Dear Chris,
> 1) We have a paradigm which is set up as:
> 60 sec. baseline, 400 sec. task, and finally 60 sec. baseline again.
> Our question is: should we set our high-pass filter at (60+400+60)*2 sec
> or something smaller?
(60+400+60)*2 sec is appropriate (and will effectively implement a
monotonic detrending of the data). Unless constrained by other design
considerations, continuous perfomance of a 400sec task would normally
be thought to be rather long. You will get a much more efficient
design if you 'chop' the activation period into little segments (e.g.
24 to 32 secs) thus placing experimental frequency in a higher range
(and less confounded with low fequency artefacts).
I hope this helps - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|