The equivalent of population mean (µ) would be the actual change in response from predictor variable (βk).
And in regression... it is assumed that βk is zero. Because unless proved otherwise through a hypothesis test, the predictor variable is assumed to not have a significant relationship with the response variable.