Introduction

Timing of action potential (AP) firing is important for the encoding and processing of information in nervous systems including the auditory system1,2,3,4,5. The first-spike latency (FSL), i.e., the time lapse between the onset of a sensory stimulus and the first AP elicited by the stimulus, is of particular importance in such encoding and processing6,7.

In the auditory system, the timing of AP firing elicited by a sound is dependent on temporal, spectral, and spatial characteristics of the sound8,9. Such dependences are especially evident for the FSL10,11,12,13,14,15,16,17,18.

Sounds in an acoustic environment can affect each other in eliciting responses. Such an effect can exist even when the sounds are spectrally, temporally, and/or spatially separated19,20,21,22,23,24,25. In our previous studies, we used the rat as an animal model and trains of stimuli to examine how the strength of firing of a neuron elicited by one sound was affected by a temporally and spatially separated second sound in a major auditory processing center, the inferior colliculus (IC)26,27. The trains included equal probability two-tone sequences, which mimicked two competing sounds that occurred at the same 50% probability. They also included oddball paradigms, which mimicked two temporally interleaved sounds with one having a low probability while the other one having a high probability (thus, mimicking a novel and a standard sound, respectively). These previous studies found that a sound elicited stronger firing when it was presented at a lower than a higher probability27. Furthermore, the firing elicited by a sound at a fixed location at the ear contralateral to the neuron being recorded was increased when the other sound was spatially separated, especially when the location-fixed sound was a novel sound26,27. These results provided important information about how spatial cues were used in the detection of a sound in the existence of another sound.

In the present study, we use equal probability two-tone sequences and oddball paradigms to examine timing of responses in the IC. Specifically, we study how the FSL of the response elicited by a sound is dependent on the probability of presentation of the sound and its spatial relationship with the other sound. Attention is paid to not only the mean but also the variation (i.e., jitter) of the FSL across responses elicited by multiple presentations of a sound.

Materials and methods

Animal Preparation

Experiments were conducted using 32 adult male Wistar albino rats (Rattus norvegicus, 250–600 g) obtained from Charles River Canada Inc. (St. Constant, QC). All but 1 had ages younger than 10 months when experiments were conducted. Normal hearing of the rats was confirmed using the clapping startle response. Surgical anaesthesia was induced by ketamine hydrochloride (60 mg/kg, i.m.) and xylazine hydrochloride (10 mg/kg, i.m.) and maintained by supplementary injections of ketamine hydrochloride (20 mg/kg, i.m.) and xylazine hydrochloride (3.3 mg/kg, i.m.).

A craniotomy was made on the right side of the skull to allow for placement of a recording electrode into the IC. The skull was cemented onto a head bar attached to a custom-made holding device. A recording electrode was held by a custom-made clamp attached to the slave cylinder of a Model 650 micropositioner, which was fitted onto a micromanipulator of a Model 900 stereotaxic instrument (Kopf Instruments, Tujunga, CA). Instruments were positioned in such a way that acoustic shadows and reflections were minimized. The rat was placed in a Model CL-15A LP acoustic chamber (Eckel Industries, Morrisburg, ON) for neurophysiological recordings. Upon completion of recordings, the rat was euthanized by pentobarbital (300 mg/kg, i.p.). Experimental protocols were approved by the University of Windsor Animal Care Committee in accordance with the guidelines of the Canadian Council on Animal Care. All experiments were performed in accordance with the relevant guidelines and regulations. The study is reported in accordance with ARRIVE guidelines.

Acoustic stimulation

Sound waveforms were generated using a System 3 real-time signal processing system controlled by a personal computer running the OpenEx software (Tucker-Davis Technologies, Alachua, FL). Sounds were presented using two Model FF1 free-field speakers (Tucker-Davis Technologies, Alachua, FL). Each speaker was held by a mounting device and could be positioned at any azimuthal location 50 cm away from the midpoint of the interaural line. Each speaker was calibrated over 100 and 65,000 Hz at five azimuths (see Fig. 1a) using a model 4135 microphone and a model 2608 measuring amplifier (Brüel & Kjaer, Dorval, QC). These azimuths included the midline of the frontal field (denoted by 0°) and 90° and 45° on the contralateral and ipsilateral side of the recording site (denoted by c90°, c45°, i45°, and i90°, in which “c” and “i” indicate contralateral and ipsilateral side, respectively).

Fig. 1
figure 1

Locations of sound calibration/presentation and trains of stimuli. (a) Locations of sound calibration/presentation. An arrow indicates that neurophysiological recordings were obtained from neurons in the right IC. (b) Top block: An oddball paradigm with both Std and Odd presented (first row), Odd omitted (second row), and Std omitted (third row). Bottom block: An equal-probability two-tone sequence with both Eqls presented (first row), and one of the two sounds omitted (second and third rows). (c) An oddball paradigm (top row) and an equal probability two-tone sequence (bottom row) were presented with two sounds colocalized at c90° (left column) or separated by a 90° angle (middle and right columns). In the figure, loudspeaker symbols indicate locations of sounds while rectangle shapes indicate sound stimuli. Red, magenta, and blue colors represent Odd, Eql, and Std, respectively. Open and filled shapes represent two different Eql sounds, respectively. A speaker with both red and blue colors indicates that Odd and Std are colocalized, while a speaker with both open and filled magenta color indicates that two Eql sounds are colocalized.

Full size image

Recording electrode and procedures

AP discharges were recorded extracellularly from single neurons in the right IC using a single-barrel glass micropipette filled with 3 M NaCl (tip diameter ~ 1.5 μm, impedance 5 ~ 10 MΩ). The electrode was within the coronal plane at a 30° angle in reference to the midsagittal plane and was 4.0 mm lateral and 0.4 mm rostral in reference to the lambda point. To search for an auditory neuron, Gaussian noise bursts at 60 dB SPL were presented from a loudspeaker at c90° while electrophysiological activities were monitored audio-visually. Neural signals were amplified by a 2400 A preamplifier (Dagan, Minneapolis, MN) and sampled at 24.4 kHz using the System 3 real-time signal processing system.

Upon isolation of a single auditory neuron, its characteristic frequency (CF, the frequency at which the neuron displayed the lowest threshold) and threshold at CF were determined using tone bursts presented at c90°. A threshold was defined as the lowest sound-pressure level at which a tone burst with 5-ms rise/fall phases and a 90-ms plateau presented at 4/s elicited AP discharges over at least 5 of 20 presentations of the tone burst.

Three trains of two-tone stimuli were created for a neuron, which included a pair of oddball paradigms (See first row of Fig. 1b upper block of diagrams) and an equal-probability two-tone sequence (See first row of Fig. 1b lower block of diagrams). Each train had 200 stimuli presented at a constant rate of 4/s. Each stimulus was a presentation of one of two tone bursts. One had a frequency lower than CF (named as fL elsewhere in the article) while the other had a frequency higher than CF (named as fH elsewhere in the article). The center frequency of fL and fH (i.e., (fL× fH)1/2) was at the CF and the difference between the two frequencies (i.e., (fH - fL)/(fH× fL)1/2) was 0.10. Each tone burst (named as TL or TH based on frequency) had 5 ms rise/fall phases and a 90 ms plateau. TL and TH had the same sound-pressure level (typically 10–30 dB above the threshold at CF). At the frequency separation and the sound-pressure level described above, both sounds could elicit AP discharges over at least 5 of 20 presentations of the tone burst (i.e., generate suprathreshold responses) at c90°. In one of the of the two oddball paradigms, TL was presented at a 10% probability (as an oddball stimulus or Odd) while TH was presented at a 90% probability (as a standard stimulus or Std). Thus, the paradigm had a combination of Odd-TL/Std-TH. In the other paradigm, the probabilities of the two sounds were reversed (thus, a combination of Odd-TH/Std-TL). In an equal-probability two-tone sequence, TL and TH were both presented at a 50% probability (as equal-probability sound or Eql). Thus, the sequence had a combination of Eql-TL/Eql-TH. Within each train of stimuli (an oddball paradigm or equal probability two-tone sequence), TL and TH were presented in a randomized order.

Responses to two oddball paradigms (Odd-TL/Std-TH and Odd-TH/Std-TL) and an equal-probability two-tone sequence (Eql-TL/Eql-TH) were first recorded when TL and TH were colocalized at c90° (Fig. 1c left column). Responses to the three trains of stimuli were then recorded when one tone burst was kept at c90° while the other one was relocated to each of the four non-c90° azimuths (i.e., c45°, 0°, i45°, and i90°, see Fig. 1c middle and right columns for a relocated sound at 0°). The sound that remained at c90°, named as a location-fixed sound elsewhere in the article, could be TL as Std or Odd of an oddball paradigm or Eql of an equal probability two-tone sequence. Correspondingly, the sound that was relocated, named as a location-changeable sound elsewhere in the article, would be TH as Odd, Std, or Eql, respectively. Alternatively, a location-fixed sound could be TH as Std, Odd, or Eql and a location-changeable sound would be TL as Odd, Std, or Eql, respectively. Effects of relocation of one sound on both FSLs of responses to the location-fixed and location-changeable sounds were examined. The response to a location-fixed sound was also examined when a location-changeable sound was omitted from presentation (Second and third rows of Fig. 1b upper and lower blocks). The response to a location-fixed sound in such a single-tone sequence was compared with the response to the same sound obtained when the location-changeable sound was presented to further study how the response to the first sound was affected by the second sound.

Data analysis

The FSL of firing elicited by each of the 200 stimuli in a train (i.e., an oddball paradigm or an equal-probability two-tone sequence) was measured in reference to the onset of the stimulus. FSLs of responses elicited by presentations of the same sound (TL or TH) in the train were used to calculate a mean and a standard deviation (SD). The SD was used to evaluate the degree of jitter of the FSL across presentations of a sound. Two sets of mean and SD values were obtained for responses to TL and TH, respectively. For responses elicited by a sound presented as Odd, the mean and SD values were calculated only if firing was elicited over 5 of a total of 20 presentations. For responses elicited by a sound presented as Eql or Std, the mean and SD values were calculated only if firing was elicited over 5 of the first 20 presentations. To minimize influences of randomly generated action potentials (e.g., spontaneous activity) on the calculation of the mean and SD FSL, those first spikes with latencies that substantially deviated from the peak of distribution of FSLs were excluded from the calculation. The criterion for such exclusion was ±3SD of the FSLs.

Responses elicited by two oddball paradigms and an equal probability two-tone sequence with two sounds colocalized at c90° were compared to find how the timing of the first spike elicited by each sound (TL or TH) was dependent on the probability of presentation of the sound (i.e., as Odd, Eql, or Std). Responses elicited by a train of stimuli obtained at different angle of separation between location-fixed (c90°) and location-changeable sounds (c90°, c45°, 0°, i45°, and i90°) were compared. Dependences of FSLs of responses to the two sounds on the angle of separation were determined.

Responses were studied in two groups of neurons that responded to a tone burst with transient and sustained patterns of firing26,27, respectively. Neurons with transient firing included onset and fast-adapting subtypes. An onset pattern had strong firing of APs only over a very brief period at the onset of a tone burst. A fast-adapting pattern had firing that attenuated gradually from a high level to the level of spontaneous firing before the offset of a tone burst. Neurons that displayed sustained firing included primary-like, pauser, and build-up/later subtypes. A primary-like pattern had strong transient firing at the onset of the sound followed by reduced firing over the rest of the tone burst without an interruption, while a pauser pattern had early transient firing and late reduced firing separated by a brief pause. Neurons with build-up/later firing had long and highly variable FSLs. These neurons were not included in the analysis, as they were unlikely important for conveying auditory information using the timing of the first spike.

Results

AP discharges elicited by trains of stimuli including oddball paradigms, equal-probability two-tone sequences, and single-tone sequences were recorded from 75 neurons in the IC. The two tone bursts that were used to create these trains of stimuli had frequencies lower and higher than the CF of a neuron, respectively (i.e., TL and TH, see “Recording electrode and procedures” for details). In each train, the two sounds had a specific combination of probabilities (Odd-Std, or Eql-Eql) and spatial relationships (location-fixed sound at c90° while location-changeable sound at c90°, c45°, 0°, i90°, i90°, or omitted). Two trains formed a pair. In one train, TL was presented at one probability and location while TH were presented at another probability and location or omitted. In the other train, the two sounds were presented at reversed probabilities and locations. Comparing responses elicited by pairs of trains indicated that timing of the first spike elicited by TL and TH at the same probability and spatial location were not different from each other. The lack of difference was confirmed by Mann-Whitney tests. Thus, throughout the rest of the article group results (shown in Figs. 3, 5, 7 and 8) are presented with two sets of data based on responses elicited by TL and TH being combined.

Fig. 2
figure 2

Responses of an example neuron elicited by a sound (TL) in an oddball paradigm or an equal-probability two-tone sequence when the other sound (TH) was colocalized at c90°. Top and bottom panels show responses to the sound presented as Odd and Std, respectively. The middle panel show the response to the sound as Eql. Each panel has a dot-raster histogram showing timing of APs elicited by all presentations of TL along with a histogram showing the distribution of FSL. First spikes elicited by Odd, Eql, and Std are indicated by red, magenta, and blue dots, respectively. Spikes elicited after the first spike (regardless of the probability of presentation of a sound) are indicated by grey dots. The mean FSL and SD FSL obtained from each response (i.e., response to Odd, Eql, or Std) are shown above a corresponding histogram. The mean value is also indicated by a vertical black line in the histogram. The color scheme used in this figure (i.e., red, magenta, and blue for first spikes elicited by Odd, Eql, and Std) is used in making other figures throughout the article. A horizonal black bar below the bottom panel indicates the duration of a sound. The inset at the top of the figure indicates the trains of stimuli and the spatial location of TL and TH (colocalized at c90°).

Full size image
Fig. 3
figure 3

Box plots showing the difference in the mean FSL (“dMean FSL” in the top row) and the difference in the SD FSL (“dSD FSL” in the bottom row) between responses elicited by the same sound (TL or TH) presented as Odd and Eql and between those elicited by the same sound presented as Std and Eql. Results were obtained when two sounds of a train of stimuli (i.e., TL and TH) were colocalized at c90°. The left, middle, and right columns are results from entire group of neurons, neurons with sustained firing, and neurons with transient firing, respectively. Two sets of data based on responses elicited by TL and TH, respectively, presented at the same probability (i.e., as Odd, Eql, or Std) were combined in making each plot. “*” and “**” indicate the statistical significance at levels of p<0.05 and p<0.005, respectively.

Full size image
Fig. 4
figure 4

An example neuron showing dependences of the mean FSL and the SD FSL of the response to a location-changeable sound on the azimuth of the sound. a) Dot-raster histograms showing responses to TH as location-changeable Odd presented at 5 azimuths (as indicated above the top panel). Red dots indicate the first spikes elicited by Odd while grey dots indicate spikes elicited after the first spikes. The distribution of the FSL is shown in each panel using a histogram. The mean and the SD of the distribution are indicated above the histogram. A vertical line in the histogram indicates the mean FSL. A horizonal black bar below each panel indicates the duration of a sound. b) Bar charts showing dependences of the mean FSL (left panel) and the SD FSL (right panel) on the azimuth of a location-changeable sound. Measurements were obtained from responses to Odd (red), Eql (magenta), and Std (blue). The inset at the top of the figure shows that an oddball paradigm with TL as location-fixed Std and TH as location-changeable Odd was used to obtain responses displayed in (a).

Full size image
Fig. 5
figure 5

Box plots showing changes of the mean (“dMean FSL” in (a)) and the SD FSL (“dSD FSL” in (b)) of responses to location-changeable Odd (left panels), Eql (middle panels), and Std (right panels) when the sound was relocated from c90° to another azimuth. Results were obtained from the entire group of neurons and neurons with sustained and transient firing (open, backward hatched, and forward hatched boxes, respectively). Two sets of data based on responses elicited by TL and TH, respectively, were combined. “*” and “**” indicate statistically significant changes at levels of p<0.05 and p<0.005, respectively. Statistical results from related-samples Friedman’s two-way analysis of variance by ranks as well as pairwise comparisons with Bonferroni correction are shown in Table 1.

Full size image

FSLs of responses elicited by two sounds colocalized at c90°

FSLs of responses elicited by two sounds (TL and TH) of a train of stimuli (either an oddball paradigm or an equal-probability two-tone sequence) were examined when the sounds were colocalized at c90°. Shown in Fig. 2 are responses of a representative neuron to TL presented as Odd, Eql, and Std. Both the mean FSL and the SD FSL were smallest when the sound was presented as Odd and largest when the sound was presented as Std.

Fig. 6
figure 6

Results from an example neuron showing dependences of the mean and SD FSL of response to a location-fixed sound on the azimuth of a location-changeable sound. a) Dot-raster histograms showing responses to TH as location-fixed Odd when TL as location-changeable Std was at 5 azimuths (as indicated above each of the 5 panels). Red dots indicate the first spikes elicited by Odd. The distribution of the FSL is shown in each panel using a histogram. The mean and the SD of the distribution are indicated above the histogram. A vertical line in the histogram indicates the mean FSL. A horizonal black bar below each panel indicates the duration of a sound. b) Bar charts showing dependences of the mean FSL (left panel) and the SD FSL (right panel) of response to location-fixed TH on the azimuth of a location-changeable TL. Measurements were obtained from responses to TH as Odd (red), Eql (magenta), and Std (blue). The inset at the top of the figure shows that an oddball paradigm with TH as location-fixed Odd and TL as location-changeable Std was used to obtain the responses displayed in (a).

Full size image

The mean FSL and SD FSL of the response elicited by each sound (TL or TH) presented as Odd, Eql, and Std were obtained from each neuron. Results from the entire group of neurons confirmed that both the mean and the SD were dependent on the probability of sound presentation (Odd vs. Eql vs. Std) (Related-Samples Friedman’s Two-Way Analysis of Variance by Ranks; χ2(2) = 15.376, p < 0.001 for mean; χ2(2) = 10.950, p = 0.004 for SD). Using measurements from the response to Eql as references, we found that a reduction in the probability of presentation of a sound (i.e., a change from Eql to Odd) significantly reduced the mean FSL and the SD FSL of the response to the sound (left boxes in left panels of Fig. 3a and b, pairwise comparisons with Bonferroni correction, p = 0.002 for mean FSL and p = 0.009 for SD FSL). However, a change from Eql to Std did not significantly change these values (right boxes in left panels of Fig. 3a and b).

Further analyses revealed that the mean FSL was different across responses to a sound presented as Odd, Eql, and Std in subgroups of neurons with sustained and transient firing (Related-Samples Friedman’s Two-Way Analysis of Variance by Ranks, χ2(2) = 6.783, p = 0.034 and χ2(2) = 9.179, p = 0.010, respectively). Pairwise comparisons with Bonferroni correction revealed that the mean FSL was significantly reduced when a sound was changed from Eql to Odd in both neurons with sustained (p = 0.037 for the left box in the middle panel of Fig. 3a) and transient firing (p = 0.018 for the left box in the right panel of Fig. 3a). The mean FSL was not significantly affected when a sound was changed from Eql to Std (Right boxes in middle and right panels of Fig. 3a). The SD FSL was different across responses to a sound presented as Odd, Eql, and Std in neurons with transient (χ2(2) = 7.663, p = 0.022) but not sustained firing (χ2(2) = 3.652, p = 0.161). In transient neurons, the SD FSL was significantly reduced when a sound was changed from Eql to Odd (left box in the right panel of Fig. 3b, p = 0.041) but not affected when a sound was changed from Eql to Std (the right box of the same panel).

Effects of spatial separation between sounds on FSLs: Response to a location-changeable sound

An oddball paradigm or an equal-probability two-tone sequence was presented when a location-fixed sound was at c90° while a location-changeable sound was at c90°, c45°, 0°, i45°, or i90°. The mean FSL and the SD FSL of the response elicited by the location-changeable sound was obtained at each azimuth. Shown in Fig. 4 are results from an example neuron. This neuron increased both the mean FSL and the SD FSL when a sound was relocated from c90° to another azimuth, especially 0° or an ipsilateral azimuth.

Fig. 7
figure 7

Box plots showing changes in the mean (“dMean FSL (ms)” in (a)) and the SD FSL (“dSD FSL (ms)” in (b)) of responses to location-fixed Odd (left panels), Eql (middle panels), and Std (right panels) when location-changeable Std, Eql, and Odd was moved from c90° to a non-c90° azimuth. Results were obtained from entire group of neurons and subgroups of neurons with sustained and transient firing (open, backward hatched and forward hatched boxes, respectively). Two sets of data based on responses elicited by TL and TH, respectively, were combined. “*” and “**” indicate statistically significant changes at levels of p<0.05 and p<0.005, respectively. Statistical results from related-samples Friedman’s two-way analysis of variance by ranks as well as pairwise comparisons with Bonferroni correction are shown in Table 2.

Full size image

Changes of the mean and the SD FSL of the response to a sound (Odd, Eql, or Std) when the sound was relocated from c90° to another azimuth were obtained from each neuron. Results from the entire group of neurons confirmed that both the mean FSL and the SD FSL of the response to Odd was increased when the sound was relocated (Open boxes in the left panels of Fig. 5a and b, Related-Samples Friedman’s Two-Way Analysis of Variance by Ranks, See Table 1 for statistical results). Pairwise comparisons with Bonferroni correction indicated that the increase was significant when Odd was at 0°, i45°, and i90°. The response to location changeable Eql or Std displayed increases in the mean FSL and SD FSL when the sound was moved away from c90° and the increase was significant at i45° and i90° (Open boxes in the middle and right panels of Fig. 5a and b, See Table 1 for statistical results). For SD FSL, the level of statistical significance at i45° and i90° was lower for the response to Std than for those to Eql or Odd (Right panel compared with the left and middle panels of Fig. 5b).

Table 1 Results from related-samples Friedman’s two-way analysis of variance by ranks and pairwise comparisons with Bonferroni correction on data presented in Fig. 5. Parts (a) and (b) of the table are related to dMean FSL (for Fig. 5a) and dSD FSL (for Fig. 5b), respectively.
Full size table

Additional analyses were conducted within subgroups of neurons with sustained and transient firing (Backward and forward hatched boxes in Fig. 5). Directional dependent increases in the mean FSL (Fig. 5a) and SD FSL (Fig. 5b) were found in both subgroups for responses to Odd, Eql, and Std (see Table 1 for statistical results). Pairwise comparisons indicate that in both subgroups of neurons directional dependent increases were generally more significant when a sound was presented as Odd than Eql or Std (Fig. 5a and b left panel compared with middle and right panels). For SD FSL, no significant increase was revealed by a pairwise comparison at any individual azimuth for the response to Eql in neurons with sustained firing (backward hatched boxes in Fig. 5b middle panel). It was not revealed in the response to Std in neurons with either sustained or transient firing (backward and forward hatched boxed in Fig. 5b right panel).

Effects of spatial separation between sounds on FSLs: Responses to a location-fixed sound

We examined whether/how the timing of the first spike elicited by a location-fixed (c90°) sound was affected by moving away a location-changeable sound. For the example neuron shown in Fig. 6, the mean FSL of the response to location-fixed Odd was reduced when location-changeable Std was at ipsilateral azimuths (Fig. 6a and red bars in Fig. 6b left panel). The mean FSL of the response to location-fixed Eql was reduced to a smaller extent by relocation of the other Eql to an ipsilateral azimuth (magenta bars in Fig. 6b left panel). Relocation of Odd didn’t substantially change the mean FSL of the response to location-fixed Std (blue bars in Fig. 6b left panel). The SD FSL of the response to a location-fixed sound varied when a location-changeable sound was moved away from c90° (Fig. 6b right panel). Such variation appeared to reflect random fluctuation as no apparent pattern of increase/decrease was observed,

Fig. 8
figure 8

Bar charts showing distributions of the difference in the mean (“dMean FSL (ms)” in (a)) and the difference in the SD FSL (“dSD FSL (ms)” in (b)) between responses elicited by the same location-fixed sound when the location-changeable sound was presented at i90° and when it was omitted from presentation. Comparisons were made in two subgroups of neurons that didn’t and did generate suprathreshold responses to a location-changeable sound presented at i90° (Open vs. filled bars). Left, middle, and right panels are based responses elicited by a location-fixed sound when it was presented as Odd, Eql, and Std. Two sets of data based on responses elicited by TL and TH, respectively, were combined in the figure. A vertical dashed line indicates no difference between responses to a location-fixed sound obtained under two conditions (i,e., location-fixed sound was at i90° and omitted). Numbers of neurons that had negative and positive differences are shown on the left and right sides of the dashed line. “NSub” and “NSup” indicate incidences in two subgroups of neurons that didn’t and did generate a suprathreshold response when stimulated by a location changeable sound at i90°. χ2 test results that compare incidences of positive and negative differences between the two subgroups of neurons: Top left panel: χ2= 0.250, p=0.617; Top middle panel: χ2= 2.788, p=0.095; Top right panel: χ2= 0.004, p=0.950; Bottom left panel: χ2= 0.293, p=0.594; Bottom middle panel: χ2= 0.133, p=0.715; Bottom right panel: χ2= 0.198, p=0.656.

Full size image

Changes of the mean and the SD FSL of the response to a location-fixed sound (Odd, Eql, or Std) caused by moving away a location-changeable sound (Std, Eql, or Odd) were obtained from each neuron. Group results confirmed that relocation of Std reduced the mean FSL but did not affect the SD FSL of the response to a location-fixed Odd (Open boxes in the left panels of Fig. 7a and b, see Table 2 for statistical results). A separation-dependent reduction of mean FSL was primarily caused by neurons with transient but not sustained firing (forward vs. backward hatched boxes). For both the entire group of neurons and neurons with transient firing, pairwise comparisons with Bonferroni correction indicated that the mean FSL was reduced when the Std was at i45° and i90° (see Table 2 for statistical results).

For responses elicited by location-fixed Eql, the mean FSL was mildly but significantly increased in the entire group of neurons when the other Eql was relocated to c45° (open boxes in the middle panel of Fig. 7a). Such an increase was due to neurons with transient but not sustained firing (Forward vs. backward hatched boxes in the middle panel of Fig. 7a). Furthermore, neurons with transient firing displayed a reduction in the mean FSL when the other Eql was relocated to i90°. The SD FSL of the response to location-fixed Eql was not affected by relocation of the other Eql (Middle panel of Fig. 7b). The mean FSL of the response to location-fixed Std was not affected by relocation of Odd in the entire group of neurons (Open box in the right panel of Fig. 7a). However, the SD FSL was mildly but significantly increased by a change in the location of Odd (Open box in the right panel of Fig. 7b, see Table 2 for statistical results). Pairwise comparisons indicated that the change was significant when the location-changeable Odd was at c45° or 0°. Additional analyses revealed that changes were primarily caused by neurons with transient but not sustained firing (Fig. 7 forward vs. backward hatched boxes).

Table 2 Results from related-samples Friedman’s two-way analysis of variance by ranks and pairwise comparisons with Bonferroni correction on data presented in Fig. 7. Parts (a) and (b) of the table are related to dMean FSL (for Fig. 7a) and dSD FSL (for Fig. 7b), respectively.
Full size table

Effect of omitting a location-changeable sound

To understand further how the response to a location-fixed sound was affected by a location-changeable sound, the response to the first sound was recorded when the second sound was omitted from presentation. The mean and the SD FSL of the response to the first sound were obtained. Differences in the mean FSL and the SD FSL were calculated by subtracting those obtained when the second sound was presented at i90° from those obtained when the second sound was omitted. These differences indicated whether the FSL of the response to the location-fixed sound was similarly/differently affected by moving a location changeable sound to i90° and omitting the sound from presentation. There was a group of neurons that didn’t pass the threshold of spiking response, i.e., generating AP over at least 5 of 20 sound presentations (see “Materials and Methods”), in response to a location-changeable sound at i90°. In this group of neurons, it was especially important to compare effects produced by moving a location-changeable sound to i90° and omitting the sound.

When a location-fixed sound was Odd, more neurons displayed longer mean FSLs and larger SD FSLs when location-changeable Std was presented at i90° than omitted which led to distributions of differences that were skewed toward the positive side (Fig. 8 top and bottom left panels). The distribution of differences of mean FSL was more skewed than that of differences of SD FSL. When a location-fixed sound was Std, similar numbers of neurons displayed positive and negative differences in mean FSL (Fig. 8 top right panel). Slightly more neurons displayed negative than positive differences in SD FSL (Fig. 8 bottom right panel). Overall, within the entire group of neurons both differences in the mean FSL and the SD FSL were dependent on the probability of presentation of a location-fixed sound (Odd vs. Eql vs. Std, χ2 test, χ2 = 30.7559, p < 0.001 for mean FSL, χ2 = 8.7577, p = 0.013 for SD FSL). The two groups of neurons that did and didn’t pass the threshold of spiking response under stimulation by a location-changeable sound at i90° displayed similar distributions of positive/negative differences at all probabilities of presentation (Fig. 8 open and filled histograms, χ2 tests, see caption for statistical results).

Discussion

The present study used oddball paradigms and equal-probability two-tone sequences presented in a free acoustic field to examine how the timing of the first spike of the response to one sound was affected by a temporally separated second sound. We found that the mean and jitter of FSL across responses elicited by multiple presentations of a sound were dependent on the probability of presentation of the sound (Figs. 2 and 3), the spatial location of the sound (Figs. 4 and 5), and the spatial relationship of the sound with the other sound (Figs. 6 and 7). The mean and jitter of the response to a sound at c90° could be affected by a sound at i90° even if the second sound did not elicit action potential firing (Fig. 8).

Dependences of FSL on characteristics of sounds

Previous studies in neural structures such as the auditory cortex and the IC have revealed that the FSL of the response elicited by a sound is dependent on spectral, temporal, and spatial characteristics as well as the intensity of the sound10–16; 28–32. Of these studies, some conducted in the auditory cortex examined the dependence of the FSL on the azimuth of a sound10,31. However, such a dependence was not systematically examined in IC neurons. The present study revealed that the mean FSL and SD FSL of response of an IC neuron elicited by a location-changeable sound (presented as Odd, Eql, or Std) became larger when the sound was moved from a contralateral to an ipsilateral azimuth (Figs. 4 and 5). The result enriched our knowledge about binaural responses of auditory neurons in the IC.

In the present study, two sounds with different frequencies (TL and TH) were used to form trains of stimuli presented under various conditions of probability combinations (Odd-Std, or Eql-Eql) and spatial relationships (location-fixed sound at c90° while location-changeable sound at c90°, c45°, 0°, i90°, or i90°). These trains formed pairs. In one train of a pair, TL was presented at one probability and location while TH were presented at another probability and location. In the other train, TL and TH were presented at reversed probabilities and locations. Comparisons between responses elicited by each pair of stimulus trains indicated that responses elicited by TL and TH presented at the same probability and location were not different from each other in the timing of the first spike. This result seemingly contradicted previous results regarding the dependence of the timing of response on frequency18. The lack of difference was likely because frequencies of the two sounds had a small disparity (10% of CF) and were on the two sides of the CF (the frequency typically leading to a peak response) with equal distances from the CF.

Our results indicated that the timing of first spike elicited by a sound (TL or TH) was dependent on the relationship of the sound with the other sound in the probability of presentation. Such a dependence was previously studied using oddball paradigms presented in a closed acoustic field33. It was found that a sound presented at the contralateral ear elicited a response with a shorter FSL when the probability of presentation was low than high (Odd vs. Std). Our results using stimuli presented in a free acoustic field agreed with this finding (Figs. 2 and 3). Comparisons of the response to Eql with those to Odd and Std further supported such a dependence. Furthermore, our results indicated that the temporal jitter of the first spike was smaller when a sound was presented as Odd than Eql or Std (Figs. 2 and 3).

The timing of the first spike elicited by a location-fixed sound was dependent on the azimuth of a location-changeable sound, especially in neurons with transient firing (Fig. 6 and 7). In these neurons, the mean FSL of the response to location-fixed Odd or Eql (but not Std) was reduced when location-changeable Std or Eql was moved from c90° to an ipsilateral azimuth (Fig. 7a left and middle panels). Interestingly, the mean FSL of the response elicited by location-fixed Eql was slightly but significantly increased when the other Eql was relocated to c45° (Fig. 7a middle panel). In the same group of neurons, the jitter of FSL of the response to location-fixed Std (but not Odd and Eql) was increased by relocating Odd from c90° to any azimuth other than i90° (Fig. 7b right panel).

Oddball paradigms and equal-probability two-tone sequences were used to examine how strengths of firing of IC neurons elicited by two sounds were dependent on the probabilities of presentation and spatial locations of the sounds26,27. Results revealed that firing elicited by a sound was weakened when the probability of presentation of the sound was increased (i.e., changed from Odd to Std) or when the sound was moved from c90° to 0° or an ipsilateral azimuth. For neurons with transient firing, the firing elicited by location-fixed (c90°) Odd or Eql was strengthened when a colocalized Std or Eql was moved to an ipsilateral azimuth. A comparison of these results with those from the present study indicated that in general for both responses to a location-changeable and a location-fixed sound the mean FSL and the strength of firing were inversely affected by changes in the probability and spatial location of a sound and spatial relationships between two sounds. The jitter and the strength of firing elicited by a sound were inversely affected by changes in the probability of presentation and the location of the sound but not the spatial relationship of the sound with the other sound.

Results from multiple sensory systems indicated that dependences of the FSL of a neural response on parameters of stimuli carry considerable information about the parameters4,8,11,34,35,36,37. Thus, the FSL can potentially be used by neurons to encode sensory information. Such latency-based encoding is especially important for neurons that respond to a stimulus with transient firing patterns. Results from the present study suggest that the timing of the first spike generated by an IC neuron can potentially be used to encode the spatial location of a sound and gauge the relationships between two sounds in probability of occurrence and spatial relationship. It should be noted that our results do not exclude a potential involvement of the strength of firing in the encoding of these characteristics of sounds.

Along with findings from our previous studies based on the strength of firing26,27, results from the present study based on the FSL provide insight into detection of a sound by IC neurons in the presence of a temporally separated interfering sound. An increase in the strength of firing and a decrease in the FSL caused by reduction of the probability of presentation of a sound likely indicate that a neuron is more sensitive to a low-probability (or a novel) than high-probability (or a standard) sound38,39. Enhanced firing and reduced FSL of the response to a location-fixed (target) sound upon spatial separation of an interfering sound may contribute to neural mechanisms that are responsible for “spatial release from masking”20,40. With effects of probability and spatial relationship combined, our results seemed to suggest that neural detection of a novel sound is enhanced when a high probability interfering sound is spatially separated.

Possible factors that affect the mean and the jitter of FSLs

The timing of an AP generated by a neuron is dependent on inputs received by the neuron and the way in which the inputs are integrated. As a key auditory processing center, the IC receives inputs from all other major central auditory structures41. Each input is associated with a specific pathway that originates from inner hair cells in the cochlea. Delays and temporal variations in neurophysiological signaling are generated by intra and intercellular processes along the pathway. Even at the auditory periphery, stochastic subcellular processes can affect the timing of depolarization of an inner hair cell, synaptic transmission between the cell and a ganglion cell, and generation of AP in an auditory nerve fibre42. Additional delays and temporal variations are generated by signaling processes within and between other neurons along central pathways leading to the IC43,44,45,46,47.

Local integration of inputs in the IC can further affect the timing of APs. Major excitatory inputs received by the structure include those from the contralateral cochlear nucleus as well as lateral superior olivary nucleus and the ipsilateral medial superior olivary nucleus48,49. Major inhibitory inputs are from the contralateral dorsal nucleus of the lateral lemniscus and the ipsilateral lateral superior olivary nucleus, superior paraolivary nucleus, and dorsal as well as ventral nuclei of the lateral lemniscus49,50,51,52,53,54,55. Due to the circuits in which these projections are involved, a sound can produce an excitatory effect on an IC neuron over the duration of the sound when it is delivered only to the contralateral ear while an inhibitory effect when it is delivered only to the ipsilateral ear56,57,58,59,60. A sound presented in a free acoustic field can reach both ears. Moving a location-changeable sound from a contralateral to an ipsilateral azimuth reduces the excitatory effect while enhances the inhibitory effect. Both changes can lead to weakening of firing and lengthening of the latency of the first spike61,62,63. These effects were likely among the major factors that caused a directional dependent increase of the FSL of the response to a location-changeable sound (Figs. 4 and 5).

When two sounds of a train were colocalized at c90°, an increase in the probability of presentation of a sound reduced the strength of firing and lengthened the FSL of the response elicited by the sound. Such effects could potentially be caused by adaptation of a neuron. Previous studies in the IC have revealed that a preceding sound can reduce firing and increase the latency of the response to a succeeding sound due to adaptation of the neuron64. The degree of adaptation is enhanced when the intensity and/or duration of the preceding sound is increased, or the strength of firing elicited by the sound is strengthened65,66,67,68,69,70,71. For a train of stimuli used in the present study, a rise in the probability of presentation of a sound (e.g., change from Odd to Std) increased the likelihood for a neuron to be repeatedly stimulated by the sound. This could have increased the degree of adaptation and reduced the mean strength of firing elicited by presentations of the sound over the train26,27. It could have also lengthened the FSL of the response (Figs. 3 and 4).

The effect of increase of the probability of presentation of a sound on the FSL could also be caused by an inhibitory aftereffect produced by the sound. Such an aftereffect might have been dependent on GABAergic inputs from the ipsilateral superior paraolivary nucleus (SPN) to the IC52,54. Neurons in the SPN receive inputs from the contralateral cochlear nucleus and generate bursts of firing at the offset of a stimulus72,73,74,75. This pathway enables a contralaterally presented sound to generate an aftereffect to inhibit the response of an IC neuron to a succeeding stimulus76,77. Results from a previous study of ours have confirmed that contralateral stimulation can elicit a GABAergic inhibitory aftereffect in the IC78. Patch-clamp recordings have revealed that a single electric shock of lemniscal fibers, which included those from the SPN, can lead to an inhibitory postsynaptic potential in an IC neuron that lasts for at least tens of millisecond79,80. Repetitive activation of inputs can cause temporal summation of such potentials80. Thus, bursts of firing generated by SPN neurons can conceivably lead to an inhibitory aftereffect on an IC neuron which lasts for somewhat longer than the silent period between two consecutive stimuli in the present study (150 ms). Such an aftereffect might have allowed the presentation of a sound to suppress the firing and lengthen the FSL of the response elicited by an immediately succeeding presentation of the sound. An increase of the probability of a sound heightened the odds of multiple consecutive presentations of the sound in a train, which could have enhanced the inhibitory aftereffect produced by the sound. Such a possibility is strongly supported by a previous study conducted in the rat’s IC showing that in many neurons blockade of GABAA receptors led to a larger increase of the firing elicited by Std than Odd81.

Adaptation and/or offset inhibition produced by a preceding acoustic stimulus could lead to suppression of the response to a succeeding stimulus even if the two stimuli had different frequencies, as long as they could interact with each other in the same frequency channel44,65,66,82. Within an oddball paradigm, both presentations of Odd and Std had a high probability to be preceded by a presentation of Std. However, differences existed between responses elicited by Std and Odd in the mean strength of firing26,27 and FSL (Figs. 2 and 3 of the present study). These differences suggested that a presentation of Std could produce a larger suppressive effect on the response to a subsequent stimulus when the stimulus had the same frequency (i.e., also Std) than different frequency (Odd).

Moving a location-changeable sound from c90° to an ipsilateral azimuth could reduce firing26,27 and consequently adaptation caused by the sound. It could also reduce contralaterally driven offset inhibition caused by the sound. Thus, the suppressive aftereffect produced by the sound on the response to a succeeding presentation of location-fixed sound would be weakened and the FSL of the response to the sound would be reduced, especially when the relocated sound was Std (Figs. 6 and 7). These results agree with a previous finding showing that the suppressive aftereffect produced by a leading sound was the strongest when the sound was at a location where it elicited the strongest excitatory responses over its duration21,22,83. Relocation of Odd from c90° to an ipsilateral angle might have also reduced the adaptation and the inhibitory aftereffect caused by the sound. However, due to the low probability of presentation of the sound, these changes could minimally affect the timing of the response to Std at c90°.

Additional insights into the influence of a location-changeable sound on the timing of the response to a location-fixed sound were gained by comparing the FSLs of the response obtained when the first sound was moved from c90º to i90º and when it was omitted (Fig. 8). When the location-fixed sound was Odd, more neurons displayed a longer FSL when Std was relocated to i90º than omitted, indicating that Std at i90º affected the response to Odd. There were two subgroups of neurons that did and didn’t generate firing in response to location-changeable Std at i90°. The two subgroups of neurons had similar distributions of differences between changes of FSL caused by relocation and omission of Std. This result indicated that even if a location-changeable sound did not elicit a suprathreshold response at i90º, it could still affect the response to a temporally separated location-fixed sound. Furthermore, adaptation was unlikely the only mechanism that were involved in generating a suppressive aftereffect on the response to a succeeding sound.

It’s less clear as to how local integration of excitatory/inhibitory inputs affected the degree of jitter in IC neurons. Similar to the mean FSL, the SD FSL of the response to a sound became larger when the probability of presentation of the sound was increased or when the sound was relocated from c90° to an ipsilateral azimuth (Figs. 3 and 5). Thus, it’s not unlikely that similar neural mechanisms (e.g., adaptation and offset inhibition) were involved in regulating both the mean and jitter of FSL of the response to a location-changeable sound. The involvement of offset inhibition in regulating jitter of FSL was supported by a previous study that used repetitive single tone burst presentations as stimuli84. Jitter of FSL could also be dependent on other mechanisms such as hyperpolarizing potentials elicited at the onset of a sound85 and rebound depolarization following hyperpolarization produced by a preceding sound86,87. The mean and the SD FSL of the response to a location-fixed sound were changed in different ways when a location-changeable sound was moved from c90º to another azimuth (Fig. 7). This result seems to suggest that mechanisms by which a location-changeable sound affected the mean and SD FSL of response to a location-fixed sound were likely different. Interestingly, the jitter of the first spike elicited by a high-probability sound (Std) was increased by relocation of a low-probability sound (Odd) to an azimuth other than i90º (Fig. 7b right panel). This result seemed to suggest that the jitter was affected by mechanisms other than adaptation and inhibitory aftereffect generated by the low-probability sound.

One thing to be kept in mind when results from the current study are interpreted is that animals from which responses were recorded were anaesthetized with a combination of ketamine and xylazine. Ketamine is an antagonist of the N-Methyl-D-aspartate (NMDA) receptor, which exists in major auditory structures including the IC80,88. It is well established that NMDA receptors in the IC can enhance the temporal summation of electrophysiological responses and transforms temporal encoding to a firing rate encoding89. Thus, antagonizing the receptor with ketamine might have enhanced the temporal precision of a response. It might also help reduce the dependence of neural sensitivity to the probability of presentation of a sound90. However, even in animals that are anaesthetised by ketamine/xylazine IC neurons still display large components of sound driven responses that are mediated by the NMDA receptors88,91. The possibility does exist that effects of probability of sound presentation and spatial location on the timing of responses were affected by ketamine in the present study. However, this possibility has yet to be verified by further experiments.

It should be noted that in the present study the location of recording was not identified for every single neuron. Thus, no maps could be made to compare neurophysiological responses with anatomical locations of neurons. It is well established that variations exist across the IC in auditory neural responses including those elicited by oddball paradigms92. Further studies should be conducted to evaluate area dependences of timing of responses elicited by temporally separated sounds including those used in the present study.

In general, results from the present study provided new insights into timing of responses to sounds that are presented in a free acoustic field. They helped us understand how the timing of response could potentially be used to encode the probability of occurrence and location of a sound and gauge how a sound is spatially related to another sound. Such knowledge can greatly help us comprehend neural bases of hearing under a binaural listening condition.