Synapse number was reduced after ouabain treatment and in old age.

Number of functional synapses per IHC, at the cochlear position corresponding to 2 kHz, for young-adult gerbils (red, N = 19), surgery-only-treated gerbils (orange, N = 3), gerbils treated with a low dose (40 µM) of ouabain (green, N = 8), gerbils treated with a high dose (70 µM) of ouabain (blue, N = 10), and old gerbils (gray, N = 14). Box plots display the median (center line), mean (white circled dot), 25th and 75th percentiles (upper and lower edges of the boxes), and maximum and minimum (whiskers). Significance levels are * p = 0.011 and ** p = 0.0022.

Numbers of gerbils of different treatment and age groups within the behavioral, electrophysiological, and histological part of this study.

Note that numbers in category 2-4 do not sum up to total numbers, since many gerbils were used in all three study parts. For the neural data, the core data for this study are the AN fibers which were tested with TFS1 stimuli. However, some additional fibers, in some animal groups also additional animals, were included for the analysis of neural tuning; these numbers are provided separately. Furthermore, both untreated gerbil groups were supplemented with data from eight young-adult gerbils and twelve old gerbils for which synapse counts were available, from Steenken et al. (2021).

Bandwidths of neural response curves were not affected by ouabain treatment or age.

Half-maximal bandwidth of response curves of single AN fibers, obtained with pure tones at a level <= 20 dB above the individual fibers’ rate threshold, as a function of BF. Color code for the different treatment groups is the same as in Fig. 1.

Mean driven spike rates did not differ between frequency shifts.

Histograms of average driven spike rates during presentation of the TFS1 stimuli (black and colored) and average spontaneous rate, without acoustic stimulation (gray). Subject groups are arranged in columns and stimulus conditions in rows. Colors code the frequency shift Δf in percent of f0, with the responses to harmonic stimuli shown in black and responses to the inharmonic stimuli grading from yellow (least inharmonic) to red (most inharmonic). The legends apply to panels above them. In each panel, the maximal number of fibers (with 40 stimulus repetitions per fiber) that the histograms are based on is indicated; note that not all fibers were tested with each frequency shift Δf. Clipped spontaneous rate bars are marked with an arrow and the maximal relative proportion beyond the axis limit.

Frequency shift representation in the neural responses differed between groups and stimulus conditions.

Vector strength frequency spectra of responses to harmonic and inharmonic stimuli, averaged across repetitions and fibers. Subject groups are arranged in columns and stimulus conditions in rows. Colors code the frequency shift Δf in percent of f0, with the same color code as in Fig. 3. Dashed lines indicate the limits of the stimulus bandpass filters. The inset panels show enlarged examples of the spectra ranging around the envelope frequency (f0, 200 or 400 Hz) and the fine structure (center) frequency (TFS peak frequency; 1000, 800, or 2000 Hz). The thin gray lines mark the y-axis scaling of the inset panels. In each panel, the maximal number of fibers (with 40 stimulus repetitions per fiber) that the spectra are based on is indicated; note that not all fibers were tested with each frequency shift Δf.

Frequency shift representation differed between envelope and TFS frequency ranges.

Frequency shift of the z-value peak versus stimulus frequency shift Δf. Subject groups are arranged in columns and stimulus conditions in double rows. Odd rows show the data at f0 and even rows at fmaxpeak (the frequency of maximal average response, cf. Fig. 11). Each circle represents four dimensions: the stimulus frequency shift Δf (position on the x-axis), the frequency shift of the z-value peak (position on the y-axis), the average vector strength (z-score) across all respective fibers (color saturation, see legend at the bottom), and the percentage of fibers with a frequency shift of the z-value peak at the corresponding stimulus frequency shift Δf (circle radius). The largest circles correspond to 100% of fibers (e.g., in panel C), medium sized circles to values around 50% (e.g., panel J). Colors code the frequency shift Δf in percent of f0, with the same color code as in Fig. 3. In each panel, the maximal number of fibers (with 40 stimulus repetitions per fiber) that the z-value peaks are based on is indicated; note that not all fibers were tested with each frequency shift Δf.

The gerbils’ discrimination performance improved with larger frequency shifts.

Sensitivity index d’ for behavioral discrimination, as a function of stimulus frequency shift Δf in percent of f0. Reference lines at d’ = 1 indicate the assumed threshold value for meaningful discrimination performance. Subject groups are arranged in columns and stimulus conditions in rows. Colors code the subject group. Circled black dots show the mean across subjects, boxes show the median and interquartile ranges, whiskers show the extrema, except for outliers, which are displayed as colored dots. Data points are considered outliers, if they lie beyond 1.5 times the distance between median and upper or lower quartile, respectively. For each panel, the maximal number of subjects in the respective group is indicated. Unfilled boxes mark conditions completed by less than half of those subjects in the respective group.

Old gerbils were typically unable to perceive frequency shifts whereas ouabain treatment did not impair discrimination.

Behavioral discrimination thresholds based on the data shown in Fig. 6. Thresholds were defined as the lowest (linearly interpolated) stimulus frequency shift Δf in percent of f0 at which the sensitivity index d’ crossed d’ = 1. The threshold was set to 100% if this criterion was never met. Colors code the subject group. Circled black dots show the mean across subjects, boxes show the median and interquartile ranges, whiskers show the extrema, except for outliers, which are displayed as colored dots. Data points are considered outliers, if they lie beyond 1.5 times the distance between median and upper or lower quartile, respectively. The y-axes at the right show the same frequency shifts in Hz. Significance levels are ** p < 0.002 for all three comparisons, young-adult vs. old p = 1.53×10-3, sham vs. old p = 1.11×10-3, ouabain high vs. old p = 0.382×10-3.

Gerbil excitation patterns for harmonic (thick black lines in panels A, C, E) and inharmonic stimuli close to the frequency shift at the TFS1 threshold (thick dark green lines in panels A, C, E) and their differences (panels B, D, F) for the three stimulus conditions.

Thin black and green lines in panels A, C, E show the physical sound spectra for the corresponding harmonic and inharmonic stimuli, respectively. Red lines in panels A, C, E show the excitation pattern elicited by the pink noise masker. Since the level of excitation produced by the pink noise is less than 30 dB below that produced by the complex tones, distortion products will be masked. Excitation patterns were calculated with the assumption that the gerbil auditory filter bandwidth is 1.8 times the human auditory filter bandwidth (see Kittel et al. 2002). The thick green lines in panels B, D, F show the differences between the excitation pattern elicited by the harmonic reference and the inharmonic stimuli close to the frequency shift at the TFS1 threshold. The thin yellow and orange lines show these differences for the different frequency shifts of the inharmonic stimuli (color codes as in Fig. 5). The grey-shaded areas in panels B, D, F indicate the 5%/95% percentiles of the amplitude statistics in the harmonic reference signals. The blue dashed lines in panels B, D, F show the gerbil threshold for detecting a change in the intensity of a 70 dB 1-kHz tone (Sinnott et al. 1992). Only for the condition 400/1600 Hz does the change in the level of the excitation pattern due to the frequency shift exceed the threshold for detecting the intensity difference. For this condition, the gerbils could solve the task, at their TFS1 threshold, by comparing the difference in excitation elicited by the harmonic and frequency-shifted inharmonic stimulus. For the other conditions, this change is less than the intensity difference limen. Thus, for these two conditions with harmonic number N of 8 the gerbils cannot rely on differences in the excitation patterns but must solve the task by comparing the temporal fine structure.

Frequency shift detection thresholds of gerbils and humans for TFS1 test stimuli in relation to the center frequency fc of the harmonic complex.

Filled symbols represent gerbil data from the present study. Human data are from published material with the limitation that the fc was in the range from 500 Hz to 4500 Hz and the fundamental f0 of the harmonic complex was in the range of 100 Hz to 400 Hz to make the range of parameter values similar to those of the present study (for references see legend, normal hearing NH, hearing loss HL). The higher and lower thresholds shown for the gerbil data reflect thresholds at fc of 1600 Hz for fundamentals f0 of 200 Hz and 400 Hz, respectively.

Examples illustrating stimulus waveforms and typical auditory-nerve responses.

Stimulus waveforms (A-D), peri-stimulus-time histograms (PSTHs, E-H), and average vector strength as a function of frequency (I-L). Columns 1 (A,E,I) and 2 (B,F,J) show data from a high (H)SR fiber, columns 3 (C,G,K) and 4 (D,H,L) show data from a low (L)SR fiber. Both fibers are from a young-adult, untreated animal and are matched in best frequency. The stimulus condition was f0 = 400 Hz and fc = 1600 Hz. A and C show examples of the harmonic stimulus (Δf = 0%, black) and its envelope (gray), B and D show examples of a strongly inharmonic stimulus (Δf = 8%, red), its envelope (gray), and the harmonic stimulus (black) as a reference.The corresponding neural responses are displayed in the same colors in the second and third row, respectively. Arrow markers are added (A-D) to highlight the differences in fine structure between harmonic and inharmonic stimuli. The PSTHs (E-H) show the instantaneous spike rate, averaged across 40 repetitions of identical stimuli. Vector strength (I to L) is shown as a function of frequency, averaged across 40 stimulus presentations. The inset panels show enlarged examples of the spectra in the envelope frequency range (400 Hz) and the fine structure frequency range (800 Hz).

Schematic representation of statistical contrasts.

Schematic of the frequency spectrum of a harmonic (black solid lines) and inharmonic (red dashed lines) TFS1 stimulus in panel A and schematic representation of the phase locking spectrum in a neuronal response in panel B. Black arrows and symbols point to the harmonic, unshifted frequencies, termed “z0” and red arrows and symbols point to the inharmonic, shifted frequencies, termed “zdf”. The annotated brackets below the frequency axis mark representative regions of the fundamental frequency, f0, and of the maximal average response, fmaxpeak. Filled connected circles show the log-z-ratio used for assessing the TFS representation, connected crosses show the log-z-ratio used for assessing the balance between f0/ENV and TFS representation.

Synapse number was reduced after ouabain treatment and in old age.

Number of functional synapses per IHC, at the cochlear positions corresponding to 2, 4, 8, and 16 kHz, for young-adult gerbils (red), sham-treated gerbils (surgery only, orange), gerbils treated with a low dose (40µM) of ouabain (green), gerbils treated with a high dose (70µM) of ouabain (blue), and old gerbils (gray). Box plots display the median (center line), mean (white circled dot), 25th and 75th percentiles (upper and lower edges of the boxes), and maximum and minimum (whiskers).

Median spontaneous rates did not differ between animal groups.

Individual spontaneous rates for all auditory-nerve fibers reported in this study (dots), group medians (box center lines), quartiles (box boundaries), minima, and maxima (whiskers). The dashed black line indicates the boundary between low- and high-spontaneous-rate fibers (18 spikes/s). Counts of included fibers within each combination of group and spontaneous rate class are indicated at to the corresponding side of the box plots. A Kruskal-Wallis test confirmed that the group medians are not significantly different at the 5% level.

Median best frequency was lower in the young-adult sample.

Individual best frequencies for all auditory-nerve fibers reported in this study (dots), and box plots in the same style as in Fig. S1. The dashed black line indicates the boundary between low and high best frequency class (1.85 kHz), as used in the results statistics. Counts of included fibers within each combination of group and best frequency class are indicated at to the corresponding side of the box plots. A Kruskal-Wallis test confirmed that the group medians are significantly different at the 5% level, with a Bonferroni-corrected post-hoc test showing that only the young-adult group median differs significantly from all other groups.