Hi AFNI experts!

I was playing around the last days with 3dFourier and tried several combinations of high AND low pass filters with 3dFourier.
I was doing something similar like:

3dFourier \
-prefix ffilt \
-highpass .0045455 \
-lowpass .11 \
'fdespr12reg+orig'

Since I am trying to understand, what I am doing I started thinking about reasonable cut offs. I also had some look on material I could find, how to set up these parameters.
Something which I found for block designs is setting the highpass filter as "2*period of cycle length." For the lowpass parameter I did not find any rational guideline.
The parameters as I understood it are very dependent of the design type you are choosing.
In our case we are using a mixed design, which can be seen somehow as combination of both a block and a rapid event-related design. There is a baseline condition (coded a zero in my stimfiles) in between "blocks" of a pure event-related presentation of the stimuli (three stimuli). The stimuli occur at 1/2 TR of 3sec and are totally randomized. The interval within "blocks" between stimuli varies from 0 to 3 TR (also subsampled as sub-TRs or 1/2TR). The mean interval is about 2 sec. Null events ("Fixation Crosses") are presented between the occurence of the 3 stimuli. During the occurence of these event-related "blocks" subjects have to perform 2 tasks (One is a control task). Only one task has to be performed for each "block". The blocks are balanced for both runs ( 183 TR each).

So formally spoken, lets say I call baseline B, the task T, and the control task C the runs would look something like this:
B T B T B C B C first run
B C B C B T B T second run

Sorry for the long introduction, but I think this has to be known to give any useful answer.

Well, I would like to filter out frequecies, which are higher than the frequency of a single event and lower than frequencies that could occur due to the appearance of these event related "blocks", which have all the same lengh of about 90 sec with 30 sec baseline periods in between.

I would assume that the frequency of a single event is probably longer than 1/6 sec (Hz). For the highpass filter I am even more puzzled, but if I understood the guidelines for block designs correct, it should be 2 time the cycle length. Half a cycle would be in my case the duration of one event-related block, which is 90 sec. Therefore, the the highpass filter should be something like 1/180 sec (Hz).

I would be glad, if anyone, who is experienced with filtering could tell me, if I am totally wrong.

Thanks,

Lukas