[Eeglablist] multiple designs in STUDY structure

[Arnaud Delorme]

Dear Agatha,

yes, you need to precompute channel and ICA measures for EVERY design. We are trying to implement mechanisms to reuse data computed in other designs.

For instance if you take ERP, and you have 3 conditions A, B, C. Design 1 is comparing A, B, and C. Design 2 is comparing B and C and design 3 is comparing A+B vs C.
Obviously data from Design 1 could be reused for the second and the third design (although I want to emphasize that this is not yet the case in EEGLAB). For the 3rd design, we would still have to recompute the ERP for A+B.

However, in some case, Design 1 is A, B, and C and design 2 is A, B and C as well. The two designs use different settings for computing the measures. In other words, it is important for the user to be able to choose or not to reuse measures from other designs, which makes it a little bit more complicated.

Now another example. Design 1 is A vs B, and Design 2 is A vs B with only trials of type X, and Design 3 is a 2x2 design that includes A vs B. EEGLAB will have to be able to tell that even if design 1, 2, and 3 deals with comparing A vs B, the data cannot be reused because some specific data trials are selected. However, in some particular cases, the data from design 2 could be reused for design 3 because the trials excluded are the same.

Finally, you have design 1 that includes all the subjects but 1, and design 2 that includes all of the subjects. Only one subject from design 2 needs to have his measures recomputed. However it is important to check that the settings (parameters) for computing these measures is compatible with the data that has already been computed in design 1.

To sum up, it is not simple but it doable. It will also require a lot of testing to be sure that there is no glitch. 
Reusing STUDY design measures is planned but not implemented yet in EEGLAB. For now, we rely on the raw power of your computer and except for ERSP/ITC, it is usually not a problem,

Best,

Arno

On May 19, 2011, at 11:19 AM, Agatha Lenartowicz wrote:

> Dear EEGLAB experts - 
> 
> I am finding the STUDY structure a little confusing. In particular my question is... do I need to precompute channel and ica measures for EVERY design separately? 
> 
> If I compute these measures are computed for design 1 (which encompasses all subjects/conditions) it seems like these files could be reused for other permutation of my conditions. But it doesn't seem that EEGLAB works that way (based on errors). The pre-computation is very lengthy in my case so it's a bit of a pain to rerun for every single possible comparison.
> 
> Could you clarify? This is not answered in the tutorials.
> 
> Thanks ~
> Agatha
> 
> 
> _______________________________________________
> 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