<div><br></div><div>Aleksander,</div>I can't speak for the history, but conceptually it makes more sense to me<div>to average across epochs before taking the difference. First, the power</div><div>spectrum computed from a single trial has high variance. Dividing by a </div>
<div>single trial spectrum could potentially involve dividing by some very </div><div>small numbers at certain time-frequency points, which would inflate </div><div>the importance of that trial in the average. Second, the statistical theory</div>
<div>of power spectral estimation states that the log of the trial-averaged</div><div>power is chi^2 distributed. So taking the difference of those averages</div><div>is taking the difference of two approximately normal distributions, </div>
<div>which makes sense. There may be other considerations that I am</div><div>overlooking. I will be interested to hear what others comment.</div><div><br>-- <br>Thomas Ferree, PhD<br>Department of Radiology<br>UT Southwestern Medical Center<br>
Email: <a href="mailto:tom.ferree@gmail.com">tom.ferree@gmail.com</a><br>Voice: (214) 648-9767<br></div><div><br><div class="gmail_quote">On Thu, Jul 30, 2009 at 6:09 AM, Aleksander Alafuzoff <span dir="ltr"><<a href="mailto:aleksander.alafuzoff@helsinki.fi">aleksander.alafuzoff@helsinki.fi</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">Hi,<br>
<br>
While (re-)reading Makeig's 1993 article, which introduced the<br>
ERSP-measure, and Delorme & Makeig's 2004 EEGLAB article, I noticed a<br>
slight discrepancy in the way ERSP is defined. In case I've failed to<br>
take something into account or misunderstood the articles, please<br>
correct me.<br>
<br>
In Makeig's original article the "mean subject ERSP" is the average of<br>
baseline normalised epoch/trial ERSP's, where each epoch's time<br>
localised spectra are normalised by that epoch's baseline. In the more<br>
recent EEGLAB article ERSP is defined as the mean square amplitude of<br>
spectral estimate F_k (averaged over epochs k) divided by the mean<br>
baseline power spectrum*.<br>
<br>
Conceptually this is a big difference: the 1993 ERSP is in principle<br>
defined for individual epochs, while ERSP according to the 2004<br>
article is only defined relative to a set of epochs. In general the<br>
two definitions would also seem to give slightly different estimates<br>
of the "spectral perturbation", although, if we assume that the<br>
underlying baseline power spectrum is constant, the definitions are<br>
equivalent.<br>
<br>
At face value, the older ERSP defintion, where each spectral estimate<br>
is normalised relative to the local baseline, would seem more<br>
appropriate, since it seems inevitable that there must be some<br>
variation in the baseline during any experiment taking more than a few<br>
minutes (if for no other reason than a drop in subject alertness). Why<br>
then does EEGLAB use the latter definition (in both the article and<br>
the code as far as I can see)? The difference between the two ERSP<br>
estimates would seem to be quite small, so is there perhaps a<br>
computational advantage in the latter formulation?<br>
<br>
* In both articles the values are also log-transformed, but as far as<br>
I can see this is irrelevant for the current concern.<br>
<br>
--<br>
<font color="#888888">Aleksander Alafuzoff<br>
Research assistant<br>
Cognitive Science Unit,<br>
University of Helsinki<br>
_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu">eeglablist-request@sccn.ucsd.edu</a><br>
</font></blockquote></div><br><br clear="all"><br>
</div>