[Eeglablist] time/frequency baseline length

[Scott Makeig]

Ayelet -

With a window size of 64 ms, you should not expect to have frequency
resolution below about 1000/64 ~ 16 Hz. Are you interested only in frequencies
higher than this?  Otherwise, you should definitely use a longer baseline
(even if your experimental events occur frequently).  We typically use -1s
to +2s for time/frequency analyses. The -1s allows a pre-event baseline
calculation using a 3-cycle window at 3-Hz and above -- and the +2s allows
use to see spectral events that occur later than 1s after the event (Makeig,
Elect. Clin. Neurophysiol., 1993).

Scott Makeig

From: "Ayelet Landau" <ayeletlandau at berkeley.edu>
To: "EEGLab List" <eeglablist at sccn.ucsd.edu>
Date: Thu, 26 Oct 2006 21:36:15 -0700
Subject: timef: baseline and window size

Dear users, I am trying to further understand the baselining that timef does
as it relates to the way the moving window FFT works. I am currently using a
prestim epoch of -200 ms with a window size of 64 ms. Using these
parameters, the output data starts around -76 ms. I was wondering what are
the parameters that determine the amount of data omitted (in this case
approximately 124ms).

In other words, when does the moving window calculation start to have enough
samples as far as timef is concerned, and should I assume that baseline is
calculated based on whatever remains before zero after an initial epoch is
omitted?

-- 
Scott Makeig, Director and Research Scientist, Swartz Center for
Computational Neuroscience, Institute for Neural Computation, University of
California San Diego, La Jolla CA 92093-0961, http://sccn.ucsd.edu/~scott
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20061027/1457d5e5/attachment.html>