In almost all electroencephalography (EEG) recording setups, there are two auxiliary electrodes in addition to the ones used to record: the ground electrode and the reference electrode.

I've so far read two explanations for this. First, the explanation provided here is that the ground electrode is used for active grounding in a driven-right-leg (DRL) setup, so as to eliminate the common mode and thus increase the common mode rejection ratio (CMRR). Second, the explanation provided here is that to measure voltage, we need a signal ground, and that an electrophysiological signal only makes sense as the difference between two sites with non-zero value.

The first explanation does make sense to me, but the second one doesn't: can't you just measure everything as the difference between the site and the ground? The two seem quite contradictory...

  • $\begingroup$ May be related: Reference signal when measuring EEG $\endgroup$
    – Arnon Weinberg
    Dec 9, 2021 at 2:55
  • 2
    $\begingroup$ @ArnonWeinberg The accepted answer of that question would be satisfactory, but if the second explanation is just plain wrong, I would appreciate someone confirming it... It would come as a bit of a shock to me to know that technical documentation for products actually sold on the market can be just plain wrong. $\endgroup$
    – David Cian
    Dec 9, 2021 at 3:37
  • 1
    $\begingroup$ I believe the second explanation is referring to things like volume conduction and non-neuronal sources of signal. $\endgroup$
    – Bryan Krause
    Dec 9, 2021 at 15:33
  • $\begingroup$ @BryanKrause could you elaborate a little bit? $\endgroup$
    – David Cian
    Dec 9, 2021 at 16:58

1 Answer 1


To record good clean EEG signals in the microvolt range, you need to eliminate as much noise as possible. Since voltages are recorded as potential differences, you need at minimum 2 electrodes to see anything at all; one reference and one signal sensor.

But you don't need a ground to use this data. You can very simply record EEG with only 1x signal and 1x ref, but it will be extremely noisy.

In EEG recordings with 1 each signal, ref, and ground, you make 2 subtractions.

  1. REF - GND

subtraction #1 removes common mode signal. The human body is an antenna, and can pick up ambient interference. This is subtracted away from the REF signal, to provide a "cleaner" baseline.

subtraction #2 provides your biopotential signal against this clean baseline REF signal.

If you record SIGNAL - GND only, you would have a potential difference which includes microvolt - volt range noise in addition to the neural signal of interest. This means the signal-noise ratio would be very low, to the point where artifact rejection and computational correction methods would be almost useless.

PyCorder, which is python based recording software from BrainProducts, actually allows you to do authentic ground-free recordings. But nobody uses this function, because nobody needs it.

The double subtraction allows for cleaner, mostly ambient-noise-free signal to be recorded at signal channels. This is also why impedance reduction is important.

This is the reasoning behind REF + GND montages in EEG recordings.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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