EN
We introduce an effective model for correlated e_g electrons which reproduces qualitatively the evolution of magnetic order in monolayer manganites when correlated wave functions are used. Here we address recent optical conductivity measurements suggesting that the splitting between the occupied and empty e_g states is very large in LaSrMnO_4, contrary to what is expected for the effective model. We argue that no contradiction was found but several simple-minded one-atom-based expectations concerning crystal-field splitting and optical conductivity are in error.