Dear Alice,

Applying the ICA before epoching aims at correcting systematic/re-occuring artifacts (saccades, blink, heart rate artifacts). As these artifacts are best displayed in chunks of your data that may not even lie within your epoch of interest, I would suggest sticking to the EEGLAB recommendation.

To be honest, I have not yet seen a paper in which the authors applied ICA (with the aim of identifying artifacts) after segmentation/epoching. But maybe I just didn't screen enough papers for exactly that ^^"

Best regards,

Pat

Hi Alice,

I have always run the ICA before Epoching. Similar to Patrick's answer, I agree that it's probably best to stick to EEG labs recommendation.

This is because random and systematic noise will occur throughout the entire recording. If the data are broken up just into Epochs, it will lose the systematic pattern of the noise and the ICA may not be able to identify the noise in the recording as well. To me this makes sense... hopefully this help you.

Best of luck,

-Elliott

Hello,

Steven Luck, who is pretty widely regarded as the god of ERP analysis methods recommends that people first visually inspect and reject highly abnormal data (like a period of extreme muscle noise), then perform ICA, then epoch.

Here is a post about it on his site with a full outline of his recommended data processing/analysis steps: (The clearest step by step outline is near the bottom of the article)

https://erpinfo.org/order-of-steps