Experimental work on light refraction in semiconductors suggests that there are two separate mechanisms by which the index of refraction becomes nonlinear: direct saturation, on the one hand, and the Burstein-Moss effect, on the other hand. We show that this picture is incorrect, since it was based on the usage of the Bloch equation. Electron systems are not Boltzmann systems and the appropriate Fermi-Dirac master equation leads to a saturation that differs from the saturation of the Bloch equation. We show, using a simple two-level system, that this result may look like the spin result in one extreme (phonon bottleneck saturation), while it shows the features of the Burnstein-Moss effect in the limit of small degeneracy of the upper level (Fermi saturation).