العربية

Epoching

Cut the continuous signal into segments around each event and prepare them for averaging.

3 lessons: 3 core

After this station you can:

  • Extract epochs around an event type with a suitable window
  • Apply baseline correction
  • Find and reject bad epochs with a suitable threshold

Lessons

  1. Extracting epochsCore12 min
  2. Baseline correctionCore10 min
  3. Rejecting bad epochsCore15 min

Test yourself

4 questions. Your score stays with you and isn't sent anywhere.

1. From the same epochs you want an ERP analysis and, later, a time-frequency analysis. Which window fits best?
Show the answer

The correct answer: From -1 to 2 seconds

Time-frequency analysis needs long epochs so that low frequencies below 10 Hz can be analyzed, and -1 to 2 seconds is enough for both. -0.2 to 0.8 works for the ERP only. A window that starts at zero has no pre-stimulus period for baseline correction.

Review the lesson: Extracting epochs

2. In the pop_rmbase window you typed -0.2 0 in Baseline latency range ([min max] in ms). What is the problem?
Show the answer

The correct answer: The field is in milliseconds, so this is a 0.2-ms period, and the correction subtracts about one point

The epoch limits in the epoching lesson were in seconds, but the baseline field is in milliseconds. So you type -200 0 here. With a 0.2-ms period only the zero point falls inside, so the "correction" subtracts a single point's value from the whole epoch.

Review the lesson: Baseline correction

3. Which statements about choosing the baseline period are true? (Select all that apply)
Show the answer

The correct answer: -200 to 0 ms is a common value in ERP studies, A period of 100 ms or less gives a noisy mean, because it has few points, If another event falls inside the period, you subtract a real response from the whole epoch

The period is chosen before you see the results and applied unchanged to every participant, because changing it after seeing the waveform changes your result. Correction subtracts a mean from each epoch and removes none, so the count stays the same.

Review the lesson: Baseline correction

4. You tried a strict ±30 µV threshold and many more epochs were rejected. What is the cost?
Show the answer

The correct answer: Fewer trials in the average, so it gets noisier, and counts may differ between conditions

A strict threshold rejects clean epochs because of normal noise and slow waves. Each rejected epoch is one trial less, and an average of fewer trials is noisier. The goal is to reject what actually harms the average, not the smallest possible number.

Review the lesson: Rejecting bad epochs