Wideband Inhibition of Dorsal Cochlear Nucleus Type IV Unitsin Cat: A Computational ModelKENNETH E. HANCOCK1and HERBERT F. VOIGT1,21Department of Biomedical Engineering and Hearing Research Center, Boston University, Boston, MA, and2Department of Otolaryngology, Boston University, Boston, MA(Received 27 May 1998; accepted 18 August 1998)Abstract—A computational model of a portion of dorsal co-chlear nucleus neural circuitry was used to investigate relation-ships between connectivity and response properties of type IVunits. The model in this study consists of four neural popula-tions. The pattern of convergence from one population to an-other and the strengths of those connections are the most im-portant model parameters. Lumped parameter electrical circuitmodels represent individual cells. Interconnections are achievedby activating variable conductances in post-synaptic cells ac-cording to spike activity in pre-synaptic cells. Auditory nervefibers are incorporated as a bank of logarithmically spacedgammatone filters that drive compartmental models of innerhair cell function. While it might be possible to configure themodel without wideband inhibition to simulate type IV unitnotch noise responses, the resulting parameters would likely bephysiologically implausible. The model with wideband inhibi-tion, however, shows the appropriate notch noise behavior. Awide variety of simulated rate versus cutoff-frequency plots areachieved varying three model parameters. The model was fit tophysiological data by finding values of these three parametersthat minimize the sum of squared errors. The results show thatwideband inhibition can quantitatively account for the re-sponses of type IV units to notch noise. © 1999 BiomedicalEngineering Society. @S0090-6964~99!00301-X#Keywords—Hearing, Neural models, Parameter estimation,Notch noise.INTRODUCTIONThe cochlear nucleus ~CN! is the first stage in theascending auditory pathway. The dorsal CN division~DCN! is a laminated structure containing several mor-phologically distinct classes of neurons that respond toacoustic stimulation with a variety of response patterns.This work is focused on type IV unit responses in thecat. Type IV units are of particular interest because theyhave been associated with cat DCN principal cells ~fusi-form and giant cells!, which project from the DCN toother brain centers.37Physiology of Type IV UnitsType IV units are characterized by their responsemaps @Fig. 1~A!#, which show excitatory and inhibitoryactivity as functions of the frequency and level of a tonalstimulus. Type IV units give excitatory responses to low-level tones near their best frequency ~BF!, the frequencyto which they are most sensitive. Responses at higherlevels and over a broad frequency range are dominatedby inhibition. In many cases, this region of inhibition canbe segregated into a central inhibitory area ~CIA! and anupper inhibitory sideband ~UIS! by an upper excitatoryregion ~UER! consisting of excitatory responses to anarrow range of frequencies above BF.29Type IV units discharge spontaneously at a medianrate of about 40 spikes/s.38Presumably, type IV unitsreceive excitatory input from spontaneously active audi-tory nerve fibers. Spontaneous activity in the DCN per-sists after cochlear ablation, however, and hence must bedue in part to intrinsic mechanisms or to noncochlearinputs to the DCN.14Whereas type IV units are mainly inhibited by tones,they are excited by broadband noise at all levels abovethreshold. Their responses to notch noise ~broadbandnoise from which a notch, centered on the type IV unitBF, has been filtered! were reported by Spirou andYoung.29Rate-level curves for notches of various widthsare shown for a typical type IV unit in Fig. 1~B!. Inhibi-tory responses are observed at lower levels; excitation isseen at higher levels. This pattern can be explained bythe fact that the stopband of the notch noise stimulus isroughly 30 dB down from the passband. At low stimuluslevels, the decreased energy in the stopband represents asignificant loss of excitation to the type IV unit, relativeto broadband noise, and the unit is inhibited. At highstimulus levels, the stopband energy is sufficient to shiftthe balance of excitation and inhibition in favor of exci-tation.Spirou and Young29further characterized notch noiseresponses using rate-versus-cutoff-frequency plots @Fig.1~C!#. Firing rates from the rate-level curves @Fig. 1~B!#Address correspondence to Herbert F. Voigt, PhD, Department ofBiomedical Engineering, 44 Cummington Street, Boston, MA 02215-2407. Electronic mail: [email protected] of Biomedical Engineering, Vol. 27, pp. 73–87, 1999 0090-6964/99/27~1!/73/15/$15.00Printed in the USA. All rights reserved. Copyright © 1999 Biomedical Engineering Society73were averaged over a range of levels where the inhibi-tion was maximal and plotted as a function of the notchcutoff frequency. Note that each notch is representedtwice: once for its lowpass cutoff and once for the high-pass. These plots typically show excitation for narrownotch widths and inhibition for wider notches. For verywide notches, the rate approaches the spontaneous rate ofthe unit as the stimulus approximates low level broad-band noise.Conceptual Model of DCN CircuitryA relatively simple conceptual model accounts formany aspects of type IV unit physiology. An importantsource of inhibition is understood to be type II units.Type II unit response maps show relatively narrow V-shaped regions of excitation. The tone thresholds of typeII units are typically about 10 dB higher than those oftype IV units, so that type II unit excitatory regions tendto overlie the central inhibitory areas of type IV units.Whereas type IV units are mostly inhibited by tones andexcited by broadband noise, type II units are vigorouslyexcited by BF tones, but respond only weakly to broad-band noise. These reciprocal response patterns led to thehypothesis that type IV units are inhibited by type IIunits,38a conclusion supported by cross-correlationresults.33,35While inhibition from type II units may account forthe central inhibitory area of type IV unit response maps,Nelken and Young20argued that inhibition of type IVunits in response to notch noise cannot arise from type IIunits. This is because the relatively narrow band of typeII units converging on a particular type IV unit is onlyweakly activated by notch noise stimuli. They concludedthat type IV units must receive inhibition from