Azimuth tuning in an anechoic environment
Figure shows our general approach to measure azimuth tuning of IC neurons. We first determined the unit's BF by delivering tone bursts as described above at 50 dB SPL. To determine the unit's threshold for VAS stimulation we set the azimuth to −75° and varied the stimulus level of a broad-band noise over a 70 dB range in 10 dB steps. We then delivered VAS stimuli at levels between 10 and 50 dB above its threshold for azimuths ranging between ±150° in 15° steps. These were delivered in random order and the response to the initial azimuth was discarded to minimize adaptation effects. Figure displays this unit's response as dot rasters to binaural (red), ipsilateral (green) and contralateral stimulation, each 100 ms in duration and separated by 300 ms. This sequence was repeated 4 times at each azimuth. It is clear that this neuron is excited by binaural and contralateral ear stimulation and inhibited by ipsilateral ear stimulation. The strongest driven activity to binaural stimulation was in the contralateral sound field (i.e., 0–150°), whereas that to contralateral ear stimulation was omnidirectional. These features are also plotted in Figure where the neural firing rates are plotted in Cartesian coordinates. These plots reflect the response during the stimulus burst. The spontaneous activity indicated was the mean of the last 100 ms of the silent period after binaural, ipsilateral and contralateral ear stimulation. In Figure , the binaural and contralateral ear alone azimuth functions are displayed in polar coordinates after subtraction of the mean spontaneous activity. If this subtraction yielded negative values, we made the azimuth function positive by adding a constant (absolute value of the minimum). Although recordings were not made at 180° we interpolated and smoothed the function over the full range. Also shown are the actual responses at each azimuth for binaural (red open circle) and contralateral ear stimulation (blue open circles) after subtraction of the spontaneous rate. This neuron displays binaural facilitation (binaural response > contralateral ear response) in the contralateral sound field and binaural suppression (binaural response < contralateral ear response) in the ipsilateral sound field. The interpolation to ±180° allowed the calculation of vector strength (VS) and vector angle (VA) as a measure of the neuron's sharpness of azimuth tuning and its preferred azimuth direction (magenta line), respectively. VS and angle measures were computed using the original definition of Goldberg and Brown (1969
). Best azimuth (BA, azimuth at the maximum response) is included as a comparison with VA.
Figure 1 Procedure for assessing a neuron's azimuth tuning. (A) IC neurons response (BF = 3.2 kHz) to wide-band noise (0.2–20 kHz) presented to both ears (red), to the ipsilateral ear (green) and to the contralateral ear, each 100 ms in duration and separated (more ...)
Figure displays the distributions of VS (red), VA (blue), and BA (black) to binaural stimulation as a function of stimulus level (upper five rows). In the bottom row is a summary of these distributions in the form of the median ± semi-interquartile range. The sharpness of azimuth tuning broadened with increasing stimulus level as evidenced by VS decreasing from 0.44 to 0.22. The direction of azimuth tuning as measured by VA shifted from −68° at 10 dB to −93° at 50 dB and its distribution became tighter with increasing stimulus level. BA shifted from −62° to −82°, but its distribution became broader with stimulus level.
Figure 2 Top 5 rows: Distributions of vector strength (VS, red), vector angle (VA, blue), and best azimuth (BA, black) at stimulus levels between 10 and 50 dB (re: neural threshold).
Bottom row: Plot of the median and semi-interquartile range of these three measures (more ...)
The degree of level tolerance in azimuth tuning varies among neurons. We investigated the hypothesis that certain neurons exhibit level tolerant azimuth tuning, i.e., no significant changes in VS and VA across stimulus levels. To test this hypothesis we first selected units that were tested at 10, 30, and 50 dB and rank ordered this sample based on VS at 50 dB. We then subdivided it into three sub populations, top 10%, top 50%, and all for stimulus levels at 10, 30, and 50 dB and averaged their rate azimuth functions (Figure , top 3 rows). For each population, VS and VA are plotted in the fourth and fifth rows, respectively. For binaural stimulation, VS and angle remain relatively stable across stimulus level in all three subpopulations. There were no statistically significant differences in vector strength across stimulus level (1 way repeated measures ANOVA, F = 4.74, df = 122, p > 0.01) for the top 37% of the units, indicating level tolerant behavior for this subpopulation. There were no statistically significant differences in vector angle across stimulus level (F = 4.93, df = 50, p > 0.01) for only the top 15% of the units. Thus, if we require no significant change in both vector strength and vector angle across stimulus levels, then the top 15% satisfied this combined criterion. For all three subpopulations, the mean VA was in the frontal contralateral quarter field (~−60°) at 10 dB and shifted backwards to ~−90° at 30 and 50 dB. This shift reflects the fact that the responses in the back and front quarter fields become more symmetrical at higher stimulus levels.
Figure 3 Population azimuth tuning to binaural (red) and contralateral ear alone (blue) stimulation as a function of stimulus level. Only those units that were tested to both binaural and contralateral ear stimulation at 10, 30, and 50 dB (n = 75) were included. (more ...)
In contrast, contralateral ear stimulation yielded VSs and angles comparable to binaural stimulation only at 10 dB and precipitously decreased with increasing stimulus level. VA at 30 and 50 dB are not plotted because the VS did not reach our minimal criterion (≥0.15). When the response was greater to binaural stimulation than to contralateral ear stimulation, we defined the difference summated over a region of azimuth to be the binaural facilitation area; when the relationship was reversed, we defined the difference to be the binaural suppression area. These measures are same as those used by Delgutte et al. (1999
). The facilitation and suppression areas for our three subpopulations are plotted as a function of stimulus level in the bottom row (Figure ). The strengths of binaural suppression and facilitation increased with increasing stimulus level. The combined consequence of this facilitation and suppression is to confine the azimuth tuning to the contralateral field over a wide range of stimulus levels.
Coding of azimuth and envelope in reverberant environments
These experiments simultaneously examine coding of location and envelope in different acoustic environments by presenting SAM noise at different azimuths and distances. The carrier sound was a 1-octave wide noise centered at the unit's BF. Using this noise band we determine the unit's threshold by presenting sounds from 0 to 70 dB SPL binaurally and separately to the ipsilateral and contralateral ear at −75° azimuth (a value that approximates the BA of most units). We then determine the unit's azimuth functions to binaural, ipsilateral, and contralateral ear stimulation at 30 dB above threshold at a distance of 80 cm in the anechoic environment. Next, we determined its MTF to modulation frequencies between 2 and 512 Hz in 1-octave steps (100% depth) at the unit's BA, at a distance of 80 cm, and a level 30 dB above its threshold. Based on its MTF, the modulation depth of its best modulation frequency was varied between 12.5 and 100% in 3 dB steps. We then selected a modulation depth that was approximately in the center of its range that produced significant (p
< 0.001) neural synchrony based on circular statistics (Mardia, 1972
Figure displays the response of an IC neuron that had a BF of 628 Hz to the above type SAM noise (444–888 Hz, 32 Hz modulation, 70.7% depth, 50 dB SPL). The top dot raster displays the unit's response to the above stimuli presented binaurally (red) and to the contralateral ear alone (blue) in the anechoic environment at our closest sound source distance (10 cm). The corresponding polar plots of azimuth represent four measures derived from the responses depicted in the dot raster. The rate azimuth function shows that the binaural response is strongest in the contralateral sound field whereas those to the contralateral ear are omnidirectional. The combination of binaural facilitation in the contralateral sound field and the binaural suppression in the ipsilateral sound field described in Figures , is also present here. The remaining 3 polar plots represent the neuron's response to the envelope. Because recordings were not made at 180° and because these measures require significant synchrony to the envelope, these polar plots have a gap at this azimuth. The syncrate measure is defined as a product of synchrony to the modulation envelope and firing rate. The significance of syncrate depends entirely on whether synchrony is significant or not (p < 0.001). The synchrony azimuth function was similar between binaural and contralateral stimulation. Consequently, the syncrate azimuth function resembled the rate azimuth function. The rightmost polar plot represents the neural modulation gain. It is defined as 20 × log (2 × neural synchrony/modulation depth in the ear) and was approximately 5 dB for binaural and monaural stimulation.
Figure 4 Response of an IC neuron (BF = 628 Hz) to a SAM 1-octave band noise delivered binaurally and to the contralateral ear alone as a function of azimuth at a fixed distance of 10 cm.
Top row: responses to different azimuths in the anechoic condition displayed (more ...)
The middle and bottom row of Figure represents the responses of the same neuron to the same stimuli but presented in our moderately reverberant and highly reverberant environments, respectively. At this close distance (10 cm), the responses in these two reverberant environments are remarkably similar to each other and to those in the anechoic condition.
In Figure we display the responses of the same neuron to the same stimuli but at a further sound source distance (80 cm). In contrast to the responses at 10 cm (Figure ), at 80 cm, the acoustic environment had noticeable effects on spike rate, syncrate, synchrony and gain azimuth functions.
Response of the same neuron in Figure to the same stimuli but at a distance of 80 cm. Same format as Figure .
The responses in the anechoic environment at 80 cm (Figure , top row) are very similar to those at 10 cm (Figure , top row). However, the responses in the two reverberant environments at 80 cm (Figure , middle and bottom rows) differed considerably from those in the anechoic condition (Figure , top row) and from those at 10 cm (Figure , middle and bottom rows). In reverberation, the firing rate to binaural stimulation in the contralateral sound field decreased relative to the anechoic condition. Additionally, the binaural facilitation and suppression in the rate azimuth function in the moderately reverberant environment was less pronounced and binaural suppression in the highly reverberant environment was absent. The firing rate to contralateral ear stimulation remained omnidirectional in all three environments. The synchrony to the modulation envelope to both binaural and monaural stimulation declined and was the weakest in the highly reverberant environment. Surprisingly, in the moderately reverberant environment, the mean neural modulation gain in the contralateral sound field to both binaural and contralateral ear stimulation reached 10 dB and further increased to 13 dB in the highly reverberant environment. These gains represent enhancements of 5 and 8 dB over the anechoic condition.
The strong firing rate of this neuron to binaural stimulation in the anechoic environment was confined to the contralateral sound field (Figures , , top rows). In order to elucidate mechanisms underlying this azimuth tuning we altered the ITD cue in the fine structure by decorrelating the sounds between the two ears. In Figure , using the same neuron as in Figures , , we show the effect of delivering decorrelated sounds to the ears. The upper row shows the dot rasters and polar plots when the same normal sound was delivered to the ears in our anechoic chamber at a distance of 160 cm. Note that as in the 10 and 80 cm distances (Figures , , top rows), the azimuth rate response is primarily confined to the contralateral sound field whereas that to monaural stimulation is omnidirectional. In all cases, binaural facilitation in the contralateral sound field and binaural suppression in the ipsilateral sound field is prominent. However, when the sound to the ears is uncorrelated (Figure , bottom row), azimuth tuning to binaural stimulation becomes omnidirectional and very similar to monaural stimulation. The fact that azimuth tuning to binaural stimulation was destroyed is strong evidence that ITD plays the key role in azimuth tuning for this neuron. ILD play a negligible role because this cue is <1 dB at 160 cm in frequency range used (1-octave centered at 628 Hz). Synchrony to binaural and monaural stimulation remained relatively unchanged under the decorrelated condition. This indicates that the envelope coding mechanism for this neuron is monaural.
Figure 6 Effect of presenting noise to each ear that is uncorrelated. Same neuron as in Figures , . Top row: responses when the normal, correlated sound was delivered to the ears in our anechoic chamber at a distance of 160 cm. Bottom (more ...)
We recorded the responses of this neuron not only at 10 cm (Figure ), 80 cm (Figure ) and 160 cm (Figure ), but also at sound source distances of 20 and 40 cm. Figure provides a summary of our findings at sound source distances between 10 and 160 cm. Plots E–L reflect the responses averaged in the contralateral sound field. The stimulus level was adjusted at each distance such that the absolute level was kept constant.
Figure 7 Azimuth tuning and envelope sensitivity as a function of distance (10-160 cm) for the same neuron in Figures , , . Plots E–L reflect the responses averaged in the contralateral sound field. (A,B) Sharpness (more ...)
The sharpness of azimuth tuning as reflected by azimuth VS is plotted as a function of distance for binaural (panel A) and contralateral ear (panel B) stimulation. The VS to binaural stimulation remained constant across distance in the anechoic environment whereas it systematically decreased with distance in reverberation and most dramatically in the highly reverberant environment. This decrease in azimuth tuning with distance to binaural stimulation in the reverberant environments is likely due to a decrease in interaural correlation of the stimulus with distance. From 10 to 160 cm, the mean interaural correlation measured in the contralateral sound field systematically decreased from 0.98 to 0.74 in the highly reverberant environment, whereas it remained constant (0.997 ± 0.002) in the anechoic condition. Recall that decorrelation destroyed azimuth tuning (Figure ). Thus, the decrease in correlation with distance in reverberation parallels the observed decrease in azimuth VS, and the constant and high interaural correlation seen in the anechoic condition parallels the observed constant and strong azimuth VS.
Although azimuth VS to binaural stimulation decreased with distance in reverberation, azimuth VA remained relatively constant with distance (panel C). For contralateral stimulation (panel B), the azimuth VS across conditions was too small to meet our criterion for azimuth tuning (VS = 0.15). Consequently, none are plotted for azimuth VA to contralateral stimulation (panel D).
The mean rates to binaural and contralateral ear stimulation in the contralateral sound field are plotted in panels E and F, respectively. The rate to binaural stimulation remained relatively constant with distance in the anechoic environment (black) but gradually declined in the two reverberant environments to a maximum decline of 30% at 160 cm in the highly reverberant condition (blue). The decrease in rate to binaural stimulation in the contralateral sound field with distance in the reverberant environments is likely due to a decrease in interaural correlation as outlined above. Specifically, this decline in firing rate is consistent with decreasing interaural correlation with distance. The response rate to contralateral ear stimulation was about half that to binaural stimulation and there was no clear relationship with distance in the two reverberant environments. However, the firing rates averaged across distance in the moderate and highly reverberant environments were 9 and 14% less, respectively, compared to that in the anechoic environment.
The syncrate functions across distance to binaural and contralateral ear stimulation are shown in panels G and H, respectively. At 160 cm in the highly reverberant environment, syncrate was not significant to contralateral ear stimulation. Thus, this point is absent in panel H. The syncrate functions show a steeper decline with distance than the rate functions because both rate and synchrony decrease with distance.
Synchrony to binaural and contralateral ear stimulation remained essentially constant in the anechoic condition. In reverberation, synchrony remained essentially constant for distances up to 40 cm and then declined (panels I and J). At 160 cm in the highly reverberant environment, synchrony was not significant to contralateral ear stimulation. Thus, this point is absent in panel J.
The neural modulation gain as a function of distance is plotted in panels K and L for binaural and contralateral ear stimulation, respectively. The mean gain across distance in the anechoic condition for both types of stimulation was 5 dB. However, at far distances it increased with reverberation to a maximum of 15 dB to binaural stimulation and 13 dB to contralateral ear stimulation.