[Eeglablist] newtimef and single trials

[Elisa Filevich]

Hi,


I am doing time frequency analysis using newtimef.
I am confused about what happens when the EEG data that is being
decomposed contains not one single trial but a vector of trials

in other words:
[ersp]  = newtimef (EEG.data ( 1, :, 2))
will give me spectral information about channel 1, trial 2;

I would expect that
[ersp]  = newtimef (EEG.data ( 1, :, [2 3]))

will output the averaged TF decomposition, and not the TF of the average.


To check that tha former is the case, I have done two things:

1. I saved the alltrials variable, converted from complex to power
values, and I got the same values as are contained in ersp, which
should indicate that newtimef computes the TF of each individual trial
and only then averages.

2. I also calculated individually

[two]  = newtimef (EEG.data ( 1, :, [2]))

[three]  = newtimef (EEG.data ( 1, :, [3]))

[two_three]  = newtimef (EEG.data ( 1, :, [2 3]))

in this case, the average of two and three does not correspond to two_three.
Instead, two_three is equal to applying newtimef to the mean of trials 2 and 3.

Sorry if there is a mistake in the logic, but I'm really confused.

Many thanks,
Elisa