Author response: Development of frequency tuning shaped by spatial cue reliability in the barn owl’s auditory midbrain

Abstract

Full text Figures and data Side by side Abstract Editor's evaluation Introduction Results Discussion Methods Data availability References Decision letter Author response Article and author information Metrics Abstract Sensory systems preferentially strengthen responses to stimuli based on their reliability at conveying accurate information. While previous reports demonstrate that the brain reweighs cues based on dynamic changes in reliability, how the brain may learn and maintain neural responses to sensory statistics expected to be stable over time is unknown. The barn owl’s midbrain features a map of auditory space where neurons compute horizontal sound location from the interaural time difference (ITD). Frequency tuning of midbrain map neurons correlates with the most reliable frequencies for the neurons’ preferred ITD (Cazettes et al., 2014). Removal of the facial ruff led to a specific decrease in the reliability of high frequencies from frontal space. To directly test whether permanent changes in ITD reliability drive frequency tuning, midbrain map neurons were recorded from adult owls, with the facial ruff removed during development, and juvenile owls, before facial ruff development. In both groups, frontally tuned neurons were tuned to frequencies lower than in normal adult owls, consistent with the change in ITD reliability. In addition, juvenile owls exhibited more heterogeneous frequency tuning, suggesting normal developmental processes refine tuning to match ITD reliability. These results indicate causality of long-term statistics of spatial cues in the development of midbrain frequency tuning properties, implementing probabilistic coding for sound localization. Editor's evaluation This research advance shows that if juvenile barn owls experience experimentally altered interaural time differences – the binaural cue used for localizing sounds in the horizontal plane – the frequency tuning properties of neurons in the space-mapped region of the midbrain undergo adaptive changes. The results therefore suggest that the statistics of sound stimulation can influence the sensitivity of auditory midbrain neurons to a fundamental stimulus feature in the developing barn owl brain. These findings will be of interest to the fields of developmental and sensory neuroscience. https://doi.org/10.7554/eLife.84760.sa0 Decision letter Reviews on Sciety eLife's review process Introduction To accurately and efficiently perceive and react to the environmental scene, the brain must rely on the sensory cues that are naturally most reliable. The ability of the brain to quickly weigh ongoing reliability of sensory cues within different modalities has been well documented (Jacobs and Fine, 1999; Rosas et al., 2005; Fetsch et al., 2011; Dacke et al., 2019). However, in instances where sensory cue reliability is anticipated to be relatively stable and predictable, predetermined weights of sensory cues may be optimal. An example of this neural operation is particularly seen in human speech processing, where reliable phonetic properties are stable across speakers (Holt and Lotto, 2006; Iverson et al., 2003; Toscano and McMurray, 2010), and may only change over the course of decades (Toscano and Lansing, 2019). While extensive research has shown changes in neural responses induced by sensory statistics (David et al., 2004; Dean et al., 2005; Fetsch et al., 2011), properties and fundamental mechanisms of this adaptive coding are important for understanding whether and how brain development is influenced by specific high-order statistics of sensory cues. In this report, we used the barn owl as a model organism of sound localization to test whether and how the brain develops and adapts to changes in stimulus reliability by altering the tuning properties of sensory neurons in a manner predictive of these relevant natural statistics. The barn owl is a highly specialized species, able to hunt in the dark solely using auditory stimuli (Payne, 1962). Barn owls use interaural time difference (ITD), the delay for a sound to reach one ear before the other, and interaural level difference (ILD), the difference in sound intensity between the two ears, to, respectively, compute sound location in azimuth and elevation (Moiseff and Konishi, 1981; Moiseff, 1989). ITD and ILD are represented in the barn owl’s external nucleus of the inferior colliculus (ICx), creating a topographic midbrain map of sound location (Knudsen and Konishi, 1978). The barn owl shows specialization in its ability to compute ITD from frequencies as high as 10 kHz (Wagner et al., 1987; Carr and Konishi, 1990; Köppl, 1997). The reliability of ITD is described as how corruptible the ITD cue is to concurrent sounds, which is determined by the acoustical properties of the head and the frequencies carrying the ITD cue (Keller and Takahashi, 2005; Cazettes et al., 2014; Fischer and Peña, 2017). These properties are summarized in the head-related transfer function (HRTF) (Wightman and Kistler, 1989; Poon and Brugge, 1993; Brugge et al., 1994; Brugge et al., 1994; Hartung and Sterbing, 1997; Keller et al., 1998), which describes the directional filtering that the external ears induce onto incoming sounds. These filtering properties vary across sound source location relative to the ears and the sound frequency. HRTF-based analysis has indicated that for sounds originating from frontal locations, ITD cues derived from the higher frequencies of the barn owl’s hearing range are less susceptible to corruption from concurrent sounds; while for sounds from peripheral space, ITD cues derived from low frequencies are less susceptible to corruption (Cazettes et al., 2014; Fischer and Peña, 2017). Because ITD is detected by ongoing phase locking within narrow frequency bands in the lower brainstem (Carr and Konishi, 1990) and that the HRTF may induce instantaneous effects on stimulus shape, the reliability of interaural phase difference (IPD) is the most precise assessment of ITD cue reliability across single-frequency channels (Fischer and Peña, 2011; Cazettes et al., 2014; Fischer and Peña, 2017). In accordance with this, ITD reliability is considered equivalent to IPD reliability for the purposes of this report. Consistent with the hypothesis of anticipated coding of sensory cue reliability, neurons in the ICx are tuned to the frequencies that are most reliable for their preferred ITD, even if the ongoing statistics are briefly changed (i.e. using earphones which bypass the head’s filtering properties) (Cazettes et al., 2014). A relationship between frequency tuning and ITD tuning has been reported in the midbrain and brainstem of mammalian models as well; however, this was proposed to be related to ITD neural coding and detection properties rather than anticipated ITD reliability (McAlpine et al., 2001; Hancock and Delgutte, 2004; Day and Semple, 2011; Bremen and Joris, 2013). In contrast, recent findings have shown that human spatial perception is driven by natural ITD statistics across frequencies, including ITD variability induced by concurrent sounds (Pavão et al., 2020), suggesting commonalities between humans and owls in the anticipation of ITD cue reliability based on acoustic properties of the head. Overall, these previous studies indicate that the predictive coding of natural ITD reliability is inherent in the brain of humans and owls, but directly testing development and causality of this statistical property on the frequency tuning of neurons representing auditory space and whether this coding is innately fixed or experience dependent remain open questions. In the barn owl, the facial ruff, a disc of stiff feathers that surrounds the head, acts as an external ear, modulating the gain and phase of sounds reaching the eardrums (Coles and Guppy, 1988; Keller et al., 1998; von Campenhausen and Wagner, 2006; Hausmann et al., 2009). Because the filtering effects of the facial ruff are direction and frequency dependent, we sought to assess whether the removal of the facial ruff would induce a change in the pattern of ITD reliability, and whether the barn owl’s auditory system adapts to these long-term changes. Previous reports studying the facial ruff removal in barn owls have focused on changes to the ILD (i.e. elevational) tuning, while noting few changes to the ITD tuning of neurons in the owl’s midbrain map of auditory space (Knudsen et al., 1984). However, to our knowledge, there have been no studies that assessed whether ITD reliability and frequency tuning of midbrain neurons changed after facial ruff removal. Based on these open questions, comparative analysis of ITD reliability was conducted using HRTFs from owls before and after facial ruff removal. Following this, the frequency tuning of neurons was measured in the ICx of barn owls where the facial ruff was removed during juvenile development. Additionally, the earliest-to-date recordings of the developing ICx were conducted in normal juvenile owls before the facial ruff develops but after hearing onset. We found frontally tuned ICx neurons of juvenile and ruff-removed owls were predominately tuned to frequencies lower than the observed frequency tuning in normal adult owls. These changes in both ruff-removed and juvenile owls are consistent with estimated differences in ITD reliability between normal and ruff-absent conditions. These results indicate that tuning to high frequencies of frontally tuned midbrain map neurons is developed and driven by the experience of natural auditory scenes during early life in the barn owl. In addition, recordings in the region immediately upstream of ICx confirmed that the ability to compute ITD from high frequencies (Carr and Konishi, 1990) was preserved in ruff-removed owls, but not inherited by ICx. Overall, this study demonstrates that the owl’s sound localization pathway implements an experience-dependent representation of anticipated ITD statistics in the midbrain map of space, supporting the idea of the brain’s adaptive anticipation of natural high-order sensory statistics. Results To first determine how ITD reliability changes after facial ruff removal, five HRTFs from barn owls before and after facial ruff removal were used, from the dataset originally published in von Campenhausen and Wagner, 2006. Given ITD is computed by comparing IPD across narrow frequency channels (Carr and Konishi, 1990; Fischer et al., 2011), ITD reliability is determined by IPD reliability. Thus, for the purposes of this analysis, IPD reliability underlies ITD reliability and will be used interchangeably. A signal’s phase can be corrupted by concurrent sounds, altering the IPD in a frequency-dependent manner (Keller and Takahashi, 2005; Cazettes et al., 2014; Fischer and Peña, 2017). For a given sound source and frequency, the amount of corruption varies with the location of the masker sound. IPD reliability can be computed as the inverse of the standard deviation of this variability (Trommershäuser et al., 2011; Cazettes et al., 2014; Fischer and Peña, 2017). To this end, we calculated the IPD for a given sound source across different locations of a second sound source using the HRTFs before and after facial ruff removal, on a frequency-by-frequency manner. Before ruff removal, the pattern of higher IPD reliability for high and lower frequencies in, respectively, frontal and peripheral locations (Figure 1a) was largely the same as previously reported (Cazettes et al., 2014). After ruff removal, there was a sharp decrease in the reliability of high frequencies coming from frontal locations, with minimal changes elsewhere (Figure 1b). Computing the difference in reliability between the two conditions highlights that after ruff removal, the changes in reliability were stronger for frequencies above 4 kHz coming from within ±40° of azimuth location (Figure 1c). As HRTFs can vary between individuals, we measured the variability in IPD standard deviation across owls for the normal (Figure 1d) and ruff-removed (Figure 1e) conditions. There was notably low variability across the owls, especially at the frequencies of interest. This suggests that this measurement, based on HRTF-based analysis, leads to largely common values among individual owls. Previous reports show that there is a uniform decrease in gain of sound level across frequencies after facial ruff removal (von Campenhausen and Wagner, 2006), nullifying the interpretation that this is merely due to a loss of gain specifically at these higher frequencies. Repeating this analysis under different conditions, where masker sounds were quieter (Figure 1—figure supplement 1a and b) and using prey vocalizations as potentially natural sounds (Figure 1—figure supplement 1c and d), produced qualitatively similar patterns of ITD reliability, suggesting these effects are stable across potential differences of acoustic environments. These acoustical simulations suggest that if ICx frequency tuning is driven by IPD reliability, and by extension ITD reliability, then we should expect lower frequency tuning in frontally tuned ICx neurons in owls raised without the facial ruff. Figure 1 with 1 supplement see all Download asset Open asset Head-related transfer function (HRTF)-based interaural phase difference (IPD) reliability in owls with and without facial ruff. Target and masker broadband sounds across varying azimuth locations were convolved with HRTFs from owls before and after facial ruff removal, then summed. IPD reliability (s.d.–1) was computed across frequencies and normalized for each location, then averaged across owls (see ‘Methods’) before (a) and after (b) the facial ruff removal (right hemifield shown). The difference between normal and ruff-removed IPDs indicates a substantial decrease in reliability for frequencies above 4 kHz at frontal locations (c). HRTF data from von Campenhausen and Wagner, 2006. To test whether frequency tuning in the barn owl’s ICx is shaped by ITD reliability, we modified ITD reliability by trimming the facial ruff from two barn owls as it grew in. While previous experiments have studied the effects of physical and virtual removal of the facial ruff on the spatial tuning of ICx neurons and sound localizing behavior (Knudsen et al., 1984; Hausmann et al., 2009), none of these reported an effect on neuronal frequency tuning in ruff-removed owls. Towards this goal, once these ruff-removed owls reached adulthood (6 mo of age), we performed electrophysiological recordings across the ICx. We recorded 117 ICx neurons across the two owls. ITD and frequency tunings were then assessed for each neuron. An example ICx unit is shown in Figure 2a–c. Typical ICx neurons display sharp tunings to ITD and ILD, with relatively broad frequency tuning width (>2 kHz) (Moiseff and Konishi, 1983; Takahashi and Konishi, 1986). Thus, individual ICx neurons recorded from ruff-removed owls had typical ITD and ILD tuning properties. The recorded population spanned ITD tunings of 0–230 µs, which equates to a range of 0–75° azimuth. However, on a population-wide scale, ILD tuning was correlated with ITD tuning in these owls (Figure 2d). This arises from the loss of bilateral asymmetry conferred by the facial ruff, which causes ILD to covary with elevation in normal conditions (Coles and Guppy, 1988; Moiseff, 1989; Keller et al., 1998). Consistent with previous reports, ILD covaries with azimuth after facial ruff removal (von Campenhausen and Wagner, 2006; Hausmann et al., 2009; Knudsen et al., 1984; Figure 2d). Figure 2 Download asset Open asset Inferior colliculus (ICx) neural responses in the ruff-removed barn owl. (a) Example interaural time difference (ITD) tuning curve. Yellow curve represents Gaussian fit to main peak, with the maximum of this curve termed best ITD. (b) Example frequency tuning curve. Black line indicates half-height used for determining half-width, low- and high-frequency bounds (black dots). Best frequency corresponds to the mean of the frequency bounds (red dot). (c) Example interaural level difference (ILD) tuning curve. (d) Best ILD plotted as a function of best ITD of neurons from ruff-removed (red dots) and normal owls (blue dots, from Cazettes et al., 2014). While there is no correlation between ITD and ILD tuning in normal owls (blue line), there is a correlation in ruff-removed owls (red line). The correlation between ITD and frequency was markedly different in the ruff-removed owls compared to normal adult owls. In normal adult owls, neurons tuned to ITDs near 0 µs, or frontal locations, are driven by high frequencies, while neurons tuned to large ITDs, or peripheral locations, are driven by low frequencies (Cazettes et al., 2014), exposing a significant correlation between ITD and frequency tuning (r2 = 0.75, p=5.6e-55). In the ruff-removed owls, there was a shallower slope of the linear correlation between ITD and frequency tuning (r2 = 0.18, p=1.6e-06, Figure 3). These two correlations were significantly different (t-test: t = 8.97, p=3.0e-19). The y-intercepts of the regression lines, which can be used to compare the frequency tuning of frontal neurons, were also significantly different (normal owls = 5.6 kHz, ruff-removed owls = 4.6 kHz; t = 7.56, p=4.0e-14). In addition, a Mann–Whitney U-test comparing the best frequency tuning of the frontally tuned neurons, defined here as preferring ITDs ≤ 30 µs (~10° eccentricity), produced consistent results (u = 63, p=7.3e-08). We defined best frequency as the median of a neuron’s frequency range (Figure 2b), so this suggests that the two groups cover different frequency ranges. In line with this, the frontally tuned neurons of ruff-removed owls showed little responsivity to tones above 6 kHz (Figure 3—figure supplement 1). ICx neurons of ruff-removed owls showed little ITD selectivity to frequencies outside their frequency range (Figure 3—figure supplement 2), suggesting that these neurons cannot use high frequencies for sound localization. The pattern of frequency tuning seen in the ruff-removed owls corresponds to the changes in ITD reliability, where the relationship between ITD and azimuth is approximately 3 µs/degrees (Keller et al., 1998; von Campenhausen and Wagner, 2006). Figure 3 with 2 supplements see all Download asset Open asset Different correlations between interaural time difference (ITD) and frequency tunings in ruff-removed and normal owls. Best frequency plotted as a function of best ITD of inferior colliculus (ICx) neurons from ruff-removed owls (red dots) and normal owls (gray dots, from Cazettes et al., 2014). Arrows indicate frequency bounds of each neuron; vertical lines denote frequency range. One potential cause of this change in frequency tuning could be due to disrupted ITD detection mechanisms induced by a change in gain of sound level, rather a change in ITD reliability. To test for this alternative, electrophysiological recordings of the lateral shell of the central nucleus of the inferior colliculus (ICCls), the immediate upstream region of ICx, were performed. ICCls neurons also display a topographic mapping of ITD, but, critically, are narrowly tuned to frequency, which increases along the dorsal-ventral axis, with tunings spanning the barn owl’s hearing range of 0.5–9 kHz (Knudsen and Konishi, 1978; Wagner et al., 2007). We recorded 43 ICCls neurons from the same two ruff-removed owls in adulthood and found the normal frequency tuning range (1–8 kHz) across ITD-selective neurons (Figure 4). In particular, ICCls neurons tuned to high frequencies and frontal ITDs in ruff-removed owls were observed (Figure 4). This suggests that the auditory system is still able to use high frequencies to compute ITDs, but this is not passed onto the ICx. Figure 4 Download asset Open asset Frequency tunings in the inferior colliculus (ICCls) of ruff-removed owls span across the owl’s normal hearing range. Frequency tuning of ICCls neurons of ruff-removed owls plotted as a function of their best interaural time difference (ITD). Best frequency denoted by blue dots; frequency range denoted by arrows. Black dashed lines indicate the upper and lower typical frequency range of the barn owl’s ICCls (from Wagner et al., 2007). In addition, changes in directional gain were considered as an alternative cause driving frequency tuning changes in the ruff-removed condition. Towards this end, the gain across frequency and location was computed from normal and ruff-removed HRTFs. In line with previous reports (Keller et al., 1998; Cazettes et al., 2014), gain in the normal condition was strongest for high frequencies in frontal space, with negative gain at all other locations and frequencies (Figure 5a). In the ruff-removed condition, gain remained strongest for high frequencies in frontal space, but expanded slightly to include stronger gain in more peripheral locations (Figure 5b). In contrast, the gain of lower frequencies had a more limited increase, to approximately 0 dB (i.e. no net gain change), likely because the facial ruff normally attenuates these sounds (Figure 5a, Coles and Guppy, 1988; Keller et al., 1998). These findings are consistent with previous analysis of changes in gain across frequencies following ruff removal (von Campenhausen and Wagner, 2006). This analysis suggests that the limited changes in gain across frequencies are insufficient to explain the changes in frequency tuning as high frequencies remain the most audible in frontal locations after facial ruff removal. Figure 5 Download asset Open asset Gain across frequencies in the barn owl. Gain computed for each frequency using normal (a) and ruff-removed (b) head-related transfer functions (HRTFs). Positive gain indicates an increase in the sound level, relative to the absence of the head to filter the sounds (0 dB). Negative gain indicates a decrease in sound level. Alternatively, the change in the frequency tuning of frontal neurons could be due to a remapping of ITD and ILD in a frequency-dependent manner following facial ruff removal. To test this possibility, we modeled the spatial tuning of frontally tuned neurons across frequencies. Based on HRTFs from von Campenhausen and Wagner, 2006, the ITD and ILD was computed for each spatial location sampled, across frequencies (described in ‘Methods’ section). These ITDs and ILDs were passed through a model ICx neuron tuned to 0 µs ITD and 0 dB ILD to simulate the neuron’s response across space (Figure 6a). This analysis was performed using the HRTFs of owls before (Figure 6b) and after (Figure 6c) facial ruff removal. In the ruff-removed condition, spatial tuning to elevation widened dramatically, as expected from previous reports (Knudsen et al., 1984; von Campenhausen and Wagner, 2006; Hausmann et al., 2009). However, we did not find any widening of modeled azimuthal spatial tuning nor any acoustical loss in the potential efficacy of high frequencies to compute ITD. In fact, after facial ruff removal, we noticed an increase in the potential efficacy of high frequencies for frontal space, likely due to ILD being correlated with azimuth after facial ruff removal (Knudsen et al., 1984; Figure 2d). These results further support the hypothesis that the barn owl’s frequency tuning is driven by ITD reliability rather than solely spatial tuning. Figure 6 Download asset Open asset Spatial tuning for frontal space following ruff removal does not show systematic alterations based on frequency. (a) Schematic methodology to determine spatial tuning, using 6 kHz as an exemplary tone. Interaural time difference (ITD) (upper left) and interaural level difference (ILD) (upper right) were estimated as a function of the spatial location of the sound source. Neural responses were predicted by entering the estimated ITD and ILD into modeled tuning curves for a simulated frontally tuned neuron (middle plots). The overall spatial tuning was calculated by combinatorial multiplication then normalization of these modeled responses (lower plot). (b, c) Spatial tuning maps for owls before (b) and after (c) ruff removal, displayed for five example tones. We next sought to determine the frequency tuning of juvenile owls before the facial ruff fully developed. Recordings were performed from two juvenile owls at 42 and 44 d post hatching. These time points corresponded to ongoing ruff development as the ruff feathers were still within the sheathed stalks, with the ruff not fully developing until approximately 60 d post hatching (Haresign and Moiseff, 1988). To our knowledge, these recordings are the earliest performed during the development of the barn owl’s midbrain. We were able to identify 39 ICx neurons, which showed typical topographic ITD tuning, as well as latencies and thresholds. However, we found that ILD tuning did not change with depth along the dorsal-ventral axis, as observed in normal adult ICx (Figure 7—figure supplement 1; Mogdans and Knudsen, 1993). Because the relationship between ILD and elevation is determined by the facial ruff (Knudsen et al., 1984; von Campenhausen and Wagner, 2006), this provided further evidence that the facial ruff was still underdeveloped in these juvenile owls. There was no correlation between ITD and best frequency in the juvenile owls (r2 = 0.18, p=0.28), although this may be attributed to the lack of recorded neurons tuned to ITDs >120 µs (Figure 7). Because this correlation was not significant, we confined our subsequent analyses to only frontally tuned neurons (best ITDs ≤ 30 µs) as the clearest differences between normal and ruff-removed conditions occurred in frontal space for both ITD reliability and frequency tunings. When we compared the best frequency tuning of frontal neurons, there was a significant difference between the juvenile owls and both the normal (u = 558, p=5.9e-04) and ruff-removed (u = 281, p=3.2e-03) adult owls (Figure 8a). The mean best frequency of the frontal neurons in juvenile owls was 4.9 kHz, which is between that of normal and ruff-removed adult owls. The size of these differences between juvenile owls and both normal and ruff-removed owls, measured by biserial rank correlation, further supported this result (r = 0.55 and r = 0.43, respectively). With a mean best frequency of 4.9 kHz for frontally tuned neurons, this suggests that the neurons in the ICx of juvenile owls do not begin development tuned primarily to high frequencies, but that this develops as the facial ruff develops. Later recordings done in these same birds’ ICx, performed in the 160–200-day-old age range, display the typical high-frequency tuning for frontally tuned neurons (data not shown), supporting this hypothesis. Figure 7 with 1 supplement see all Download asset Open asset Tuning properties of the juvenile owl’s inferior colliculus (ICx) neurons. ICx neurons recorded from juvenile owls, before the facial ruff developed, plotted by their best interaural time difference (ITD) and frequency tuning range. Weak correlation between ITD and frequency (green line). Figure 8 Download asset Open asset Comparisons of frequency and spatial tunings of frontally tuned inferior colliculus (ICx) neurons between normal, ruff-removed and juvenile owls. (a) Box plots indicating the distribution of best frequencies for neurons tuned to frontal interaural time differences (ITDs) (±30 µs, equivalent to approximately ±10° azimuth) for normal (blue, n = 24), ruff-removed (red, n = 33) and juvenile (green, n = 30) owls . (b) Box plots for each group, denoting the high and low bounds of frequency tuning curves. (c) Box plots representing the responsive frequency range for neurons of each group. (d) Box plots representing the width of ITD tuning curves of each group. Individual neurons and Mann–Whitney U-test p-values used to compare groups are shown over each box plot. While there were clear differences in best frequency across juvenile, ruff-removed and normal adult owl groups, we found additional differences when comparing the high and low boundaries of each neuron’s frequency tuning (see Figure 2b for definitions). While there was a significant difference in the neurons’ high-frequency bounds between the juvenile and ruff-removed owls (u = 206, p=6.2e-05), there was no significant difference between the juvenile and normal adult owls (u = 409, p=0.4). In contrast, we saw the opposite for the low-frequency bounds: there was a difference between the juvenile and normal adult owls (u = 600, p=3e-05), but no significant difference between the juvenile and ruff-removed owls (u = 466, p=0.7). These results are summarized in Figure 8b and suggest that the juvenile owls’ ICx neurons are broader and more heterogeneously tuned to frequency than normal or ruff-removed adult owls, which then gets refined during development. In accordance with this, we found significant differences in the responsive frequency range of ICx neurons between the juvenile owls and both the normal adult owls (u = 190, p=0.003) and ruff-removed adults (u = 221, p=1.6e-04), but not between the normal and ruff-removed adults (u = 418, p=0.73) (Figure 8c). In addition, because there was no difference in the responsive frequency range between the adult groups, we can deduce that their neurons are sensitive to different frequencies based on the results from Figure 8a. Analysis of ITD tuning width of frontal neurons showed significantly broader tuning width in ruff-removed (u = 72, p=1.2e-07) and juvenile owls (u = 188, p=0.0028), consistent with their tuning to lower frequencies and the known relationship between frequency tuning and ITD tuning width of midbrain neurons (Takahashi and Konishi, 1986; Wagner et al., 2007; Figure 8d). These results indicate that