Monaural Spectral Contrast Mechanism for Neural ... · Monaural Spectral Contrast Mechanism for Neural Sensitivity to Sound Direction in the Medial Geniculate Body of the Cat THOMAS

Monaural Spectral Contrast Mechanism for Neural Sensitivity to SoundDirection in the Medial Geniculate Body of the Cat

THOMAS J. IMIG, PIERRE POIRIER, W. ANDREW IRONS, AND FRANK K. SAMSONDepartment of Physiology, Kansas University Medical Center, Kansas City, Kansas 66160-7401

Imig, Thomas J., Pierre Poirier, W. Andrew Irons, and Frank dlebrooks and Green 1991; Yin and Chan 1988), pinna (orK. Samson. Monaural spectral contrast mechanism for neural sen- spectral) cues are also important for localization of broad-sitivity to sound direction in the medial geniculate body of the cat. band sounds (Blauert 1983). Distortion or elimination ofJ. Neurophysiol. 78: 2754–2771, 1997. Central auditory neurons pinna cues in humans causes a substantial decrease in accu-vary in sound direction sensitivity. Insensitive cells discharge well racy of vertical localization and an increase in front/backto all sound source directions, whereas sensitive cells discharge

reversals in the horizontal plane (Fisher and Freeman 1968;well to certain directions and poorly to others. High-frequencyGardner and Gardner 1973; Musicant and Butler 1984; Old-neurons in the latter group are differentially sensitive to binauralfield and Parker 1984).and monaural directional cues present in broadband noise (BBN).

Acoustical measurements in a variety of species (e.g.,Binaural directional (BD) cells require binaural stimulation fordirectional sensitivity; monaural directional (MD) cells are sensi- cats: Musicant et al. 1990; Rice et al. 1992; humans: Batteautive to the direction of monaural stimuli. A model of MD sensitivity 1967; Hebrank and Wright 1974b; Searle et al. 1975; Shawwas tested using single-unit responses. The model assumes that 1974) demonstrate that diffraction of high-frequency soundMD cells derive directional sensitivity from pinna-derived spectral waves by the head and pinna produces a free field to tym-cues (head related transfer function, HRTF). This assumption was panic membrane transformation in gain that is both fre-supported by the similarity of effects that pinna orientation pro- quency and direction dependent. The spectral transformationduces on locations of HRTF patterns and on locations of MD cell

is referred to as the head-related transfer function (HRTF)azimuth function peaks and nulls. According to the model, MD(Wightman and Kistler 1989), and Fig. 1 shows examplesneurons derive directional sensitivity by use of excitatory/ inhibi-that were obtained in a cat by Musicant et al. (1990). Surgi-tory antagonism to compare sound pressure in excitatory and inhib-cal removal of a cat’s pinna completely changes the HRTFitory frequency domains, and a variety of observations are consis-

tent with this idea. 1) Frequency response areas of MD cells consist (Musicant et al. 1990) and HRTFs shift with respect to theof excitatory and inhibitory domains. MD cells exhibited a higher head when pinna orientation changes (Young et al. 1996),proportion of multiple excitatory domains and narrower excitatory showing that the HRTF is determined mainly by the pinna.frequency domains than BD cells, features that may reflect special- Unilaterally deaf humans can localize broadband high-ization for spectral-dependent directional sensitivity. 2) MD sensi- frequency sounds but not tonal stimuli (Butler 1975; Hauslertivity requires sound pressure in excitatory and inhibitory fre- et al. 1983; Slattery and Middlebrooks 1994), showing thatquency domains. Directional sensitivity was evaluated using stim-

directional information can be derived from monaural spec-uli with frequency components confined exclusively to excitatorytral cues. Monaural localization also has been studied indomains (E-only stimuli) or distributed in both excitatory andnormal hearing individuals in which unilateral deafness isinhibitory domains (E/I stimuli) . Each of 13 MD cells that weresimulated by unilateral ear occlusion. They localize alltested exhibited higher directional sensitivity to E/I than to E-

only stimuli; most MD cells exhibited relatively low directional sounds toward the side of the functional ear regardless ofsensitivity when frequency components were confined exclusively actual location. Some residual localization capacity may re-to excitatory domains. 3) MD sensitivity derives from excitatory/ main (Butler 1986; Butler et al. 1990; Fisher and Freemaninhibitory antagonism (spectral inhibition). Comparison of re- 1968; Hebrank and Wright 1974a; Oldfield and Parker 1986;sponses to best frequency and E/I stimuli provided strong support Slattery and Middlebrooks 1994) although whether this isfor spectral inhibition. Although spectral facilitation conceivably due to monaural mechanisms is controversial (Wightmancould contribute to directional sensitivity with direction-dependent

and Kistler 1997). The capacity of other species to localizeincreases in response, the results did not show this to be a signifi-sound monaurally has not been extensively studied. Monau-cant factor. 4) Direction-dependent decreases in responsivenessrally deafened cats can localize the elevation of noise burststo BBN reflect increased sound pressure in inhibitory relative towith normal accuracy (Sutherland 1991, 1994). By compari-excitatory frequency domains. This idea was tested using theson, accuracy of azimuthal localization by monaurally deaf-strength of two-tone inhibition, which is a function of stimulus

levels in inhibitory relative to excitatory frequency domains. The ened cats is better than chance but poor as compared withfinding that two-tone inhibition was stronger at directions where binaural localization (Jenkins and Masterton 1982; Neff andBBN responses were minimal than at directions where they were Casseday 1977).maximal supports the model. Single units in the superior colliculus (SC) derive direc-

tionality in part from monaural cues. Spatially selective neu-rons form an auditory space map in the deep layers of the

I N T R O D U C T I O NSC in a number of species including cat (Middlebrooks and

Although azimuthal (horizontal) localization of tonal Knudsen 1984), ferret (King and Hutchings 1987), guineapig (Palmer and King 1985), and barn owl (Knudsen 1984).stimuli depends exclusively on binaural disparity cues (Mid-

2754 0022-3077/97 $5.00 Copyright q 1997 The American Physiological Society

J0197-7/ 9k20$$no41 10-29-97 18:30:51 neupa LP-Neurophys

by 10.220.33.4 on May 9, 2017

http://jn.physiology.org/D

ownloaded from

http://jn.physiology.org/

SPECTRAL CONTRAST AND DIRECTIONAL SENSITIVITY 2755

sensitivity (e.g., Neti and Young 1992; Rhode andGreenberg 1992, 1994; Shamma and Symmes 1985;Shamma et al. 1993; Young et al. 1988; Zakarauskas andCynader 1993). The hypothetical operation of such a mecha-nism is shown in Fig. 1. HRTFs associated with two sounddirections (0 and 367 elevation) (redrawn from Musicant etal. 1990) are represented by light and heavy lines. A single-unit frequency response area composed of excitatory (/)and inhibitory (0) domains also is shown. Given a broad-band sound with a flat spectrum in the free field, the HRTFswill represent the spectral distribution of sound pressure ateach direction. A stimulus presented at 07 would producerelatively greater pressure in the excitatory than in the inhibi-tory domain thus causing a net excitation and neural dis-

FIG. 1. Spectral contrast model for monaural directional sensitivity.charge. At 367, pressure would be relatively greater in theShaded areas represent a hypothetical frequency response area of a monaural

directional (MD) cell that consists of excitatory (/) and inhibitory (0) inhibitory than in the excitatory domain, producing a netfrequency domains. Line graphs show head related transfer functions for 2 inhibition and silencing of the cell. Overall, neural respon-sound directions (0 and 367 elevation, redrawn from Musicant et al. 1990). siveness to different sound directions would reflect the netThese functions depict sound pressure gain at the tympanic membrane

excitatory/ inhibitory interactions resulting from the differ-relative to the free field. Neural responsiveness reflects excitatory/ inhibitoryential pressure distribution in excitatory and inhibitory fre-antagonism resulting from the differential pressure distribution in excitatory

and inhibitory domains. A broadband stimulus located at 07 produces greater quency domains. Although this illustration depicts how ele-sound pressure in the excitatory domain than in the inhibitory domain, vation sensitivity might be derived, the same principle ap-causing the cell to discharge. The same stimulus located at 367 produces plies to azimuth sensitivity.greater sound pressure in the inhibitory than in the excitatory domain and

Experiments were performed to test five predictions. 1)thus no discharge occurs.HRTFs are postulated to provide directional cues to whichMD cells are sensitive. Changes in pinna orientation produce

In the guinea pig and ferret, both monaural spectral and displacement of HRTF patterns that should cause similarbinaural cues contribute to directional tuning. At stimulus displacements of MD cell azimuth function peaks and nulls.levels within Ç20 dB of a unit’s threshold, azimuth and 2) MD cells should have frequency response areas that con-elevation tuning to broadband noise-burst (BBN) stimula- sist of excitatory and inhibitory domains. 3) MD sensitivitytion is determined largely by monaural mechanisms; at should require sound pressure in excitatory and inhibitoryhigher levels, binaural mechanisms are most important frequency domains. 4) MD sensitivity should reflect excit-(King et al. 1994; Moore et al. 1993; Palmer and King atory/ inhibitory antagonism (spectral inhibition) that pro-1985). duces direction-dependent reductions in response. 5) De-

Monaural and binaural mechanisms also contribute to neu- creases in responsiveness to BBN should occur at directionsral directionality in the lemniscal auditory system of the where there is an increase in sound pressure in inhibitorycat. Azimuth sensitivity is a characteristic of many central relative to excitatory frequency domains. The results of theseauditory neurons, i.e., they respond well at some azimuths experiments are largely consistent with the predictions, sug-and poorly at others (azimuth function peaks and nulls, re- gesting that MD cells use antagonistic inputs from excitatoryspectively) . Responsiveness is poor regardless of sound and inhibitory frequency domains to derive directional sensi-pressure level (SPL) at null directions [auditory cortical tivity from HRTF spectral cues. Some of these results havefield (AI): Imig et al. 1990; Rajan et al. 1990; medial genicu- been reported in preliminary form (Imig et al. 1994; Ironslate body (MGB): Irons 1989; inferior colliculus (IC): Ait- 1989).kin and Martin 1987; Aitkin et al. 1984; Moore et al. 1984]. This work represents a part of the PhD research conductedComparison of binaural and monaural (unilateral ear occlu- by W. Andrew Irons.sion) responses to free-field BBN stimulation shows thathigh-frequency neurons may derive azimuth sensitivity from

M E T H O D Sbinaural disparity cues [binaural directional (BD) cells]and/or monaural spectral cues [monaural directional (MD) Eighteen healthy young-adult cats with clean external ears, trans-cells] at midbrain (IC: Poirier et al. 1996), thalamic (MGB: lucent tympanic membranes, and low-threshold single-unit re-Irons 1989; Samson et al. 1996), and cortical levels (AI: sponses were used in the experiments. All husbandry and experi-

mental procedures were carried out using protocols approved bySamson et al. 1993, 1994). Sensitivity to monaural spectralthe Institutional Animal Care and Use Committee of the Kansascues also contributes to elevation sensitivity. PreliminaryUniversity Medical Center. Chronic recording procedures werereports suggest that MD cells in the IC, MGB, and AI areused in the MGB of 14 cats, and acute recording procedures weresensitive to elevation, whereas BD cells are more likely toused for the remainder. Acute recording procedures have beenbe broadly tuned and/or insensitive (Imig et al. 1995; Poirierdescribed in detail elsewhere (Barone et al. 1996).et al. 1995; Samson et al. 1991). Surgical implant of a chronic recording chamber was carried

How do neurons derive directional sensitivity from mon- out using inhalation anesthesia (0.8–1.5% isoflurane in O2) andaural spectral cues? Antagonistic interactions between excit- standard aseptic procedures under veterinary supervision. Postoper-atory and inhibitory domains in frequency response areas ative discomfort was ameliorated with analgesics. A stainless steel

recording chamber was positioned stereotaxically over a craniot-may provide a spectral contrast mechanism for directional


by 10.220.33.4 on May 9, 2017


ownloaded from


T. J. IMIG, P. POIRIER, W. A. IRONS, AND F. K. SAMSON2756

omy and fastened to the skull with dental acrylic and stainless steel was calibrated by placing a microphone (B&K type 4133 1/2-in)at the center of the loudspeaker array, aiming it at the loudspeaker,screws. A stainless steel tube, used to secure the animal’s head to

the support frame during recording sessions, was attached to the and performing a fast Fourier transform (FFT) on the impulseresponse. Tables of maximum SPLs attainable at different frequen-chamber. The chamber was sealed with a stainless steel cap at times

other than during recording sessions to protect the craniotomy from cies were derived from FFT data and stored in a computer diskfile for use during experiments. High-frequency drivers (Radiomechanical damage and infection. The dura mater was covered

with a broad spectrum antibiotic ophthalmic ointment (bacitracin, Shack 40–1310B) with a usable frequency range between 4 and40 kHz were used for these studies. Output increased from 4 kHzneomycin, and polymyxin) before sealing the chamber.to a peak at 8 kHz at 20 dB/octave, decreased by 5 dB/octave upCats were allowed to recover fully from the surgery before theto 35 kHz, and then decreased at 60 dB/octave. Loudspeakers withfirst recording session, and thereafter recording sessions werelower frequency output were also available for use to characterizescheduled at ¢2-wk intervals. Animals were anesthetized duringfrequency tuning õ4 kHz.chronic recording sessions to prevent movements. Anesthesia was

A random number generator produced BBN waveforms with ainduced with isoflurane in O2 and atropine (0.1 mg/kg im) wasflat spectrum (0–50 kHz) and random amplitude distribution. Theinjected. Isoflurane was replaced with pentobarbital sodium (initialactual spectrum of the BBN delivered to the animal was shapeddose, 10 mg/kg iv) , which was maintained throughout the re-by the sound system (mainly the loudspeaker) . Band-pass stimulicording session with an intravenous infusion pump at a rate suffi-were synthesized by summing together random-phase, equal-am-cient to eliminate pinna reflexes and spontaneous movements (Ç2–plitude sine-waves in 5 Hz steps between the frequency limits of4 mgrkg01

rh01) . Dexamethasone (2 mg/kg iv) was injected tothe band-pass. Band-pass stimuli with band limits of 1–40 kHzreduce the possibility of cerebral edema. Application of ophthalmicwere sometimes used (also referred to as BBN). Although the fineointment prevented corneal drying. Tracheal intubation ensured astructure of the two BBN stimuli differed, their effective band-patent airway. Breathing and heart rates were monitored duringwidths were similar, and there was no difference in responses ofthe recording session, and temperature was maintained using acells that were tested with both stimuli. Stimulus envelopes forthermostatically controlled heating pad. After the recording ses-all stimuli were 50 ms in duration and had linear rise/fall timession, animals were kept warm until they awakened from the anes-of 5 ms.thesia. Marking lesions were placed during terminal recording ses-

Ear plugging was used to infer the responses of single neuronssions for histological localization of recording sites. Each animalto monaural stimulation. Unilateral ear occlusion was effected bywas given a lethal dose of anesthetic at the end of the terminalinjecting a viscous liquid ear mold compound (Ear Mold Impres-recording session and perfused through the heart with a 10% solu-sion Material, All American Mold Lab) into the concha and eartion of formol saline. Histological processing has been describedcanal. Attenuations produced by ear plugs varied between 32 andpreviously (Barone et al. 1996).70 dB in the range of 4–32 kHz, between 35 and 55 dB at 2 kHz,Single-unit recordings were carried out in an electricallyand 25 dB at 1 kHz (Samson et al. 1993).shielded, anechoic, sound-isolation chamber. The anesthetized cat

Data sets consisting of single-unit responses to BBN bursts thatrested in a sling with its head rigidly fixed by clamping the head-varied in azimuth and SPL were compared statistically to determinesupport tube. The head was positioned with the horizontal Horsley-whether ear plugging had a significant effect on the cell’s response.Clarke plane tilting forward and down at an angle of Ç187 fromNonparametric statistics were used because cells discharged a smallhorizontal, which approximates the head position of an alert catnumber of spikes and therefore spike counts are not normally dis-looking forward. The ears were pulled to an upright position usingtributed. One of two methods was used depending on whether orstrings that were attached to the outer surfaces of each pinna.not repeat data sets were obtained using one treatment condition.Sterility was maintained within the recording chamber throughoutIf replications were available, an analysis of variance (ANOVA)the procedure. Single-unit activity was recorded using paralene-was used on the ranked responses. Otherwise, a x 2 test or a Fisherinsulated, tungsten electrodes (Frederick Haer) with nominal im-exact probability test was used (Samson et al. 1993, 1994; Siegelpedances of 1–5 MV measured at 1 kHz in the brain. Details1956). Elevation data were treated in the same manner.concerning single-unit recording, computer control of data collec-

tion, and data analysis have been described previously (Barone etal. 1996; Samson et al. 1993).

R E S U L T SAn array of loudspeakers with similar frequency response char-

acteristics allowed the free-field presentation of sounds the direc- These findings are based on recordings from well-isolatedtions of which could be varied in azimuth (direction in the hori- single units in two tonotopic subdivisions of the MGB (ven-zontal plane passing through the interaural line) and elevation. tral nucleus, VN, and lateral part of the posterior groupLoudspeakers were aimed at the interaural midpoint and located at

of thalamic nuclei, PO) (Imig and Morel 1985a,b) witha distance of 0.79 m from it. An aluminum tubing frame supportedsupplemental data from one unit in cortical field AI. Theloudspeakers along a vertical meridian (vertical array) of an imagi-search for single-unit responses was carried out using BBNnary sphere centered on the cat’s head. The vertical array consistedbursts that were presented sequentially from loudspeakersof seven loudspeakers spaced at 22.57 intervals ({67.5, {45,spaced at 307 intervals along the horizontal array throughout{22.5, and 07) . A horizontal array of 13 loudspeakers spaced at

157 intervals along a 1807 arc of the equator was attached to the the frontal field. At each location, SPL was varied from 0vertical support. Sound direction could be varied in elevation by to 80 dB in 20-dB steps before changing to the next loud-presenting sound from different loudspeakers and could be varied speaker. Single and multiunit low-threshold frequency selec-in azimuth the same way or by rotating the frame about its vertical tivity was assessed regularly to ensure that the electrode wasaxis. located in a high-frequency (ú4 kHz) representation.

Auditory waveform synthesis, acoustic calibration, stimulus tim- Once a single unit was encountered, a data set was ob-ing and sequencing, and data collection were controlled by a PDPtained to assess azimuth sensitivity. This consisted of re-11/73 computer. Stimulus waveforms were generated at an outputsponses to 10–20 repetitions of BBN bursts that were typi-sample rate of 100 kHz using a 16-bit D/A converter (Boys Towncally presented at 307 intervals throughout the frontal fieldNational Research Hospital) , low-pass filtered at 40 kHz (Kemoand between 0 and 80 dB SPL in 20-dB steps at each azi-VBF/8, 0180 dB/octave) to prevent aliasing, attenuated with

computer controllable attenuators, and amplified. Each loudspeaker muth. If a unit responded well at some azimuths and poorly


by 10.220.33.4 on May 9, 2017


ownloaded from



at others ( i.e., was azimuth sensitive) , it was studied further Fig. 2C) indicating that directional tuning was determinedentirely by monaural input. The cell was unresponsive toto characterize its sensitivity to monaural and binaural direc-

tional cues. Additional data sets were obtained using monau- ipsilateral monaural BBN stimulation. Thus this cell wasclassified as MD-E0, MD indicating that it derived azimuthral stimulation of the contralateral ear, monaural stimulation

of the ipsilateral ear, and finally repeating binaural stimula- sensitivity from monaural stimulation and E0 indicating thatit received excitatory (E) input from one ear and that stimu-tion to assess response stability (i.e., contra- and ipsilateral

with respect to the recording site in the left MGB). Monaural lation of the other was ineffective (0) .This cell was also sensitive to the elevation of monauralstimulation was simulated by the use of unilateral ear plug-

ging, which attenuated sound reaching one ear. Rather com- contralateral BBN stimuli (CS, Fig. 2D) . The elevation-level response area (ELRA) represents the cell’s responsesplete response profiles to monaural and binaural stimulation

were obtained for 103 azimuth-sensitive cells in the MGB at elevations distributed along the vertical meridian passingthrough a peak azimuth (457) . The unit was most responsive(Samson et al. 1996). Best frequencies (BFs, center fre-

quency of excitatory response domains) of the sample at 07 elevation and less responsive above and below thehorizontal plane. It was entirely unresponsive at 457 eleva-ranged between 4.2 and 37 kHz.

The role of monaural and binaural mechanisms in de- tion regardless of SPL. There was no significant statisticaldifference between monaural and binaural elevation func-termining a unit’s directional sensitivity was inferred by

comparison of responses to monaural and binaural BBN tions (CS, BS, Fig. 2F) . This is consistent with the previousresults and shows that directional tuning was determinedstimulation as described by Samson et al. (1993, 1994).

Units were classified as BD (n Å 69) if they were azimuth entirely by monaural input.Unlike MD-E0 cells that are strictly monaural, MD-EIsensitive to binaurally presented BBN but were insensitive

to monaural stimulation, i.e., responded well at each direc- cells receive binaural inhibition and an example of such acell is shown in Fig. 14B. The binaural azimuth functiontion, were unresponsive to monaural stimulation, or exhib-

ited completely different response borders (defined by adja- (BS) shows a peak located in the contralateral rear quadrant.The peak is defined by nulls at 30 and 1807, and thesecent peaks and nulls) under binaural and monaural condi-

tions. Units were classified as MD (n Å 30) if they were features also were seen in response to monaural contralateralBBN stimulation (CS), thus earning the cell a MD classifi-sensitive to the azimuth of monaural BBN stimulation and

exhibited some or all of the same response borders under cation. Compared with the binaural function, the contralat-eral monaural function revealed increased responsiveness onmonaural and binaural conditions. MD cells are estimated

to represent Ç17% of the high-frequency neurons in VN the ipsilateral side of the head, leading to the conclusionthat the cell received inhibition from the ipsilateral ear. Al-and PO based on a previous report that 57% of the units in

VN and PO were azimuth sensitive (Barone et al. 1996) though they differ with respect to binaural inhibition, bothMD-E0 and MD-EI cells derive directional sensitivity fromand on the proportion of MD cells (29%, 30/103) in the

azimuth-sensitive sample used for this report. All MD cells monaural spectral cues, and thus their monaural responsesare treated together in the remainder of the report.exhibited short-latency onset responses (minimum 1st spike

latencies ranged between 6.7 and 32.7 ms, mean 10.2 ms), MD cells are sensitive to the direction of monaural BBNstimuli but are insensitive to the direction of monaural tonaland these are the responses described in detail below. Addi-

tionally, some units (4/30) also exhibited long-latency re- stimuli (and also to binaural tonal stimuli in the case of MD-E0 cells). This characteristic is illustrated by the MD-E0 cellsponses in the range of 43–200 ms. These usually occurred

at higher levels, although the long-latency response of one in Fig. 2. An ALRA was obtained to binaural stimulation with20-kHz BF tone bursts (Fig. 2B). Unlike its response to noiseunit was related reciprocally to its short-latency response

over a wide range of SPL. stimulation, the neuron responded well to tone bursts at eachazimuth from which they were presented. The azimuth func-tion (BS 20 kHz, Fig. 2C) shows that discharge rate, averagedMonaural directional sensitivityover SPL, varies to some degree as a function of sound direc-tion but it exhibits less modulation (42%) than the BBN func-Figure 2 compares the monaural and binaural responses

of a MD cell. Its responses to monaural contralateral BBN tions and does not meet the 75% criterion for azimuth sensitiv-ity. Similar differences in elevation sensitivity are seen instimulation (CS) are displayed in the form of an azimuth-

level response area (ALRA), a normalized iso-response con- response to BBN and 20-kHz BF tones (Fig. 2, D–F). Thesefindings suggest that azimuth and elevation sensitivity dependtour plot (Fig. 2A) . The cell was most responsive at 45

and 607 of azimuth and was completely unresponsive at 907 on monaural cues that are present in BBN but not in tonebursts. According to the model, these monaural cues are pinnaregardless of SPL. These data are replotted as an azimuth

function (CS, Fig. 2C) in which responses are averaged spectral cues, and MD cells derive directional sensitivity fromthem by use of excitatory/inhibitory antagonism to compareover SPL and normalized. Responses are near optimal at the

function peak (45 and 607) and minimal at the null (907) . sound pressure in excitatory and inhibitory frequency domains.The remainder of this reports describes the results of experi-The terms ‘‘peak’’ and ‘‘null’’ are used rather loosely

throughout this report to refer to peaks and valleys in azi- ments designed to test the validity of this model.muth and elevation functions. Cells were classified as azi-

MD receptive field location follows changes in pinnamuth sensitive if azimuth function modulation was ¢75%orientation( in this case, modulation was 100%). There was no signifi-

cant statistical difference between monaural contralateral Changing pinna orientation from sideways to forward fac-ing causes a corresponding movement in prominent featuresand binaural azimuth functions (CS and BS, respectively,


by 10.220.33.4 on May 9, 2017


ownloaded from



FIG. 2. MD cells are sensitive to the direction of monaurally presented broadband noise (BBN) bursts and insensitive tothe direction of best frequency (BF) tone bursts (8815-12) . A : azimuth-level response area shows the cell’s response tomonaural contralateral BBN stimulation (CS). Iso-response contour lines and shading represent 5, 25, 50, and 75% ofmaximum (maxima are indicated by large diamonds). Cell responded optimally at 45–607 azimuth, and it was unresponsiveat 907. B : azimuth-level response area shows the cell’s response to binaural BF (BS 20 kHz) tone bursts. Cell responds wellat each azimuth. C : azimuth functions were obtained by averaging responses over sound pressure level (SPL) and normalizing.Responses obtained using contralateral monaural (CS) and binaural (BS) BBN stimulation were statistically indistinguishable.In both cases, functions exhibited 100% modulation (i.e., high azimuth sensitivity) . In response to BF tone bursts presentedbinaurally (BS 20 kHz), the cell exhibited 42% modulation (lower azimuth sensitivity) . D : elevation-level response areashows the cell’s response to monaural contralateral BBN stimulation (CS). Data in panels D–F were obtained at 457 azimuth.E : elevation-level response area shows the cell’s response to binaural BF (20 kHz) tone bursts. Cell responds well at eachelevation. F : elevation functions were obtained by averaging responses over SPL and normalizing. Those obtained usingmonaural (CS) and binaural (BS) BBN stimulation were statistically indistinguishable. In both cases, they exhibited ¢99%modulation (high elevation sensitivity) . In response to BF tone bursts presented binaurally (BS 20 kHz), the cell exhibitedmuch lower modulation (39%). Recording site was located in the lateral part of the posterior group of thalamic nuclei.

of the HRTF with respect to the head (Young et al. 1996). Pinna orientation was controlled by use of positioningstrings that were glued to the medial and lateral margins ofIf MD directional tuning depends on pinna-derived spectral

cues, then the locations of azimuth-function peaks and nulls each pinna close to the scalp. Adjustment of the tension onthe strings allowed independent positioning of each pinna.should follow changes in pinna orientation on the side of

the ear that provides excitatory input to the cell. By placing knots in the strings and using a slotted holder to


by 10.220.33.4 on May 9, 2017


ownloaded from



engage the knots, different orientations could be reliablyreproduced. Figure 3 shows both pinnae in the forward andside positions. Pinna orientation can be described roughlyby reference to an imaginary line oriented perpendicular tothe plane defined by the free margins of the pinna. In bothforward and side positions, this line projected nearly hori-zontally. In the forward position, it formed an angle ofÇ257with midsaggital plane; in the side position it formed anangle ofÇ40–457. Thus pinna orientation in these two posi-tions differed by 20–257 of angular rotation.

By independently changing the position of each pinna, itwas possible to show that the location of peaks and nulls ina MD cell’s azimuth function followed changes in orienta-tion of the pinna on the side of the excitatory ear. Theresponses of a thalamic MD-E0 unit that received contralat-eral excitatory input are illustrated in Fig. 4A. Azimuth func-tions are identified according to pinnae orientations (e.g.,IsCf, Ipsi side Contra forward) and are grouped accordingto the orientation of the contralateral pinna. Those plottedwith continuous lines were obtained with the contralateralpinna in the forward position (Cf); those plotted with inter-rupted lines were obtained with the contralateral pinna inthe side position (Cs). Azimuth function peaks in the contra-lateral pinna forward group are displaced toward the left ofthose in the contralateral pinna side group. The mean bestazimuth (midpoint of the 75% range) for the two Cf func- FIG. 4. Effect of pinna orientation on azimuth tuning of 2 MD cells.

Orientation of each pinna was changed independently to assess its effecttions was 16.57, the mean for the five Cs functions was 37.47,on azimuth tuning. Peaks of azimuth functions obtained with the contralat-a difference of Ç217 which corresponds closely with theeral pinna in the forward orientation (continuous lines) are displaced to theestimated angular rotation of the pinna.left of those obtained with the contralateral pinna in the side orientation

There was no systematic relationship between ipsilateral (interrupted lines) . Orientation of the ipsilateral pinna has no significantpinna orientation and best azimuth. With the contralateral effect on the locations of azimuth function peaks. Pinna orientations are

described by the following notation: I, ipsilateral pinna; C, contralateralpinna in the forward position, similar best azimuths werepinna; f, forward orientation; s, side orientation. In both cases, binauralfound for forward and side orientations of the ipsilateralBBN stimulation was used. A : unit 8606-19. Recording site was located inpinna (IsCf, 167, IfCf, 177) . With the contralateral pinna the lateral part of the posterior group of thalamic nuclei. Data were obtained

oriented to the side, there was greater variation in best azi- in the following sequence: IsCs, IsCf, IfCf, IfCs, IfCs, IfCs, IsCs. Sequen-tial order is presented to show that azimuth function peak location is relatedmuth (34–437) , but there was no systematic relationship toto pinna orientation and not to some unrelated changes in the cell’s responseipsilateral pinna orientation. Average best azimuths for Isproperties that occurred over time. B : unit 9215-9. Recording site was(38.57) and If functions (36.77) were similar. located in cortical field AI. Data were obtained in the following sequence:

Although the effect of pinna orientation was tested using IsCf, IsCs, IfCs, IfCf, IsCf, IsCs, IfCs, IfCf.only one thalamic MD cell, confirming observations alsowere obtained from a MD-E0 cell located in cortical field AI. pinna forward group are displaced toward the left of thoseFigure 4B shows eight azimuth functions that were obtained in the contralateral pinna side group. In neither the contrausing different pinna orientations. As was the case with the forward or contra side group does ipsilateral pinna orienta-thalamic cell, azimuth function peaks in the contralateral tion bear a systematic relationship to any features of the

azimuth function. In both MD units, directional tuning fol-lowed the orientation of the pinna on the side of the earthat provided excitatory drive and was unaffected by theorientation of the other pinna. Although these data are lim-ited to the responses of two MD cells, the results are unam-biguous and consistent with the hypothesis that HRTFs pro-vide the monaural spectral cues from which MD cells derivedirectional sensitivity.

Frequency response areas of MD cells exhibit excitatoryand inhibitory domains

The model predicts that MD cells receive excitatory andinhibitory inputs from different frequency domains. Fre-

FIG. 3. Two symmetrical pinna orientations (forward and side) viewed quency response areas of MD cells were mapped to deter-from the front ( left) and top (right) . Pinnae were pulled to forward andmine if they exhibited this property. Excitatory frequencyside positions by adjusting the tension on strings glued to their lateral

margins. domains were determined by presenting tone bursts at a peak


by 10.220.33.4 on May 9, 2017


ownloaded from



tone stimulation paradigm. Two-tone stimuli consisted ofconstant (F1) and variable (F2) frequency components thatwere presented simultaneously. F1 was centered in an excit-atory domain Ç15 dB above threshold. F2 was presented atdifferent frequencies and SPLs. If the two-tone response wassmaller than the F1 response, this was taken as evidencethat the F2 stimulus was inhibitory. Figure 5B shows thetwo-tone response area that delimits regions of two-toneinhibition. The 50% iso-response contour was used to indi-cate the approximate locations of inhibitory domains andwas combined with the excitatory response (Fig. 5A) toshow both inhibitory (‘‘-’’ and stippled) and excitatoryfrequency domains in a frequency response area (FRA,Fig. 5C) .

FRAs consisting of flanking excitatory and inhibitory do-mains were characteristic of all MD cells that were testedusing two-tone stimulation. FRAs consisting of one (Fig. 6,A and B) or two low-threshold excitatory domains (Figs.5C, 6C, and 14A) were encountered commonly. Figures 6Dand 11A show examples with three excitatory domains. Thecell in Fig. 6D was lost before the existence of inhibitorydomains could be verified. The finding that FRAs of MDcells exhibit distinct excitatory and inhibitory domains isconsistent with the spectral contrast model.

Detailed excitatory FRAs were available for a sample of65 BD and MD cells. FRAs for BD cells that exhibitedbinaural inhibition were obtained using monaural stimula-tion. FRAs for the remaining BD cells were obtained usingbinaural stimulation because they exhibited binaural facilita-tion and were less responsive to monaural stimulation. Thecombined MD/BD sample included 15 units that exhibitedtwo or more excitatory frequency domains. In each FRA,excitatory domains had thresholds within 20 dB of eachother and were separated by a frequency range throughoutwhich the cell was not excited by tones at intensities õ60dB above threshold (Figs. 5C, 6, C and D, 11A, and 14A) .Some cells in both the MD and BD samples exhibited multi-ple excitatory domains, but this feature was more commonin the MD (9/24, 37%) than in the BD (6/41, 15%) sample.

FIG. 5. Construction of a frequency response area showing excitatoryThese proportions were significantly different (Fisher exactand inhibitory domains (9320-5) . A : an excitatory frequency response area

shows a cell’s responses to tone-burst stimulation as a joint function of probability test, P õ 0.037) (Siegel 1956).frequency and SPL. Iso-response contour lines and shading represent 5, 25, There was not any obvious harmonic relationship between50, and 75% of maximum (maximum is indicated by a large diamond). B : center frequencies of the excitatory domains in the2-tone frequency response area. Darker shading indicates greater response

multipeaked FRAs. The center frequency of an excitatorymagnitude, and the diamond indicates the maximum response. Inhibitorydomains are indicated by regions of lighter shading. The 50% response domain was measured at 30 dB above its lowest threshold.contour is indicated by a bold line. C : a frequency response area showing Only adjacent peak spacing was determined in the case ofexcitatory and inhibitory domains is constructed by adding the 50% contour FRAs with three peaks. The ratio of the center frequenciesfrom B to the excitatory frequency response area (A) . Inhibitory domains

of the higher frequency peak divided by the lower rangedare stippled and identified by circled minus signs. Recording site was locatedin the lateral part of the posterior group of thalamic nuclei. from 1.45 to 2.5. The average ratio was 1.82 and was slightly

higher for BD cells (average 1.98) than for MD cells (aver-age 1.75).azimuth. Tones varied in frequency (one-quarter or one-

Many MD cells had narrower excitatory frequency do-eighth octave steps between 4 and 40 kHz) and SPL (10–mains than did BD cells. To make a quantitative comparison,20 dB steps from near threshold to 80 dB SPL). Monauralwidths of excitatory domains were measured at 30 dB abovestimulation was used to ensure that binaural inhibition, ifthreshold in those cases for which thresholds were deter-present, did not influence the observations. Responses weremined to a precision of 10 dB (n Å 35). In the case of unitsplotted as iso-response contours (e.g., Fig. 5A) . In this case,with two or more separate excitatory frequency domains,the contours delimited two, narrow, low-threshold, excit-widths were obtained for each. Figure 7 shows the distribu-atory domains that were separated from each other by ations of tuning widths for the MD and BD samples. Widthsregion in which tone bursts did not excite the cell at any

tested SPL. Inhibitory domains were identified using a two- of excitatory domains for MD cells averaged 0.52 octaves,


by 10.220.33.4 on May 9, 2017


ownloaded from



FIG. 6. Examples of frequency response areas of MD cells showing excitatory and inhibitory frequency domains. Cellin D was lost before the existence of inhibitory areas could be tested. A : 9314-2, recording site located in the lateral part ofthe posterior group of thalamic nuclei. B : 9317-7, recording site in the ventral nucleus. C : 9314-5, recording site in theventral nucleus. D : 9314-10, recording site was located in the lateral part of the posterior group of thalamic nuclei.

whereas for BD cells averaged 1.20 octaves ( t Å 05.73, exhibit directional sensitivity to stimuli with spectral compo-nents that engage both excitatory and inhibitory domainsdf Å 51, P õ 0.0001).(E/I stimuli) and should be insensitive to the direction ofstimuli with spectral components limited only to excitatoryMD sensitivity usually requires stimulation of bothdomains (E-only stimuli) . To test the prediction, azimuthexcitatory and inhibitory frequency domainsand elevation function modulation was measured for E-only

The model hypothesizes that directional sensitivity results and E/I stimuli. Useful data were obtained from 13 MDfrom comparison of sound pressure in excitatory and inhibi- cells (BF range: 6–37 kHz). In all cases, excitatory andtory frequency domains. If this is true, then MD cells should inhibitory frequency domains initially were delimited as de-

scribed above.The elevation sensitivity of the MD cell the FRA of which

appears in Fig. 6C was tested using a variety of stimulusspectra. Its excitatory and inhibitory domains are representedin Fig. 8A (bottom) . Data were obtained using the E-onlyand E/I stimuli the spectral compositions of which are shownin Fig. 8A, top (r, 1 or 2 frequency components; ,indicate BBN or band-pass stimuli) . ELRAs obtained usingE/I stimuli exhibited narrow elevation tuning (e.g., Fig.8C) ; those obtained using E-only stimuli exhibited insensi-tive responses (e.g., Fig. 8D) . Elevation function modula-tion bears a systematic relationship to stimulus type (Fig.8B) . Relatively low modulation (õ45%) occurred in re-FIG. 7. Excitatory domains of MD cells are narrower than those of

binaural directional (BD) cells. Widths of 25% contours were measured sponse to E-only stimuli that consisted of one frequency30 dB above threshold (determined to a resolution of 10 dB). For units component (22.2 kHz), two frequency components thatwith multiple excitatory domains with thresholds that differed by õ30 dB, were located in different excitatory domains (15.5 / 22.2the width of each was measured. Frequency response areas were obtained

kHz), or many hundreds of frequency components in a sin-using monaural stimulation for MD cells and for BD cells that show binauralinhibition. Binaural stimulation was used for BD facilitatory cells. gle excitatory domain [band-pass (BP) 22–24 kHz, BP 22–


by 10.220.33.4 on May 9, 2017


ownloaded from



FIG. 8. Elevation sensitivity of a MD-E0 cell (9314-5 ; see text for definition) requires stimulation of both excitatory andinhibitory frequency domains. A : spectral composition of test stimuli is shown in relationship to excitatory and inhibitoryfrequency domains. Bottom : location of excitatory and inhibitory domains (from data in Fig. 6C) . , cell’s response totone-burst stimulation averaged over SPL and represents the location of the excitatory domains. Location of the inhibitorydomain is indicated ( – – – and r) . Spectral composition of stimuli used to test elevation sensitivity is shown ( top) ; r,individual frequency components; , band-pass stimuli composed of discrete frequency components spaced in 5-Hzintervals. Stimuli are divided into 2 groups: those with energy confined only to excitatory domains (E-only stimuli) andthose with energy in both excitatory and inhibitory domains (E/I stimuli) . B : elevation functions (responses averaged overSPL) exhibited higher modulations for E/I stimuli ( ) than for E-only stimuli ( – – – ). C : example of an elevation-level response area (ELRA) obtained using an E/I stimulus (BBN). D : example of an ELRA obtained using an E-onlystimulus (22.2 kHz).

35 kHz]. Relatively high modulation (ú90%) was obtained both excitatory and inhibitory frequency domains. IndividualMD units each showed greater modulation to E/I stimulifrom E/I stimuli that consisted of two frequency components

(20 / 22.2 kHz) or many frequency components (BP 13– than to any E-only stimulus. Data for the sample are summa-rized in Fig. 10, which shows the distribution of elevation24 kHz, BBN). The cell was tested with an additional E/

I stimulus (BP 13–20 kHz) but showed no short-latency and azimuth function modulations for E-only and E/I stim-uli. The distribution is bimodal. The high modulation moderesponse so no elevation function appears for this stimulus

in Fig. 8B. Lack of a response to this stimulus is discussed is composed mainly of modulations for E/I stimuli thatranged from 64 to 100%. The low modulation mode is com-later.

Similar results were obtained for azimuth function modu- posed exclusively of modulations for E-only stimuli thatranged from 21 to 79%. Overall, the mean modulations forlation. One example is shown for the MD cell the FRA of

which appears in Fig. 6A. Spectral compositions of test stim- the two types of stimuli were significantly different (EI,91.4%; E-only, 37.9%, t Å 17.7, df Å 58, corrected foruli are shown in relationship to the cell’s excitatory and

inhibitory frequency domains in Fig. 9A. In response to a heterogeneous variance, P õ 0.0001).Stimuli were classified according to number of frequencybroadband E/I stimulus, the cell showed a focal response in

the contralateral rear quadrant (Fig. 9C) . It was insensitive components, i.e., one, two, or hundreds in the case of band-pass stimuli and BBN. Modulations for most E-only stimulito the location of a narrow band E-only stimulus (Fig. 9D) .

Azimuth functions (Fig. 9B) reveal relatively little modula- were lower than those for any E/I stimulus regardless ofwhether the E-only stimulus was a single frequency (E, meantion in response to E-only stimuli and greater modulation to

E/I stimuli. 35%), two frequency components located in different excit-atory domains (E / E, mean 49%), or hundreds of fre-These examples support the hypothesis that directional

sensitivity of MD cells requires frequency components in quency components confined to a single excitatory domain


by 10.220.33.4 on May 9, 2017


ownloaded from



FIG. 9. Azimuth sensitivity of an MD-E0 cell (9314-2) requires stimulation of both excitatory and inhibitory frequencydomains. A : spectral composition of test stimuli in relationship to excitatory and inhibitory frequency domains. Bottom :location of excitatory and inhibitory domains (from data in Fig. 6A) . See Fig. 8 legend for details. B : azimuth functions(responses averaged over SPL) exhibited higher modulations for E/I stimuli ( ) than for E-only stimuli ( – – – ). Someazimuth functions are averages of several azimuth-level response area (ALRA) data sets: BBN (n Å 3), band-pass (BP)1–15.5 (n Å 2), BP 13–40 (n Å 4), BP 13–15.5 (n Å 2). C : example of an ALRA obtained using an E/I stimulus (BBN).D : example of an ALRA obtained using an E-only stimulus (BP 13–15.5 kHz).

(BP E, mean 37%). E/I stimuli resulted in relatively highmodulation regardless of whether they were broadband(BBN, mean 92%), narrower band-pass (BP E/I, mean88%), or two frequency components (E / I, mean 88%).This shows that in most cases stimulation of excitatory andinhibitory domains was necessary to produce MD sensitivity.Furthermore, MD sensitivity resulted from stimulation ofexcitatory and inhibitory domains regardless of whether thisinvolved two frequency components or hundreds of fre-quency components.

Directional sensitivity to an E-only stimulus: anexceptional unit

The distributions of E/I and E-only modulations wereFIG. 10. Directional response modulation is greater to E/I than to E-only nonoverlapping except for the responses of two cells that

stimuli. This graph shows the azimuth and elevation function modulations exhibited relatively high modulation to E-only stimuli. Oneobtained to all E/I and E-only stimuli that were used in the sample of 13 unit contributed modulations of 52% (BF tone) and 66%MD cells. Each cell was tested with several stimuli, and for each individual

(narrow band-pass) for two E-only stimuli. Its responsescell, modulations were greater to E/I stimuli than to E-only stimuli. Forwere somewhat variable between data set repetitions so wethe sample, modulation was significantly greater to E/I than to E-only

stimuli. BP EI, band-pass stimulus that included both excitatory and inhibi- don’t attribute much significance to these relatively hightory domains; E / I, 2 frequency components, one of which was located modulations. But the other cell was more interesting becausein an excitatory domain, the other in an inhibitory domain; E, a single it exhibited very consistent responses and a high modulationfrequency component located in an excitatory domain; BP E, a band-pass

to the azimuth of a two-tone stimulus in which both compo-stimulus that was limited to an excitatory domain; E / E, 2 frequencycomponents located in different excitatory domains. nents were located in excitatory domains. Its frequency re-


by 10.220.33.4 on May 9, 2017


ownloaded from



sponse area (Fig. 11A) consisted of three excitatory domains to a tonal BF stimulus. Greater responsiveness to a multifre-quency stimulus indicates facilitation; less responsivenesscentered at 6, 10, and 18 kHz. There was an inhibitory

domain between the 6- and 10-kHz excitatory domains, but indicates inhibition. Figure 12A shows a unit’s dischargerate (DR, averaged over SPL) plotted as a function of azi-the remainder of the frequency range was not explored using

the two-tone paradigm. This MD-E0 cell responded only to muth for BF tones and three different E/I stimuli. At eachazimuth, the DRs to E/I stimuli were smaller than those tostimulation of the contralateral ear. Figure 11B shows azi-

muth functions in which responses are plotted as spikes/ the BF stimulus, suggesting that E/I stimuli produced a netinhibition at each direction. Figure 12B shows another unit’sstimulus (averaged over SPL) rather than normalized re-

sponses as was done in previous figures. In response to BBN DR elevation functions for a BF stimulus and a variety ofstimulation, the peak response occurred behind the head E/I stimuli. DRs for E/I stimuli ( interrupted lines) were(1807) with nulls at 030 and 0607. Considering only the considerably smaller than those to the BF stimulus (continu-{907 azimuth range over which measurements were ob- ous lines) at all directions except 0457. Here, the responsestained for E-only and E/I stimuli, modulation was greatest to BBN and the BF tone were approximately equal. Theseto BBN (93%). Relatively low modulation was obtained for data are typical of the sample in showing that DRs to E/I6-kHz (37%) and 10-kHz tones (43%), consistent with the stimuli were less than or equal to those to BF stimuli. Thisresponses of most other MD cells to tonal stimuli. Higher shows that spectral inhibition is a major contributor to themodulation (79%) occurred in response to a two-tone stimu- directional tuning of MD cells in response to E/I stimuli.lus consisting of 6- and 10-kHz components. This was the Although responses to E/I stimuli revealed only spectralhighest modulation seen to any two-tone E-only stimulus in inhibition, this does not completely rule out the possibleour sample. existence of spectral facilitation. The above-mentioned re-

Directional sensitivity to the two-tone stimulus appears to sults are based on DRs that were averaged over SPL. Facili-reflect suppression of the 10-kHz response by the 6-kHz tation could be present at certain SPLs but not be evidentresponse. One interpretation is that the tone-evoked excit- in the averaged response because of a predominance of inhi-atory response was followed by a period of decreased re- bition at other SPLs. An analysis was performed to determinesponsiveness. At030 and0607 where the two-tone response whether facilitation was present to any stimulus within ashows a null, the first spike latency to the 6-kHz tone was data set. This entailed selecting the maximum spike countshorter than that to the 10-kHz tone for equal SPL stimuli.At these azimuths, the magnitude of response to the two-tonestimulus was similar to that of the 6-kHz tone. Decreasedresponsiveness after the 6-kHz excitation could eliminatethe excitatory response to the 10-kHz tone. At directionswhere the two-tone response was near maximal (e.g., 0–907) , the response to 10-kHz stimulation occurred at shorterlatency than the response to 6-kHz stimulation for equal SPLstimuli, and the response to the two-tone stimulus was simi-lar in magnitude to that of the 10-kHz stimulus.

Azimuth-dependent differences in first spike latencies tothe two tones are understandable because of threshold differ-ences; at 030 and 0607, threshold was considerably higherfor 10-kHz than that for 6-kHz tones, whereas thresholdsfor 6- and 10-kHz tones were more nearly equal at otherazimuths. At each azimuth, first spike latencies showedmonotonic decreases with increasing SPL reaching asymp-totic minima of 11 ms (10 kHz) and 20 ms (6 kHz). Atdifferent azimuths, threshold differences shifted the 6- and10-kHz latency functions with respect to each other alongthe SPL axis, causing the 6-kHz response to occur earlier atsome azimuths, and the 10-kHz response to occur earlier atothers.

Directional sensitivity derives predominantly fromdirection-dependent strength of spectral inhibition

The model postulates that antagonistic excitatory-inhibi-tory spectral interaction (spectral inhibition) causes dis- FIG. 11. A MD-E0 cell (9404-11) that exhibited azimuth sensitivity tocharge rate modulation. Alternatively, facilitatory interac- a 2-tone E-only stimulus. A : frequency response area obtained at 0907

azimuth using contralateral monaural stimulation. Cell was unresponsive totions also could produce modulation. The relative contribu-stimulation of the ipsilateral ear. B : azimuth functions obtained using antions of spectral facilitation and spectral inhibition toE/I stimulus (BBN) and 3 different E-only stimuli (6, 10, and 6 / 10directional sensitivity was evaluated. Spectral facilitation kHz). Unlike most MD cells, this one showed a high degree of azimuth

and inhibition can be assessed by comparing a unit’s respon- sensitivity to the 2-tone E-only stimulus. Recording site was located in thelateral part of the posterior group of thalamic nuclei.siveness to a multifrequency stimulus with its responsiveness


by 10.220.33.4 on May 9, 2017


ownloaded from



that differs among units to E-only stimuli or other effects(e.g., Fig. 11) cannot be determined from these data.

Effect of sound direction on strength of two-toneinhibition

The model postulates that directional variation of respon-siveness to BBN is a result of changes in the level of soundpressure in inhibitory relative to excitatory domains. Thisidea was tested using two-tone stimuli. The strength of two-tone inhibition is a function of the relative stimulus levels inexcitatory and inhibitory frequency domains; e.g., increasedlevel in an inhibitory domain relative to an excitatory domaincauses stronger two-tone inhibition. According to the model,two-tone inhibition should be relatively strong at a BBNresponse null and relatively weak at a response peak.

Four MD cells were tested with monaural, two-tone stimu-lation presented from peak and null directions. The responsesof a MD-EI cell are illustrated in Fig. 14. Azimuth functionswere obtained for the full 3607 circle around the head usingbinaural and monaural contralateral BBN stimulation (BSand CS, respectively, Fig. 14B) . The cell’s monaural FRAexhibited two excitatory domains and flanking inhibitorydomains (Fig. 14A) . Figure 14C shows normalized re-sponses to monaurally presented two-tone stimuli that con-sisted of a constant frequency F1 (15.5 kHz, centered in thelower excitatory domain) and a variable frequency F2. The

FIG. 12. Directional response modulation is produced by spectral inhibi- amplitudes of F1 and F2 in each two-tone stimulus weretion. A : azimuth functions for responses to BF tonal (14.3 kHz) and E/I adjusted so that they were identical in SPL (in the free field) .stimuli (9314-2) . These same data are shown in Fig. 9B except responses

Two-tone responses were obtained at a null direction (307are plotted as spikes/stimulus. B : elevation functions for BF tonal (22.2kHz) and E/I stimuli (9314-5) . These same data are shown in Fig. 8B azimuth) using four different SPLs (continuous lines) . Atexcept responses are plotted as spikes/stimulus. Cells are less responsive each SPL, the two-tone stimuli produced strong inhibitionto E/I than to BF stimuli revealing spectral inhibition. when F2 was located within frequency ranges adjacent to

F1. Two-tone inhibition was minimal at a peak direction(1207 azimuth, interrupted lines) . These results are consis-

response from each data set, regardless of stimulus direction tent with the hypothesis. At a null direction, increased soundor SPL (in those cases in which replicated data sets were pressure within inhibitory domains relative to the 15-kHzavailable, maximum responses were averaged). Maximumresponse values were obtained from data sets for E/I stimuli(RE/I ) and BF stimuli (RBF) , and these were used to expressmaximum responsiveness to an E/I stimulus as a percentageof the maximum responsiveness to the BF stimulus (100∗RE/I /RBF) . Figure 13 shows the ratios that were obtainedfrom the MD sample. If facilitation occurred in response toa particular stimulus in an E/I data set, then the value ofthe ratio should be ú100%. In most cases, the ratio was°100%. In only a few cases was it larger, and in these cases,the relative response difference between E/I and BF stimuliwas small. Thus if there is a net facilitation of maximalresponses to E/I stimuli, it occurs only in a small proportion

FIG. 13. There is negligible net spectral facilitation in the responses ofof the cases and it is of relatively small magnitude. MD cells to E/I stimuli. Responses to EI stimuli were compared with

responses to BF tones (E-only stimulus) . In the case of units that exhibitedAlthough negligible net facilitation appears to be presentú1 excitatory domain, BF was taken as the center frequency of the domainin the responses of MD cells to E/I stimuli, it is possibleto which the cell was most responsive. Comparisons were obtained fromthat MD cells may exhibit a net facilitation to E-only stimuli.ALRA (12 units) and ELRA data sets (2 units) . To make the comparison,

Responses to 14 multicomponent E-only stimuli were ob- the maximum response (spike counts) was taken from EI (REI ) and BFtained from eight MD cells. Responsiveness ratios to multi- (RBF) data sets, regardless of the direction or SPL to which it occurred. If

ú1 data set was available for a particular stimulus, maximum responsesple-component E-only stimuli varied from 38 to 147%, butwere averaged. These data were used to express responsiveness to an EIon average were nearly identical to that for BF stimulistimulus as a percentage of the responsiveness to the BF stimulus (100∗REI /(100 { 25.7%, SD). Whether this variability represents ex- RBF) . Frequency histogram shows that, in most cases, the ratio was°100%.

perimental error (e.g., uncontrolled changes in unit respon- This shows that the greatest response to EI stimuli was usually less thanor equal to the greatest response to BF stimuli.siveness) , a mixture of spectral inhibition and facilitation


by 10.220.33.4 on May 9, 2017


ownloaded from



FIG. 14. Magnitude of 2-tone inhibition is greater at a null than at a peak azimuth (9318-1) . A : a frequency responsearea showing excitatory and inhibitory frequency domains. B : azimuth functions (responses averaged over SPL) obtainedusing binaurally (BS) and monaurally (CS) presented BBN. Responses were averaged over SPL. C : responses to 2-tonestimuli using a 15.5 kHz constant frequency (F1), and equal SPL F1 and variable frequency (F2) components. Two-toneinhibition is stronger at a null (307) than at a peak (1207) azimuth. D : responses to 2-tone stimuli using a 26.3 kHz F1.Strong 2-tone inhibition is present at both 30 and 1207. Recording site was located in the ventral nucleus.

excitatory domain could account for decreased respon- using BBN stimulation. The unit was most responsive atelevations of 0 and 22.57 and least responsive at directionssiveness to BBN stimulation. At a peak direction, stimulus

levels were too low in inhibitory relative to excitatory do- below the horizontal plane. Tone burst stimulation shows anexcitatory domain centered at 20 kHz (Fig. 15B) .mains to suppress the cell’s response.

Interestingly, two-tone stimulation using a F1 centered in Responses to two-tone stimulation are displayed in Fig.15C . These are averages of responses that were obtained atthe upper excitatory range (26.3 kHz, Fig. 14D) showed

strong and more or less equivalent inhibition at both null six different levels over a 50-dB range beginning Ç10 dBabove F1 threshold. The response to the F1 stimulus aloneand peak directions. This suggests that the excitatory drive

produced by the upper excitatory domain was suppressed at was nearly identical at the two locations. The two-tone func-tion for the null (022.5%) shows prominent inhibitoryboth peak and null locations and thus may not have contrib-

uted much, if any, to the cell’s BBN response. Although this troughs on both the high- and low-frequency sides of the20-kHz F1 frequency. The two-tone function for a peakis only speculation in the case of this unit because it was

not tested using an E/I stimulus that included only the high- elevation (22.5%) shows little evidence for inhibition on thehigh-frequency side. There is a trough on the low-frequencyfrequency excitatory domain, there is other evidence that

suggests that excitatory drive from one of a cell’s multiple side, but it is not as deep as that obtained at the null elevation.Thus two-tone inhibition appears to be present at both peakexcitatory domains can be suppressed completely at both

peak and null locations. The unit shown in Fig. 8 was unre- and null directions, but it is more prominent at the null.Comparable two-tone data also were obtained on two othersponsive to an E/I stimulus (BP 13–20 kHz) that included

one of two excitatory domains. MD cells in which two-tone inhibition was stronger at nullthan at peak directions. These data suggest that MD sensitiv-Figure 15 shows the responses of another MD cell to two-

tone stimulation presented from null and peak elevations. ity is a result of changes in the level of sound pressure ininhibitory relative to excitatory frequency domains.This cell was classified as MD-EI because it received excit-

atory input from the ipsilateral ear and inhibitory input fromDirectional tuning varies with pattern of E/I stimulationthe contralateral ear. All responses shown here were obtained

using ipsilateral monaural stimulation. Figure 15A shows an FRAs of MD cells typically exhibit multiple excitatorydomains, multiple inhibitory domains, or both multiple excit-elevation function (averaged over SPL) that was obtained


by 10.220.33.4 on May 9, 2017


ownloaded from



stimulation of the excitatory domain and the low-frequencyinhibitory domain (BP 1–15.5 kHz) resulted in a responsenull at 1807. Because response nulls presumably occur atdirections where spectral inhibition is strongest, this suggeststhat different inhibitory domains produce different direc-tional patterns of spectral inhibition.

Different local minima in the BBN azimuth function maybe the result of spectral inhibition produced by differentinhibitory domains. A null in the BBN function at 307 corre-sponds to the null in the BP 13–40 kHz function but not inthe BP 1–15.5 kHz function. A null in the BBN function at1807 corresponds to the minimum of the BP 1–15.5 kHzfunction but not one in the BP 13–40 kHz function. A moresystematic analysis of this problem must await additionaldata.

D I S C U S S I O N

MD cells in the MGB derive directional sensitivity frommonaural cues that are present in BBN but not tone bursts,as is the case for MD cells in cortical field AI of the cat(Clarey et al. 1995; Samson et al. 1993). Our results suggestthat MD neurons derive directional sensitivity from monau-ral HRTF cues using antagonistic excitatory/ inhibitory inter-actions to compare sound pressure in excitatory and inhibi-tory frequency domains.

Monaural spectral cues

MD cells were distinguished from other cells in the MGBby virtue of their directional sensitivity to monaurally pre-sented BBN bursts. Although we made no attempt to mea-sure the acoustic cues from which MD cells derived direc-tional sensitivity, we assume that they were high-frequencyspectral cues as described by the HRTF (Musicant et al.1990; Rice et al. 1992). A critical reader might questionthis interpretation, however, because the use of multipleloudspeakers to present sounds from different azimuths in-troduces a potential source of error. Loudspeakers had simi-lar but slightly different frequency response characteristics,

FIG. 15. Magnitude of 2-tone inhibition is greater at a null than at a and these introduce loudspeaker-dependent variations inpeak elevation (9404-21) . This was a MD-EI cell that received excitatory noise spectra in addition to the HRTF-dependent spectralinput from the ipsilateral ear and inhibitory input from the contralateral variation. Nevertheless, it seems very unlikely that fre-ear. This figure shows responses to monaural ipsilateral stimulation. A : an

quency-response differences among loudspeakers accountedelevation function that was obtained using BBN stimulation at 0307 offor anything more than a minor component of differencesazimuth. Responses are averaged over SPL. Cell was most responsive at 0

and 22.57 and was relatively unresponsive at directions below the horizontal in unit responsiveness that we interpreted as direction depen-plane. B : frequency response area showing a single excitatory domain cen- dent. The azimuth tuning of a few MD cells was studiedtered at 20 kHz. C : responses to 2-tone stimuli using a 20 kHz F1. F1 and using only a single moveable loudspeaker (Figs. 2 and 4A) .F2 were presented using equal attenuations, and responses represent the

In these cases, there can be no question that response modu-average of 6 different attenuations (50 dB range). At the ‘‘null’’ direction,the 2-tone response function shows deep troughs on either side of the F1 lation was azimuth dependent not loudspeaker dependent.stimulus, indicating the presence of 2-tone inhibition. High-frequency In a previous study of MD cells in cortical field AI (Samsontrough is absent at the ‘‘peak’’ elevation, and the low-frequency trough is et al. 1993), azimuth functions for a given MD cell wereshallower and shifted toward lower frequencies.

found to be similar whether they were obtained using asingle moveable loudspeaker or a multiloudspeaker array.Furthermore, similar azimuth functions were obtained withatory and multiple inhibitory domains. Stimuli that engage

different combinations of excitatory and inhibitory domains the array rotated to different positions so that different loud-speakers were used to present sound from the same direction.may produce different patterns of directional tuning, as seen

in the case of one cell the FRA of which consists of one These observations show that neural responsiveness of MDcells depends on azimuth and are not the result of differencesexcitatory domain flanked by two inhibitory domains (Fig.

9A) . Azimuth functions are shown in Fig. 9B . Stimulation in loudspeaker characteristics. Thus the HRTF spectral cuesseem to be the main determinant of neural responsivenessof the excitatory domain and the high-frequency inhibitory

domain (BP 13–40 kHz) resulted in a response null at 307, to different sound directions.


by 10.220.33.4 on May 9, 2017


ownloaded from



Effect of pinna orientation low in response to stimuli with frequency components lim-ited to excitatory frequency domains but high in response

Young et al. (1996) measured HRTFs for pinna orienta- to stimuli with frequency components that are distributed intions similar to those used in the present study. They found both excitatory and inhibitory frequency domains.that the spatial distribution of spectral notches shifted in As a consequence of integration of excitatory and inhibi-location with respect to the head but maintained a constant tory inputs from different frequency domains, MD cells ex-relationship to the pinna. Thus although movement of the hibit greater response modulation to HRTF spectral cuespinna does produce some changes in HRTFs, movement than would be expected from auditory eighth nerve fibers.may not introduce sufficient distortion of spectral cues as to It is not uncommon for MD cells to be completely unrespon-greatly change MD cell directional sensitivity. sive to certain directions regardless of SPL. Eighth nerve

The directional preference of single units has been shown fibers reveal the presence of spectral notches as increasedto be dependent on pinna position in a number of different thresholds and decrements in responsiveness (Poon andspecies with mobile pinnae. A unit’s directional preference Brugge 1993a,b) but would be expected to respond to anymay be shaped by monaural mechanisms for sounds pre- sound direction (or any HRTF spectrum) given a sufficientlysented at near threshold levels (Middlebrooks and Pettigrew high SPL. Thus for responses averaged over a broad range of1981; Sun and Jen 1987). In such cases, a unit’s receptive SPLs, MD cells show greater direction-dependent responsefield follows the pinna’s acoustic axis (direction of maxi- modulation (i.e., greater directional sensitivity) than eighthmum amplification). For stimuli presented at higher levels nerve fibers.that might reasonably be assumed to produce suprathreshold Some MD cells exhibit focal directional selectivity tostimulation at both ears, the influence of pinna position on monaural stimulation, a pattern of response to HRTF spectraneuronal directional preference has been interpreted as re- that does not appear to be present in peripheral auditorysulting from changes in monaural acoustic axis and binaural neurons. Spectral notches may be represented as responsedisparity cues (Aitkin et al. 1984; Middlebrooks and Knud- minima in receptive fields of eighth nerve fibers (Poon andsen 1987). As is the case in previous studies, the present Brugge 1993 a,b; Rice et al. 1995). Thus in response to afindings show that directional preference of MD cells closely broad range of SPLs, eighth nerve fibers would be expectedfollows the position of the pinna on the side of the head to respond well at all directions with a local minima infrom which it receives excitatory input. Nevertheless, the discharge rate reflecting the presence of a spectral notch. Inmechanism responsible for this is different from those that contrast, a spectral notch in theory could produce either ahave been described previously because it is monaural but response null or peak in a MD cell. A MD cell with annot strictly dependent on changes in directional amplifica- appropriate configuration of excitatory and inhibitory do-tion. Changes in directional amplification might correlate mains might respond to all directions except those at whichclosely with changes in location of a response peak, as has a spectral notch center frequency corresponds with the CFbeen demonstrated in the cat’s superior colliculus (Mid- of the excitatory domain because only at this location doesdlebrooks and Knudsen 1987), but this does not account for spectral inhibition reach a sufficient level to cancel excitatorythe corresponding movement of response nulls on both sides drive. The monaural azimuth function of the cell illustratedof the azimuth function peak. Spectral patterns present in in Fig. 14B shows some of these characteristics in that it isthe HRTF produce response nulls by a mechanism of spectral responsive to a broad range of azimuths. This pattern isinhibition. somewhat similar to that expected from eighth nerve fibers,

although the decrement in discharge rate at the null locationis more pronounced. On the other hand, a MD cell mightCentral versus peripheral mechanismsreceive more powerful and/or lower threshold inhibitoryinput than excitatory input, such that at most directions spec-Mechanisms by which pinna-derived spectral cues are rep-

resented in neural discharge rates appear to differ in central tral inhibition predominates and the cell does not respond.Only when a spectral notch is centered on an inhibitoryand in peripheral auditory neurons. Poon and Brugge (1993)

studied the responses of auditory eighth nerve fibers to syn- frequency domain does the excitatory drive predominate andthe cell respond. In this case, azimuth and elevation functionsthetic spectral notches in filtered noise bursts that mimic

naturally occurring spectral notches. They concluded that would appear as focal response peaks surrounded by re-sponse nulls and a number of examples are illustrated herethe responses of eighth nerve fibers reflect the amount of

sound pressure within their excitatory response areas. This (e.g., Figs. 2, 4, 8, and 9).conclusion is also consistent with the findings of Rice et al.(1995), who studied the responses of auditory eighth nerve Frequency response areas of MD cellsfibers to noise bursts that were filtered to mimic the HRTFin the cat. Although two-tone suppression can be demon- All MD cells that were tested exhibited frequency re-

sponse areas that consisted of excitatory and inhibitory do-strated in the responses of eighth nerve fibers (Sachs andKiang 1968), Poon and Brugge (1993) concluded that en- mains. FRAs of MD and BD cells show some differences

that might be related to a specialization for responding toergy outside the excitatory response area does not appear toplay a major role in a fiber’s response to the notch stimuli. HRTF spectra. First, excitatory domains of MD cells were,

on average, narrower than those of BD cells. This could beIn contrast to eighth nerve fibers, the azimuth and elevationsensitivity of MD cells to monaural BBN stimulation appears a result of flanking inhibitory domains that restrict the width

of excitatory domains. Second, response areas consisting ofcritically dependent on integration of input from excitatoryand inhibitory frequency domains. Directional sensitivity is multiple excitatory domains were more common in the MD


by 10.220.33.4 on May 9, 2017


ownloaded from



than the BD sample, suggesting that this may reflect a mech- siveness, threshold, and latency of responses in the differentdomains. Postexcitatory decrement in responsiveness mayanism for recognition of HRTF spectra. The pattern of multi-

ple excitatory and inhibitory domains is reminiscent of fre- play a role in the directional sensitivity of this cell, butthe mechanism responsible for the decrement is unknown.quency response areas generated by a three-layer neural net-

work model that was designed to transform spectral Significant differences in the minimum latency in differentexcitatory domains of multipeaked cortical neurons haverepresentation of pinna filtered stimuli into a space-mapped

representation of sound direction at the output (Neti and been reported by Sutter and Schreiner (1991). In most cases,the high-frequency domain exhibited a longer latency thanYoung 1992). Under conditions of monaural input, many of

the neurons in both the hidden and output layers had FRAs the low-frequency domain, but in other cases, the oppositewas true as was the case of one cell that is included in thethat consisted of interdigitated excitatory and inhibitory fre-

quency domains. In some cases, the excitatory-inhibitory– present report ( i.e., Fig. 11).excitatory domain architecture could be interpreted as aspectral notch detector with the center frequency of the notch Spectral cues and sound localizationcorresponding to the center frequency of the inhibitory do-

Cats derive different types of directional information frommain, although in other cases, such a simple interpretationspectral cues in different frequency ranges. HRTFs show awas not possible.single spectral notch that systematically changes in fre-Single units with multiple excitatory and interleaving in-quency between 5 and 18 kHz as a function of azimuth andhibitory domains have been found in cortical field AI of theelevation, and a more complex pattern of peaks and notchescat. Sutter and Schreiner (1991) discovered that neuronsat higher frequencies (18–50 kHz) (Rice et al. 1992). Accu-with multiple excitatory frequency domains were limitedracy of vertical and horizontal orientation in the frontal fieldalmost entirely to the dorsal third of the field. Whether oris nearly the same to BBN and 5–18 kHz band-pass noise,not neurons with multipeaked response areas in dorsal AIand orientation accuracy to high-pass noise (ú18 kHz) isusually have MD properties is unknown, although one exam-poor, suggesting that spectral information in the 18–50 kHzple of an AI MD cell with two excitatory domains has beenrange is irrelevant for this task (Huang and May 1996). Indocumented (Samson et al. 1993). The finding thatcontrast, removal of high-frequency components has littlemultipeaked response areas are present in parts of the MGBeffect on minimal audible angles in the horizontal plane butthat project to AI (ventral nucleus and the lateral part of thecauses an increase in minimum audible angles at positiveposterior group of thalamic nuclei) (Morel and Imig 1987)and negative elevations in the median plane (May and Huangleaves open the possibility that such cortical response areas1996). Many MD cells have excitatory and inhibitory fre-may reflect subcortical processes.quency domains that span both the 5- to 18-kHz and 18- to50-kHz frequency ranges. Thus their directional sensitivity

Relationship of spectral inhibition and facilitation to has a potential role in both of these behavioral tasks.directional sensitivity

We thank C. Bailey for careful preparation of histological materials,Our results suggest that direction-dependent strength dif-data analysis, and preparation of illustrations and H. Cheng for computerferences in spectral inhibition produce MD sensitivity. Re-programming.

sponses to E/I stimuli clearly reveal a net spectral inhibition; Support for this work was provided by the National Institute on Deafnessthere is little evidence for a net spectral facilitation in these and Other Communicative Disorders (DC-00173); BRSG S07 RR 05373

and 554855 awarded by the Biomedical Research Support Grant Program,responses. On the other hand, these data do not rule out theDivision of Research Resources, National Institutes of Health; Fonds de lapossibility that spectral facilitation may play some role inRecherche en Sante du Quebec (P. Poirier) ; and Fonds pour la Formationthe responses of these cells. It is possible that responses to E/ de Chercheurs et l’Aide a la Recherche (P. Poirier) .

I stimuli represent a sum of spectral facilitation and spectral Address reprint requests to T. J. Imig.inhibition to all frequency components of the stimulus, but

Received 7 March 1997; accepted in final form 7 July 1997.that spectral inhibition is the dominant interaction so thatthe result is a net spectral inhibition. Our findings certainly

REFERENCESleave open the possibility that spectral facilitation may bepresent in the responses to E-only stimuli. Spectral facilita- AITKIN, L. M., GATES, G. R., AND PHILLIPS, S. C. Responses of neurons in

the inferior colliculus to variations in sound-source azimuth. J. Neuro-tion has been demonstrated in the responses of cortical neu-physiol. 52: 1–17, 1984.rons in the cat (Nelken et al. 1994; Sutter and Schreiner

AITKIN, L. M. AND MARTIN, R. L. The representation of stimulus azimuth1991). Nevertheless, even if spectral facilitation is present by high best-frequency azimuth selective neurons in the central nucleusin the responses of some MD cells, most are insensitive to the of the inferior colliculus of the cat. J. Neurophysiol. 57: 1185–1200,

1987.direction of E-only stimuli, suggesting that by itself spectralBARONE, P., CLAREY, J. C., IRONS, W. A., AND IMIG, T. J. Single unit re-facilitation does not play a major role in their directional

sponses in the primary auditory cortex and medial geniculate body ofsensitivity.the cat: evidence for cortical synthesis of azimuth receptive fields with

Whereas direction-dependent variation in spectral inhibi- nonmonotonic level tuning. J. Neurophysiol. 75: 1206–1220, 1996.tion seems to be an important mechanism of MD sensitivity, BATTEAU, D. W. The role of the pinna in human localization. Proc. R. Soc.

Lond. B Biol. Sci. 168: 158–180, 1967.another mechanism was encountered that was capable ofBLAUERT, J. Spatial Hearing . Cambridge, MA: MIT Press, 1983.producing directional sensitivity to monaural stimulation.BUTLER, R. A. The influence of external and middle ear on auditory discrim-Interactions between frequency components in excitatory ination. In: Handbook of Sensory Physiology, Auditory System, edited by

frequency domains may contribute to MD sensitivity in some W. D. Keidel and W. D. Neff. Berlin: Springer-Verlag, 1975, vol. 2, p.247–260.cells (e.g., Fig. 11) by virtue of differences in respon-


by 10.220.33.4 on May 9, 2017


ownloaded from



BUTLER, R. A. The bandwidth effect on monaural and binaural localization. MUSICANT, A. D. AND BUTLER, R. A. The psychophysical basis of monaurallocalization. Hear. Res. 14: 185–190, 1984.Hear. Res. 21: 67–73, 1986.

BUTLER, R. A., HUMANSKI, R. A., AND MUSICANT, A. D. Binaural and mon- MUSICANT, A. D., CHAN, J.C.K., AND HIND, J. E. Direction-dependent spec-tral properties of cat external ear: new data and cross species comparisons.aural localization of sound in two-dimensional space. Perception 19:

241–256, 1990. J. Acoust. Soc. Am. 87: 757–781, 1990.NEFF, W. D. AND CASSEDAY, J. H. Effect of unilateral ablation of auditoryCLAREY, J. C., BARONE, P., IRONS, W. A., SAMSON, F. K., AND IMIG, T. J.

Comparison of noise and tone azimuth tuning of neurons in cat primary cortex on monaural cat’s ability to localize sound. J. Neurophysiol. 40:44–52, 1977.auditory cortex and medial geniculate body. J. Neurophysiol. 74: 961–

980, 1995. NELKEN, I., PRUT, Y, VAADIA, E., AND ABELES, M. Population responses tomultifrequency sounds in the cat auditory cortex: one- and two-parameterFISHER, H. G. AND FREEMAN, S. J. The role of the pinna in auditory localiza-

tion. J. Aud. Res. 8: 15–26, 1968. families of sounds. Hear. Res. 72: 206–222, 1994.NETI, C. AND YOUNG, E. D. Neural network models of sound localizationGARDNER, M. B. AND GARDNER, R. S. Problem of localization in the medial

plane: effect of pinnae cavity occlusion. J. Acoust. Soc. Am. 53: 400– based on directional filtering by the pinna. J. Acoust. Soc. Am. 92: 3140–3156, 1992.408, 1973.

HAUSLER, R., COLBURN, S., AND MARR, E. Sound localization in subjects OLDFIELD, S. R. AND PARKER, S.P.A. Acuity of sound localization: a topog-raphy of auditory space. II. Pinnae cues absent. Perception 13: 601–617,with impaired hearing. Acta Otolaryngol Suppl. (Stockh.) 400: 1–62,

1983. 1984.OLDFIELD, S. R. AND PARKER, S.P.A. Acuity of sound localization: a topog-HEBRANK, J. AND WRIGHT, D. Are two ears necessary for localization of

sound sources on the median plane? J. Acoust. Soc. Am. 56: 935–938, raphy of auditory space. III. Monaural hearing conditions. Perception 15:67–81, 1986.1974a.

HEBRANK, J. AND WRIGHT, D. Spectral cues used in the localization of PALMER, A. R. AND KING, A. J. A monaural space map in the guinea pigsuperior colliculus. Hear. Res. 17: 267–280, 1985.sound sources on the median plane. J. Acoust. Soc. Am. 56: 1829–1834,

1974b. PHILLIPS, D. P., ORMAN, S. S., MUSICANT, A. D., AND WILSON, G. F. Neu-rons in the cat’s primary auditory cortex distinguished by their responsesHUANG, A. Y. AND MAY, B. J. Sound orientation behavior in cats. II. Mid-

frequency spectral cues for sound localization. J. Acoust. Soc. Am. 100: to tones and wide-spectrum noise. Hear. Res. 18: 73–86, 1985.1070–1080, 1996. POIRIER, P., SAMSON, F. K., AND IMIG, T. J. Elevation sensitivity of single

neurons in the medial geniculate body (MGB) of the cat. Soc. Neurosci.IMIG, T. J., IRONS, W. A., AND SAMSON, F. R. Single-unit selectivity toazimuthal direction and sound pressure level of noise bursts in cat high- Abstr. 21: 1176, 1995.frequency primary auditory cortex. J. Neurophysiol. 63: 1448–1466, POIRIER, P., SAMSON, F. K., AND IMIG, T. J. Directional mechanisms and1990. spatial preferences of single units in the cat’s inferior colliculus (IC).

Soc. Neurosci. Abstr. 22: 889, 1996.IMIG, T. J. AND MOREL, A. Tonotopic organization in ventral nucleus ofmedial geniculate body in the cat. J. Neurophysiol. 53: 309–340, 1985a. POON, P.W.F. AND BRUGGE, J. F. Sensitivity of auditory nerve fibers to

spectral notches. J. Neurophysiol. 70: 655–666, 1993a.IMIG, T. J. AND MOREL, A. Tonotopic organization in lateral part of posteriorgroup of thalamic nuclei in the cat. J. Neurophysiol. 53: 836–851, 1985b. POON, P. W. AND BRUGGE, J. F. Virtual-space receptive fields of single

auditory nerve fibers. J. Neurophysiol. 70: 667–676, 1993b.IMIG, T. J., POIRIER, P., AND SAMSON, F. K. Single unit directional sensitivityto monaural stimulation requires spectral components in both excitatory RAJAN, R., AITKIN, L. M., IRVINE, D.R.F., AND MCKAY, J. Azimuth sensitiv-and inhibitory frequency domains in the high-frequency representation ity of neurons in primary auditory cortex of cats. I. Types of sensitivityof the cat’s medial geniculate body. Soc. Neurosci. Abstr . 20: 322, 1994. and the effects of variations in stimulus parameters. J. Neurophysiol. 64:

872–887, 1990.IMIG, T. J., POIRIER, P., AND SAMSON, F. K. Elevation sensitivity in the cat’sthalamo-cortical auditory system: role of monaural spectral and binaural RHODE, W. S. AND GREENBERG, S. Physiology of the cochlear nuclei. In:disparity cues (Abstract) . J. Acoust. Soc. Am. 97: 3398, 1995. The Mammalian Auditory Pathway: Neurophysiology, edited by A. N.

Popper and R. R. Fay. New York: Springer-Verlag, 1992, chapt. 3, p.IRONS, W. A. Directional Selectivity of Single Units in the Auditory Thala-mus of Cats (PhD dissertation). Kansas City, KS: Univ. of Kansas, 1989. 94–152.

RHODE, W. S. AND GREENBERG, S. Lateral suppression and inhibition in theJENKINS, W. M. AND MASTERTON, R. B. Sound localization: effects of uni-lateral lesions in central auditory system. J. Neurophysiol. 47: 987–1016, cochlear nucleus of the cat. J. Neurophysiol. 71: 493–514, 1994.1982. RICE, J. J., MAY, B. J., SPIROU, G. A., AND YOUNG, E. D. Pinna-based spec-

tral cues for sound localization in cat. Hear. Res. 58: 132–152, 1992.KING, A. J. AND HUTCHINGS, M. E. Spatial response properties of acousti-cally responsive neurons in the superior colliculus of the ferret: a map RICE, J. J., YOUNG, E. D., AND SPIROU, G. A. Auditory-nerve encoding ofof auditory space. J. Neurophysiol. 57: 596–625, 1987. pinna-based spectral cues: rate representation of high-frequency stimuli.

J. Acoust. Soc. Am. 97: 1764–1776, 1995.KING, A. J., MOORE, D. R., AND HUTCHINGS, M. E. Topographic representa-tion of auditory space in the superior colliculus of adult ferrets after SACHS, M. B. AND KIANG, N.Y.S. Two-tone inhibition in auditory-nervemonaural deafening in infancy. J. Neurophysiol. 71: 182–194, 1994. fibers. J. Acoust. Soc. Am. 43: 1120–1128, 1968.

KNUDSEN, E. I. Synthesis of a neural map of auditory space in the owl. In: SAMSON, F. R., BARONE, P., CLAREY, J. C., AND IMIG, T. J. Elevation sensi-Dynamic Aspects of Neocortical Function, edited by G. M. Edelman, tivity of azimuth sensitive neurons in cat primary auditory cortex (AI) .W. M. Cowan, and W. E. Gall. New York: Wiley, 1984, p. 709–723. Soc. Neurosci. Abstr. 17: 303, 1991.

MAY, B. J. AND HUANG, A. Y. Sound orientation behavior in cats. I. Local- SAMSON, F. K., BARONE, P., CLAREY, J. C., AND IMIG, T. J. Effects of earization of broadband noise. J. Acoust. Soc. Am. 100: 1059–1069, 1996. plugging on single-unit azimuth sensitivity in cat primary auditory cortex.

II. Azimuth tuning dependent upon binaural stimulation. J. Neurophysiol.MIDDLEBROOKS, J. C. AND GREEN, D. M. Sound localization by humanlisteners. Annu. Rev. Psychol. 42: 135–159, 1991. 71: 2194–2216, 1994.

SAMSON, F. K., CLAREY, J. C., BARONE, P., AND IMIG, T. J. Effects of earMIDDLEBROOKS, J. C. AND KNUDSEN, E. I. A neural map of auditory spacein the cat’s superior colliculus. J. Neurosci. 4: 2621–2634, 1984. plugging on single-unit azimuth sensitivity in cat primary auditory cortex.

I. Evidence for monaural directional cues. J. Neurophysiol. 70: 492–MIDDLEBROOKS, J. C. AND PETTIGREW, J. D. Functional classes of neurons511, 1993.in primary auditory cortex of the cat distinguished by sensitivity to sound

location. J. Neurosci. 1: 107–120, 1981. SAMSON, F. K., POIRIER, P., IRONS, W. A., AND IMIG, T. J. Effects of earplugging on responses of azimuth-sensitive neurons in medial geniculateMOORE, D. R., HUTCHINGS, M. E., ADDISON, P. D., SEMPLE, M. A., AND

body and primary auditory cortex of barbiturate-anesthetized cats. Soc.AITKIN, L. M. Properties of spatial receptive fields in the central nucleusNeurosci. Abstr. 22: 889, 1996.of the cat inferior colliculus. II. stimulus intensity effects. Hear. Res. 13:

175–188, 1984. SLATTERY, W. H. AND MIDDLEBROOKS, J. C. Monaural sound localization:acute versus chronic unilateral impairment. Hear. Res. 75: 38–46, 1994.MOORE, D. R., KING, A. J., MCALPINE, D., HUTCHINGS, M. E., AND MARTIN,

R. Functional consequences of neonatal unilateral cochlear removal. SEARLE, C. L., BRAIDA, L. D., CUDDY, D. R., AND DAVIS, M. F. Binauralpinna disparity: another auditory localization cue. J. Acoust. Soc. Am.Prog. Brain Res. 97: 127–133, 1993.57: 448–455, 1975.MOREL, A. AND IMIG, T. J. Thalamic projections to fields A, AI, P and VP

in the cat auditory cortex. J. Comp. Neurol. 265: 119–144, 1987. SHAMMA, S. A., FLESHMAN, J. W., WISER, P. R., AND VERSNEL, H. Organi-


by 10.220.33.4 on May 9, 2017


ownloaded from



zation of response areas in ferret primary auditory cortex. J. Neurophys- WIGHTMAN, F. L. AND KISTLER, D. J. Headphone stimulation of free-field listening. I. Stimulus synthesis. J. Acoust. Soc. Am. 85: 858 –iol. 69: 367–383. 1993.

SHAMMA, S. A. AND SYMMES, D. Patterns of inhibition in auditory cortical 867, 1989.WIGHTMAN, F. L. AND KISTLER, D. J. Monaural sound localization revisited.cells in awake squirrel monkeys. Hear. Res. 19: 1–13, 1985.

SHAW, E.A.G. Transformation of sound pressure level from the free field J. Acoust. Soc. Am. 101: 1050–1063, 1997.YIN, T.C.T. AND CHAN, J.C.K. Neural mechanisms underlying in-to the eardrum in the horizontal plane. J. Acoust. Soc. Am. 56: 1848–

1861, 1974. teraural time sensitivity to tones and noise. In: Auditory Function:Neurobiological Bases of Hearing, edited by G. M. Edelman, W. E.SIEGEL, S. Nonparametric Statistics for the Behavioral Sciences. New York:

McGraw-Hill, 1956, p. 96–101. Gall, and W. M. Cowan. New York: Wiley, 1988, chapt. 13, p. 385 –430.SUN, X. AND JEN, P.H.S. Pinna position affects the auditory space represen-

tation in the inferior colliculus of the FM bat, Eptesicus fuscus. Hear. YOUNG, E. D., SHOFNER, W. P., WHITE, J. A., ROBERT, J.-M., AND

VOIGT, H. F. Response properties of cochlear nucleus neurons inRes. 27: 207–219, 1987.SUTHERLAND, D. P. Elevation Discrimination in Cats: Role of Central Audi- relationship to physiological mechanisms. In: Auditory Function:

Neurobioligical Bases of Hearing, edited by G. M. Edelman, W. E.tory Structures (PhD dissertation). Talahassee, FL: Florida State Univer-sity, 1994. Gall, and W. M. Cowan. New York: Wiley, 1988, chapt. 9, p. 277–

312.SUTHERLAND, D. P. A role of the dorsal cochlear nucleus in the localizationof elevated sound sources . Assoc. Res. Otolaryngol. Midwinter Meet. 14: YOUNG, E. D., RICE, J. J., AND TONG, S. C. Effects of pinna position on

head-related transfer functions in the cat. J. Acoust. Soc. Am. 99: 3064–33, 1991.SUTTER, M. L. AND SCHREINER, C. E. Physiology and topography of neurons 3076, 1996.

ZAKARAUSKAS, P. AND CYNADER, M. S. A computational theory of spectralwith multipeaked tuning curves in cat primary auditory cortex. J. Neuro-physiol. 65: 1207–1226, 1991. cue localization. J. Acoust. Soc. Am. 94: 1323–1331, 1993.


by 10.220.33.4 on May 9, 2017


ownloaded from


Monaural Spectral Contrast Mechanism for Neural ... · Monaural Spectral Contrast Mechanism for Neural Sensitivity to Sound Direction in the Medial Geniculate Body of the Cat THOMAS

Documents