Should ICA be performed on the entire EEG recording or only on the task period?

Hi everyone,

I have an EEG recording that consists of two periods: a task period and a resting-state period. I am only interested in analyzing the EEG data collected during the task.

I am planning to use ICA for artifact removal, and I am wondering about the best approach in this situation.

Should I first separate the task period from the resting-state period and then run ICA only on the task data? Or is it better to perform ICA on the entire continuous recording, including both the task and rest periods, and then apply the resulting ICA decomposition to the task data?

Since my final analysis will only focus on the task data, I would appreciate any advice on which approach is preferable and why.

Thank you!

whole recording, definitely. ica gets more stable the more data you feed it, so theres no reason to throw away the rest period. just run the decomposition on the continuous data and apply it to the task segment after. only exception would be if part of the recording is totally unusable, but rest data is usually fine.

1 Like

Thank you so much for your response. I really appreciate it, and I found it very helpful.