[Mne_analysis] Basic Question (I think)

[Pablo Brusco]

Hi,

We have an experiment that we started with EEGLAB and now we are migrating
to MNE.
Something like building a classifier to distinguish subjects that are
"listening" vs "speaking".

At this point, we have one file (.set and .fdt) by condition
(listening/speaking) that contains several epochs for that condition.

In each epoch, we have only one event that matters for the time serie at
certain time, always the same (t=1.5 for example).

The thing is, I'm trying to create an Epochs object (maybe an
`_BaseEpochs`) to visualize / experiment with the data.

So, after using `mne.io.read_epochs_eeglab(set_filename)` I can see that
the events have no sense to me..

The question is,

Is it reasonable to have 1 event per epoch? or should it be 1 event for the
hole "Epochs" object?
And then, whats the way I can combine all this data? (I'm currently using
`mne.concatenate_epochs([listening_epochs, speaking_epochs])`)

For example, imagine I have 100 epochs for the class listening, and 200
epochs for the class speaking.

After I concatenate the epochs, should I have this 2 events?

[(1000 0 1), (1000 0 2)]
(lets say 1000 is the frame corresponding to the event at time 1.5 secs, 1
is the event id for speaking and 2 for listening)

or should I have 300 events?
[(1000 0 1)]*100 + [(1000 0 2)]*200

I'm new at EEG so maybe I'm missing a simple detail.

Thanks a lot.

-- 
Pablo Brusco
http://habla.dc.uba.ar/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20160527/6b585f54/attachment.html