Saturday, December 25, 2010

P300 and Emotiv EPOC: Does Emotiv EPOC capture real EEG?

I’ve received a great response for my blog discussion about Emotiv EPOC and using it to capture P300 responses. However, considering changes that have been taken place during the period, I’ve now improved my article a little bit. You can access the new article by following this link. I would like to thank all who posted comments during that time and welcome further comments about my new article.


  1. Hi Hiran,
    Nice work here!
    A few suggestions and comments:
    1. Maybe you can use the capture features of OpenVIBE to make doubly sure the timing markers do not suffer any lag from passing around the serial port loop. OpenVIBE has driver support for EPOC and you should be able to capture data directly into your scenario.
    2. I strongly recommend using the alternate reference locations for P300 studies. These are the locations behind the ears, where the rubber comfort pads are usually located. Given the strong tendency of rear sensors to display the P300 signal, the default reference locations at approximately P3/P4 will have a significant P300 component which will be subtracted from all the other channels - in other words the P300 signal will be bigger and more detectable on more channels if you use a reference away from the parietal area.
    3. We found that the central locations can be more or less modelled by averaging the left and right sides, especially for more central pairs such as O1/O2, F3/F4 and AF3/AF4, and to a lesser extent P7/P8. This suggests it may be worthwhile to look for P300 signals in combined (O1+O2), (P7+P8), (F3+F4) and so on.
    4. It's not such a surprise that the P300 potentials are visible since many other users have reported successful use of the P300 spellers in both OpenVIBE and BCI2000 packages and the Neurophone app from Dartmouth College. However it's great to see the signal distributions and definite independent verificaton!
    Great work - please feel free to post links on the Emotiv user forum!
    Best regards,
    Research Manager and CTO, Emotiv

  2. Hi Geoff,

    Thanks for the comment. As you've mentioned I've tried to read EEG data directly into OpenViBE, but soon I found that OpenViBE suffers from higher system load and as a result sometimes there are glitches in the signals. That's why I decided to record using TestBench, because it seemed more robust than OpenViBE in recording. I've checked the lags resulting from serial port loop as well. There is a small random delay (doubt its progressive). But I thought that doesn't affect that much in this basic validation.
    Thanks for highlighting the reference electrode. Indeed I used the ones near the ears. As you've mentioned I can try to improve the detection by using combined electrodes. That is again a good way to improve the robustness of sensing, because sometimes electrode contacts fail during experiments.
    All the best in 2011!


  3. Dear Hiran,

    The link to the pdf file is not working, and I am starting a project with ERPs and Epoc and your work could really help me out.

    Could you re-upload the pdf please.



  4. Hi Hiran,

    i am one of the authors from (German Fanpage). And we are really interested in your study about the neuroheadset. But the link does not work. The Headset works good for us, but we want to know how and why :)

  5. Hi Heiner, Kirill, and others,
    Sorry for the long delay in fixing the broken link. Now it should work.

  6. Hi - good research! Are you aware of anyone using a P300 speller with the Emotiv headset please for a person with Locked In Syndrome or simlar? (see )

  7. Hello Hiran,

    I am doing a research related to BCI area and I need to get some information about how to manage my EEG experiment. I am using the Emotiv EPOC+ device and I have to record user's response to a visual stimulus. I have read your great article and I appreciate if you can give me some guidelines about how to do the experiment in detail.

    I use Emotiv TestBench to record and save data in the format of ".edf".
    At first, I ask my user to close their eyes and after almost 10s I ask them to open their eyes and look at a specific video. After the playing is finished I will stop recording. I repeat this trial three times per user. I have about 13 number of users.

    I am not sure if this way I am doing the experiment correctly?
    I have also started to use EEGLab to remove artifacts, the baseline, DC part, applying ICA,..

    I appreciate if you can any help me with this matter.

    Thank you.