[Eeglablist] Design for Buffer and Baseline for Time-frequency Analysis

[u3564255 at connect.hku.hk]

Dear EEGLABLIST,

I would like to consult you guys about the design for the buffer zone and baseline for time-frequency analysis (TF analysis).

The first question concerns the similarity between the buffer zone and the baseline. According to my understanding, temporal smoothing may cause some of the signals in the buffer to spread to the baseline. If it is true, would it be better for the buffer zone and the baseline to be the same (e.g., both show only a cross on the screen) instead of having a buffer zone that includes some part of the previous trial? I think it might be better because in this case, the baseline and hence the resulting post-stimulus time-frequency analysis will not be contaminated by signals from the previous trial.

The second question concerns the length of the pre-stimulus period (buffer zone + baseline). I hope to analyse delta, which canonically starts from 1 Hz, as the lowest frequency band of interest. According to Cohen (2014), the rule of thumb is that the sufficient length of the buffer is three cycles of the lowest frequency of interest. This means that I want to have a buffer lasting 3 seconds. Also, I hope to have a baseline of -400 to -100 ms (i.e., a period from 400 ms to 100 ms before the stimulus). Does this mean that the length of my pre-stimulus period should be 3400 ms?

I am sorry if the answers to these questions are obvious. Thank you so much in advance!

Best regards,
Samuel