[Eeglablist] difficulties in re-epoching data

[Clayton Hickey]

Hi Julie, Hi Arno,

Thanks for your responses. I think my troubles stemmed from my own 
oversight; I forgot that EEGLAB conserves all events after epochs are 
created (unlike the other packages I work with). As both of you pointed 
out, this means that some of the epochs I initially created could 
contain more than one of the event on which the epochs were defined. 
When I re-epoched, epoch.m initially selected all of those defining 
events... and then skipped those that were not 'centers' of the 
first-round epochs.

thanks, much appreciated,

clayton


Julie Onton wrote:
> Clayton, is it possible that some of your epochs overlap with each other such that event '223' is
> sometimes not the time-locking event? In that case, you could not epoch on that event without
> going over the time limits of the existing epochs.
>
> Julie
>
>   

> Dear Clayton,
>
> you are doing the right thing. You could also run ICA on the 
> continuous data before extracting data epochs and then you would have 
> this problem.
>
> First, it seems normal that you will lose some epochs because most of 
> them correspond to different conditions than the one you are 
> interested in.
>
> Second if you have a large window for the original epochs (let say 3 
> seconds from -1 to 2 seconds) and your time-locking events are every 1 
> second for instance, then over the course of a 3-second epoch you will 
> have several events (for instance, one at -1s, one at 0s obviously, 
> one at 1 second and one at 2 seconds). When you re-epoch the data, not 
> only the events at 0s are considered but also all events at -1s,1s, 
> and 2s. However, for these non-0 events, your new epoching window (let 
> say still -1 to 2 seconds) will not allow to extract a full epoch and 
> EEGLAB will return the "out of boundary" message you mentionned.
>
> Hope this helps,
>
> Arno
>
> Clayton Hickey wrote:
>> Hello all,
>>
>> I'm having some difficulty in re-epoching a dataset.
>>
>> I'm doing vision research, and want to correct for blink artifacts 
>> but discard trials in which eye movements took place. I'm currently 
>> starting analysis by modifying event codes such that relevant 
>> conditions are identified. I then epoch the data such that all 
>> relevant trials are selected, and run infomax ICA on this 'grand' 
>> dataset. Later, I identify components associated with blinks, and 
>> components associated with eye movements. I remove the components 
>> associated with blinks, but discard epochs associated with eye 
>> movements. I then re-epoch the clean dataset such that each condition 
>> is stored in its own datafile.
>>
>> When I re-epoch, I lose a number of trials, sometimes the majority. 
>> For example, here's the output I receive on one try:
>>
>> [*code*]
>>  >> test = pop_epoch(EEG, {'223'}, [-0.5 1.0]);
>> pop_epoch():212 epochs selected
>> Epoching...
>> Warning: event 7 out of data boundary
>> Warning: event 12 out of data boundary
>> Warning: event 53 out of data boundary
>> Warning: event 66 out of data boundary
>> Warning: event 69 out of data boundary
>> Warning: event 84 out of data boundary
>> Warning: event 87 out of data boundary
>> Warning: event 104 out of data boundary
>> Warning: event 107 out of data boundary
>> Warning: event 124 out of data boundary
>> Warning: event 137 out of data boundary
>> Warning: event 140 out of data boundary
>> Warning: event 147 out of data boundary
>> Warning: event 152 out of data boundary
>> Warning: event 157 out of data boundary
>> Warning: event 162 out of data boundary
>> Warning: event 165 out of data boundary
>> Warning: event 168 out of data boundary
>> Warning: event 193 out of data boundary
>> Warning: event 208 out of data boundary
>> pop_epoch():192 epochs generated
>> pop_epoch(): time limits have been adjusted to [-0.500 1.000] to fit 
>> data points limits
>> eeg_checkset: recomputing the ICA activation matrix ...
>> pop_epoch(): checking epochs for data discontinuity
>> [*end code*]
>>
>> I'm not sure what it means that the events are out of the data 
>> boundary, as I'm epoching based on a single event code. I have traced 
>> this error in epoch.m, but the code is reasonably complicated and I 
>> can't quite figure out what's going wrong. Has anyone else had 
>> similar troubles? Or perhaps does anyone have any insight?
>>
>> Thanks very much,
>>
>> Clayton Hickey
>> Vrije Universiteit Amsterdam
>>
>>
>> _______________________________________________
>> eeglablist mailing list eeglablist at sccn.ucsd.edu
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to 
>> eeglablist-unsubscribe at sccn.ucsd.edu
>>
>>
>>   
>