[Eeglablist] Doing ERP with Emotiv EPOC (Marco Sim?es)

[Jacob Jolij]

Dear Marco,

I've managed to get reasonable to actually quite good quality ERPs with 
the Emotiv in a visual target detection task, but I used Matlab 
(Psychtoolbox) and the Research SDK to directly integrate data 
acquisition (and online preprocessing) into my stimulus presentation 
software. It was the only way I could get the timing right, but it seems 
to work reasonably well. Before trial onset I clear the Emotiv buffer by 
reading it out, and discarding the data, except for the value of the 
counter. At the time of stimulus presentation, I send a marker to the 
marker channel, and after the trial I read out the buffer again. Per 
trial I check for dropped samples, and if no samples are dropped, I 
align my trials by means of the stimulus marker, and check the timing 
using the counter value I read out at trial onset. I typically record 
epochs of approx. 2s:  1s baseline and 1s after stimulus presentation - 
my intertrial interval is now somewhat longer than my average 
experiment, but it works quite well. I do not filter the data, but 
re-reference the data by subtracting median activity of all electrodes 
per sample, then do a linear detrend, and finally smooth the data. This 
approach gives me pretty good data - it's not a substitute for a 
traditional EEG system, but data quality is good enough to do 
single-trial detection of an N200 with fair accuracy, for example. No 
papers out yet, but should be underway spring 2013 if my students do well...

Best,

Jacob