Over the past few years, time-dependent ultrafast fluorescence spectroscopy method has been applied to the study of protein dynamics. However, observations from these experiments are in a controversy with other experimental studies. Participating of theoretical methods in this debate has not reconciled the contradiction, because the predicted initial relaxation from computer simulations is one-order faster than the ultrafast fluorescence spectroscopy experiment. In those simulations, pairwise force fields are employed, which have been shown to underestimate the roughness of the free energy landscape. Therefore, the relaxation rate of protein and water molecules under pairwise force fields is falsely exaggerated. In this work, we compared the relaxations of tryptophan/environment interaction under linear response approximation employing pairwise, polarized, and polarizable force fields. Results show that although the relaxation can be slowed down to a certain extent, the large gap between experiment and theory still cannot be filled.