The spatial buildup of compression and suppression in the mammalian cochlea

Abstract

We recorded responses of the gerbil basilar membrane (BM) to wideband tone complexes. The intensity of one component was varied and the effects on the amplitude and phase of the others were assessed. This suppression paradigm enabled us to vary probe frequency and suppressor frequency independently, allowing the use of simple scaling arguments to analyze the spatial buildup of the nonlinear interaction between traveling waves. Most suppressors had the same effects on probe amplitude and phase as did wideband intensity increments. The main exception were suppressors above the characteristic frequency (CF) of the recording location, for which the frequency range of most affected probes was not constant, but shifted upward with suppressor frequency. BM displacement reliably predicted the effectiveness of low-side suppressors, but not high-side suppressors. We found "anti-suppression" of probes well below CF, i.e., suppressor-induced enhancement of probe response amplitude. Large (>1 cycle) phase effects occurred for above-CF probes. Phase shifts varied nonmonotonically, but systematically, with suppressor level, probe frequency, and suppressor frequency, reconciling apparent discrepancies in the literature. The analysis of spatial buildup revealed an accumulation of local effects on the propagation of the traveling wave, with larger BM displacement reducing the local forward gain. The propagation speed of the wave was also affected. With larger BM displacement, the basal portion of the wave slowed down, while the apical part sped up. This framework of spatial buildup of local effects unifies the widely different effects of overall intensity, low-side suppressors, and high-side suppressors on BM responses.

FIG. 1

BM responses to wideband tone complexes. Left column amplitude, normalized to stapes motion. Right column phase re stapes. In A and B, the intensities of all 40 components of an equal-amplitude complex were varied together in 10-dB steps as indicated in the graph. In the other panels, all components except one (the suppressor) were kept at 20 dB SPL, and suppressor intensity was varied in 10-dB steps as indicated in the graph. Black triangles indicate CF (14.8 kHz) and black circles indicate suppressor frequency: 4.5 kHz (C, D), 14.8 kHz (E, F), and 25.1 kHz (G, H). Datapoints in the high-frequency plateau are shown in gray. Experiment RG12420.

FIG. 2

Amplitude change (left panels) and phase change (right panels) induced by increasing overall intensity (A, B) or the intensity of a suppressor below CF (C, D), near CF (E, F), or above CF (G, H). The curves were obtained by referencing the data of Figure 1 to the response to the 20-dB-SPL equal intensity data, thus emphasizing deviation from (near) linearity. Layout and symbols as in Figure 1. Suppressor components are not shown in C–H.

FIG. 3

Contour plots of amplitude and phase change. The data of Figure 2 are displayed as filled contour plots. Color bars above A and B quantify the changes in amplitude (left panels) and phase (right panels). Layout and symbols as in Figures 1 and 2. Vertical dashed lines mark the data for CF probes.

FIG. 4

Dependence of suppression on probe frequency and suppressor frequency. Amplitude changes (left panels) and phase changes (right panels) induced by 50-dB-SPL suppressors (upper panels) and 80-dB-SPL suppressors (lower panels) are shown as filled contour plots. Black triangles indicate CF (14.8 kHz). Color bars quantify the changes. Experiment RG12420.

FIG. 5

As Figure 4, but for a different cochlea. CF was 14.7 kHz. Experiment RG12449.

FIG. 6

Examples of large phase shifts from two cochleae (RG12420, CF = 14.7 kHz, triangles; RG12449, CF = 14.8 kHz, squares). A Phase changes induced by single-tone suppressors (open symbols) and wideband intensity increments (solid symbols), compared to their 20-dB reference, and plotted against normalized probe frequency. Suppressor intensity is indicated in the graph. Normalized suppressor frequency is indicated by the solitary symbols close to the abscissa. B The phase change of the probes associated with the peak values of A as a function of suppressor intensity. Symbols of A and B are matching; probe frequency is indicated in the graph. Peak values in A and B are identical.

FIG. 7

Dependence of amplitude and phase of probe responses on suppressor intensity. For a below-CF (A, B), near-CF (C, D), and above-CF (E, F) suppressor the amplitude (left panels) and phase (right panels) of the entire collection of probes is shown. Suppressor frequency is indicated in the graphs. The gradual transition from blue to green curves corresponds to increasing probe frequencies, a subset of which is indicated in A. Thick lines are used for the curves of CF probes (14.7 kHz). The dashed black lines indicate a slope of −1 dB/dB. Experiment RG12449.

FIG. 8

Phase change against amplitude change for different suppressors. Wideband intensity increments (A); below-CF (B), near-CF (C), and above-CF (D) suppressor as indicated in the graphs. Different colors correspond to different probe frequencies and a subset is indicated in A. The f _probe = CF (14.7 kHz) line is thicker than the others. Same data as shown in Figure 7.

FIG. 9

Maximum rate of suppression (ROS) derived from the reduction of probe response amplitude with suppressor intensity (see text). A, B ROS versus suppressor frequency. Different curves represent different probe frequencies, a subset of which is indicated in kHz by the numbers enclosed by circles. Below-CF probes are shown in blue, CF probes are shown in black and marked by thicker lines, and above-CF probes are shown in red. C ROS of CF probes for five cochleae, plotted against normalized suppressor frequency. D Contour plot showing ROS as a function of both probe frequency and suppressor frequency. Suppressor intensity was 80 dB SPL. ROS is quantified by the color bar to the right of the graph.

FIG. 10

The role of BM displacement magnitude of suppressors. Panels A–F show how the response amplitude (left) and phase (right) of a given probe vary with the magnitude of BM displacement evoked by suppressors. Different curves within each panel correspond to different suppressor frequencies. A selection of suppressor frequencies is indicated in C. With increasing suppressor frequency, line color gradually changes from blue to yellow. Different rows of panels correspond to different probe frequencies: a below-CF (A, B), near-CF (C, D), and above-CF (E, F) probe as indicated in the graphs. Black dashed curves indicate equality of probe-alone and probe + suppressor BM displacement (see text). In all panels, thick lines indicate CF (14.7 kHz) suppressors. G Total RMS displacement against suppressor intensity. Different lines represent different suppressor frequencies, a selection of which is indicated in the graph. Line color gradually changes from blue to red with increasing suppressor frequency. H Suppression by nondominant suppressors. The gray area marks the combinations of probe and suppressor frequencies for which at least 3 dB of suppression was found for any suppressor intensity. The black area is the subset of frequency pairings for which probe-induced BM displacement exceeded suppressor-induced displacement. Experiment RG12449.

FIG. 11

Performance of a simple gain control model based on local BM displacement (see text). Measured and predicted BM response amplitudes for two below-CF suppressors (A, B), and an above-CF suppressor (C), each presented at 80 dB SPL. Suppressor frequency is indicated by solid black circles (918 Hz, 14.7 kHz, and 24.6 kHz, respectively). The different lines in each plot show: the unsuppressed (reference) response at 20 dB SPL (black dashed line); measured suppressed response (blue solid line); prediction of suppressed response from the model (red solid line). D and E show the model mismatch (predictions minus data) for all suppressor frequencies, with suppressor intensity equal to 50 and 80 dB SPL, respectively. Black triangles indicate CF (14.7 kHz). Experiment RG12449.

FIG. 12

Schematic spatial profiles of BM displacement magnitude. The vertical distance from the baseline indicates the amplitude of BM displacement in response to a stimulus component. A log scale of displacement is implied, meaning that parallel curves correspond to proportional variations in amplitude. A Displacement profile of unsuppressed probe wave. The peak portion is sensitive to suppressive effects. The basal tail and most apical portion are insensitive. The black arrow marks the transition between the insuppressible tail and the suppressible peak. B Adding a low-side suppressor does not affect the tail, but causes a progressive suppression of the probe wave, visible as a divergence between suppressed and unsuppressed probe profiles starting near the black arrow (see text). C High-side suppressors have limited overlap with the probe. Apical to the overlap region, the probe partially rebounds owing to its lowered amplitude (see text). At point P, suppression is observed but not the excitation of the suppressor (“phantom suppressor”). If the suppressor frequency is much higher than the probe frequency (red dashed profile), it will not affect the probe wave.

FIG. 13

Probe wave subjected to different amounts of suppression. Black curve unsuppressed probe wave. Red curve probe wave of reduced amplitude caused by a suppressor. Blue curve further reduction caused by increasing the suppressor intensity. The arrows mark the different growth of suppression at two locations labeled a and b.

FIG. 14

The spatial buildup of high-side suppression. A Spatial profile of high-side suppression derived from the schematic wave envelopes of Figure 12C, using the same color coding. B The suppressor-induced amplitude change as a function of probe frequency, with probe and high-side suppressor varied in a fixed frequency ratio. Using a scaling argument (see text), these data illustrate the spatial buildup of high-side suppression, with low frequencies corresponding to the basal locations. C Suppressor/probe ratio of BM displacement corresponding to the data in panel B. Positive values indicate suppressor dominance. Experiment RG12449; 60-dB-SPL suppressor.

FIG. 15

Quantification of suppression strength by loss gradients. The loss gradient is the estimated amplitude loss per length unit of a suppressed wave relative to a reference wave (see text). The estimates were derived from responses to the wideband, equal-amplitude stimuli using the 50-dB-SPL data as a reference. Data from five cochleae are shown as indicated in the graph along with the CF of the recording sites.

FIG. 16

The effect of suppression on wave velocity and phase. A Velocity profiles of traveling waves subject to varying levels of suppression. The low-intensity reference (black curve), shows a rapid transition (marked T) from the fast basal portion of the wave to the slow apical portion. Increasing amounts of suppression cause a progressive smoothing of the transition (brown and ochre lines), causing a pivoting around T (see text). The ×s mark the effect of a high-side suppressor having a spatial range that is restricted to locations basal to T. B The corresponding wave number profiles show a transition from small to large values and a suppression-induced pivoting. C The local phase, obtained from spatial integration of wave number. The unsuppressed wave (black curve) shows the transition as a sharp kink, which is rounded by suppression (see text). The point B at which the phase is unaffected by mid-level suppressors lies apical to T. D Suppressor-induced phase shifts obtained by using the unsuppressed phase profile of panel C as a reference.