[Eeglablist] On the usage of eegfilt.m and employing epochs

[Kaushik J L]

Hello,
I intend to determine the extent of correlation between different channels
in the usual delta, theta, alpha and beta bands. In order to filter out the
bands of interest, I tried using eegfilt.m. But I notice that the function
takes a transition bandwidth of 0.15*cut-off frequency. For the theta band
(8-13), this could mean that only frequencies above 15 Hz. experience
significant attenuation. Is this acceptable in practice? Or should I
consider using a filter with a steeper transition band? Can IIR butterworth
filter be used for this purpose?

Question 2:
As a general rule, is it sensible to first filter out the required band by
operating on the complete data and then split the output into epochs before
further processing (such as calculating the correlation between
corresponding epochs of different channels)?

Thanks,
Kaushik
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20110104/b3c5ba08/attachment.html>