Abstract
Interaural time differences (ITDs) are a major cue for sound localization and change with increasing head size. Since the barn owl's head width more than doubles in the month after hatching, we hypothesized that the development of their ITD detection circuit might be modified by experience. To test this, we raised owls with unilateral ear inserts that delayed and attenuated the acoustic signal, and then measured the ITD representation in the brainstem nucleus laminaris (NL) when they were adults. The ITD circuit is composed of delay line inputs to coincidence detectors, and we predicted that plastic changes would lead to shorter delays in the axons from the manipulated ear, and complementary shifts in ITD representation on the two sides. In owls that received ear inserts starting around P14, the maps of ITD shifted in the predicted direction, but only on the ipsilateral side, and only in those tonotopic regions that had not experienced auditory stimulation prior to insertion. The contralateral map did not change. Thus, experience-dependent plasticity of the ITD circuit occurs in NL, and our data suggest that ipsilateral and contralateral delays are independently regulated. As a result, altered auditory input during development leads to long-lasting changes in the representation of ITD.
Significance Statement The early life of barn owls is marked by increasing sensitivity to sound, and by increasing ITDs. Their prolonged post-hatch development allowed us to examine the role of altered auditory experience in the development of ITD detection circuits. We raised owls with a unilateral ear insert and found that their maps of ITD were altered by experience, but only in those tonotopic regions ipsilateral to the occluded ear that had not experienced auditory stimulation prior to insertion. This experience-induced plasticity allows the sound localization circuits to be customized to individual characteristics, such as the size of the head, and potentially to compensate for imbalanced hearing sensitivities between the left and right ears.
Introduction
Upon hearing onset, connections in brainstem nuclei are rapidly refined, and at this time appear most vulnerable to disruption by deafening and noise rearing (Kapfer et al., 2002; Werthat et al., 2008; Kandler et al., 2009; Sonntag et al., 2009; Popescu and Polley, 2010; Butler and Lomber, 2013; Polley et al., 2013; Keating et al., 2015; Clarkson et al., 2016; Babola et al., 2018; Persic et al., 2020). Unilateral hearing loss, in particular, has profound effects on binaural circuits (Moore et al., 1999; Moore and King, 2004; Sanes and Bao 2009; Popescu and Polley, 2010; Persic et al., 2020). Recent reviews have outlined the effects of hearing loss on auditory brainstem plasticity (Sanes and Bao 2009; Tzounopoulos and Kraus, 2009; Anderson et al., 2010; Keating and King, 2013; Friauf et al., 2015; Lauer et al., 2019; Persic et al., 2020; Rubio, 2020) and upon localization behavior in rats (Clements and Kelly, 1978), ferrets (Moore et al., 1999; Keating and King, 2013; Kumpik and King, 2019), gerbils (Maier et al., 2008), and barn owls (Knudsen, 2002; Bergan and Knudsen, 2008; Keuroghlian and Knudsen, 2007). Previous studies have shown clear refinement of inputs to the interaural time difference (ITD)-sensitive neurons of the mammalian medial superior olive (MSO) during development (Kapfer et al., 2002; Chang et al., 2003; Magnusson et al., 2005; Werthat et al., 2008; Winters and Golding, 2018).
In the present study, we examined the effects of unilateral manipulation of auditory experience on ITD coding in nucleus laminaris (NL), the first binaural nucleus in the barn owl. Barn owls (genus Tyto) are altricial—blind and deaf when they hatch (Köppl and Nickel, 2007; Kraemer et al., 2017). Young barn owls begin to acquire auditory sensitivity around day 4 posthatch (P4), their eyes open by about P10, their heads grow to adult width by about P30, and their facial ruff and sensitivity to high-frequency sounds are fully developed by about P60 (Haresign and Moiseff, 1988; Rich and Carr, 1999; Köppl et al., 2005; von Campenhausen and Wagner, 2006; Köppl and Nickel, 2007; Kraemer et al., 2017). The barn owl's NL contains large, well-ordered, reproducible maps of ITD (Carr et al., 2015), which we exploited to test the hypothesis that neural delays might be modified by experience (Cheng and Carr, 2007). The ITD maps are created by afferent delay lines whose responses sum in NL to create a binaural ITD-sensitive extracellular potential termed the neurophonic (Weinberger et al., 1970; Kuokkanen et al., 2013). We reasoned that, if delays adjusted to temporally altered acoustic input, then the maps of ITD would be shifted compared to normal.
To test our hypothesis, we first described the maps of ITD in adult owls (McColgan et al., 2014; Carr et al., 2015). In pilot experiments, with preliminary results reported in abstract form (Köppl et al., 2012), we then fitted 8 young owls (P20 or older) with unilateral ear-canal inserts designed to selectively introduce a time delay to inputs from that ear (Gold and Knudsen, 1999; Köppl et al., 2012), and later compared their ITD maps with those of normal adult owls. The manipulated owls did not show altered maps of ITD (Köppl et al., 2012). Although abnormal auditory experience starting around P25 induced adjustments in tuning for binaural localization cues in the owl midbrain (Gold and Knudsen, 1999, 2000a,b), we reasoned that P20 might be past a critical period for the adjustment of delays in the brainstem NL. We therefore raised 6 owls from about P14 with unilateral earmold plugs that introduced both a time delay and significant attenuation to inputs from that ear, and report on the results of both studies here.
Materials and Methods
All experiments were carried out on barn owls of both sexes. Recordings from American barn owls (Tyto furcata; previously classified as Tyto alba pratincola) in Maryland conformed to procedures in the NIH Guidelines for Animal Research and were approved by the Animal Care and Use Committee of the University of Maryland. Experiments in Oldenburg were carried out on European barn owls (Tyto alba) following the protocols and procedures approved by the authorities of Lower Saxony, Germany (permit no. AZ 33.9-42502-04-11/0337).
Acoustic filtering devices and earmold plugs
In our first pilot experiments (Köppl et al. 2012), eight owls (2 T. furcata and 6 T. alba) were raised with acoustic filtering devices (Gold and Knudsen, 1999). Two owls (T. furcata) were first binaurally occluded with foam rubber earplugs (E.A.R. Cabot, Knudsen et al., 1984a) to limit the auditory experience from 20 to 28 days of age (P20–P28), corresponding to the time that their ear canals were open but too narrow to accommodate the device. Around 29 days of age, the binaural foam plugs were removed, and the filtering device was sutured into an ear canal. The foam earplugs and the acoustic filtering devices were sutured into place while the owls were anesthetized with isofluorane (1–2%). The acoustic filtering device was designed by Gold (Gold and Knudsen, 1999) to alter auditory experience, and custom-made from acetal delrin (Plastics SRT). A small, circular flange fit tightly to the inner walls of the ear canal, and the rest of the device was located just behind the preaural flap and in front of the facial ruff feathers. The device was designed to increase the path length of sound reaching the affected ear and to change the resonance properties of the ear canal while still providing a low-impedance pathway to the tympanic membrane. The other six owls (T. alba) were raised beginning with a modified acoustic filtering device that could fit a younger ear canal (5.5–7.5 mm diameter of the ear-canal part), which was then replaced with a full size (8.5 mm diameter) device when the ear canals were large enough (Table 1). One of the other six owls (T. alba) was raised with an early version of the filtering device that had a shorter and smaller diameter chamber; the acoustic properties of this device were not measured. All 8 owl chicks were raised to minimum age of 3 months wearing the devices (Table 1).
Ear plug manipulations
Since the acoustic filtering device did not appear to affect the map of ITD in the NL, a further 6 owl chicks (T. furcata) were raised with unilateral impression earmold plugs (Gold Velvet, Oklahoma City) inserted into the left ear canal under 1% isofluorane anesthesia. These earmold plugs could be inserted at an earlier age (starting at P14) to limit the auditory experience. At P14, the average diameter of the ear canal was about 1.8 mm, increasing to about 6.2 mm at P30 (Haresign and Moiseff, 1988). During this first month after hatching, the ear-opening diameter increases fairly linearly (diameter y (in mm) at time x (in days) was y = 0.275x − 2.05), and it was possible to insert an earmold plug after about P14. The 6 owls, plus 2 age-matched controls, were hand-raised in a Brinsea TLC-4 incubator, with temperature adjusted every few days as their thermoregulation ability increased (Rich and Carr, 1999). The earmold plugs were examined daily and replaced every few days as the ear canals grew until the owls were about 3 months old. The plugs were then removed, and the owls were returned to the aviary to reach adulthood (>6 months), after which their maps of ITD were measured. Block of the middle ear was not possible because of the large interaural canal that connects the two ears (Gans et al., 2012; Kettler et al., 2016).
The effects of the foam plug (E.A.R., Cabot Corp.) and the acoustic filtering device have been published previously (Knudsen et al., 1984a; Gold and Knudsen, 1999). The differences in attenuation and phase caused by the earmold plug used in the second set of experiments were measured using the same type of earmold plug and a miniature microphone (Knowles EM3068) placed adjacent to the tympanum in 2 owl cadaver heads to compare signals from plugged and unplugged ears. Over the 3–8 kHz range of frequencies, the median attenuation varied from 12 to 24 dB, and the median ongoing time delay (converted from phase delay) increased in the range from 17 to 84 µs. These data indicate that the level and temporal properties of sound reaching the eardrum were altered substantially at all frequencies.
Auditory brainstem responses
To determine if plug rearing led to long-lasting changes in sensitivity, we used auditory brainstem responses (ABRs) to measure the sensitivity to free-field sound in 4 of the 6 owls that had been raised to adulthood with an earmold plug and in 4 age matched controls (Fig. 1A and Table 1). The plugs were removed prior to ABR recordings. For comparison with previous measurements, we used the same protocol and equipment as in Kraemer et al. (2017). Owls were anesthetized by intramuscular injection of 3 mg/kg/h xylazine (“Xyla-ject”) and 16 mg/kg/h ketamine hydrochloride (“Ketavet”). Supplemental doses of ketamine and xylazine were administered to maintain a suitable plane of anesthesia. Owls were placed in a large anechoic chamber (I.A.C.), wrapped in a heating blanket (38°C) with their head positioned 30 cm away from the speaker. Lidocaine was applied to the skin prior to inserting the electrodes subcutaneously. BioSigRP software (Tucker-Davis Technologies (TDT)) was used to present stimuli and record from the three platinum subdermal needle electrodes: the negative electrode behind the ear facing directly towards the speaker, the ground electrode behind the opposite ear, and the active electrode directly on the midline vertex of the barn owl's head. We used SigGenRP software (TDT) to create tone-burst stimuli, which were fed through an RP2.1 (TDT). The electrode signal was processed using a Medusa Digital Biological Amplifier System (RA4L Head stage and RA16PA PreAmp, RA16BA Medusa Base station). After data collection, the signals were notch filtered at 60 Hz, low-pass filtered at 3,000 Hz, and high-pass filtered at 30 Hz using BioSigRP software. Thresholds for clicks and individual frequencies were established as the decibel level between the lowest stimulus level where any recognizable feature of the waveform was discernible, using visual inspection (Kraemer et al., 2017), and the trace without a detectable peak. Because the stimulus level was changed in intervals of either 10 dB or 5 dB, the threshold was either 5 dB or 2.5 dB SPL below the level at which a detectable peak was recognized. At least two thresholds (from 300 stimulus repetitions each) were determined for each frequency per trial. Thresholds were then averaged for each bird. ABR audiograms from owls raised with an earplug only differed from the age-matched control group at 0.5 kHz (Fig. 1A).
ABR evaluation of earmold plugs. A, Plug rearing does not affect the overall hearing sensitivity, following the removal of plug inserts. Average ABR audiograms for free-field tone-burst stimuli, for plug reared (inserts removed) and age-matched control adults (n = 8, 4 plugged and 4 control). Error bars show the standard deviation in dB SPL. B, Earmold plugs induce reliable sound attenuation. ABR audiograms for P21 animals (mean of two), with no plug, a single plug, or both ears plugged. Thresholds (for tone-burst stimuli) were the lowest with no plug insert, higher with a single plug, and highest with bilateral plugs. The adult audiogram is shown for comparison (Kraemer et al., 2017). Click thresholds for no plug or a single plug at P21 are shown at 0.1 kHz. C, Representative ABR traces from a single P21 animal before (blue) and after (gray) acute bilateral ear plug insertion. Traces were evoked by a free-field 45 dB SPL click stimulus at t = 0.
We also measured the effect of acute earmold plug insertion (unilateral and bilateral) in 2 of the 6 earmold plug owls at P21 to confirm the degree of attenuation caused by the earmold plugs around the time of insertion and provide exemplar traces of ABR responses to 45 dB stimuli to plugged and normal ears (Fig. 1B,C). We used the same ABR setup as before, with free-field stimulation. We measured the effect of both monaural and binaural earplug insertion and compared normal ABRs with ABRs recorded with bilateral and unilateral plugs (Fig. 1B). Above 2 kHz, the mean bilateral attenuation for P21 animals was 16 ± 3 dB, and that for adult owls was 17 ± 4 dB. An unbalanced one-way ANOVA showed significant differences in detection thresholds between normal and earplugged conditions for frequencies between 0.5 and 8 kHz in the adult owl (F(1,13) = 19.47, p < 0.0007, not shown), and for frequencies between 0.5 and 6.3 kHz in the P21 chick (F(1,10) = 9.16, p < 0.013, data in Fig. 1B, compare P21 bilateral and no plug).
Surgery and stereotaxis and electrophysiological recording
To measure neurophonic potentials in NL, animals were examined over 2 or 3 separate 8 h experiments, spaced approximately a week apart, and designed to follow the protocols as proposed in Köppl et al. (2012) and Carr et al. (2013, 2015). Anesthesia was induced by intramuscular injections of 16 mg/kg/h ketamine hydrochloride (“Ketavet” Phoenix) plus 3 mg/kg/h xylazine (“Xyla-ject”, Phoenix). Supplementary doses of ketamine and xylazine were administered to maintain a suitable plane of anesthesia. The body temperature was measured using a cloacal probe and kept constant at 39°C by a feedback-controlled heating blanket wrapped around the owl's body (Harvard Instruments). An electrocardiogram was used to monitor the plane of anesthesia.
The owl's head was held in a controlled position by a custom-designed stereotaxic apparatus, using ear bars and a beak holder designed to position the head at 70° to the visual axis (Pettigrew, 1979). Then a metal head plate and marker of a standardized zero point was permanently glued to the skull. After this, the ear bars and beak holder were removed, and the head held by the head plate alone. A craniotomy was made around the desired area relative to the zero point and a small hole made in the dura mater, taking care to avoid blood vessels. Each electrode was positioned at defined rostrocaudal and mediolateral locations, before being advanced into the brain. In some cases, the electrode was angled to facilitate access to the most medial regions of the brainstem.
Owls were placed on a vibration-insulated table within a sound-attenuating chamber (IAC) that was closed during all recordings. Generally, commercial, Epoxylite-coated tungsten electrodes (Frederick Haer Corporation) were used, with impedances between 2 and 20 MΩ and generally 250 µm diameter shank; Figure 8G in Kuokkanen et al. (2010). A grounded silver chloride pellet, placed under the animal's skin around the incision, served as the reference electrode. Extracellular electrode signals were amplified and filtered by a custom-built headstage and amplifier (mA 2000, Walsh Electronics). Recordings were passed in parallel to an oscilloscope, a threshold discriminator (SD1, TDT) and an analog-to-digital converter (DD1, TDT) connected to a personal computer via an optical interface (TDT). Digitized waveforms were saved for the offline analysis.
Acoustic stimulus generation and calibration
For neurophysiological recordings, acoustic stimuli were digitally generated by custom-written software (“Xdphys” written in Dr. M. Konishi's lab at Caltech, CA) driving a signal-processing board (DSP2, TDT). After passing a digital-to-analog converter (DD1, TDT) and an anti-aliasing filter (FT6-2, corner frequency 20 kHz, TDT), the signals were variably attenuated (PA4, TDT), impedance-matched (HB6, TDT) and attenuated by an additional fixed amount before being fed to commercial miniature earphones. Two separate channels of signals could be generated, passing through separate channels of associated hardware and driving two separate earphones (Yuin PK2, China). The earphones were housed in custom-built, calibrated from 0.1 to 10 kHz, closed sound systems inserted into the owl's left and right ear canals, respectively. Sound pressure levels and phase were calibrated individually at the start of each experiment, using built-in miniature microphones (Knowles EM3068). The analog stimulus waveforms were saved in most experiments.
Acoustic clicks, noise bursts, and tone stimuli were digitally created with a sampling rate of 48.077 kHz (sampling interval: 20.8 µs). Clicks were digitally produced at a rate of 4/s and then processed by the TDT-systems (Wagner et al., 2005, 2009). Several parameters of the click stimulus could be varied: the intensity (maximally 0 dB attenuation, corresponding to the 65 dB SPL spectrum level at 5 kHz (34 dB SPL overall level)), the duration (1–4 samples equivalent to 20.8–83.2 µs), and the polarity. A condensation click at 0 dB attenuation and 2 samples duration served as a reference. The click stimulus was repeated 128 times. Tone bursts of different frequencies (500–10,000 Hz range, 50–500 Hz steps) were presented monaurally and binaurally at levels from 20 to 60 dB SPL and averaged over 5 repetitions. The tone burst duration was 200 ms, including a 5 ms delay, 5 ms rise/fall times, constant starting phase, 400 ms epoch length, and an inter-stimulus interval of 750 ms. The noise stimuli had a spectrum between 0.1 and 13 kHz, 5 ms rise/fall times, and a duration of 200 ms. The overall level of the noise varied between 20 and 60 dB SPL. Stimulus ITDs generally covered about two periods at the BF of a location, and the ITD step size was 30 µs.
Recording protocol
While lowering the electrode, noise bursts were presented as search stimuli. Once auditory neurophonic responses were discernable, tonal stimuli of varying frequencies were applied from both the ipsi- and contralateral side to judge the position of the electrode. To map the best ITD within an isofrequency band, we made multiple stereotaxically controlled penetrations through the overlying cerebellum into NL. Apart from low-frequency NL, which has a lateral bend where frequencies from 500 to 2,000 Hz are represented (Takahashi and Konishi, 1988a; Palanca-Castan and Köppl, 2015), NL is sufficiently large and ordered for 3–4 serial penetrations along the same isofrequency slab within an 8 hour experimental period.
At a given recording site, the following protocol was tested to obtain a data set. To determine the best frequency (BF) both ipsi- and contralaterally, average threshold crossings of the neurophonic potential in response to 3–5 repetitions of each tone burst stimulus were calculated within the stimulus time-window (Kuokkanen et al., 2010). This paradigm was chosen to derive iso-intensity frequency response curves. Best ITD for each recording location was determined using two assumptions: first, that ITD tuning changed continuously with depth. Second, that peak responses to ITDs at or close to 0 µs would occur within a penetration (Carr et al., 2015). Best ITDs were calculated from responses to tone bursts presented binaurally at the estimated BF (Kuokkanen et al., 2013). During the experiment, the tuning to ITD was judged by a Rayleigh test (p < 0.05). We recorded from both sides of the brain, with left ear leading ITDs described as negative, and right positive.
Electrolytic lesions and histology
Lesions (10 µA, 3–10 s) were made at locations closest to 0 µs best ITD using a constant current device (CS3, Transkinetics, or Lesion Making Device). After a survival time of 5–14 days, anesthetized owls were perfused transcardially with saline, followed by 4% paraformaldehyde in phosphate buffer. Brains were blocked in the same stereotaxic apparatus as for in vivo recordings, marked on either the left or right side, then cut in the transverse plane orthogonal to the electrode penetrations. Forty micron sections were mounted on gelatin-coated slides and stained with cresyl violet. A trained observer, blinded to condition, examined all sections at low magnification to identify electrode tracks and lesions, and then visualized and reconstructed every third section using a Neurolucida neuron tracing system (Microbrightfield Bioscience) that included an Olympus BX61 light microscope, a video camera (MBF-CX9000, Microbrightfield), a motorized stage controller, and a personal computer. Sections were traced and digitized using a 10× objective on the microscope, with the Z-position defined based on the section thickness. Each section was aligned with the previous using rotation, reflection, or translation only to generate 3D reconstructions of NL. Lesions, lesion tracks, and NL borders were marked using the tracing function. The ruler function was used to measure distances from the lesion center to the borders of the nucleus. The distance of each lesion along its isofrequency band was measured from 3D reconstructions of NL. Isofrequency bands traverse NL and are composed of inputs from the nucleus magnocellularis with similar BF (Carr et al., 2015). Sections were not corrected for shrinkage (Kuokkanen et al., 2010, 2013) since measurements were normalized. Lesions varied in size, from 60 to 270 µm diameter (mean ± SD = 190 ± 64 µm, n = 61, Fig. 3).
Maps of ITD in NL. A, Exemplar ITD sensitive recording series through the left-hand side (LHS = plug side) NL. At each recording position (depth of the electrode denoted in µm), the response shows local maxima at the best ITD. There is a shift in best ITD from contralateral (positive best ITDs) to frontal ITDs (near zero best ITD) with increasing recording electrode depth (from dorsal to ventral). In this dorso-ventral series of recordings, with depth denoted by the increasing shades of blue, lightest blue line shows the most dorsal recording at 10,100µm, thickest blue line has the best ITD closest to 0 µs, and the darkest blue line shows the most ventral recording at 10,600 µm depth. Note that ITD tuning curves are largely periodic, and only one cycle of each ITD tuning curve is shown in blue (rest in gray). Neurophonic responses in NL were evoked by binaural tones (5.6 kHz) that varied in ITD, and ITD-sensitive responses were recorded only within NL. B, Nissl-stained transverse 40 µm brain section at the level of central NL showing the top half of the brainstem, with NL, NM and nucleus angularis (NA). Asterisk marks a lesion near the ventrolateral edge of NL (95% along isofrequency band). Scale bar = 1,000 μm. The dorsal (at the top) and ventral (at the bottom) borders of NL are outlined by a glial border that surrounds NL, with the tracts made up of NM axons above and below the borders. C, Transverse brain section at the level of caudal NL showing 3 exemplar lesions from penetrations through both LHS (earplug) and RHS NL. Note the incoming auditory-nerve tract (N8). Lesions on both sides in NL mark 0-µs best ITD: on the left, a lesion 43% along the isofrequency band, and 18% up from the ventral edge of NL (black-ringed symbol in F) and on the right, two lesions, one at 50% along the isofrequency band and 20% from the ventral edge, and the other at 76% of the isofrequency band and 11% from the ventral edge, both black ringed symbols in E. Scale bar = 1,000 µm. D, Ipsilateral and contralateral ITD maps in device reared owls (blue symbols and lines) are not different to the normal maps as characterized in Carr et al. 2015 (black solid line). Regression for the line ipsilateral to the device is y = −0.53x + 0.70, R2 = 0.19, closed symbols and solid blue line. Regression contralateral to the device is y = −0.50x + 0.57, R2 = 0.39, open symbols and dashed blue line. Each data point marks the dorso-ventral and isofrequency axis coordinates of a 0 µs best ITD recording site. The dorso-ventral position of lesions marking 0 μs best ITD in NL decreased with increasing position along an isofrequency band (more lateral). E, ITD maps in contralateral NL are robust to unilateral earmold plug-rearing. Locations of lesions in NL contralateral to the earmold plug show no systematic shift in the mapping of frontal space. Blue symbol color denotes recordings from regions of NL with BFs ≤5 kHz, and green symbols recordings >5 kHz. Regression line colors match symbol color. F, Unilateral earmold plug-rearing is correlated with a compensatory shift in the ipsilateral ITD map, supporting the prediction in Figure 2D. Locations of lesions in NL (green dots) revealed a systematic shift in the mapping of frontal space, around 0 µs best ITD, with respect to the normal map (black solid line, data from Carr et al. (2015)), in NL ipsilateral to the earmold plug. Blue symbol colors denote recordings from regions of NL with best frequencies ≤5 kHz, and green symbols recordings >5 kHz. Regression line colors match symbol color. Note D, E and F share a common x-axis.
Data analysis
A brief signal analysis was done during the experiment to determine the frequency and ITD tuning. After the experiment, the responses were quantitatively analyzed with custom-written software (MATLAB; BEDS scripts from G. B. Christianson). Frequency and ITD tuning were quantified using the variance and the signal-to-noise ratio (SNR) (Kuokkanen et al., 2010) of the ongoing neurophonic response; onset effects were excluded from the analysis, and we averaged the variance and SNR over several trials. Only those recordings with both frequency and ITD tuning of the response variance (Kuokkanen et al., 2013; Carr et al., 2015) were accepted. The vector strength of the ITD tuning of the response variance was evaluated at each recording depth, and required to be >0.02 and statistically significant at p < 0.05 to be included in the subsequent analyses (Kuokkanen et al., 2013). The frequency tuning of sites with weaker than the above ITD tuning was analyzed if there was a strong ITD tuning of the cyclic-mean amplitude (Kuokkanen et al., 2010, 2013) of vector strength >0.1 and p < 0.005. Additionally, the monaural response curves needed to have maximum SNR (SNRmax) > 10 dB each. This second criterion was added so that some clearly ITD-tuned recording locations, however with a strong overlaid noise, would not need to be excluded from the analysis.
Apart from the direct comparison of the best frequencies (as defined in Kuokkanen et al. (2010) by the half-width at the half-height, derived from its peak value), we also cross-correlated the frequency response curves to ipsilateral and contralateral stimulation at each recording site to determine if there were a relationship between interaural mismatches in frequency tuning and ITD tuning. The BF response curves were linearly interpolated between stimulus frequencies to about 1–5 Hz resolution to ensure a smooth cross-correlation curve. The frequency lag at the maximum of the cross-correlation was used to describe the shift between the tuning curves. The statistical significance of the BF shift of each group was tested with Student's t test with respect to the null hypothesis (no BF shift) and Šidák-corrected for multiple comparisons.
We evaluated the effects of monaural occlusion on BF and best ITD using linear mixed-effects modeling in R statistical software (lme4 package). Linear mixed-effects models were chosen because they are robust to missing data and unbalanced designs. All earmold plugs had been placed to the left ear, which simplified analysis. Thus, Lbrain Lstimulus means recording from the side with plug (Lbrain), with ipsilateral stimulation (Lstimulus), while Lbrain Rstimulus means recording from the side with plug, with contralateral stimulation. The data were divided into low (≤5 kHz) and high (>5 kHz) BF groups based on the stimulus frequency of the maximum SNR and stimulus frequency of the ITD tuning curve. The SNRmax of the neurophonic potential was incorporated as the response variable in each model, and recording side, BF-population (high or low BF frequency regions of NL), and their interaction (side × BF-population) were included as fixed effects. We incorporated subject ID as a random effect in each model to account for individual variation. Post-hoc analyses of significant fixed effects were performed by computing the estimated marginal means (i.e., least-square means) of linear mixed-effects models and examining pairwise comparisons among groups using the R package emmeans. We tested for significant differences in the slopes and intercepts of ITD maps measured in the ipsilateral and contralateral NL using ANCOVA tests followed by post-hoc pairwise comparisons of significant effects.
Results
Since the development of stable ITD cues coincides with development of brainstem auditory circuits, we hypothesized that ITDs might be modified by experience (Fig. 2). We used the accessible map of ITDs in the NL to test this hypothesis in young owls raised with unilateral ear inserts of different types (Knudsen et al., 1984a; Gold and Knudsen, 1999; Popescu and Polley, 2010; Keating et al., 2015; Anbuhl et al., 2017, 2022). All inserts attenuated as well as delayed the acoustic stimulus to the manipulated ear (Knudsen et al., 1984a; Gold and Knudsen, 1999; this study, see the “Materials and Methods” section). It is important to note that ITD coding by the brainstem NL is based on phase-locked neural inputs whose preferred phase is basically invariant with sound level (for review, see Heil and Peterson (2015); barn owl: Köppl, 1997b, Wagner et al., 2005). Thus, although acoustic attenuation will additionally delay neural onset responses (Wagner et al., 2005), the effect of attenuation on ongoing responses is predicted to be negligible here, and the dominant effect of the ear inserts is assumed to be the delay that results in a unilateral, frequency dependent phase shift of the neural inputs to the developing NL on both sides of the brainstem. We predicted that, if compensation were to occur, the manipulated-ear inputs should become “faster”, i.e., show a decrease in delay (compared to unmanipulated ears) when the plug is removed. If compensation were comprehensive, changes of up to 50 µs are to be expected, which corresponds to the typical acoustic delay effected by the ear inserts for frequencies up to 5–6 kHz (Knudsen et al., 1984a; Gold and Knudsen, 1999; see the “Materials and Methods” section). At still higher frequencies, the introduced acoustic delays are smaller.
We were able to test the hypothesis that delays might be modified by experience because we had previously characterized the normal ITD map (McColgan et al., 2014; Carr et al., 2015). In a typical recording through NL in an unmanipulated animal, an electrode penetration from dorsal to ventral in the brainstem reveals an ordered representation beginning with best ITDs to far contralateral stimuli. These best ITDs progress towards more frontal ITDs, then shift to a smaller representation of ipsilateral space, covering a total range of ∼200 µs ITD (Carr et al., 2015) (Figs. 2, 3A). The transition from contralateral to ipsilateral ITDs is associated with recordings of best ITDs of about 0 µs. In our experiments, this specific location was always marked by a small electrolytic lesion for later reconstruction (Fig. 3B,C). On both sides, we identified the 0 µs points as precisely as possible using online analyses (mean ± SD = −3.8 ± 5.5 µs, n = 61) prior to making a lesion. The best ITDs were confirmed off-line using the measures of SNR (Kuokkanen et al., 2010) and found to be close to the online measurements (mean ± SD = −5.7 ± 10.0 µs, mean difference 1.8 ± 7.3 µs, n = 61). We then reconstructed the lesion locations to determine if there were changes in the map of ITDs.
Predicted shift in ITD representation. A, Schematic connections from the auditory nerve (AN) to the first-order nucleus magnocellularis (NM) and nucleus angularis (NA), and to the NL, on the right hand side (RHS) of the brainstem, from (Ashida and Carr, 2011). B, A schematic map of NL (RHS of the brainstem), with delay line axons from NM interdigitating in NL to create the maps of ITD. The dashed red line marks the position of 0 µs best ITD. Medial in the brainstem is to the left. C, D, Diagrams of the RHS ITD maps in NL. Blue shading shows the locations of the best ITDs corresponding to sound sources in the ipsilateral hemifield (i.e., originating from the right side), red shading shows contralateral ITDs, and the pale diagonal band shows 0 µs or frontal ITD, from (Carr et al., 2015). C, Normal map also predicted if there is no compensation for the altered auditory experience. D, Predicted shift in ITD representation after insert removal if there were compensation for ipsilateral plug insert rearing.
NL is tonotopically organized, with low best frequencies mapped caudally, and progressively higher frequencies rostrally. A band across the mediolateral dimension of NL, oriented along a diagonal of about 45° to the midline, represents a single frequency and is referred to as an isofrequency band (Takahashi and Konishi, 1988b; Carr and Konishi, 1990; Carr et al., 2013, 2015). For each bird, we made multiple penetrations within each isofrequency band to characterize the ITD maps both ipsilateral and contralateral to the brain side that had experienced the earmold plug (Figs. 2, 3B,C). In Carr et al. (2015), we modeled the normal ITD map (red dashed line in Fig. 2B), obtaining the regression shown as black lines in Figure 3D–F. This normal ITD map is characterized by a steady shift in the position of 0 µs best ITD along the isofrequency axis (Carr et al., 2015), which can be described by a regression of y = −0.74x + 0.78, R2 = 0.64. These data enabled us to look for a change in the map of ITD after raising an owl with an ear insert. When tested with the ear insert removed, we hypothesized that a “faster” ipsilateral input should cause the representation of frontal space, around 0 µs ITD, to be mapped significantly more ventrally within NL (Fig. 2D).
Owls raised with foam plugs and unilateral acoustic filtering devices inserted after P20 did not show altered ITD maps
In pilot experiments, to assess the role of experience in the development of sensitivity to ITDs in NL, young owls were fitted with unilateral acoustic filtering devices, from P20 at the earliest (for details, see the “Materials and Methods” section and Table 1) (Gold and Knudsen, 1999; Köppl et al., 2012). The acoustic filtering devices increased the path length of sound reaching the affected ear and changed the resonance properties of the ear canal while still providing a low-impedance pathway to the tympanic membrane (Gold and Knudsen, 1999). A disadvantage of these filtering devices is that their large size did not permit earlier insertion in the ear canal. After we had used these unilateral acoustic filtering devices in 8 barn owl chicks, we found maps of ITD (Fig. 3D, blue lines) on both brain sides that were similar to the normal maps (Fig. 3D, black line). In the owls raised with these unilateral acoustic filtering devices from P20 at the earliest, the maps of ITD were not statistically different to the normal owls. An ANCOVA with a z-test for slopes showed that the two ITD maps were not significantly different from each other or from the normal map (slopes: F2,42 = 0.029, p = 0.972; y-intercepts: F2,44 = 0.862, p = 0.425). The ITD maps did, however, appear to be less organized, i.e., the regressions explained less of the variance (Fig. 3D) (Köppl et al., 2012).
Owls raised with a unilateral earmold plug inserted at P14–17 had altered maps of ITD
To further test for the developmental windows of auditory plasticity of brainstem and midbrain auditory circuits, we mapped ITDs in an additional 6 owls raised with earmold plugs first inserted around P14–17 (Fig. 3E,F). In the age-matched control owls, as described above, ITD maps show a steady shift in the position of 0 µs best ITD in the mediolateral dimension within any isofrequency slab (black reference lines in Fig. 3E,F). In the owls raised with the unilateral earmold plugs, the maps of ITD were, again, not statistically different to the normal owls when we combined data from all isofrequency bands. An ANCOVA test indicated that neither the slopes (F1,40 = 0.228, p = 0.635) nor the intercepts (F1,41 = 1.56, p = 0.094) of the maps on the earplugged brain side (Fig. 3F) and the contralateral side (Fig. 3E) were different from those of the normal owls.
It is important to note that the absence of a significant effect of an earmold plug on the ITD map depends on our combining results from all isofrequency bands. We had combined these results because, in normal birds, the range of ITDs does not change with frequency, at least within the 3–8 kHz range examined (Carr et al., 2015). However, the age at which we inserted the earmold plugs (around P14–18) was around the age when the owl chicks could already hear frequencies between 500 Hz–5 kHz; owls only begin to hear frequencies above 5 kHz after P16 (Köppl and Nickel, 2007; Kraemer et al., 2017). We therefore separated the results by the BF, i.e., position along the tonotopic axis. We found that the earplugged ITD maps from the 3–5 kHz regions of NL (Fig. 3F, solid blue line) were not different to the normal ITD map (Fig. 3F, black line). Interestingly, the ITD maps from 5 to 7.5 kHz were shifted ventrally (Fig. 3F, solid green line). For earplug side recordings, there was a significant effect of the intercept (F2, 40 = 16.243, p = 6.855 × 10−6), but not the slopes (F2, 38 = 0.6852, p = 0.5105). Post-hoc pairwise comparisons revealed statistically significant differences between the intercepts in which the high-BF ITD map was significantly lower than the normal ITD map (p < 0.0001) and the low-BF ITD map (p = 0.0016). There was no significant difference between the normal and low-BF population ITD maps (p = 0.4519). The ventral-ward shift of about 30% for the high-BF population earplugged ITD map in Figure 3F (solid green line), as indicated by its deviant y-intercept, should correspond to a shift in best ITD of approximately 50 µs (Carr et al., 2015). These results were consistent with the prediction that only a brief period of auditory experience was required for the neural delay lines in the ITD circuit to mature. These results were also consistent with the predicted ventral-ward shift in the map of ITD on the side with the plug (Fig. 2D). In summary, the maps ipsilateral to the plug shifted depending on the tonotopic region examined (Fig. 3F); only ITD maps in the high-BF region (>5 kHz) of NL showed plastic shifts.
In contrast to our results showing a shift of ITD maps in NL on the earplugged side of the brain, in the contralateral NL the maps of ITD were not statistically different from the normal maps, even when we segregated the low BF region from the high BF region (Fig. 3E, solid lines). There was neither statistically significant difference in the slopes (F2,39 = 0.366, p = 0.696) nor the intercepts (F2,41 = 1.408, p = 0.256) of the regression lines for high-BF and low-BF populations compared to the normal ITD map of the control owls (Carr et al., 2015).
In summary, we hypothesized that the maps of ITD in the ipsilateral and contralateral NL would shift in opposite directions, but instead observed a shift confined to only one side of the brainstem.
Balance of ipsi- and contralateral inputs
It might be possible to create ITD shifts of the kind we observed through imbalanced inputs to the coincidence detectors from each side. We therefore compared both the BF and the ITD sensitivity of the neurophonic responses in low-BF (≤5 kHz, unshifted maps) and high-BF (>5 kHz, shifted map) regions of the NL ipsilateral to the plug with matching responses from the contralateral NL, and with measurements from age-matched controls (Fig. 4). Computational modeling and analysis of monaural neurophonic responses have shown that the neurophonic potential in NL is generated by hundreds of statistically independent sources, suggesting the afferent delay lines from the nucleus magnocellularis are its primary contributors (Kuokkanen et al., 2010, 2013, 2018). We therefore used the neurophonic to quantify the strength of the response from each ear, using the SNR of the monaural neurophonic responses to tones (Fig. 4B–E). These analyses helped to separate the role of location in brain (left NL, ipsilateral to the plug vs right NL, contralateral to the plug) from experience (recording from ipsi- and contralateral inputs within each NL, where one input originated from the plug side, and the other from the unmanipulated side).
Plug effects on the strength of the inputs to NL. A, Diagram of earmold plug placement in the left ear, with schematic transverse section through the brainstem showing left (plug) ear inputs (in red color) to the left nucleus magnocellularis (NM). NM projects to both ipsilateral (red, Lbrain Lstimulus) NL and to the contralateral (magenta, Rbrain Lstimulus) NL. The right (no plug) ear inputs to the right NM are shown in blue. Right NM projects to ipsilateral (blue) NL, and to contralateral (cyan) NL. Color codes are maintained in the figure to facilitate comparisons. B, Exemplar iso-level neurophonic response curves (solid lines) recorded at a single location in right NL to stimulation of either the left ear (magenta, Rbrain Lstimulus), or right ear (blue, Rbrain Rstimulus). SNR (dashed lines) computed from each recording (SNRmax of 17.6 dB for Lstimulus, 18.2 dB for Rstimulus). C, SNRmax for all BF responses recorded on both the left (plug; red and cyan) and right (no plug; magenta and blue) sides of the brain. Responses to stimuli on the same (=left) side as the earmold plug use filled symbols. Responses to stimulation of the right ear (no plug) use open symbols. Data on age-matched controls are shown in gray. The vertical dashed line divides the plot by frequency into low-BF responses (≤5 kHz) and high-BF responses (>5 kHz). D, Box plots of the same SNRmax data as in C (same color code) but separated for low- and high-BF populations. E, a comparison of estimated marginal means computed from the linear mixed-effects model analysis of all neurophonic SNRmax data from animals with earmold plugs and age-matched controls, across BF groups. Responses recorded on the brain side ipsilateral to the plugged ear evoked by stimuli presented to the plugged ear, LbrainLstimulus, were significantly different from RbrainRstimulus (p < 0.0001), from RbrainLstimulus (p = 0.0004) responses and from age-matched controls (p < 0.0001). Additionally, the SNRmax of NL neurophonics recorded from the left side of the brain (ipsilateral to the plugged ear) and evoked by ipsilateral stimulation (Lbrain Lstimulus) were significantly different to those recorded from the left side of the brain and evoked by contralateral stimulation (Lbrain Rstimulus, p = 0.0142 ). Plots of estimated marginal means represent group means adjusted for individual variation via the random effect incorporated in the linear mixed-effects model. Confidence intervals for estimated marginal means are indicated by blue bars; pairwise comparisons are shown as red arrows with significant differences between groups indicated by nonoverlapping arrows. F, Difference (in octaves) in BF for neurophonic responses to monaural ipsi- or contralateral stimulation (example in Fig 4B, solid lines), separately for NL on left (plug, red) side and right (no plug, blue) side of the brain, for all recordings, and age matched controls (gray line marks median). Positive difference corresponds to BF (Lstimulus) > BF(control, Rstimulus) independent of the recording side. No BF differences were significantly different from zero. G, To measure the difference in the frequency tuning, we correlated ipsi- and contralateral SNR tuning curves (example in Fig. 4B, dashed lines). The peaks of the correlation curve show a significant positive shift at the population level of high BF left brain (>5 kHz, Lbrain) such that BF (LbrainLstimulus)−BF(LbrainRstimulus) = 0.029 ± 0.008 octaves (mean ± SEM), p = 0.004 (Student's t test, Šidák-corrected for multiple comparisons). No other BF differences were significantly different from zero.
We assessed the influence of rearing with a unilateral earmold plug on the SNRmax of neurophonic responses to monaural stimulation at BF using a linear mixed-effects model incorporating recording side (control, plug, and no plug, with plug/no plug groups each divided into recordings from ipsi- and contralateral inputs on both sides of the brainstem as described in the “Materials and Methods” section), BF-population (above or below 5 kHz), and the interaction of side and BF-population as fixed factors, and subject ID as a random effect. The linear mixed-effects model showed a strong effect of the earmold plug on SNRmax (p < 0.0001, Table 2). Post-hoc pairwise comparisons between groups with a Tukey adjustment of p-values for multiple comparisons showed that ipsilateral inputs to NL on the same side as the earmold plug (Lbrain Lstimulus in Fig. 4A) were significantly weaker than the ipsilateral inputs to NL on the side contralateral to the earmold plug (Rbrain Rstimulus; p < 0.0001) and also weaker than the age-matched controls (p < 0.0001; Fig. 4D,E). Similarly, ipsilateral inputs to NL on the plug side were weaker than contralateral inputs (Lbrain Rstimulus; p = 0.0142). Although NM neurons project bilaterally, recordings from their axonal arbors showed reduced SNRs on the plug side (Lbrain Lstimulus vs control, p < 0.0001), but not on the unmanipulated side (Rbrain Lstimulus vs control; p = 0.467) (Fig. 4E). In summary, the strength of inputs from the manipulated ear to coincidence detectors in NL on the plug side was reduced. This contribution was most apparent in the high BF regions of NL (Fig. 4C–E).
Main effects of Side (plug or no plug) and BF-Population (above or below 5 kHz) and their interaction on SNRmax
The stereausis hypothesis predicts that ITD tuning may be altered if ipsilateral and contralateral inputs originate from different cochlear positions, with different frequency tuning (Shamma et al., 1989; Pena et al., 2001; Carr et al., 2009; Plauška et al., 2017). We therefore investigated the relationship of interaural mismatches in frequency tuning and plug side (Fig. 4F,G). There was no significant difference in the ipsi- and contralateral BFs themselves within any group (Fig 4F). We also assessed the respective skewness of the ipsi- and contralateral tuning curves with their cross-correlation—if the curves had similar shapes, their cross-correlation would peak at zero frequency shift. However, the peaks of the cross-correlation curves were not always centered at zero: The ipsilateral tuning was significantly higher than contralateral tuning in high-BF plug side recordings (Fig. 4G). Such a shift might indicate a frequency imbalance of the inputs from NM to NL. In principle, such a frequency mismatch could be large enough to affect ITD tuning. However, the observed BF mismatches, based on NM latencies in Köppl (1997a), only could account on average for −5.6 µs ITD shifts (±3.8 µs SD, range: −9.4 µs to 0.3 µs, Student's 2 population t test with respect to the high-BF control population, p = 0.0003, N = 9). Our data thus suggest that frequency tuning mismatches could only contribute to, but never fully explain, the 50 µs shifts in ITD tuning associated with rearing with a monaural earmold plug.
We also quantified the binaural ITD tuning at both plug and no-plug sides, using SNRmax to determine if there were a loss of synchrony associated with plug-rearing (Fig. 5E,F). The best ITD of each recording location was determined from the variance of the cyclic-mean signal (Kuokkanen et al., 2010, 2018), which corresponded to the circular-mean direction (mean phase) of the variance as a function of ITD (Fig. 5B,D). As before, we used a linear mixed-effects model incorporating recording side (control, plug, or no plug), BF-population (above or below 5 kHz), the interaction of side and BF-population as fixed effects, and subject ID as a random effect. A statistical test of the model showed significant effects of side (p = 0.0176) in which the SNRmax values of recordings from the plugged side were significantly lower than those from the unmanipulated side (p = 0.0273, Fig. 5G). Thus, rearing with a unilateral earmold plug could be associated with a decrease in the SNR (Table 3).
Plug effects on the ITD responses in NL on both sides of the brain. A, The mean power spectral density (PSD) of the binaurally driven neurophonic response in NL, for recording from the unmanipulated side, at favorable (blue curve) and unfavorable (gray curve) ITDs. Note the large “signal” peak (1.1 × 10−2 mV2/kHz) at the stimulus frequency for the favorable ITD (arrow at 6.1 kHz) and the average level around the peak (“noise”: 7 × 10−5 mV2/kHz). Note that lines in the plot are 3-point smoothed. For the PSD of the response to unfavorable ITD, the signal peak height was 11 × 10−5 mV2/kHz and the noise level was 3 × 10−5 mV2/kHz. B, Exemplar neurophonic ITD curve (solid blue line, amplitude of the cyclic mean response) and SNR (gray dashed line), with BF tones as the stimulus, and responses recorded on the right (unmanipulated) side of the brain. The ITD curve was obtained from the variance of the sustained activity (onset excluded) across trials, where 100% corresponds to 0.79 mV2 in this example. The SNR is based on the signal and noise levels of the response PSDs to respective stimulus ITDs (examples in A), maximum SNR corresponds to 22.4 dB. Stimulation at 50 dB SPL; stimulus frequency 6.1 kHz. C, Exemplar mean PSD of the binaurally driven neurophonic response in NL, for recording from the side ipsilateral to the plug, at favorable (red curve) and unfavorable (gray curve) ITD. Note the large “signal” peak (7.9 × 10−2 mV2/kHz) at the stimulus frequency for the favorable ITD (arrow at 5.8 kHz) and the average level around the peak (“noise”: 6 × 10−4 mV2/kHz). For the PSD of the response to unfavorable ITD, the signal peak height was 8 × 10−4 mV2/kHz and the noise level was 3 × 10−4 mV2/kHz. D, Exemplar neurophonic ITD curve (solid red line, amplitude of the cyclic mean response) and SNR (gray dashed line), with BF tones as the stimulus, and responses recorded on the left (plugged) side of the brain. 100% corresponds to 6.3 mV2 in this example (two example PSDs are shown in C), and the maximum SNR corresponds to 21.1 dB. Stimulation at 50 dB SPL; stimulus frequency 5.8 kHz. E, Maximum SNRs for all ITD-sensitive neurophonic recordings from animals raised with an earmold plug, separated according to brainstem side relative to the manipulated ear (Lbrain: ipsilateral to the plugged ear, red; and Rbrain: contralateral to the plugged ear, blue). Age-matched control measures of maximum SNR are also shown (gray). The vertical dashed line divides the plot into high (>5 kHz) and low (≤5 kHz) BF regions. F, Box and whisker plots of all maximum SNR values computed from ITD-sensitive neurophonic responses from animals raised with a monaural earmold plug. Data recorded from the left (plugged) brainstem side (red, Lbrain), right side (blue, Rbrain), and age-matched controls (gray). Lbrain recordings were significantly different from Rbrain (p = 0.027, indicated on figure by asterisks). G, Comparison of the estimated marginal means for the main effect of recording side on SNRmax, derived from linear mixed-effects modeling. Confidence intervals for estimated marginal means are represented as blue bars; red arrows indicate pairwise comparisons between groups where overlap of arrows indicates statistical nonsignificance.
Main effects of Side (plug or no plug) and BF-Population (above or below 5 kHz), and their interaction on ITD
Discussion
The sensitivity to binaural delays can be modified when an ear is plugged prior to auditory experience (Knudsen et al., 1984a; Popescu and Polley, 2010;; Kumpik and King, 2019); Thornton et al., 2021. We examined the effects of these manipulations at the site of the first binaural comparisons in barn owls because of the reproducible nature of the ITD maps in NL (Carr et al., 2015). There were long lasting changes in these maps, associated with a brief window for plasticity around the onset of hearing. Furthermore, changes were restricted to the side of the hearing loss, suggesting independent regulation of ipsilateral and contralateral circuits for the detection of ITD.
Effects of monaural occlusion
Young owls learn to associate ITDs with corresponding locations in space (Keller and Takahashi, 1996; Poganiatz and Wagner, 2001). If an ear is occluded during the first two months of life, owls were able to recover normal localization accuracy while plugged (Mogdans and Knudsen, 1993; Brainard and Knudsen, 1995; Bergan and Knudsen, 2008). When the plug was removed, owls showed errors, but recovered localization (Knudsen et al., 1984b). These behavioral changes are associated with significant remodeling of inferior colliculus circuits (Knudsen, 2002).
Knudsen's experiments showed different sites and time constants for plasticity in the circuits for detecting both ITDs and interaural level differences (Mogdans and Knudsen, 1993; Gold and Knudsen, 2000b). Gold developed a phase-shifting acoustic filtering device to examine the plasticity in the barn owl ITD pathway. Rearing with this device inserted monaurally caused remodeling of the circuits within the inferior colliculus (Gold and Knudsen, 2000a,b). When we raised young owls, using their paradigm, and with device insertion from P20, we found no effect of altered auditory experience in the brainstem NL (Köppl et al., 2012). These results left unanswered the question of whether the brainstem representation of ITD could be modified by experience. Although the ear canals of younger owls were too small for the acoustic filtering device, we were able to induce a monaural conductive hearing loss and a temporal (phase) shift, beginning around P14–18 using an earmold plug. The extent of the shift in the ITD map was about 50 µs, consistent with the delay imposed by these plugs (Knudsen et al., 1984a; this study). Thus, our data suggest different critical periods for binaural integration in the brainstem and midbrain.
Multiple critical periods for binaural integration have been reported in other animal models of monaural occlusion (Popescu and Polley, 2010; Rosen et al., 2012; Polley et al., 2013; Keating et al., 2015; Mowery et al., 2015; Anbuhl et al., 2022). Monaural deprivation typically weakens the effect of inputs from the deprived ear and disrupts binaural integration. Varying the age of hearing loss onset revealed brief critical periods for membrane and firing properties, as well as inhibitory synaptic currents in gerbil auditory cortex (Mowery et al., 2015), while in the pallid bat, auditory cortex recordings revealed different maturation rates for the inhibitory inputs that shape binaural responses (Razak and Fuzessery, 2007). In ferrets, unilateral conductive hearing loss produced long-lasting deficits in binaural hearing (Moore et al., 1999; Kumpik and King, 2019).
The effects of monaural occlusion studied here are distinct from deafferentation, or ear removal, which is associated with rapid and dramatic neuron death after the removal of afferent input to the cochlear nucleus in both young mammals and birds [for review, see Parks et al. (2004)]. Monaural occlusion has been studied in chicks (Tucci and Rubel, 1985), barn owls (Knudsen et al., 1984a; Keuroghlian and Knudsen, 2007) and in mammals, principally in ferrets and gerbils (Moore and King, 2004; Moore, 2009; Popescu and Polley, 2010; Kral et al., 2013; Caras and Sanes, 2015; Sinclair et al., 2017; Rubio, 2020). In ferrets, like barn owls, monaural occlusion during infancy led to compensatory changes in auditory spatial tuning that tended to preserve the alignment between the neural representations of visual and auditory space in the superior colliculus (King et al., 1988). The key finding was that experience-induced plasticity allows the sound localization circuits to be customized to individual characteristics, such as the size and shape of the head and ears, and to compensate for natural conductive hearing losses, including those associated with middle ear disease in infancy (King et al., 2000).
In gerbils, monaural occlusion has well-defined effects upon the brainstem circuits for the detection of ITD. Experiments with earplugs, monaural deafening and noise rearing in gerbils support a role for the experience-dependent segregation of inhibition to MSO cell bodies (Kapfer et al., 2002; Maier et al., 2008; Werthat et al., 2008; Sinclair et al., 2017). Recordings from the MSO's target, the nucleus of the lateral lemniscus, related ITD sensitivity to the maturation of inhibitory inputs to the MSO (Seidl and Grothe, 2005). Thus, both in vivo recordings from the dorsal nucleus of the lateral lemniscus and studies of MSO support refinement in the days after hearing onset, like that observed for barn owl NL.
Development of auditory sensitivity
To understand the effects of conductive hearing loss, it is important to know when animals can hear. In both birds and mammals, auditory responses typically begin with a restricted range of low to mid-frequencies (Romand, 1979; Brugge, 1983; Arjmand et al., 1988; Rubel et al., 1988; Sarro and Sanes 2010). The rates of threshold maturation differ, depending upon where the animal falls along the altricial to precocial spectrum. Some mammals, such as guinea pigs (Dum, 1984) and humans (review in Werner (2007)), are born with functioning auditory systems. Other mammals, such as gerbils (McFadden et al., 1996), ferrets (Moore, 1982), and cats (Walsh et al., 1986a,b), are considered deaf at birth and altricial with respect to hearing (Pujol and Hilding, 1973; Cant, 1998; Wess et al., 2017).
Birds show a low- to high-frequency pattern of auditory maturation [review in Kubke and Carr (2000, 2006)]. Precocial birds respond to sound in the egg (Konishi, 1973; Gray and Rubel, 1985; Jones and Jones, 1995; Sato and Momose-Sato, 2003; Mann and Kelley, 2011; Rivera et al., 2018) with adult-like thresholds to low and middle frequencies by hatching (Saunders et al., 1973). Altricial bird hearing matures after hatching (Aleksandrov and Dmitrieva, 1992; Brittan-Powell and Dooling, 2004; Köppl and Nickel, 2007). When barn owl chicks hatch, they are insensitive to sound, and first show auditory responses to 1–2 kHz around P4–P7 (Köppl and Nickel, 2007; Kraemer et al., 2017). Sensitivity increases with age, with neural responses to 12 kHz appearing 2–3 months after hatching. Thus, owls experience a prolonged period of development that coincides with the maturation of the skull and ruff (Haresign and Moiseff, 1988; Hausmann et al., 2010).
The slow maturation of barn owl hearing allowed us to identify a critical period for auditory experience in the NL; we were able to insert an earmold plug when the young owls could hear frequencies up to 1–5 kHz (about 2 weeks after hatching) but were insensitive to higher frequency sounds (Köppl and Nickel, 2007; Kraemer et al., 2017). Since the maps of ITD were only shifted in regions of NL with BFs above about 5 kHz, this suggests that experience is required for final maturation of the ITD circuit. These findings are consistent with studies of the development of phase locking in the barn owl's auditory nerve, in which phase locking emerged with hearing onset, in a low-to-high BF progression, over a prolonged period (Köppl, 2007). There are also strong parallels with results from the work on the gerbil brainstem, where activity-dependent refinement occurs in the first days after hearing onset. Delays in maturation with white noise exposure suggest that not only activity, but patterned activity, is required for the maturation of ITD tuning (Seidl and Grothe, 2005).
Modification of ITD circuits
Our data suggest that the inputs to NL are plastic within a narrow time window around the onset of normal auditory experience, and furthermore, that the effects of monaural manipulation of acoustic experience appear predominantly ipsilateral. We had predicted that an acoustic delay caused by conductive hearing loss in one ear might be compensated for by a reduction in neural delay in the inputs to NL. We reasoned that these “faster” inputs could be detected after ear insert removal by a change in the ITD map, as a downward shift of the location of 0 µs best ITD in the ipsilateral map, and a matching upward shift on the contralateral side. The results were more complex than predicted. Those parts of NL due to respond to frequencies that the owl could not hear at the time of plug insertion (i.e., those with best frequencies above 5 kHz) proved to be most plastic. Furthermore, only the ITD maps in the NL ipsilateral to the manipulated ear changed. One possible explanation for the prominence of ipsilateral changes is that ipsilateral conduction times may be more readily adjusted by shortening axonal projections, for example. In contrast, speeding up of the contralateral projection from the plugged side might be harder to achieve because these axons are already thicker to compensate for the longer distance (Seidl et al., 2010).
Identification of the mechanisms underlying modification of ITD circuits will require further study. Our analyses support limited compensation from misaligned frequency tuning or stereausis (Shamma et al., 1989; Pena et al., 2001; Carr et al., 2009; Plauška et al., 2017). According to the stereausis hypothesis, differences in wave propagation along the cochlea can provide delays necessary for coincidence detection if the ipsilateral and contralateral inputs originate from different cochlear positions, with different frequency tuning (Shamma et al., 1989). The stereausis model has not been supported in barn owls (Pena et al., 2001), alligators (Carr et al., 2009), and gerbils (Plauška et al., 2017), but could apply to circuit development. Models show that the necessary coherence in the signal arrival times could be attained during development, allowing learning to select connections with matching delays from among a broad distribution of axonal delays (Gerstner et al., 1996; Kempter et al., 2001). Our observation of a decreased SNR (i.e., decreased coherence of the population activity of NM axons in NL) on the plug side may indicate a suboptimal selection process.
The effect of side further suggests the action of both or either of two mechanisms previously observed in gerbils: altered inhibition and/or changes in myelination. Gerbils show local and experience dependent changes in the myelination and conduction velocity of inputs to the superior olive (Lehnert et al., 2014; Ford et al., 2015; Sinclair et al., 2017; Stange-Marten et al., 2017; Dutta et al., 2018), and similar mechanisms may apply to the developing barn owl circuit. Changes in inhibition could potentially compensate for the small phase delays measured in the barn owl NL, especially because inhibitory circuits allow for monaural effects; the superior olivary nuclei are the sole source of unilateral input to the NL (Carr et al., 1989; Burger et al., 2005; Kraemer and Carr, 2019). In chicken, the avian superior olivary nuclei mutually inhibit each other (Burger et al., 2005, 2011). Thus, loud sounds in one ear should increase inhibition on the ipsilateral side, decrease inhibition on the contralateral side, and potentially alter response latency (Burger et al., 2011).
Footnotes
We acknowledge Dr. E. Brittan-Powell for help with calibration and ABR recordings, and Dr. E. Smith for help with equipment and calibration. We thank Dr. Hermann Wagner and Sandra Brill for their help in measuring the neurophonic and Dr. Sahil Shah for establishing the lesion measurement paradigm. This research was sponsored by the National Institute on Deafness and Other Communications Disorders (NIDCD) grant DC-000436 and DC-019341 (CEC). It was also supported by the US-American Collaboration in Computational Neuroscience “Field Potentials in the Auditory System” as part of the National Science Foundation/NIH/French National Research Agency/German Ministry of Education and Research/United States-Israel Binational Science Foundation Collaborative Research in Computational Neuroscience Program, Research Grant 01GQ1505A. Research at the Carl von Ossietzky University of Oldenburg was supported in part by awards from the Hanse-WissenschaftKolleg, and the Alexander von Humboldt foundation.
↵* Dual senior authors.
The authors declare no competing financial interests.
- Correspondence should be addressed to Catherine Carr at cecarr{at}umd.edu or Paula Kuokkanen at p.kuokkanen{at}biologie.hu-berlin.de.
