Neurons in the dorsal cochlear nucleus (DCN) exhibit strong nonlinearities in spectral processing. Low-order models that transform the stimulus spectrum into discharge rate using a combination of 1^st- and 2^nd-order weighting of the spectrum (quadratic models) usually fail to predict responses to novel stimuli for principal neurons in the DCN, even though they work well in ventral cochlear nucleus (Yu and Young 2000). Here we investigate the effects of spectral contrast on the performance of such models. Typically, the models fail for stimuli with natural-sound-like spectral contrasts (~12-dB), but have good prediction performance at small (3-dB) contrasts. The weights also typically increase substantially in amplitude at smaller spectral contrast. These changes in weight size with contrast are partly inherited from similar effects seen in auditory nerve fibers, but there must be additional effects from inhibitory circuits in the DCN. These results provide insight into the reasons for the poor performance of spectro-temporal receptive field (STRF) models in predicting responses of auditory neurons. Because the general shapes of the weights do not change between low and high contrast, they also suggest that STRFs may capture meaningful properties of neural receptive fields, even though they do not do well at predicting responses.