ERPLAB Studio Panels: Extract Epochs - ucdavis/erplab GitHub Wiki

The Extract Bin-Based Epochs panel allows you to convert continuous EEG to epoched EEG (and perform baseline correction). It will be inactive unless BINLISTER has been run on the EEGset.

You should not use EEGLAB's built-in epoching routine (Tools > Extract epochs) when you plan to use ERPLAB to average the data.

ERPLAB will convert the starting and ending times of the epoch that you specify in milliseconds into the corresponding sample points, but these may not be at exactly the millisecond values you specify. For example, if you sampled at 250 Hz (4 ms per sample), and you specified an epoch of -25 to +150 ms, the actual starting and ending points cannot be at -25 and +150 ms because there are no samples at these exact times (i.e., neither -25 nor +150 is an integer multiple of 4). ERPLAB will find the nearest sample points, which in this case would be -24 and +152 ms. Also, the number of sample points may not be what you expect. If, for example, you sample at 1000 Hz and specify an epoch of 0 to 1000 ms, this will actually end up being 1001 data points (because there are sample points at 0 ms, 1 ms, 2 ms, and so on up to and including 1000 ms). Information about exactly how ERPLAB selects a time window in this case can be found on the Timing Details page.

Note that you cannot run BINLISTER on epoched data. However, there is a trick that often works if you need to re-bin data after epoching. Specifically, you can go to the Utilities panel and click Convert to Continuous EEG, which will create a continuous dataset, with boundary events between the epochs. This doesn't always solve the problem, but it does in many cases. You can see an extended description of this in the tutorial page on Converting Epoched to Continuous EEG.

You should ordinarily perform baseline correction when you epoch the data. You simply specify the time period to be used as the baseline period (or None if you do not want to perform baseline correction).

The baseline correction process subtracts the mean voltage during the baseline period from the entire waveform for each channel in each epoch. This eliminates any overall voltage offset from the waveform in each epoch. If you do not perform baseline correction, your waveforms may be shifted away from zero when being plotted (both for epoched EEGsets and the resulting averaged ERPs). In addition, the Simple Voltage Threshold artifact detection algorithm may not work properly if you have not performed baseline correction.

There are special circumstances where you will not want to perform baseline correction during the epoching process. You can perform baseline correction later using the Baseline Correction & Linear Detrend panel (for either epoched EEGsets or averaged ERPsets). You can also use this panel to re-baseline data that have already been baselined.