additive, just like ordinary ERP baselines are. This approach does not make much sense if you think what ratio you are computing there...
> 
> > After some searching we are still not clear on what this "additive" option 
> 
> It's not clear to me either.
> 
> > Specifically, we would like to know what is happening with the baseline when we select, or don't select the additive baseline option?   And does it matter if our epochs already have a baseline created during epoching
>  (our baseline is the first second of the three seconds before the trigger ).
> 
> If my guess is correct, then after doing it the mean baseline value in each condition could be non-zero, while the mean of the mean baseline values across condition should be zero. Personally I don't feel like testing it because I don't see a point in doing it.
> 
> > ERSP always compares post trigger data to that of the baseline, right? 
> 
> That's the default setting (all negative latency is counted as baseline). But you can also specify the baseline period explicitly.
> 
> Makoto
> 
> On Tue, Jul 17, 2018 at 10:35 PM Kathleen Van Benthem <kathy_vanbenthem at carleton.ca> wrote:
>> Dear List,
>> Can someone explain when ​one would want to ​select the "assume additive baseline" option​ when in the Study Function: ERSP​. 
>> After some searching we are still not clear on what this​ "additive" option​ does in the Stud​y ERSP​ function.
>> We are interested comparing power in frequency bands in the few seconds before the "responded to" vs. the "ignored" triggers (auditory cues)​ (N=30, two conditions per participant)​.
>> 
>> ​Specifically, we would like to know what is happening with the ​baseline when we select, or don't select the additive baseline option?   And does it matter if our epochs already have a baseline​ created during epoching​ (​our baseline is ​the first second of the three seconds before the trigger​​).
>> 
>> ERSP always compares post trigger data to that of the baseline, right? 
>> Thanks for your help in this matter,
>> 
>> Kathleen Van Benthem 
>> kathy.vanbenthem at carleton.ca
> 
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu
> 
> 
> -- 
> Makoto Miyakoshi
> Swartz Center for Computational Neuroscience
> Institute for Neural Computation, University of California San Diego
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu