[Eeglablist] Choice of LORETA-ICA "Scaling"?

[Scott Luisi]

Hello,

I'm a graduate student at the University of Minnesota working on a project
involving resting state EEG networks. After preprocessing in EEGLAB I
started exploring the use of the LORETA implementation in the LORETA-KEY
software package and I was wondering if there was anybody on this mailing
list with the expertise of the program or perhaps functional ICA in general.

In any of the 'functional ICA' modules ("transposed fICA Networks" for
instance) there is a choice of "Scaling" with a drop-down that allows you
to choose between 'None, Subject-Wise, Frequency-Time-Wise,
Voxel/Electrode-wise, and VoxThenFreq-wise':
• What does the "Scaling" choice do functionally/mathematically? The
tutorial for fICA guides you to choose "Frequency/Time-wise" however the
output I see appears much like resting state networks I'd expect when
choosing "VoxThenFreq-wise" rather than "frequency/time-wise," so I can't
tell if my setup is wrong or my type of data is just different than the
tutorial.
○ Are there specific situations one choice would be explicitly superior to
another (e.g. Frequency/Time-wise for time-locked ERP data)?
○ How does the scaling choice alter the way the results should be
interpreted?

Thank you very much for any assistance you can provide!
Scott Luisi