parametric modulation of power

[Eric Maris]

Dear Floris,


> I have a (sort of conceptual) question on how to test for parametric
> modulations in power as a function of e.g. dificulty.
>
> I have a task which has 5 difficulty levels, so from 1=easy to 5=hard.
> I made 5 separate power-spectra, which could eg be named subj1_TFR{1} ..
> subj1_TFR{5}. I can of course inspect each power spectrum separately,
> and try to eye-ball whether there may be parametric increases or
> decreases in power with increasing difficulty level, but it would be
> nice if I could get a direct view on how 'linearly in/decreasing' the
> power spectrum per sensor and frequency is.
> Is this possible within Fieldtrip? Are there scripts around for doing
> such a linear regression analysis?


This is possible using using CLUSTERRANDANALYSIS. However, at present, only
independent samples regression analyses are possible. In practice, this
means that you can only analyse your data for each of the subjects
separately
(treating the trials as replications).

Using CLUSTERRANDANALYSIS implies that you evaluate your effect using
randomization inference. The nice thing about CLUSTERRANDANALYSIS is that it
controls your family-wise error rate.


Eric Maris