5.9: Exercise- Global Field Power

Another useful reference-free transformation is mean global field power (MGFP). This transformation is easiest to explain visually. To begin, reset everything with the Reset button, then load Grand_N400_diff.erp in the ERPsets panel. View the difference wave as a butterfly plot by selecting Bin 5 and all the EEG channels (channels 1–28) in Bin & Channel Selection and then selecting BINS with CHANNELS overlay in Plot Settings. Once you click Apply, it should look like Screenshot 5.5. Notice that all the waveforms are relatively close together until the start of the N400 at ~200 ms, and they come together again at the end of the N400 at ~600 ms.

In general, when the amplitude of a given ERP component increases at one site, there will also be an increase in the spread of the voltages across all the electrode sites. Mean global field power quantifies the amount of spread as the standard deviation (SD) across electrode sites at each point in time. However, it is only valid when all the electrode sites share the same reference, which is why Screenshot 5.5 shows only the EEG channels (the EOG channels were referenced differently).

To compute the mean global field power for the EEG electrodes, go to the Channel Operations panel and make sure that the Mode is set to Modify existing ERPset. We’re going to create a new channel (channel 31) with the mean global field power for the EEG channels (channels 1–28). To do this, put the following equation in the Channel Operations text box: ch31 = mgfperp(1:28) label MGFP.

Now click Run. If you go to the Bin & Channel Selection panel, you should now see that there is a channel 31 labeled MGFP. Select this channel along with channels 1-28 and click Apply. The MGFP channel will be visible as a positive value at all time points (because the standard deviation of a set of values is always positive). You should be able to see that the height of the MGFP channel tracks the spread across the EEG channels.

Mean global field power has several useful characteristics. First, the reference electrode has no impact on MGFP (because it contributes equally to each channel and is effectively a constant). Second, it aggregates all the scalp signals into a single channel, which can avoid the need to select which channels to include in an analysis (Hamburger & Van der Burgt, 1991). Third, it is typically cleaner than the individual-channel waveforms (because noise is typically minimized by transformations that combine the data from multiple sites). However, you should keep in mind that the standard deviation across channels will increase as the noise level increases, so special methods are necessary to compare mean global field power amplitudes across conditions that differ in the number of trials or any other factor that might impact the noise level (Files et al., 2016).

In my lab, we like to use mean global field power to ask when information about a stimulus has reached the cortex. That is, we look at the first point at which this signal starts increasing above the baseline level (see, e.g., Kiat et al., 2021). We then use this information to determine the start of the measurement window for our primary analyses. By using mean global field power for this purpose, we get a very clean signal, and we don't have to choose a specific electrode site.

---