Mechanisms underlying azimuth selectivity in the auditory cortex of the pallid bat

Abstract

This study focused on mechanisms underlying azimuth selectivity in the primary auditory cortex (A1) of pallid bats. The pallid bat listens to prey-generated noise (5-35 kHz) to localize and hunt terrestrial prey. The region of A1 tuned between 5 and 35 kHz consists of two clusters of neurons distinguished by interaural intensity difference (IID) selectivity: binaurally inhibited (EI) and peaked. The first aim of this study was to use sequential dichotic/free-field stimulation to test the hypothesis that IID is the primary cue underlying azimuth selectivity in neurons tuned in the prey-generated noise frequency band. IID selectivity and ear directionality at the neuron's characteristic frequency (CF) were used to predict azimuth selectivity functions. The predicted azimuth selectivity was compared with the actual azimuth selectivity from the same neurons. Prediction accuracy was similarly high for EI neurons and peaked neurons with low CF, whereas predictions were increasingly inaccurate with increasing CF among the peaked neurons. The second aim of this study was to compare azimuth selectivity obtained with noise and CF tones to determine the extent to which stimulus bandwidth influences azimuth selectivity in neurons with different binaural properties. The azimuth selectivity functions were similar for the two stimuli in the majority of EI neurons. A greater percentage of peaked neurons showed differences in their azimuth selectivity for noise and tones. This included neurons with multiple peaks when tested with tones and a single peak when tested with noise. Taken together, data from the two aims suggest that azimuth tuning of EI neurons is primarily dictated by IID sensitivity at CF. Peaked neurons, particularly those with high CF, may integrate IID sensitivity across frequency to generate azimuth selectivity for broadband sound. The data are consistent with those found in cat and ferret A1 in that binaurally facilitated neurons depend to a greater extent (compared to EI neurons) on spectral integration of binaural properties to generate azimuth selectivity for broadband stimuli.

Figure 1

(A) The IID selectivity of a binaurally inhibited (EI) neuron determined using broadband noise as stimulus. Positive IIDs denote a louder sound in the contralateral ear. Negative IIDs denote a louder sound in the ipsilateral ear. The response magnitude (ordinate) in this and subsequent figures is the number of spikes for 20 repetitions of each stimulus. (B) The azimuth sensitivity function (ASF) of the same EI neuron as in (A) obtained using broadband noise as stimulus. Positive azimuths denote contralateral locations. Negative azimuths denote ipsilateral locations. The vertical arrow points to the azimuth at which response declines to 50% of maximum. This is termed the ‘50% azimuth’, and quantifies the medial boundary of the ASF. (C) The IID selectivity of a peaked neuron determined with noise as stimulus. (D) The ASF of the same peaked neuron determined with noise as stimulus. The short dashed line indicates the range of azimuth eliciting greater than 80% of maximum response. The arithmetic center of this range (short vertical arrow) is termed the ‘peak azimuth’. The long dashed line marks the range of azimuth eliciting greater than 50% of maximum response. This range is termed ‘50% width’. In peaked neurons, the term ‘50% azimuth’ is used to indicate azimuth at which response declines to 50% of maximum on the ipsilateral side of the peak.

Figure 2. IID-Azimuth functions for five different frequencies

These data were adapted from Fuzessery (1996).

Figure 3. Predicting azimuth sensitivity of an EI neuron with a CF of 21 kHz using ear directionality, frequency tuning and IID sensitivity

(A) The IID sensitivity of the neuron was recorded using dichotic stimulation. The vertical arrow shows the IID at which response decreases to 50% of maximum. The table in the inset shows predicted number of spikes for each azimuth location based on the IID generated by a 20 kHz tone at that azimuth (using the IID-azimuth function for 20 kHz from Figure 2), and the IID sensitivity function of the neuron. The predicted azimuth function is plotted in (B) along with this neuron’s actual azimuth function that was determined with free-field stimulation. The vertical arrows indicate the actual and predicted 50% azimuths, which were almost identical. IID and azimuth sensitivity functions were determined with broadband noise. The predicted and actual ASF were similar (correlation coefficient, r = 0.96). Predicted ASFs in this and subsequent figures go up to only ±60° because the IID-azimuth relationship is only available for this range of azimuth.

Figure 4. EI neurons can show different forms of azimuth selectivity depending on the CF and ear directionality at CF

(A) The CF of this EI neuron was 20 kHz and its 50% IID was - 15 dB (vertical arrow). (B) The predicted azimuth tuning indicated a sigmoid function with a 50% azimuth ~ −32° (vertical arrow). The actual azimuth function recorded was similar to the predicted curve (correlation coefficient, r = 0.93). (C) The CF of this EI neuron was 10 kHz, but the IID selectivity was similar to the EI neuron in ‘A’ (50% IID = −14 dB). (D) The predicted azimuth tuning was omnidirectional in that the response did not decline below 50% of maximum across the azimuths tested. The actual azimuth function was similar to the predicted function (r = 0.8).

Figure 5. Predicting ASF of peaked neurons

(A) The peaked neuron shown here had a CF of 20 kHz. The 50% IID width (IID range between the two vertical arrows) and 50% IID are shown in the panel. (B) The predicted and actual azimuth functions show a peak at midline, 50% azimuth bandwidth (range of azimuths between the two vertical arrows) of ~50° and a 50% azimuth (left vertical arrow) ~ −25°. (C) This peaked neuron had similar IID selectivity as the neuron in (E), but a CF of 10 kHz. (D) The predicted azimuth function indicated a broader tuning compared to the previous neuron. The actual azimuth function matched these predictions.

Figure 6. Peaked neurons with CF between 23–30 kHz were predicted to have multiple peaks

(A) IID selectivity of a neuron with CF=25 kHz. (B) The ASF is predicted to show two peaks, but the actual ASF tested with noise exhibited a single peak. (C) IID selectivity of a neuron with CF=30 kHz. (D) Using the ear directionality for 25 kHz predicts three peaks in the ASF. When tested with noise, only one peak was seen.

Figure 7. Azimuth tuning across the population of recorded neurons was predicted by interactions between IID, frequency tuning and ear directionality

(A) The predicted and actual 50% azimuth for EI neurons. The dashed lines in all panels indicate unity slope. (B) The predicted and actual 50% azimuth of peaked neurons. (C) The predicted and actual peak azimuth for peaked neurons. (D) The predicted and actual 50% azimuth bandwidth for peaked neurons.

Figure 8. Predictability of noise-ASF in EI and peaked neurons

(A) Distributions of correlation coefficients in EI and peaked neurons. (B) There was no relationship between prediction accuracy in the EI neurons. (C) Correlation coefficients were high for most low-CF peaked neurons. There was a decline in prediction accuracy of high-CF peaked neurons.

Figure 9. Azimuth sensitivity functions of EI neurons in response to broadband noise (BBN) and CF tone

(A) A neuron in which the ASFs were similar for noise and tones. The vertical arrows indicate the 50% azimuth, a measure of the medial boundary of ASF. The response magnitude shown in this and subsequent figures is for 20 repetitions of each stimulus. (B) A neuron which responded better to noise than to CF tone.

Figure 10. ASF of peaked neurons in response to noise and CF tones

(A) A neuron with almost identical ASF for the two stimuli. (B, C) Neurons with peaked response to both stimuli, but the response to CF tone was at least 60% below maximum responses to noise. (D) A neuron with peaked ASF to noise and sigmoid ASF to tone. (E, F) Neurons with single peaked ASF for noise and multi-peaked response to tones.