Between-ear sound frequency disparity modulates a brain stem biomarker of binaural hearing

Abstract

The auditory brain stem response (ABR) is an evoked potential that indexes a cascade of neural events elicited by sound. In the present study we evaluated the influence of sound frequency on a derived component of the ABR known as the binaural interaction component (BIC). Specifically, we evaluated the effect of acoustic interaural (between-ear) frequency mismatch on BIC amplitude. Goals were to 1) increase basic understanding of sound features that influence this long-studied auditory potential and 2) gain insight about the persistence of the BIC with interaural electrode mismatch in human users of bilateral cochlear implants, presently a limitation on the prospective utility of the BIC in audiological settings. Data were collected in an animal model that is audiometrically similar to humans, the chinchilla (Chinchilla lanigera; 6 females). Frequency disparities and amplitudes of acoustic stimuli were varied over broad ranges, and associated variation of BIC amplitude was quantified. Subsequently, responses were simulated with the use of established models of the brain stem pathway thought to underlie the BIC. Collectively, the data demonstrate that at high sound intensities (≥85 dB SPL), the acoustically elicited BIC persisted with interaurally disparate stimulation (click frequencies ≥1.5 octaves apart). However, sharper tuning emerged at moderate sound intensities (65 dB SPL), with the largest BIC occurring for stimulus frequencies within ~0.8 octaves, equivalent to ±1 mm in cochlear place. Such responses were consistent with simulated responses of the presumed brain stem generator of the BIC, the lateral superior olive. The data suggest that leveraging focused electrical stimulation strategies could improve BIC-based bilateral cochlear implant fitting outcomes.NEW & NOTEWORTHY Traditional hearing tests evaluate each ear independently. Diagnosis and treatment of binaural hearing dysfunction remains a basic challenge for hearing clinicians. We demonstrate in an animal model that the prospective utility of a noninvasive electrophysiological signature of binaural function, the binaural interaction component (BIC), depends strongly on the intensity of auditory stimulation. Data suggest that more informative BIC measurements could be obtained with clinical protocols leveraging stimuli restricted in effective bandwidth.

Keywords: auditory brain stem response; binaural hearing; binaural interaction; cochlear implants.

Fig. 1.

A: a midline electrode montage was used to record auditory brain stem response (ABR) waveforms elicited by unilateral or bilateral stimulation with one-third-octave Gabor clicks. The stimulus to the left ear was always 4 kHz; the stimulus to the right ear was varied in center frequency (and thus duration) parametrically. GND, ground electrode. B: left (L), right (R), and binaural waveforms were recorded independently. The monaural ABR sum [mono sum (L + R)] was then computed and subtracted from the binaural ABR (bin) to derive the binaural interaction component (BIC). A standard index of BIC amplitude is the value of the most prominent negative deflection (peak DN1); we instead computed root-mean-square amplitude (see text).

Fig. 2.

Example auditory brain stem response waveforms from a single animal for 9 different frequencies (rows) across 3 different intensities (columns). All waveforms were elicited by monaural (right ear) stimulation. Vertical dashed line indicates the nominal stimulus onset. Vertical scale bar, 2 µV. CF, center frequency.

Fig. 3.

A: auditory brain stem responses (ABRs) and derived binaural interaction components (BIC) for each animal in the present study (S1–S6) at 65 dB SPL in the frequency-matched condition (4-kHz Gabor click to both ears). Blue traces are the ABR for monaural presentation to the left ear (L); red traces are the ABR for monaural presentation to the right ear (R); magenta traces are the binaural ABR (B); dashed red and blue traces are the sum of left and right ABRs (L+R); black trace is the BIC; and gray shaded region indicates the temporal window over which BIC amplitude (root mean square) was computed (3–9 ms; see methods). Vertical scale bar, 1 µV. B: BIC across frequency-mismatch conditions at 65 dB SPL for a single animal (S1); each panel shows the L, R, B, and L+R ABRs and the derived BIC for presentation of a Gabor click at the given frequency in the right ear and presentation of a 4-kHz Gabor click in the left ear. C: BIC across 3 levels of frequency mismatch (columns) and 3 nominal sound intensities (rows) in a single animal (S1). Vertical scale bar, 2 µV.

Fig. 4.

A: BIC amplitudes (root mean square, RMS) across frequency for 3 intensities (rows) for each animal (S1–S6). Larger downward values indicate greater BIC amplitude. Vertical scale bar, 0.2 µV. B: BIC amplitude across all frequencies and intensities normalized within subjects to the value obtained at 65 dB in the frequency-matched condition (open circle at center). Shaded points represent cross-subject averages, with darker shades representing greater BIC amplitudes. The size of each point reflects the number of animals tested and reflected in the average (n = 2, 4, or 6 animals; see inset legend). Contour lines reflect the cross-subject weighted BIC tuning function [weights according to number of animals per data point; contour interval = 0.25 normalized units, with the lightest shaded contour approximating the values yielding 0.25 (25%) of the BIC amplitude observed at 65 dB in the frequency-matched condition]. C: because BIC modulation appeared sharpest at intermediate levels (B), “tuning” at 65 dB SPL was evaluated for each animal by normalizing amplitudes at each frequency to the amplitude recorded for the frequency-matched condition (black trace is mean; gray shading indicates ±SD).

Fig. 5.

A: auditory brain stem response (ABR) for an 8,000-Hz Gabor click across 3 interaural time differences (ITDs) for a single animal (S6). Blue traces are the ABR for monaural presentation to the left ear (L); red traces are the ABR for monaural presentation to the right ear (R); magenta traces are the binaural ABR (B); dashed red and blue traces are the sum of left and right ABRs (L+R); black traces are the BIC. B: BIC amplitude across 9 ITDs as computed via cross-correlation (x-corr.) and normalized to the value obtained at 0 µs for 2 animals (S5 and S6; thin lines); the mean is given by the thick line.

Fig. 6.

A, top: Gabor click waveform (black) and envelope (gray) for the −1/2-octave frequency-mismatch condition (L, left ear, 4,000 Hz; R, right ear, 2,828 Hz). Whereas envelope peaks are temporally aligned with a nominal interaural time difference (ITD) of 0 µs, the envelope flanks are temporally misaligned. Here the envelope slope (red traces) reaches a maximum 172 µs earlier in the right (lower frequency) channel (R′) than in the left channel (L′). Bottom: the binaural interaction component (BIC) as measured via peak cross-correlation (x-corr.; as in Fig. 5B) for 2 animals (S5 and S6; thin lines) reached a maximum at a nominal ITD of 0 µs, although amplitudes were skewed to negative (left-leading) ITD values that produced greater alignment of the rising envelope slopes (vertical dashed line demarcates maximum slope alignment at −172 µs; thick line indicates mean BIC). B: same as in A, but for the +1/2-octave frequency-mismatch condition, where the left ear (lower frequency) slope maximum preceded the right. Correspondingly, BIC amplitudes were skewed to the right, with the trace for 1 animal peaking at a nonzero (right leading) ITD.

Fig. 7.

A: schematic illustration of simulation procedure. From bottom to top: Gabor click stimuli (fixed at 4,000 Hz in left ear, varied over a 3-octave range in right ear) were presented to “left ear” and “right ear” populations of model auditory nerve (AN) fibers (Zilany et al. 2009, 2014; Table 1), which were relayed as excitatory ipsilateral projections (solid lines) and contralateral inhibitory projections (dashed lines) to simulated lateral superior olive (LSO) neurons (Ashida et al. 2016). In practice, monaural stimulation and binaural interaction were simulated within 11 parallel channels spanning the range of frequencies tested empirically (see methods). B: simulated LSO responses and resultant binaural interaction, derived as for empirical auditory brain stem response measurement, at a 65-dB SPL stimulus intensity. Blue traces are monaural presentation to the left ear (L); red traces are monaural presentation to the right ear (R); magenta traces are binaural presentation (B); dashed red and blue traces are the sum of left and right (L+R); black traces are the binaural interaction component (BIC). Vertical scale is arbitrary. C: response map across frequency mismatch and Gabor click intensity; as in Fig. 4B, values are normalized to the frequency-matched condition at 65 dB SPL. Contour interval = 0.25. D: normalized BIC amplitude via simulated responses (red line) vs. empirical measurements (black line and shading; from Fig. 4C) at 65 dB SPL. CF, characteristic frequency; RMS, root mean square.

Fig. 8.

A: measurements of the intrinsic matching of the inputs to lateral superior olive (LSO; in gerbil, replotted from data in Sanes and Rubel 1988) demonstrate that inputs are closely, but not exactly, matched in characteristic frequency (CF). The distribution of excitatory (Exc)-inhibitory (Inh) disparity across 85 neurons (fit with a Gaussian) yields a half-width of 0.42 octave. B: binaural interaction component (BIC) simulations (see Fig. 7 and text) were repeated with variation of Exc and Inh input number (columns) using 3 different distributions of spontaneous rate inputs (rows; L, low; M, medium; H, high). For each input number-by-type distribution, simulations were completed using Exc and Inh integration windows in the model neurons of 0.8 and 1.6 ms, respectively (red curves; Ashida et al. 2016) and “short” windows of 0.4 and 0.8 ms, respectively (blue curves; see text). All simulations were completed using exact matching of Exc and Inh input CFs (solid curves) and variable mismatching within an ~0.4-octave window (E/I CF var.; dashed curves). FWHM, full width at half maximum; integ., integration; oct., octave.