I saw your book mentioned that to do time-frequency analysis, we need a long baseline. In our lab, we run mismatch negativity experiments with relatively short ISI, and the baseline of a segment is usually only 200 ms. Is time-frequency analysis inappropriate for those experiments?
The ISIs can be short. But is there a longer ITI (inter-trial-interval)? In many MMN experiments, there is a gap of, e.g., 1-2 seconds between each string of tones. That would be the baseline period.
If you don’t have ITIs, then you can either not use any baseline normalization, or use the average over all tones to be the baseline.
Thank you for your advice, Mike. The ITI is longer. I’ll look more into this.