bci2000.org BBS

Hi there,
As a part of my research, I have run the classic DF P300 speller paradigm and collected the EEG at Fz, Cz, C3, C4, Pz, Oz, O1 and O2 recording sites referred to the left mastoid. Unfortunately, I couldn't see any P300 in response to the target by using regular Monopolar montages (ref. to all and CAR) but surprisingly when I used any occipital channel (Oz/O1/O2) as my reference (e.g. Cz-O1, Fz-O1,....), P300 was so clear and visible with very high amplitude. Afterwards, I run the experiment for few more times with the same and different reference spot. In all of the cases, the target was simply found by my program using that weird montage with 100% accuracy. As far as i know P300 can be seen in the occipital channels too; thus they cannot be the good reference spots.
So now Im completely stuck and really need your help to find the solution to this problem.
I'd greatly appreciate your help and comments,
Kamyar

Kamyar,

you don't need to re-run your experiments to try a different montage. Re-referencing in software will suffice -- unless there is a physical problem with your recording.

To avoid physical extinction of signals by amplifier saturation, make sure that all your electrodes are the same material, including the ground electrode.

Hope this helps,
Juergen

Juergen,
I'd really appreciate your reply,
I should tell you that all of the electrodes we're using (including the ground one) are the same material. My question here is can we get p300 while our reference is one of the three occipital channels (O1,O2,Oz) and if so do you have any idea why we couldn't see the p300 when we picked left mastoid as a new reference?
Note that I'm talking about only one experiment and a software does the re-referencing job.

Best,
Kamyar

Kamyar,

Your intuition is correct; using occipital channels as a reference will typically attenuate the P300 response. However, in some cases, it is possible to see a P300 using an occipital channel as a reference. In general, a mastoid reference will produce a robust P300 response.

Given that you have already re-referenced this data set to several different montages, I would suggest that you collect a new data set with a different subject in an attempt to replicate this result. I suggest this because individual differences can be substantial in ERP data.

Regards,
Eric