A computational model of a portion of dorsal cochlear nucleus neural circuitry was used to investigate relationships between connectivity and response properties of type IV units. The model in this study consists of four neural populations. The pattern of convergence from one population to another and the strengths of those connections are the most important model parameters. Lumped parameter electrical circuit models represent individual cells. Interconnections are achieved by activating variable conductances in post-synaptic cells according to spike activity in pre-synaptic cells. Auditory nerve fibers are incorporated as a bank of logarithmically spaced gammatone filters that drive compartmental models of inner hair cell function. While it might be possible to configure the model without wideband inhibition to simulate type IV unit notch noise responses, the resulting parameters would likely be physiologically implausible. The model with wideband inhibition, however, shows the appropriate notch noise behavior. A wide variety of simulated rate versus cutoff-frequency plots are achieved varying three model parameters. The model was fit to physiological data by finding values of these three parameters that minimize the sum of squared errors. The results show that wideband inhibition can quantitatively account for the responses of type IV units to notch noise.