[Eeglablist] Binning different delay durations

Tarik S Bel-Bahar tarikbelbahar at gmail.com
Mon Oct 23 16:27:56 PDT 2017

Hello Alexandrea, here are some brief notes for you below. Thanks for
sharing about your progress and solutions, so that other users on the list
can benefit from your experiences.


My understanding is that eeglab requires epochs to be the same length, both
for single-subject analysis, and for a specific eeglab STUDY design. If you
are not familiar with eeglab, remember to work with the eeglab tutorial and
tutorial data first, and attempt some of your tests on that data.

I would generally just recommend that you do analyses separately for each
of the three kinds of epochs,

One of your possible paths is to simply make three different kinds of
single-subject files, with all epochs for one kind of epoch length in one
file, the other epoch length in a different file, and the other epoch
length in a different file. And then show  the results for each of three
different kinds of delays side by side. Once you get to that point, you
could investigate a way to compare for differences across the epoch lengths
on whatever EEG metric you are focused on.

Alternatively you can just analyze all epochs just as -200 to 500, and just
keep track of which ones come from different length epochs. So that means
cutting down the longer epochs. In this way to you take a "common chunk"
from within each kind of epoch.

Alternatively, you can select a period of about, for example, ~500 ms from
within each kind of epoch, and do baselining of that common-length period
via code rather than the eeglab gui. In this way to you take a "common
chunk" from within each kind of epoch.

My understanding is that the current study allows you to load continuous
single-subject files and to make different kinds of epochs there. See past
eeglablist comments on that. Makoto has also mentioned that several times.
I usually don't use that method, but it could work for you. Most likely you
would still need to have different epoch lengths be associated with
different study designs. Check in the current online study documentation
and eeglab developers for more info about that too.

There are various ways to extract epochs with variable time-length, you
would have to attempt that via your own matlab code. For example, you would
just have a matrix of your epochs, and some of the epochs would be longer
and some would be shorter. So for the shorter epochs there would be no data
in the timebins after 500 ms. Via this method (or similar ones), you would
at least be able to get ERPs/ERSPs across a common time period across all
epochs. This path requires that you have matlab expertise and can access
eeglab functions outside of eeglab, calling them as you need to, and
feeding them matrices of data that you create with your own code.

If you're having trouble conceptualizing how to do this analyses, your best
bet is to read articles that have used the same protocol and have done
analyses of these different epoch lengths. If you review ~10 articles that
have done this, you should have a better idea regarding how to go about
analyses, and how they dealt with variable-length epochs.

In the end, you might just have a single number that reflects an EEG
metrics at a particular band for each kind of epoch.

For baselining, you should not have a problem, just use the -200 to 0
baseline for each kind of epoch. If you've already collected the data, you
can't change the baseline. Yes, it is possible that you may want different
baselines for the three different kinds of epochs, but I don't think so.

another option, though I would not recommend it, would be to "somehow"
compress the data from the longer periods (perhaps by downsampling) and
thus "squishing them all to the same size.

If you are having trouble just doing basic binning/epoching, you should be
able to create new events in the continuous data that signify the onset of
each of the three different kinds of epochs. Then you would be able to
create single-subject epoched files for each of the three different kinds
of epochs (3 different epoched files), and/or epoch on those three
different kinds of events, but with the same start/end length of the epoch.

One may also want to see if erplab or fieldtrip or brainstorm allow
variable epoch lengths.

On Fri, Oct 20, 2017 at 12:09 PM, Alexandrea Kilgore-Gomez <
akilgoregomez at nevada.unr.edu> wrote:

> Dear eeglablist,
> I have been having issues binning trials of an experiment with different
> delay durations. The experiment deals the maintenance period during working
> memory by presenting a set size of three items and asking whether or not
> the the probe is old or new after a delay. The delay periods are variable,
> with durations of 500, 900, and 1500 ms and randomized within a single
> session. The time period of interest is -200 ms to 500, 900, or 1500 ms.
> I want to be able to analyze the EEG data from the same file instead of
> recording and saving each delay duration separately, a task I have been
> unsuccessful at thus far. I am also concerned about baseline corrections
> with different delay durations.
> Is there a way to extract epochs with variable time-length from the same
> session? I have two overall questions: How do I bin this
> experiment? and How do I go about epoching the continuous data with
> different delay durations? Any help would be greatly appreciated.
> Thanks in advance,
> Alex
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.
> ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20171023/0c73658a/attachment-0001.html>

More information about the eeglablist mailing list