The Journal of Neurophysiology Vol. 79 No. 2 February 1998, pp. 1053-1069
Copyright ©1998 by the American Physiological Society

Altered Spectral Localization Cues Disrupt the Development of the Auditory Space Map in the Superior Colliculus of the Ferret

Jan W. H. Schnupp, Andrew J. King, and Simon Carlile

University Laboratory of Physiology, Oxford OX1 3PT, United Kingdom

	ABSTRACT

Abstract Introduction Methods Results Discussion References

Schnupp, Jan W. H., Andrew J. King, and Simon Carlile. Altered spectral localization cues disrupt the development of the auditory space map in the superior colliculus of the ferret. J. Neurophysiol. 79: 1053-1069, 1998. Spectral localization cues provided by the outer ear are utilized in the construction of the auditory space map in the superior colliculus (SC). The role of the outer ear in the development of this map was examined by recording from the SC of anesthetized, adult ferrets in which the pinna and concha had been removed in infancy. The acoustical consequences of this procedure were assessed by recording outer ear impulse responses via a probe-tube microphone implanted in the wall of the ear canal. Both monaural and binaural spectral cues normally show a number of asymmetric features within the horizontal plane, which allow azimuthal locations on either side of the interaural axis to be discriminated. These features were eliminated or altered by chronic pinnectomy. The responses of auditory units in the SC to noise bursts presented in the free field were examined at sound levels of ~10 and 25 dB above unit threshold. After bilateral pinnectomy, the representation of auditory space was severely degraded at both sound levels. In contrast to normal ferrets, many units had bilobed azimuthal response profiles, indicating that they were unable to resolve sound locations on either side of the interaural axis. There was also much less order in the distribution of best azimuths or elevations of those units that were tuned to a single direction. Some units were tuned to locations that extended much further into the hemifield ipsilateral to the recording side than the normal range of best azimuths. Unilateral removal of the outer ear, which disrupts the monaural spectral cues for one side only, had a much smaller effect on the development of the auditory representation. At supra- and near-threshold sound levels, the representation of sound azimuth in the SC on both sides of the brain was less scattered than that found after bilateral pinna removal. Nevertheless, units with bilobed responses, broader tuning, and inappropriate best azimuths were observed in both the left and right SC of ferrets in which the left pinna and concha had been removed in infancy. These data illustrate that the localization cues provided by the outer ear play a critical role in the development of the auditory space map in the SC. In contrast to other manipulations of either auditory or visual inputs, the map does not appear to adapt to the changes in spectral cues brought about by pinna removal, suggesting that residual binaural cues are, by themselves, insufficient for its normal maturation.

	INTRODUCTION

Abstract Introduction Methods Results Discussion References

The deeper layers of the superior colliculus (SC) contain a map of auditory space: the large majority of auditory neurons show a clear preference for particular sound source directions, which vary systematically with anatomic position within the nucleus (King and Hutchings 1987; King and Palmer 1983; Knudsen 1982; Middlebrooks and Knudsen 1984). Early studies of the physiological basis for the auditory spatial selectivity of SC neurons focused primarily on binaural mechanisms. In cats, these neurons were found to be sensitive to interaural level difference (ILD) cues as a result of excitatory/inhibitory (EO/I) type interactions for units tuned to lateral positions, or facilitatory (OO/F and EO/F) interactions for units tuned to positions on or closer to the anterior midline (Middlebrooks 1987; Wise and Irvine 1984, 1985). Tuning to interaural time difference (ITD) cues underlies the representation of sound azimuth in the SC or optic tectum of the barn owl (Olsen et al. 1989) but appears to contribute less in the mammalian SC. ITD-sensitive neurons have been observed in the SC of the cat (Hirsch et al. 1985), but these neurons are typically sensitive to time differences outside the normal physiological range. Because the response latencies of neurons in the SC as well as other parts of the auditory system vary with sound level, it has been proposed that the ITD sensitivity of mammalian SC units provides a mechanism for ILD detection (Yin et al. 1985), rather than acting as a detector for ITD localization cues per se.

Although ILDs are central to the functioning of the SC auditory space map, the early studies cited above used experimental designs (dichotic broadband noise or tonal stimulation) that ignore the fact that ILDs generated naturally through filtering by the outer ear and head-shadowing are highly frequency dependent. The additional contribution of spectral localization cues to the construction of the mammalian auditory space map was demonstrated in free-field studies by Palmer and King (1985) and by Carlile and King (1994). Auditory neurons in the SC are typically tuned unambiguously to single sound locations. However, when spectral cues are disrupted in adult ferrets by acute removal of the structures of the outer ear, many units in the SC become tuned to two sound azimuthal locations, one on each side of the interaural axis (Carlile and King 1994). Apparently no longer able to resolve front-back confusions, they appear nevertheless able to exploit the remaining binaural cues to maintain a systematically organized but spatially ambiguous map of sound azimuth. Conversely, when binaural cues are eliminated in ferrets or guinea pigs by unilateral deafening, many units in the SC either lose their spatial selectivity or show inappropriate spatial tuning when stimulated at high sound levels, but form a map of auditory space when stimulated at sound levels close to unit threshold (King et al. 1994; Palmer and King 1985). This last result has led to the notion that spatial selectivity at near-threshold sound levels for neurons in all but the most rostral regions of the SC is based on monaural spectral cues, whereas binaural mechanisms also operate to maintain tuning at suprathreshold sound levels.

Together these studies show that the spatial selectivity of SC neurons is due not only to a sensitivity to interaural differences in overall sound pressure level, but, in addition, incorporates mechanisms that exploit pinna-derived spectral cues, both monaurally and binaurally. However, the neural circuits implementing these mechanisms, and many important details of their operation remain to be identified. Accordingly, the correct merging of congruent binaural and monaural localization cues corresponding to a given region of space can be thought of as a critical step in the development of the map. Previous studies have shown, by rearing ferrets (King et al. 1988) and barn owls (Knudsen 1985) with one ear occluded, that this system can compensate for the introduction of aberrant ILDs during early life. Recordings made in adult animals with the plugs still in place revealed the presence of near normal maps of auditory space. In barn owls, removal of the outer ear structures, the facial ruff and preaural flap, also alters the binaural cue values corresponding to particular sound locations and thereby disrupts the correspondence of visual and auditory representations in the optic tectum. However, units in the tectum of juvenile owls respond to this treatment by an adaptive change in their ILD and particularly their ITD tuning, thereby achieving a partial realignment of the representations of visual space and of auditory spatial cues (Knudsen et al. 1994).

No attempts have been made to investigate the role of spectral localization cues in the development of the neural systems underlying auditory localization in mammals. Given the apparently greater dependence of the representation of auditory space in the mammalian SC on outer ear spectral cues, we might expect that early disruption of those cues would have a more profound effect on the development of the auditory space map than it does in barn owls. As filling the cavities of the outer ear or inserting a tube into the external auditory meatus are not practical possibilities for small mammals reared with a number of siblings and very attentive mothers, we altered the spectral localization cues associated with different regions of space by removing the pinna and concha just before the onset of hearing, which, in the ferret, occurs at 27-28 days after birth (Moore and Hine 1992; Morey and Carlile 1990). We found that this procedure degraded both the spatial tuning parameters and topographic organization of preferred sound directions of auditory units in the SC. Our results provide further evidence for the importance of postnatal experience and reveal limits to the adaptive plasticity of the auditory space map.

Preliminary results have been published in abstract form (Schnupp et al. 1995a).

	METHODS

Abstract Introduction Methods Results Discussion References

Neonatal surgery

We removed the pinna and concha bilaterally in five ferrets, and on the left side only in seven ferrets. The surgery was performed between postnatal days (P) 24 and 28 (mean age P26). During surgery the animals were anesthetized with alphaxalone/alphadolone acetate (Saffan) at a dose of 4 ml/kg ip. The cartilage surrounding the opening of the external auditory meatus was left intact to ensure that it did not collapse. A long acting local anesthetic was applied, and the animals were administered a systemic postoperative analgesic (Temgesic), and a prophylactic broad spectrum antibiotic (Ampicillin; 0.1 ml im). The animals recovered and were allowed to mature to adulthood. At the start of the recording session, they were between 301 and 955 days (mean 501 days) old.

Ferrets have deep, slightly curved ear canals, which renders a thorough otoscopic examination difficult. We therefore performed regular tympanometric examinations to ensure that the surgery did not lead to an occlusion of the meatus or to impaired middle ear function. In one unilaterally operated animal, auditory brain stem responses to free-field pure tone stimulation were measured 24 h before the single-unit recordings in the SC. The brain stem-evoked responses suggested normal sensitivity in both ears.

Preparation for recording

For full details of the surgical preparation, see King and Hutchings (1987) and King et al. (1994). Briefly, the animals were anesthetized with an initial intramuscular injection of Saffan at a dose of 2 ml/kg. During surgery, additional doses of Saffan were administered intravenously as required. Body temperature was monitored rectally and held constant at 39°C. A craniotomy was performed to expose the cortex overlying the SC bilaterally, and the skull was attached to a minimal head holder that was clamped to a supporting post behind the animal. A series of measurements across the head was made before the initial incision to ensure that, as far as possible, the natural position of the pinnae or the ear canals had been restored after the cranial surgery. The eyelids were trimmed and the eyes protected with a zero refractive power contact lens to allow the mapping of visual receptive fields in the superficial layers of the SC, which were used as landmarks to guide electrode penetrations through the intact cortex. To prevent eye movements, the animals were paralyzed with 12 mg iv gallamine triethiodide (Flaxedil) and artificially ventilated with room air supplemented with 95% O₂-5% CO₂. During recording, paralysis and anesthesia were maintained by a continuous infusion of 20 mg·kg¹·h¹ Flaxedil and 1 mg·kg¹·h¹ pentobarbital sodium (Nembutal) in Locke's solution. The heart rate, electrocardiogram, electroencephalogram, and end-tidal CO₂ were monitored continuously as a means of assessing that the animals were maintained in an adequate and stable level of anesthesia.

Single-unit recording

Extracellular recordings were made in a sound-proof anechoic chamber. Visual stimuli consisted of 100-ms light flashes delivered from a 1-cm-diam light-emitting diode (LED). Digitally generated broadband noise bursts (30-30,000 Hz, 100-ms duration with a 5-ms rise/fall time) were delivered from a KEF T27 loudspeaker, which was mounted on a vertical hoop with a radius of 0.65 m (Annetts et al. 1987). Movement of the speaker along the hoop allowed the vertical angle of the stimulus to be varied, whereas rotation of the hoop around the animal changed the azimuthal angle. The speaker position was varied either directly by a remote control unit or via the data acquisition software. The center of the animal's head, defined by the intersection of the interaural axis and the midsagittal plane, was located at the center of the imaginary sphere described by the hoop. The locations of both visual and auditory stimuli were defined in spherical coordinates in which elevations were measured along the meridians extending between the poles above and below the animal's head, and azimuths were measured along circles parallel to the horizontal plane. An azimuth of 0° corresponded to the anterior midsagittal plane and 0° elevation to the audiovisual plane. Azimuthal locations contralateral to the recording site and elevations below the audiovisual horizon were denoted by negative angles.

Single-unit recordings were made using conventional tungsten-in-glass microelectrodes. The electrode was advanced stereotaxically into the SC through the overlying occipital cortex using a remote-controlled stepper motor microdrive, which allowed movement of the electrode to be controlled from outside the chamber. The electrode signal was amplified (10,000 times), band-pass filtered (500-5,000 Hz), and monitored on a digital storage oscilloscope and an acoustic monitor. As the electrode was lowered vertically through the cortex, visual stimuli were presented from an LED positioned directly in front of the eye. This flashing stationary visual stimulus evoked a strong, characteristic response when the electrode entered the superficial layers of the SC. The center of the visual multiunit receptive field was then mapped with either a hand-held or hoop-mounted LED. Subsequently, the electrode was advanced into the deeper layers of the SC as auditory stimuli were presented from the position of the center of the visual multiunit receptive field recorded in the overlying superficial layers. In the earlier recording experiments on the bilaterally operated animals, auditory single units were isolated with a conventional level discriminator. In the later experiments on the unilaterally pinnectomized animals, single units were isolated from the digitized signal (25 kHz, 8 bit) using our own spike-sorting software, which quantified the duration and amplitude of different components of the action-potential waveform.

The noise bursts were presented with an interstimulus interval of 1.5 s. We first estimated the threshold for each auditory single unit with the loudspeaker positioned near the receptive-field center of the visual responses obtained in the superficial layers of the same electrode penetration. We then determined the unit's azimuth response profile by measuring the response in 20° azimuth steps from positions 160° contralateral to 160° ipsilateral to the recording site. The vertical coordinates of the speaker were held constant at the elevation of the visual receptive-field center for the overlying superficial layers. Elevation response profiles were also obtained for some units in the bilaterally pinnectomized animals by varying the vertical position of the speaker from 60° below to 80° above the audio-visual horizon while maintaining the azimuth at the value that evoked the maximum response. Response profiles were usually determined at two sound levels relative to unit threshold, typically 5-10 dB (near-threshold condition) and 25-30 dB (suprathreshold condition). These values were chosen because previous studies have suggested that the spatial tuning of SC neurons may be derived from different cues at these sound levels relative to unit threshold (Carlile and King 1994; King and Carlile 1994; King et al. 1994; Palmer and King 1985). The largest ILDs generated by the ferret's head are in the order of 25 dB (Carlile 1990a; Carlile and King 1994). If units receive binaural inputs and the threshold is equivalent for stimulation of either ear, then the suprathreshold stimuli will provide binaural cues at all frequencies and positions. For the near-threshold stimuli, input to the ear contralateral to the sound source is likely to be subthreshold for relatively high frequencies and lateral sound source locations, so the system should become at least partly dependent on monaural cues.

The discharge of the unit was measured in response to 20 stimulus presentations at each speaker position. After the first spatial profile was obtained, we usually redetermined the unit threshold at the speaker position that gave the strongest response. Most units were onset units with latencies of typically just <10 ms, whereas others showed multipeaked profiles in their poststimulus time histograms (PSTHs). To measure the strength of the response, spikes were counted in a window determined individually for each unit by inspection of the PSTH pooled over all the data for that unit. For the onset units, the duration of this window was typically 20 ms, whereas the window used to assess the response of units with multipeaked temporal response profiles was typically around 110 ms in duration. In rare cases we set much longer windows because the increase in spike discharge lasted for ~350 ms. A "spontaneous period," arbitrarily set for all units at 500-1,000 ms after stimulus onset, was used to derive a running estimate of the unit's spontaneous activity. The number of sound-evoked spikes was calculated as the mean number of spikes per presentation in the response period minus the number of spikes during an equivalent length of time in the spontaneous period. The number of evoked spikes per stimulus presentation was then plotted against speaker azimuth in polar coordinates, and the resulting azimuth profiles were classified objectively by the data acquisition software into one of five categories: "tuned," "bilobed," "broad," "complex," or "omnidirectional." Tuned profiles contained a single peak, whereas bilobed profiles had two peaks and complex cells more than two. A peak in the response profile was defined as a region of 80% of the maximal response, flanked by regions with 40% of the maximal response, with the flanking regions not more than 160° apart. Broad cells exhibited a response peak that was wider than 160°; these were typically hemifield responses. In omnidirectional profiles the response remained above 40% of the maximum at all speaker positions tested.

Within our spherical coordinate system for defining speaker location, elevation profiles describe meridians, between +90° and 90°, rather than circles. In a previous study in normal ferrets (King and Hutchings 1987), we found that the responses of SC units were always reduced when the elevation profiles were extended beyond +90°, above the animal's head, and into the opposite hemifield (King and Hutchings 1987). Within the range of elevations tested in this study, a peak in the response profile was identified if the response reached at least 80% of maximum, bordered either on one or on both sides by a position where the response fell to 40% of maximum. Elevation response profiles were characterized as broad if the response remained above 40% of maximum over the full 140° tested. We were therefore able to classify the elevation profiles as tuned, broad, bilobed, or complex. These classification criteria have been used in several studies in this laboratory and therefore allow a direct comparison of the results reported here with those from previous studies.

Histological reconstructions of recording sites

Recording sites were marked with small electrolytic lesions (5 µA for 5 s). At the end of the recording session, typically after ~40 h of recording, the animal was overdosed with pentobarbital sodium (Euthatal, iv) and perfused through the heart with phosphate-buffered saline followed by 10% formal saline. The brain stem was removed, cryoprotected with 30% sucrose, cut into50-µm coronal sections, and Nissl stained to allow reconstruction of the recording sites. The histological coordinates were expressed as a fraction of SC length and width and then plotted on a standardized SC template. The best azimuths and elevations were plotted against the distance of each unit from the rostrolateral and rostromedial borders, respectively, on this standardized template in directions normal to the isoazimuth and isoelevation contours previously described in normal, adult ferrets (King and Hutchings 1987).

Outer ear spectral transfer functions

In three of the bilaterally pinnectomized animals, outer ear spectral transfer functions (STFs) were recorded to assess the acoustical consequences of chronic pinna removal. The methods for measuring ferret STFs have been described in detail elsewhere (Carlile 1990b; Carlile and King 1994). Briefly, probe microphone assemblies comprising a Brüel & Kjær ¹/₂in. microphone (B & K 4143) with a conical adapter (B & K UA0040) and a 12-mm polythene extension tube (1 mm ID, 1.5 mm OD) were damped with steel wool and placed alongside the animal's body. The probe tube extension was then surgically inserted into the ear canal from behind the pinna and glued in place with cyanoacrylate, so that it protruded <1 mm into the ear canal. Rectangular pulse click stimuli (16.5 µs, Wavetek 182A) were delivered through the hoop-mounted speaker assembly described above. The signal recorded by the probe microphone was band-pass filtered (10 Hz to 30 kHz) and digitized at 83.3 kHz with a Cambridge Electronic Design 1401 laboratory interface. Measurements were made at 10° azimuth intervals, with the exception of a 30° wide region right behind the animal, which our robotic hoop could not reach. Azimuths ipsilateral to the ear in which the microphone was implanted were denoted by negative numbers. The stimulus-recording system was calibrated at the start of each experiment by measuring the transfer function of the probe microphone assembly alone, which was placed in the anechoic chamber at the location normally occupied by the ipsilateral ear of the animal. The plastic extension of the probe microphone was embedded in plasticine to reduce cross talk across its walls. The STF of the outer ear was determined from the fast Fourier transforms of the digitized signal, after subtracting the microphone's STF.

The outer ear STFs comprise location-independent and location-dependent components. The location-independent components reflect those aspects of the STF that are common to all spatial locations and are principally due to the acoustics of the auditory canal (Carlile 1990b; Middlebrooks et al. 1989). The location-independent components will reflect the precise location of the recording microphone within the auditory canal because of the frequency-dependent pattern of longitudinal standing waves over the frequency range of interest. Because the precise recording location will vary from animal to animal, it is important to remove these components so as to facilitate comparison across recordings. To illustrate the location-dependent features, the acoustical data are presented in the form of directionality transfer functions (DTFs). These were calculated by subtracting the mean STF for all sound source positions from the STF for each individual position. This removes the above-mentioned features that depend on microphone placement, because these do not vary with speaker position. DTFs are very similar, both in the way they are calculated and in what they show, to location dependency functions (LDFs), which are often reported in the literature (Carlile 1990b; Carlile and King 1994; Wightman and Kistler 1989). For the calculation of LDFs, the STF from a fixed reference point, rather than the averaged STF, is subtracted from the STFs for individual positions. DTFs or LDFs both measure how the filtering of the outer ear varies about some reference value, and it is these location-dependent variations that are most likely to be of functional significance as cues for sound location.

We also examined how the interaural spectral difference (ISD) cues varied with spatial location by subtracting the DTF for a particular azimuth from the DTF for the corresponding mirror-image position on the other side of the midline. This procedure assumes that the filter properties of each ear are equivalent and symmetric. Removal of the pinna and concha on one side only will greatly alter the DTFs measured in the ipsilateral hemifield for that ear. However, unilateral pinnectomy should have little effect on the DTFs measured for locations in the contralateral hemifield because, for the ferret, the head and body are the principal determinants of the directional properties of the auditory periphery for these locations (Carlile 1990a). Therefore, for convenience, we used the measurements made in normal and bilaterally pinnectomized ferrets to assess the effect of unilateral pinnectomy on the pattern of ISD cues. In so far as the location-independent components also capture some acoustical features that are not ear-canal specific, there may be some differences in these nondirectional components for the intact and pinnectomized ear. This may introduce a frequency-dependent but location-independent bias into the ISDs calculated for unilateral pinnectomy. Nevertheless, as with the DTFs, these calculations illustrate faithfully how interaural spectral differences vary with location, and how this variation differs between the normal, unilaterally and bilaterally pinnectomized ferrets.

	RESULTS

Abstract Introduction Methods Results Discussion References

Acoustical consequences of chronic pinna removal

The consequences of neonatal pinna removal on the spectral localization cues generated by the adult ferret outer ear are very similar to those described by Carlile and King (1994) for acute pinna removal in adult animals. A marked effect of the removal of the pinna and concha is a reduction of front-back asymmetric features in the directional transfer functions. Figure 1A shows a representative DTF recorded in a previous study (Carlile 1990b; Carlile and King 1994) from a normal ferret outer ear. The side contralateral to the recording microphone (positive azimuth coordinates) shows attenuations of the signal due mainly to head shadowing effects, although it is likely that the body will play a role for posterior contralateral locations. On the side ipsilateral to the recording microphone, one can observe a number of features that are asymmetric with respect to the interaural axis. For example, the pinna gain reaches a peak of around +12 dB for frequencies between 19 and 23 kHz in front of the interaural axis. Several spectral notches, where the transmission gain was reduced by ~10 dB over a narrow frequency range, were also present in the normal horizon DTF. These notches, which are indicated by the small green regions in Fig. 1A, were particularly apparent at higher frequencies for ipsilateral locations close to the anterior midline. The clearest notch in the anterior quadrant increased in frequency from ~23 to 27 kHz as the stimulus location was moved from the midline to 45°. A broader drop in transmission gain (~15 dB) was measured at ~16 kHz for positions well behind the interaural axis (azimuth 150°).

View larger version (82K): [in this window] [in a new window]   FIG. 1. Horizon directionality transfer functions (DTFs) shown as contour plots. Warmer colors indicate higher spectrum level gains. An azimuth of 0° denotes the position in front of the animal, 90° the position on the interaural axis on the side where the probe microphone was implanted. A: normal ferret DTF shows a high-frequency gain (19-23 kHz), which is most pronounced in the ipsilateral anterior quadrant. Frequencies around 16 kHz are attenuated for positions behind the animal (150 to 180°). B: DTF after pinna and concha removal in infancy shows fewer asymmetries than the normal DTF. The maximum gain now symmetrically straddles the interaural axis (90°).

Figure 1B shows the DTF for one of the adult ferrets in which the pinna and concha had been removed bilaterally in infancy. Instead of the asymmetric high-frequency gain found in normal animals, we observed a smaller peak (around +9 dB) centered at 25 kHz, which straddled the interaural axis symmetrically. The reduction in midfrequency (~16 kHz) gain observed near 150° azimuth in Fig. 1A was no longer apparent. Instead, a much smaller (3 dB) trough was found at higher frequencies (~22 kHz). The small high-frequency notches present in the anterior hemifield for the normal ferret were also missing after chronic pinnectomy. Thus the narrow green areas close to the anterior midline on the ipsilateral side in Fig. 1A are no longer visible in Fig. 1B. Removal of the pinna and concha therefore resulted in a marked loss of the asymmetric features that characterize the horizon DTFs in the normal ferret, presumably making the monaural spectral cues less useful for localizing sounds and, in particular, for resolving front-back confusions.

We observed some variations in the STFs recorded for the three bilaterally pinnectomized ferrets, which probably reflected a combination of factors. However, as described above, the largest component of this variation is due to slight differences in the location of the recording microphone within the canal, and this is subsequently removed in the calculation of the directional transfer function. The patterns of DTFs produced by pinnectomy in these other animals were accordingly very similar to the one illustrated in Fig. 1B. Furthermore, the DTFs of all three animals were significantly different from normal and very closely resembled those reported by Carlile and King (1994) for five ferrets following acute pinnectomy in adulthood. This suggests that early pinna removal alters the monaural spectral cues in ways that are consistent and reproducible.

Figure 2A shows the pattern of ISDs for a normal ferret. Amplitude spectra obtained at stimulus positions contralateral to the recording microphone (shown along the positive side of the azimuth axis in Fig. 1) reflect the shadowing effects of the head and, to a lesser extent, the torso, and are relatively symmetric with respect to the interaural axis (90°). Subtracting these spectral patterns from those recorded ipsilaterally to calculate ISD functions therefore preserves many of the asymmetric features seen on the ipsilateral side of the DTF. ISD patterns for normal animals contain a prominent peak for frequencies from ~18 to 22 kHz, which extends from the interaural axis well into the anterior quadrant and a sharp trough at ~16 kHz and 150° azimuth. A number of other, smaller asymmetric features are also apparent in the ISDs for the horizontal plane.

View larger version (75K): [in this window] [in a new window]   FIG. 2. A: normal adult ferret horizon interaural spectral difference function (ISD). These values combine pinna directionality effects from the ipsilateral side and head shadowing effects from the contralateral side. The ISDs were calculated by subtracting from the DTF measured in one ear at each azimuth location the DTF for the mirror image location on the other side of the midline. Consequently, every peak or trough on one side is mirrored by the opposite feature on the contralateral side. Many of the asymmetric features seen in the ipsilateral DTF (Fig. 1A) are also apparent in the pattern of ISDs. B: Adult ISD after bilateral removal of the pinna and concha in infancy. The peaks and troughs in this function are not as steep and, more importantly, are far more symmetric with respect to the interaural axis (90°) than those seen in the normal animal.

Figure 2B shows the ISD patterns after bilateral pinnectomy. As with the DTFs, these binaural spectral cues exhibited fewer and less pronounced asymmetries relative to the interaural axis than in normal animals. The negative ISDs at ~16 kHz for posterior locations and other, smaller troughs have been eliminated, and the peak at high frequencies (now ~24 kHz) is smaller, but still appears to be slightly higher in the anterior quadrant. As with the DTF, the ISD pattern in Fig. 2B closely resembles that described after acute bilateral pinnectomy in the adult ferret (Carlile and King 1994).

Figure 3 shows the ISD patterns after unilateral removal of the pinna and concha in infancy. The ISDs relative to the intact right ear (Fig. 3A) feature large positive values for most of the ipsilateral hemifield. A pronounced high-frequency peak (18-23 kHz) for the anterior quadrant and a trough near 15 kHz in the posterior quadrant are present. The pattern of ISDs therefore shows several clear similarities to that observed in the normal animal. This is consistent with the view that the ISDs for sound locations ipsilateral to the intact ear are dominated by the acoustical effects of the near (intact) ear. Figure 3B illustrates the ISDs relative to the left, operated ear. The ISDs in the hemifield ipsilateral to the pinnectomized ear exhibit several pronounced asymmetries, such as a high-frequency negative gain in the posterior quadrant, but differ considerably from the normal pattern.

View larger version (80K): [in this window] [in a new window]   FIG. 3. Horizon ISD functions after unilateral removal of the pinna and concha in infancy. A: ISD relative to the right, intact ear (i.e., DTF in the right ear for a stimulus position to the right, minus DTF in the pinnectomized, left ear for the same stimulus position to the right). This pattern is similar to that seen in normal animals (Fig. 2A), but with a greater gain present at 90° for frequencies from 10 to 18 kHz. B: ISD function relative to the left ear after removal of the left pinna and concha in infancy. The ISD function differs far more from the normal pattern than that seen in A. Nevertheless, there are some asymmetric features (for example, larger ISDs for posterior than for anterior positions at 10-18 kHz).

In summary, bilateral pinna removal severely reduces the asymmetric features normally present in both monaural and binaural spectral cues, whereas unilateral pinna removal disrupts the DTFs primarily on one side and produces ISD patterns that differ considerably for the two hemifields.

SC physiology

We recorded azimuth response profiles for 39 auditory units at near-threshold sound levels and for 35 units at suprathreshold levels in the SC of 5 bilaterally pinnectomized animals. These animals also yielded 26 elevation response profiles at both sound levels. Only azimuth data were collected from the seven animals in which the left pinna and concha had been removed in infancy. We recorded 41 suprathreshold and 34 near-threshold profiles from units in the right SC and 29 suprathreshold and 25 near-threshold profiles from units in the left SC of these ferrets. All the animals in each group gave qualitatively similar data, regardless of the animal's sex or the age at which the recordings were made.

Given that the filtering properties of the pinna and concha produce a gain in amplitude for stimuli at certain frequencies and locations, the removal of these structures might be expected to lead to an increase in unit thresholds. Indeed, acute pinna removal has been shown to lead to an increase in behavioral thresholds in cats (Flynn and Eliot 1965). However, we found that average unit thresholds in the SC were not affected by either unilateral or bilateral pinna removal in infancy. The mean threshold in the bilaterally operated animals was 28.6 ± 15.9 (SD) dB SPL, compared with28.7 ± 14.0 dB SPL in the left SC and 29.5 ± 9.6 dB SPL in the right SC of the unilaterally operated animals and 32.6 ± 14.8 dB SPL in normal ferrets. The small differences in mean values are not statistically significant (t-tests, P > 0.2 in all cases).

Azimuth receptive-field profiles

Figure 4 compares the relative proportions of each of the azimuth profile classes in normal, control animals and in each of the various pinnectomized conditions. Fewer auditory units were tuned to a single azimuthal angle in the SC of ferrets that were reared after uni- or bilateral pinna removal. This decline in the proportion of tuned response profiles was accompanied by a corresponding increase in the proportion of bilobed profiles. In some cases, pinnectomy also led to the appearance of a small number of units with complex or omnidirectional response profiles, which were absent in normal animals.

View larger version (74K): [in this window] [in a new window]   FIG. 4. Proportions of azimuth receptive-field response profiles in normal animals and in ferrets that had undergone pinna removal in infancy. Suprathreshold refers to data collected at 25-30 dB above unit threshold. Near-threshold refers to data collected at ~10 dB above threshold.

Using a statistical test for comparing proportions (Bailey 1981), we found that the reduction in the proportion of tuned units in the bilaterally pinnectomized animals was significant both at supra- and near-threshold sound levels (P < 0.02 in both cases). Unilateral removal of the left pinna also significantly reduced the proportion of tuned units in the right SC (P < 0.005 for near-threshold, P = 0.04 for suprathreshold data). However, in the left SC, contralateral to the intact outer ear, the proportion of tuned units was reduced significantly only at near-threshold (P = 0.014), but not at suprathreshold sound levels (P = 0.23).

Azimuth bandwidths

Figure 5 shows histograms of the 50% bandwidths of the azimuth response profiles measured in normal animals and in each of the experimental conditions. As in previous studies (e.g., King and Hutchings 1987; King et al. 1994), we have defined the 50% bandwidth of an azimuth response profile as the angular extent of all parts of the profile where the stimulus evoked a response of 50% of the maximum. In all groups of animals, the 50% bandwidths were distributed over a wide range and tended to increase with sound level. The azimuth profiles in the pinnectomized conditions were, on average, broader than in ferrets with intact pinnae. This is consistent with our observation that a smaller proportion of units was classed as tuned in these conditions.

View larger version (53K): [in this window] [in a new window]   FIG. 5. Histograms showing the distribution of 50% bandwidths (in degrees) of the azimuth response profiles observed in the different conditions. Bandwidths tend to increase with sound level. At corresponding sound levels relative to unit threshold, bandwidths were on average larger in the pinnectomized conditions than in the unoperated controls.

We used Wilcoxon rank-sum tests to ascertain whether the increases in mean bandwidth caused by pinnectomy are statistically significant. Table 1 shows the significance levels obtained. Near-threshold azimuth profiles were significantly broader than those obtained from normal ferrets, but there were no significant differences between the different pinnectomized conditions. With suprathreshold stimulation, we observed a significant increase in the mean bandwidth with respect to the normal population after bilateral pinna removal and in the left, but not the right, SC after unilateral removal of the left pinna.

Azimuth topography

In Figs. 6 and 7, the azimuthal best positions are plotted against the rostrocaudal distance of each recorded unit from the rostrolateral border of the nucleus. Data from units with tuned () and bilobed receptive-field profiles ( and ) are shown. No attempt was made to assign best positions to the other types of receptive-field profiles. For each bilobed profile the angles of both the "appropriate" () and the "inappropriate" peak () in the response profile are indicated. Appropriate peaks are defined as those that lie closer to the visual best azimuth observed in the superficial layers in the same vertical electrode penetration, and therefore correspond better to the expected azimuth preference for that unit, given the close alignment of visual and auditory maps in the normal SC.

View larger version (27K): [in this window] [in a new window]   FIG. 6. Topographic order in the distribution of auditory unit best azimuths at suprathreshold sound levels. Data are shown for tuned units () and bilobed units ( and ) only. The best azimuth refers to the loudspeaker position where the broadband noise stimulus evoked the maximal response. If adjacent stimulus positions evoked responses of >90% of the maximum, the best azimuths were calculated by interpolation. For each bilobed unit both the "appropriate" () and the "inappropriate" () best azimuths are shown. Appropriate values are those that correspond better to the expected topography based on the normal data. Data in C are adapted from Carlile and King (1994).

View larger version (19K): [in this window] [in a new window]   FIG. 7. Topographic order in the distribution of auditory unit best azimuths at near-threshold sound levels. Layout and symbols identical to those of Fig. 6.

Figure 6 shows the data collected using suprathreshold stimulation. In normal, unoperated animals (Fig. 6A) there is a clear relationship between a unit's best azimuth and its position within the nucleus: anterior sound source locations are represented rostrally and more contralateral locations are represented more caudally. In animals that had undergone bilateral pinnectomy in infancy (Fig. 6B), by contrast, the relationship between unit best azimuth and recording position along the rostrocaudal axis of the SC was much more scattered, with many more units tuned to locations in the ipsilateral hemifield. Some systematic variation in azimuth tuning was still apparent: most of the tuned and bilobed units recorded in the rostral half of the SC had ipsilateral best positions, whereas the majority of the units recorded in caudal SC responded maximally to contralateral sound source locations. However, as described below, the azimuth representation is severely disrupted compared with that found in normal ferrets.

This result differs from that reported by Carlile and King (1994) for adult animals that had both pinnae removed just before recording. Acute pinna removal also resulted in a marked decrease in the proportion of tuned response profiles, the great majority of the others displaying bilobed profiles. However, in these animals, the best azimuths of units that remained tuned and the "appropriate" lobes of bilobed units still displayed a clear topographic order that was statistically indistinguishable from normal. Because of slight differences in the criteria used to classify response profiles, we have reanalyzed the data obtained by Carlile and King (1994) and present them in Fig. 6C to allow a direct comparison between the effects of acute pinnectomy in adulthood and chronic pinnectomy in infancy. The two auditory best positions for most (16/17, 94%) of the bilobed units in the acutely pinnectomized group were distributed on either side of the interaural axis. This is consistent with and predicted by the spatially ambiguous ISDs available to these animals. Nevertheless, by considering the best positions associated with the remaining tuned units and with just one lobe of the bilobed response profiles, it is apparent that a topographic, albeit ambiguous, representation of sound azimuth is present in the SC. This contrasts with the data obtained from animals in which the pinnae were removed bilaterally in infancy, where the topographic order in the best azimuths of both tuned and bilobed units was disrupted, and where less than half (4/9, 45%) of the bilobed units straddled the interaural axis in a way that would be consistent with tuning to ambiguous spatial cues.

Unilateral pinna removal in infancy (Fig. 6, D and E) had a less severe effect on the distribution of azimuthal best positions. The large majority of tuned units had suprathreshold best azimuths within the normal range. Nevertheless, an analysis of topography errors (see Fig. 8 and Tables 2 and 3) indicated that the auditory representation in both the left and the right SC was less precise after chronic unilateral pinnectomy.

View larger version (46K): [in this window] [in a new window]   FIG. 8. Distribution of topography errors (difference between observed and expected azimuth preferences) in the auditory representation for supra- (left) and near-threshold (right) stimulation. The bar under each histogram is centered on the mean error and extends 2 SDs to either side, illustrating the scatter in these distributions.

Figure 7 shows the azimuth topography data collected using near-threshold stimuli. We found essentially the same result for each condition as with the data obtained at suprathreshold sound levels. The azimuth representation in each of the pinnectomized groups was more scattered than the normal data, and some ipsilateral best azimuths were found in the bilaterally pinnectomized animals. The increase in scatter was considerable in the bilaterally operated animals, although, like the suprathreshold data, the best azimuths of the tuned and bilobed units still exhibited a tendency to become more contralateral as the recording electrode was moved from the rostral to the caudal end of the SC (Fig. 7B). Fewer units were tuned to aberrant azimuths after unilateral pinna removal, although the auditory representations in both the left (Fig. 7C) and the right SC (Fig. 7D) of these animals were again less precise than normal.

We have previously shown that linear regression coefficients are not the optimal tool for describing the auditory representation in the ferret SC. The visual and auditory maps in the SC do not present uniform magnification factors as a greater proportion of the nucleus is devoted to the representation of anterior space than to more peripheral regions. Accordingly, King and Carlile (1993) found that the polynomial Y = 15.9  13X  24.2X², where Y is unit best azimuth and X the distance from the rostrolateral pole, provides a better fit for the normal topography plots than a linear regression equation. We therefore adopted a method used in a recent study (Schnupp et al. 1995b) to quantify the scatter in the topography: we defined a "topography error" for each spatially tuned auditory unit as the difference between its best azimuth and the value predicted by the polynomial for the normal data set. The variance of the distribution of topography errors provides a quantitative measure for the scatter in the representation.

Figure 8 shows the distribution of topography errors for the supra- and near-threshold conditions in histogram form. Apart from the differences in the width (variance) of these distributions, some of the histograms shown are not centered on zero. A shift in the mean error toward positive values could be interpreted as a systematic shift in the representation toward more anterior or ipsilateral positions, whereas negative mean errors suggest a shift toward more contralateral positions. Because removal of the pinna and concha leads to an overall reduction in gain, zero ILDs no longer represent positions directly in front of the animal in the unilaterally pinnectomized group (Fig. 3). Instead, the azimuth associated with zero ILD has become highly frequency dependent, and, for most frequencies, zero ILDs are found at positions closer to the operated ear (compare Figs. 2 and 3). Removal of the left pinna might therefore result in an overall contralateral shift in the representation of sound azimuth (i.e., a negative shift in mean topography error) in the right SC and an ipsilateral (positive) shift in the left SC. Tables 2 and 3 show the result of t-tests to compare the mean topography errors in each condition against the normal control group (variances were not assumed to be equal). Although three of the four histograms for the unilaterally pinnectomized ferrets (Fig. 8) show shifts in mean error in directions expected from the overall change in ISD pattern, only one of the differences in mean error reached significance. In the bilaterally pinnectomized ferrets, the mean topography error was significantly different from normal at supra- but not at near-threshold sound levels. This is consistent with the presence in the suprathreshold condition of many units that were abnormally tuned to positions within the ipsilateral hemifield.

Tables 2 and 3 also depict the variances in topography errors and the results of F-tests comparing the values obtained for the different conditions. The rostrocaudal spread of recording sites in both the left and right SC of the unilaterally pinnectomized ferrets was smaller than that in the other two groups. Each pairwise comparison was therefore restricted to an equivalent region of the nucleus, i.e., we compared the data from the bilaterally pinnectomized ferrets with the full normal data set, whereas only those units recorded in the caudal 70 or 80% of the SC of these animals were used in the comparison with the data obtained from right and left SC, respectively, of the unilaterally pinnectomized group.

The variance in topography errors obtained at both supra- and near-threshold sound levels was significantly larger in all of the chronic pinnectomized groups than in the normal control animals. The variance was also larger in the bilaterally pinnectomized group than in the unilaterally pinnectomized ferrets, but no difference was found in the scatter of the auditory representation between the left and right SC in the ferrets in which the pinna and concha had been removed on the left side only. These F-tests operate on the assumption that the topography errors are normally distributed. Many of the topography error distributions shown in Fig. 8 have irregular shapes, raising doubts about the assumption of normality. Although the irregular appearance of the histograms may simply reflect a relatively low yield of auditory units in the experimental animals, a series of Lilliefors tests on these data suggested that the distribution of the suprathreshold topography errors in the right SC after unilateral pinnectomy may indeed differ significantly from normal (P = 0.03). However, the presence of a nonnormal distribution can in itself be taken as evidence that the auditory topography in these animals has been altered by the modified spectral cues.

We obtained the same overall picture when we measured the misalignment between the auditory best positions and the center of the visual receptive fields of multiunit responses recorded in the superficial layers. As with the auditory topography errors, the variance in auditory-visual misalignment was significantly greater after early pinna removal than in the normal control group. Bilateral pinna removal again resulted in a significantly larger scatter than unilateral surgery, and there was no difference in the variance for the left and right SC in the ferrets in which only the left pinna and concha had been removed.

Elevation receptive-field classes

Of the 26 units recorded in the bilaterally pinnectomized ferrets that yielded suprathreshold elevation profiles, 12 (46%) were tuned, 6 (23%) bilobed, 7 (27%) broad, and 1 (4%) was complex. Of the near-threshold profiles, 19 (73%) were tuned, 3 (11.5%) bilobed, 3 (11.5%) broad, and 1 (4%) complex. An elevation data set recorded under comparable conditions in previous studies using normal adult ferrets (King and Carlile 1993; King et al. 1994) contained exclusively (18/18) tuned suprathreshold profiles. A sample of 20 near-threshold elevation profiles from this normal control group comprised 18 (90%) tuned and 2 (10%) broad units. Thus, as with the azimuth response profiles, bilateral pinna removal in infancy led to more ambiguous elevation tuning in SC auditory units.

Elevation topography

Figure 9 illustrates the effect of early bilateral pinna removal on elevation topography. Data from normal adult ferrets are shown at supra- and near-threshold sound levels, respectively, in Fig. 9, A and B. The best elevations shift from superior to inferior values in a clear monotonic progression from the medial to the lateral side of the SC. The relationship between best elevation and recording site within the SC for the tuned units in the pinnectomized data set is shown in Fig. 9, C and D. In contrast to the azimuth topography data, which require a curvilinear regression, the normal elevation data are adequately described by a linear regression model. Regression equations and values of R² for each condition are also shown in Fig. 9. Although the R² values for the normal data are highly significant (P < 0.01 for both supra- and near-threshold data), those obtained after pinna removal failed to reach statistical significance, indicating that the topographic order in the elevation representation had been disrupted.

View larger version (24K): [in this window] [in a new window]   FIG. 9. Representation of sound elevation in normal (A and B) and bilaterally pinnectomized ferrets (C and D). Best positions of units with tuned elevation profiles are plotted against unit location along the elevation axis described in King and Hutchings (1987).

	DISCUSSION

Abstract Introduction Methods Results Discussion References

We have examined the neurophysiological consequences of altering monaural and binaural spectral localization cues by surgical removal of the pinna and concha in infancy. Acoustical measurements made once the animals had reached adulthood revealed that this procedure produced essentially the same change in the pattern of cue values as acute pinna removal in adulthood (Carlile and King 1994). Thus many of the acoustical features apparent in both the monaural and binaural spectral cues, which may allow corresponding locations on either side of the interaural axis to be discriminated in normal animals, are eliminated after chronic pinnectomy. On the other hand, ITDs and ILDs will still vary with sound source location due to the separation of the ears. Pinna removal therefore allows us to examine the relative contribution of different localization cues to the elaboration of the neural map of auditory space in the SC. The results of this study suggest that spectral cues play a pivotal role in the development of the representation of auditory space in the mammalian SC.

Adaptive plasticity in the auditory space map in the SC

Both visual and auditory experience are used to guide the development of the auditory space map (reviewed by King and Carlile 1995; Knudsen and Brainard 1995). In most studies, this neural representation was reported to show considerable potential for compensatory plasticity in response to experimental manipulations of either visual or auditory inputs. For example, the auditory space map can adapt to a large degree to changes in localization cue values introduced by raising animals with one ear occluded (King et al. 1988; Knudsen 1985). An ear plug will attenuate and, to some extent, change the phase of incoming sounds, so plugging one ear will produce shifts in ILD and ITD cues. A monaural ear plug will also lead to highly abnormal ISDs due to the frequency-dependent attenuation produced by the plug and changes in the DTF of the occluded ear. In the ferret, the SC ipsilateral to the plug shows considerable adaptation to these altered binaural cues, provided the plug is introduced in infancy, but not in adulthood (King and Carlile 1995; King et al. 1988).

The only other study to investigate the effects of manipulating the localization cues provided by the outer ear on the development and maintenance of an auditory representation in the brain was carried out on barn owls by Knudsen et al. (1994). Barn owls have asymmetric outer ears that cause ILDs to vary systematically with sound elevation, whereas ITDs vary with azimuth. The auditory space map in the optic tectum appears to result from a two-dimensional array of ILD and ITD sensitivity (Olsen et al. 1989). Removing the facial ruff and the preaural flap, structures that are homologous to the mammalian pinna and tragus, caused the ears to become essentially symmetric and changed the spatial locations that correspond to particular ILD and ITD values. Recordings from auditory units in the optic tectum of these animals revealed a change in ITD and, to a lesser extent, ILD tuning. The tuning changed in such a way as to reestablish at least a partial correspondence between the representation of these auditory spatial cue values and the retinotopic map of visual space in the optic tectum (Knudsen et al. 1994). Moreover, these adaptive changes in the neural representation of the binaural cues were observed in owls that had undergone facial ruff and preaural flap removal either in infancy or adulthood.

Consequences of chronic bilateral pinna removal in ferrets

In contrast to the results reported for the barn owl, we found that changing the spatial pattern of auditory localization cues by outer ear modification does not lead to changes in auditory spatial tuning that could be considered to be adaptive in nature. The substantial difference in the topographic organization of the auditory representation in the SC of ferrets after acute bilateral pinna removal in adulthood and chronic bilateral pinna removal in infancy certainly suggests that the tuning of auditory neurons to localization cues has been altered in the latter group, but not in a way that preserves the normal topographic order of the auditory map and its registration with the visual map.

The capacity of the auditory space map in the owl's optic tectum to undergo a compensatory change extends to adulthood in response to facial ruff and preaural flap removal (Knudsen et al. 1994) but not for monaural occlusion (Knudsen 1985). Knudsen et al. (1994) suggested that this might be because ruff removal, unlike monaural occlusion, does not change the cues associated with sound sources in front of the animal and therefore requires no change in the tuning properties of rostral neurons that represent this region of space. However, this is also true for neonatal bilateral pinnectomy in ferrets, and yet we found that the auditory representation in these animals differed more from normal than that found in animals that had been reared with one ear plugged from the same age (King and Carlile 1995; King et al. 1988) or in which only one pinna had been removed.

Recent studies in cats have emphasized the importance for sound localization of spectral notches, which are prominent features in the midfrequency range of the head-related transfer function in this species (Huang and May 1996; May and Huang 1996, 1997; Rice et al. 1992). For frontal positions, the notch frequency varies systematically with both azimuth and elevation, which may explain why sound localization is most accurate in this region of space (Huang and May 1996; May and Huang 1996). It has been suggested that a binaural comparison of the notch frequency in the two ears of the cat could provide a basis for identifying the location of the sound source (Rice et al. 1992). Less pronounced location-dependent spectral notches are also present near the anterior midline in the ferret's head-related transfer function and are eliminated by pinna removal (Fig. 1) (Carlile and King 1994). However, the map of auditory space is disrupted throughout the SC after chronic pinnectomy, not just in the rostral region that represents anterior sound locations, suggesting that the loss of other spectral features may be equally important.

In the only other study in which altered auditory experience has been reported to lead to a disruption of the auditory space map, Withington-Wray et al. (1990) raised guinea pigs in an environment of constant, omnidirectional noise in an attempt to deprive them of all localization cues. Bilaterally pinnectomized ferrets, in contrast, still have access to a wide range of ITD and ILD information. The disruption of spectral cues caused by pinna removal should not be a sufficient acoustical basis for the degraded topographic order in the auditory representation. Acute pinna removal in adulthood leads to very similar changes in the spatial pattern of DTFs and ISDs. But although the suprathreshold representation of auditory space is subsequently highly ambiguous, the azimuthal best positions of one lobe of the directional responses are mapped along the rostrocaudal axis of the SC in a manner that is statistically indistinguishable from normal (Carlile and King 1994). The preservation of auditory topography in these animals can be explained in terms of the topographic variation in sensitivity to interaural differences in overall sound pressure level, demonstrated by Wise and Irvine (1985) in the cat SC using flat, dichotically delivered noise stimuli.

A sharp increase in the number of bilobed response profiles is seen after both acute and chronic bilateral pinna removal, indicating that spectral cues are used to resolve the spatial ambiguities that exist for the remaining binaural cues. This is consistent with the increase in front-back confusions exhibited by human listeners when the cavities of the outer ear are filled (Oldfield and Parker 1984) and by adult chinchillas after bilateral pinna removal (Heffner et al. 1996). However, despite the availability of these residual, albeit ambiguous, binaural cues, the highly scattered representation in Fig. 6B suggests that the systematic variation in sensitivity to interaural differences in sound level has failed to develop, or has somehow become obscured, after bilateral pinna removal in infancy. Wise and Irvine (1985) observed that binaural interactions in rostral SC of the cat were predominantly facilitatory (OO/F or EO/F) and almost exclusively inhibitory (EO/I) in the caudal part of the nucleus. EO/I interactions should suppress responses to ipsilateral stimulation, and our data suggest that ipsilateral best positions are rarer in the caudal SC of the pinnectomized animals. There may therefore be at least a tendency for the binaural interactions to be of the appropriate type in different regions of the SC, although beyond that they do not appear to have developed appropriately.

The disruption of the elevation map by bilateral pinnectomy is more predictable, given the greater dependence of both human listeners (Middlebrooks and Green 1991) and other mammals (Heffner et al. 1996) on spectral cues for judging the vertical location of a sound source. Although the sensitivity to different localization cues of auditory neurons located at different sites across the mediolateral axis of the SC has not so far been measured dichotically, the consequences of unilateral deafening suggest that tuning for both azimuth and elevation relies on monaural and binaural cues (King et al. 1994). It remains to be determined whether the more ambiguous elevation tuning and degraded topography observed at suprathreshold sound levels can be explained in terms of an altered pattern of ISDs or, as seems to be the case for the azimuth representation, reflects abnormal development in sensitivity to those cues.

Consequences of chronic unilateral pinna removal

The disruptive consequences of unilateral pinna removal in infancy on auditory neurons in the SC were considerably milder than those caused by bilateral pinnectomy. This is also the case for the effects of pinna removal on sound localization in adult chinchillas (Heffner et al. 1996). To some degree this result is expected because the ISD functions for the unilaterally pinnectomized animals retained a number of asymmetric features. This implies that unambiguous binaural spectral information as well as one set of essentially normal monaural spectral cues were available throughout development.

Given the differences in the spatial distribution of spectral cues in each hemifield after unilateral pinna removal, we expected to find greater differences between the responses recorded in the left SC and those recorded in the right SC. However, as discussed above for the neural consequences of acute pinnectomy, topographic order in an otherwise ambiguous representation of sound azimuth is still present after the spectral cues have been disrupted. This may explain why we failed to find a significant difference in suprathreshold topography errors between the left and the right SC despite the marked differences in the two horizon ISD functions (Fig. 3). Nevertheless, the presence of a more precise representation of sound azimuth on both sides of the brain compared with that found in the SC after bilateral pinna removal suggests that the spatial tuning of auditory neurons in the unilaterally pinnectomized ferrets is much closer to that found in normal ferrets. This may reflect the greater availability of instructive experiential cues, presumably provided by the intact ear, in these animals.

The normal SC represents the contralateral side of space together with a small region of the frontal ipsilateral hemifield. After removal of the left pinna, the monaural spectral cues available in the hemifield contralateral to the right SC (the DTF in Fig. 1B) are severely impaired, whereas those available to the left SC are much less affected (Fig. 1A). In the cat, auditory units in rostral SC, which are tuned to ILDs close to zero, exhibit OO/F binaural interactions and therefore require a binaural input (Hirsch et al. 1985; Wise and Irvine 1985). However, recordings from unilaterally deafened guinea pigs (Palmer and King 1985) and ferrets (King et al. 1994) indicate that monaural pinna cues provide sufficient directional information for the formation of a normal near-threshold map of space. Furthermore, there is evidence for a systematic variation in sensitivity to monaural spectral cues along the rostrocaudal axis of the nucleus (Carlile and Pettigrew 1987). It is therefore not surprising that disrupting these cues in the bilaterally pinnectomized ferrets produces a degraded representation of sound azimuth at near-threshold sound levels (Fig. 7B). On the other hand, we found that the bandwidth of the tuning, the proportion of tuned units, and the variance of the topography errors were not significantly different between the two sides after unilateral pinna removal despite the apparent differences in the localization cues available.

Carlile and King (1994) found that the near-threshold representation of sound azimuth still shows some topographic order after acute bilateral pinna removal in adult ferrets. We attempted to explain this in terms of the binaural sensitivity of rostral units to very small interaural differences in overall sound level, together with possible tuning to residual monaural cues arising from lateral locations for caudal units. Clearly, the weak topography observed in the right SC after neonatal removal of the contralateral pinna and concha could be explained in a similar way. Moreover, the relatively minor disruption of the near-threshold map in the left SC, opposite the intact outer ear, may have been caused by the abnormal ISDs available, which included a frequency-dependent lateral shift in the azimuthal locations corresponding to zero ILDs as a result of the asymmetry in gain provided by the two ears.

We are still left with the question of why the near-threshold topography in the animals operated bilaterally in infancy was so much poorer than that in the right SC after left pinna removal. One possibility is that the SC maps on either side of the brain may develop as a unit, linked through the intercollicular commissure. In the ferret, this reciprocal projection arises from the deeper layers, is topographically organized, and, although strongest in the rostral half of the SC, extends throughout the rostrocaudal extent of the nucleus (Jiang et al. 1997). In hamsters some intercollicular neurons have sensory receptive fields (Rhoades et al. 1981), and in cats both excitatory and inhibitory connections appear to exist between the two colliculi (Behan and Kime 1996). Moreover, Withington et al. (1994) found that monocular enucleation in the guinea pig resulted in a bilateral impairment of auditory map topography, even though the retinocollicular projection in this species is almost entirely crossed. It is therefore conceivable that functional interactions via the intercollicular commissure may contribute to the development of very similar auditory representations in the left and right SC after unilateral pinna removal, which are much closer to the normal pattern than is the case after bilateral pinnectomy.

The possibility of intercollicular cross talk may also be relevant to the question of why the developing auditory representation apparently shows a greater capacity to adapt to monaural occlusion than to monaural pinna and concha removal. The ear plug should remove most of the near-threshold input to the contralateral SC, leaving normal monaural spectral cues available to the other nucleus. On the other hand, although the transmission gain provided by the pinnectomized ear should be reduced, different, and potentially conflicting, monaural spectral cues will be provided by the two ears after unilateral removal of the pinna and concha.

Although we have attempted to interpret our findings in terms of the acoustical consequences of pinna removal, it is important to consider the possibility that denervation of the outer ear, with the loss of both cutaneous and proprioceptive afferents, may be involved. In the cat, somatosensory signals from the dorsal column nuclei, and particularly those regions that receive inputs from the pinna and surrounding skin, appear to modulate the activity of auditory neurons in the dorsal cochlear nucleus (DCN) (Davis et al. 1996; Young at al. 1995). Because the DCN may play a role in processing spectral localization cues (Sutherland and Masterton 1992; Young et al. 1992), the loss of somatosensory inputs after pinna removal could potentially influence the auditory responses of neurons in higher areas, including the SC, that encode those cues. Although the function of nonauditory inputs to the DCN is unclear, they may be concerned with pinna motion (Young et al. 1995). As the ferret's pinnae are essentially immobile, there would appear to be little need for inputs that signal pinna motion or position to reach the SC in this species. Moreover, the similarity between the auditory responses recorded in the left and the right SC after unilateral pinna removal in infancy would also appear to argue against an explanation for our data in terms of nonacoustic factors.

Role of the outer ear in the development of the auditory space map

We have found that the topographic order of the auditory representation gradually develops during the second and third postnatal months (King and Carlile 1991, 1995). Initially, this process seems to follow the maturation of the acoustical values that occurs as the head and pinnae grow in size (Carlile 1991; King and Carlile 1995). This suggests that the spatial tuning observed at any given age is determined by the pattern of localization cues available at that time, and that a topographically ordered map emerges only once those cues have attained their adult values. However, the present data, along with earlier studies in which auditory or visual inputs were altered developmentally, indicate that the processes underlying the formation of the auditory space map are plastic and capable of adjusting in response to altered cues. Moreover, we have recently shown that N-methyl-D-aspartate-type glutamate receptors are involved in the normal maturation of the map, highlighting the importance of neural activity in this process (Schnupp et al. 1995b).

Whereas other studies have focused on the dominant role of vision in establishing and maintaining map alignment (King and Carlile 1995; Knudsen and Brainard 1995), visual cues were clearly not sufficient in the pinnectomized ferrets to guide the development of a matching auditory topography. The disruptive effects of chronic pinna removal suggest that spectral cues play an essential role in establishing the sensitivity of SC neurons to combinations of monaural and binaural cues that correspond to the same region of space. Given the evidence for a possible Hebbian mechanism of synaptic plasticity (Schnupp et al. 1995b), our findings are consistent with the possibility that the monaural representation, thought to be based on pinna cues, acts as a template for the development of sensitivity to binaural inputs conveying information about the same region of space. Alternatively, the lack of cues available for resolving the front-back ambiguities inherent in the binaural cues may be responsible for the failure of the system to construct a map of auditory space. In contrast to most other manipulations of sensory experience, removal of the outer ear structures alters the auditory environment in ways that appear to exceed the capacity of the developing space map to undergo adaptive changes.

	ACKNOWLEDGEMENTS

  We are grateful for the support provided by the Wellcome Trust. J.W.H. Schnupp was a Wellcome Prize Student, A. J. King is a Wellcome Senior Research Fellow, and S. Carlile was a Beit Memorial Fellow.

	FOOTNOTES

   Present address of S. Carlile: Dept. of Physiology, University of Sydney, NSW 2006 Australia.

  Address for reprint requests: A. J. King, University Laboratory of Physiology, Parks Rd., Oxford OX1 3PT, United Kingdom.

  Received 22 January 1997; accepted in final form 4 September 1997.

	REFERENCES

Abstract Introduction Methods Results Discussion References

• ANNETTS, C.S.J., HOWSE, B.P.A., HUTCHINGS, M. E., KING, A. J. A digitally controlled vertical hoop system for the presentation of free-field auditory and visual stimuli (Abstract). J. Physiol. (Lond.) 382: 21P, 1987.
• BAILEY, N.T.J. In: Statistical Methods in Biology, , 2nd ed.. Cambridge, UK: Cambridge Univ. Press, 1981
• BEHAN, M., KIME, N. M. Spatial distribution of tectotectal connections in the cat. Prog. Brain Res. 112: 131-142, 1996.[Medline]
• CARLILE, S. The auditory periphery of the ferret. I. Directional response properties and the pattern of interaural level differences. J. Acoust. Soc. Am. 88: 2180-2195, 1990a.[Medline]
• CARLILE, S. The auditory periphery of the ferret. II. The spectral transformations of the external ear and their implications for sound localization. J. Acoust. Soc. Am. 88: 2196-2204, 1990b.[Medline]
• CARLILE, S. Postnatal development of the spectral transfer functions and interaural level differences of the auditory periphery of the ferret. Soc. Neurosci. Abstr. 17: 232, 1991.
• CARLILE, S., KING, A. J. Monaural and binaural spectrum level cues in the ferret: acoustics and the neural representation of auditory space. J. Neurophysiol. 71: 785-801, 1994.[Abstract/Free Full Text]
• CARLILE, S., PETTIGREW, A. G. Distribution of frequency sensitivity in the superior colliculus of the guinea pig. Hear. Res. 31: 123-136, 1987.[Medline]
• DAVIS, K. A., MILLER, R. L., YOUNG, E. D. Effects of somatosensory and parallel-fiber stimulation on neurons in dorsal cochlear nucleus. J. Neurophysiol. 76: 3012-3024, 1996.[Abstract/Free Full Text]
• FLYNN, W. E., ELLIOT, D. N. Role of the pinna in hearing. J. Acoust. Soc. Am. 38: 104-105, 1965.[Medline]
• HEFFNER, R. S., KOAY, G., HEFFNER, H. E. Sound localization in chinchillas. III. Effect of pinna removal. Hear. Res. 99: 13-21, 1996.[Medline]
• HIRSCH, J. A., CHAN, J.C.K., YIN, T.C.T. Responses of neurons in the cat's superior colliculus to acoustic stimuli. I. Monaural and binaural response properties. J. Neurophysiol. 53: 726-745, 1985.[Abstract/Free Full Text]
• HUANG, A. Y., MAY, B. J. Sound orientation behavior in cats. II. Mid-frequency spectral cues for sound localization. J. Acoust. Soc. Am. 100: 1070-1080, 1996.[Medline]
• JIANG, Z. D., MOORE, D. R., KING, A. J. Sources of subcortical projections to the superior colliculus in the ferret. Brain Res. 755: 279-292, 1997.[Medline]
• KING, A. J., CARLILE, S. Maturation of the map of auditory space in the superior colliculus of the ferret. Soc. Neurosci. Abstr. 17: 231, 1991.
• KING, A. J., CARLILE, S. Changes induced in the representation of auditory space in the superior colliculus by rearing ferrets with binocular eyelid suture. Exp. Brain Res. 94: 444-455, 1993.[Medline]
• KING, A. J., CARLILE, S. Responses of neurons in the ferret superior colliculus to the spatial location of tonal stimuli. Hear. Res. 81: 137-149, 1994.[Medline]
• KING, A. J., CARLILE, S. Neural coding for auditory space. In: The Cognitive Neurosciences, edited by and M. S. Gazzaniga . Cambridge, MA: MIT Press, 1995, p. 279-293
• KING, A. J., HUTCHINGS, M. E. Spatial response properties of acoustically responsive neurons in the superior colliculus of the ferret: a map of auditory space. J. Neurophysiol. 57: 596-624, 1987.[Abstract/Free Full Text]
• KING, A. J., HUTCHINGS, M. E., MOORE, D. R., BLAKEMORE, C. Developmental plasticity in the visual and auditory representations in the mammalian superior colliculus. Nature 332: 73-76, 1988.[Medline]
• KING, A. J., MOORE, D. R., HUTCHINGS, M. E. Topographic representation of auditory space in the superior colliculus of adult ferrets after monaural deafening in infancy. J. Neurophysiol. 71: 182-194, 1994.[Abstract/Free Full Text]
• KING, A. J., PALMER, A. R. Cells responsive to free-field auditory stimuli in guinea-pig superior colliculus: distribution and response properties. J. Physiol. (Lond.) 342: 361-381, 1983.[Abstract/Free Full Text]
• KNUDSEN, E. I. Auditory and visual maps of space in the optic tectum of the owl. J. Neurosci. 2: 1177-1194, 1982.[Abstract]
• KNUDSEN, E. I. Experience alters the spatial tuning of auditory units in the optic tectum during a sensitive period in the barn owl. J. Neurosci. 5: 3094-3109, 1985.[Abstract]
• KNUDSEN, E. I., BRAINARD, M. S. Creating a unified representation of visual and auditory space in the brain. Annu. Rev. Neurosci. 18: 19-43, 1995.[Medline]
• KNUDSEN, E. I., ESTERLY, S. D., OLSEN, J. F. Adaptive plasticity of the auditory space map in the optic tectum of adult and baby barn owls in response to external ear modification. J. Neurophysiol. 71: 79-94, 1994.[Abstract/Free Full Text]
• MAY, B. J., HUANG, A. Y. Sound orientation behavior in cats. I. Localization of broadband noise. J. Acoust. Soc. Am. 100: 1059-1069, 1996.[Medline]
• MAY, B. J., HUANG, A. Y. Spectral cues for sound localization in cats: a model for discharge rate representations in the auditory nerve. J. Acoust. Soc. Am. 101: 2705-2719, 1997.[Medline]
• MIDDLEBROOKS, J. C. Binaural mechanisms of spatial tuning in the cat's superior colliculus distinguished using monaural occlusion. J. Neurophysiol. 57: 688-701, 1987.[Abstract/Free Full Text]
• MIDDLEBROOKS, J. C., GREEN, D. M. Sound localization by human listeners. Annu. Rev. Psychol. 42: 135-159, 1991.[Medline]
• MIDDLEBROOKS, J. C., KNUDSEN, E. I. A neural code for auditory space in the cat's superior colliculus. J. Neurosci. 4: 2621-2634, 1984.[Abstract]
• MIDDLEBROOKS, J. C., MAKOUS, J. C., GREEN, D. M. Directional sensitivity of sound-pressure levels in the human ear canal. J. Acoust. Soc. Am. 86: 89-108, 1989.[Medline]
• MOORE, D. R., HINE, J. E. Rapid development of the auditory brainstem response threshold in individual ferrets. Dev. Brain Res. 66: 229-235, 1992.[Medline]
• MOREY, A. L., CARLILE, S. Auditory brainstem of the ferret: maturation of the brainstem auditory evoked response. Dev. Brain Res. 52: 279-288, 1990.[Medline]
• OLDFIELD, S. R., PARKER, S.P.A. Acuity of sound localisation: a topography of auditory space. II. Pinna cues absent. Perception 13: 601-617, 1984.[Medline]
• OLSEN, J. F., KNUDSEN, E. I., ESTERLY, S. D. Neural maps of interaural time and intensity differences in the optic tectum of the barn owl. J. Neurosci. 9: 2591-2605, 1989.[Abstract]
• PALMER, A. R., KING, A. J. A monaural space map in the guinea-pig superior colliculus. Hear. Res. 17: 267-280, 1985.[Medline]
• RHOADES, R. W., FISH, S. E., VONEIDA, T. J. Anatomical and electrophysiological demonstration of tectotectal pathway in the golden hamster. Neurosci. Lett. 21: 255-260, 1981.[Medline]
• RICE, J. J., MAY, B. J., SPIROU, G. A., YOUNG, E. D. Pinna-based spectral cues for sound localization in the cat. Hear. Res. 58: 132-152, 1992.[Medline]
• SCHNUPP, J.W.H., KING, A. J., CARLILE, S., BARON, J. Role of the outer ear in the development of the auditory space map in the ferret superior colliculus. Soc. Neurosci. Abstr. 21: 671, 1995a.
• SCHNUPP, J.W.H., KING, A. J., SMITH, A. L., THOMPSON, I. D. NMDA-receptor antagonists disrupt the formation of the auditory space map in the mammalian superior colliculus. J. Neurosci. 15: 1516-1531, 1995b.[Abstract]
• SUTHERLAND, D. P., MASTERTON, R. B. Effects of unilateral lesions on discrimination of changes in sound source elevation. Abstr. Assn. Res. Otolaryngol. 15: 51, 1992.
• WIGHTMAN, F. L., KISTLER, D. J. Headphone simulation of free-field listening. I. Stimulus synthesis. J. Acoust. Soc. Am. 85: 858-867, 1989.[Medline]
• WISE, L. Z., IRVINE, D.R.F. Interaural intensity difference sensitivity based on facilitatory binaural interaction in cat superior colliculus. Hear. Res. 16: 181-187, 1984.[Medline]
• WISE, L. Z., IRVINE, D.R.F. Topographic organization of interaural intensity difference sensitivity in deep layers of cat superior colliculus: implications for auditory spatial representation. J. Neurophysiol. 54: 185-211, 1985.[Abstract/Free Full Text]
• WITHINGTON, D. J., BINNS, K. E., INGHAM, N. J., THORNTON, S. K. The effects of monocular enucleation on the representation of auditory space in the superior colliculus of the guinea-pig. Brain Res. 636: 348-352, 1994.[Medline]
• WITHINGTON-WRAY, D. J., BINNS, K. E., DHANJAL, S. S., BRICKLEY, S. G., KEATING, M. J. The maturation of the superior collicular map of auditory space in the guinea pig is disrupted by developmental auditory deprivation. Eur. J. Neurosci. 2: 693-703, 1990.[Medline]
• YIN, T.C.T., HIRSCH, J. A., CHAN, J.C.K. Responses of neurons in the cat's superior colliculus to acoustic stimuli. II. A model of interaural intensity sensitivity. J. Neurophysiol. 53: 746-758, 1985.[Abstract/Free Full Text]
• YOUNG, E. D., NELKEN, I., CONLEY, R. A. Somatosensory effects on neurons in dorsal cochlear nucleus. J. Neurophysiol. 73: 743-765, 1995.[Abstract/Free Full Text]
• YOUNG, E. D., SPIROU, G. A., RICE, J. J., VOIGT, H. F. Neural organization and responses to complex stimuli in the dorsal cochlear nucleus. Philos. Trans. R. Soc. Lond. B Biol. Sci. 336: 407-413, 1992.[Medline]

0022-3077/98 $5.00 Copyright ©1998 The American Physiological Society