In a brain-computer interface paper, I found the following paragraph:

Effective assessment of BCI performance requires two levels of evaluation: the user and the system. The user must control the signal features ..."

My question is how can a person control the signal features of his/her brain signal? Does this mean, for example, that a person can give more attention to increase signal to noise ratio for the P300 signal, and be able to look more carefully at the screen that shows flashes in an SSVEP experiment?


1 Answer 1


Yes, and no. Biofeedback has been an active field of research since the 70s, but has been a bit "fringified" as a cure for ADHD, etc.

It is possible to fine tune the brain activity for specific tasks. To borrow an example from BCI, the Rolandic Mu Rhythm can be used to "train" a user to refine the movement of a cursor or other pointing device.

Jonathan Wolpaw has been doing BCI work long before it was "cool", so his work on BCIs for ALS patients is definitely worth checking out.

As to whether you could train the P300, this signal is seen most prevalently in the "oddball" paradigm, which by nature relies on the stimulus that evokes it being of low probability. Letting the user know that a low probability stimulus is about to occur would eliminate most of the response.

  • $\begingroup$ i have two question on your answer , first , what is "low probability stimulus" , is the stimulus a natural event ? and the second question is how the response will eliminate if the user know that stimulus is going to evoke ? $\endgroup$
    – Learner
    Aug 4, 2014 at 17:54
  • $\begingroup$ @Learner Haven't forgotten about your question, sorry $\endgroup$ Aug 6, 2014 at 4:27
  • $\begingroup$ @Learner It's laid out in the Wikipedia article, but if you have an auditory stimulus that's going "beep, beep, beep, beep", and every few seconds there is a "boop" mixed in, the "boop" is a low-probability stimulus (less likely to occur) and will generate a more pronounced P300 wave. This phenomenon can occur naturally as well if we are surprised by something (a siren on a quiet day, etc.). I don't quite know how such an effect manifests for a visual stimulus, but I imagine that the "oddball" effect is somewhat similar. $\endgroup$ Aug 6, 2014 at 22:43
  • $\begingroup$ The P300 response will not be as pronounced, but won't be completely eliminated in the case of "if the user knows that stimulus..." that you cite in the comment. Remember that we analyze these EEG patterns over multiple epochs (short periods of time) and can time-lock them to the stimulus, so we can "line up" the responses and average them. $\endgroup$ Aug 6, 2014 at 22:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.