Eye color is not discrete Mendelian. That's only correct to the first order.

Also Mendelian has little to do with promotor sequences or differential transcription in deviations.