Signal scaling for FirstLevel models

I’m doing seed based connectivity analysis and I’m getting the following warning:

UserWarning: Mean values of 0 observed.The data have probably been centered.Scaling might not work as expected
  warn('Mean values of 0 observed.

I have indeed centered my data before running FirstLevelModel. Should I not do it so that nilearn can properly scale the data? Or should I not use signal_scaling at all for seed based connectivity analysis, in which I only have regressors with (demeaned) time series?


Mean centering is fine to do, but you only need to do it once.


Simply use signal_scaling=False