Dear Andre,
The spectral decomposition methods usually give you a complex number
per frequency. This complex number is a coefficient telling by how
much you should scale a sine at that frequency (that's the amplitude)
and by how much you should shift it (that's the phase) to provide
best fit to the data. The point is that shifting is with respect to
something and that's where there might be inconsistencies. I'm not
sure whether phase is defined the same way for all methods and all
settings. You can try experimenting and looking if you get consistent
results. But it should definitely be the case that if you use the same
settings for different data, then what phase means will be consistent.
Best,
Vladimir
On Mon, May 17, 2010 at 7:02 PM, Andre C. <[log in to unmask]> wrote:
>
> Hello Vladmir,
>
> I have made you this question in the SPM workshop, but I am still a bit
> confused about the answer. When SPM calculates the phase estimates for the
> time-frequency decomposition, what exactly do the values (from pi to -pi)
> relate to?
>
> Specifically, are the phase values consistent between subjects? For example,
> does the value of pi in two different subjects mean the same thing? I know
> that for PLV values that is not a big issue, since they are calculated
> within subjects, but, if I am extracting the phase values to do some
> phase-power coupling it is really important that similar values indicate
> similar phases, is that the case?
>
>
> Thanks again,
>
> --
> Andre M. Cravo
>
> Visiting Student
> Brain & Cognition Laboratory
> Department of Experimental Psychology
> Oxford University
>
|