Skip to main content
NCBI home page
Journal of Neurophysiology logoLink to Journal of Neurophysiology
. 2011 Dec 7;107(5):1301–1312. doi: 10.1152/jn.00222.2011

What is the role of the medial olivocochlear system in speech-in-noise processing?

Jessica de Boer 1,2,✉, A Roger D Thornton 2, Katrin Krumbholz 1
PMCID: PMC3311680  PMID: 22157117

Abstract

The medial olivocochlear (MOC) bundle reduces the gain of the cochlear amplifier through reflexive activation by sound. Physiological results indicate that MOC-induced reduction in cochlear gain can enhance the response to signals when presented in masking noise. Some previous studies suggest that this “antimasking” effect of the MOC system plays a role in speech-in-noise perception. The present study set out to reinvestigate this hypothesis by correlating measures of MOC activity and speech-in-noise processing across a group of normal-hearing participants. MOC activity was measured using contralateral suppression of otoacoustic emissions (OAEs), and speech-in-noise processing was measured by measuring the effect of noise masking on performance in a consonant-vowel (CV) discrimination task and on auditory brain stem responses evoked by a CV syllable. Whereas there was a significant correlation between OAE suppression and both measures of speech-in-noise processing, the direction of this correlation was opposite to that predicted by the antimasking hypothesis, in that individuals with stronger OAE suppression tended to show greater noise-masking effects on CV processing. The current results indicate that reflexive MOC activation is not always beneficial to speech-in-noise processing. We propose an alternative to the antimasking hypothesis, whereby the MOC system benefits speech-in-noise processing through dynamic (e.g., attention- and experience-dependent), rather than reflexive, control of cochlear gain.

Keywords: auditory, efferent, otoacoustic emissions, auditory brain stem responses, corticofugal

the auditory system contains extensive efferent pathways. One component of the auditory efferent system, referred to as the olivocochlear bundle, projects from the superior olivary complex right back to the cochlea. The medial part of the olivocochlear bundle innervates the outer hair cells, which drive the active amplification mechanism within the cochlea. Stimulation of the medial olivocochlear (MOC) efferents by sound occurs through direct (i.e., reflexive) input from the cochlear nucleus (Guinan 1996) and reduces the gain of the cochlear amplifier (Cooper and Guinan 2006). This enables the MOC system to reflexively turn down the cochlear response to sound. Together with the middle-ear muscle (MEM) reflex, the MOC system might thus play a role in protecting the ear against damage from overexposure (Maison and Liberman 2000; Rajan 2000).

Whereas MOC activation decreases the response to signals presented in quiet (Guinan and Gifford 1988), the response to signals presented in masking noise can be increased in relation to the noise-evoked response (e.g., Kawase et al. 1993; Kawase and Liberman 1993). In the auditory nerve, noise stimulation decreases the range within which the signal can produce a response (dynamic range) by raising the baseline rate of firing and lowing the firing rate at which fibers saturate. It has been shown that MOC activation can ameliorate these effects and thus partially restore the signal response (Guinan 2006). This finding has led to the hypothesis that one of the functions of the MOC system might be to enhance the perception of signals, such as speech, in noisy environments (Giraud et al. 1997; Liberman and Guinan 1998). Whereas this “antimasking” hypothesis remains speculative, there is at least some experimental evidence supporting it. This evidence mainly comes from studies that have exploited the fact that MOC activity can be measured by measuring the suppression of otoacoustic emissions (OAEs) by contralateral acoustic stimulation (Guinan 2006). Most of these studies found a positive correlation between the amount of OAE suppression and performance in tasks involving signal-in-noise perception (de Boer and Thornton 2008; Giraud et al. 1997; Micheyl and Collet 1996; Micheyl et al. 1997). de Boer and Thornton (2008), for instance, found that participants with stronger OAE suppression, and thus greater MOC activity, showed better performance (before training) in discriminating the [bi:]/[di:] consonant-vowel (CV) contrast in noise compared with participants with lesser OAE suppression.

The aim of the current experiment was to reinvestigate the antimasking hypothesis by retesting the correlation between OAE suppression and speech-in-noise performance. Like de Boer and Thornton (2008), we measured OAE suppression and performance in a CV discrimination-in-noise task for the same set of participants. In addition, we measured the effect of noise on the neural processing of speech sounds at the brain stem level. In humans, auditory brain stem responses (ABRs) can be measured noninvasively using EEG (for a review, see Burkard and Don 2007). Traditionally, ABRs have been elicited using click or tone-pip stimuli. More recently, however, Kraus and others (for a review, see Johnson et al. 2005) used simple speech sounds (vowels or CV syllables) to evoke “speech ABRs”. Noise masking has been shown to reduce the amplitude and increase the latency of speech ABRs, and there is some evidence that the size of these effects might be predictive of performance in speech-in-noise tasks (Parbery-Clark et al. 2009). This suggests that interindividual variability in speech-in-noise perception is at least partly due to variability in the effect of noise on the brain stem processing of speech. Thus, if the correlation between OAE suppression and CV discrimination-in-noise performance observed by de Boer and Thornton (2008) reflects antimasking by the MOC system, a similar correlation would be expected to be reflected in the effects of noise on speech ABRs. We also used a different CV contrast than de Boer and Thornton (2008) to test whether their positive correlation between OAE suppression and speech-in-noise performance would generalize across acoustic differences among speech sounds. Some of the studies that measured correlation between OAE suppression and nonspeech signal-in-noise performance found either no, or even negative, correlation, depending on the acoustic properties of the signal used (Micheyl and Collet 1996; Micheyl et al. 1995).

METHODS

Each participant underwent three different experiments, all conducted within a single session, lasting ∼3 h. The first experiment measured discrimination performance of the CV syllables [da] and [ga] in a background of masking noise. The syllables were derived from natural speech tokens and presented at a signal-to-noise ratio (SNR) of 10 dB. This experiment was intended to assess participants' speech-in-noise performance. The second experiment measured suppression of click-evoked OAEs by stimulation with contralateral noise. Previous findings indicate that contralateral OAE suppression reflects reduction in cochlear gain through MOC activation by the contralateral stimulus (Guinan 2006). The third experiment measured ABRs to the CV syllable [da], generated synthetically in this case, and presented either in quiet or in two different levels of masking noise corresponding to SNRs of 10 and 20 dB.

All measurements were conducted in a sound-attenuating booth. During the OAE and ABR measurements, participants watched a silent movie with subtitles to remain alert.

CV discrimination-in-noise task.

This task required participants to discriminate speech sounds from a 96-step continuum between the CV syllables [da] and [ga], taken from the commercial auditory training package Phonomena (MindWeavers, Oxford, UK). The continuum was created by analyzing recordings of the syllables [da] and [ga], spoken by an adult male, into spectral (15 reflection coefficients) and source [including voicing and fundamental frequency (F0)] parameters using linear-predictive coding. The endpoints of the continuum were then resynthesised using the source parameters of the original [da] syllable, together with the appropriate spectral parameters for the respective syllable ([da] or [ga]). The 94 intervening sounds were synthesized using the same source parameters (from the original [da] syllable) and intermediate spectral parameters, derived by linear interpolation between the two endpoints ([da] and [ga]). With the use of the same source parameters, all sounds had the same duration (337 ms) and root-mean-square (RMS) level, and their F0 followed the same trajectory, decreasing gradually from 150 to 105 Hz over the syllable duration.

The spectral variation along the continuum affected only the consonant portion of the syllables (within approximately the first 50 ms). Figure 1 indicates that the most salient difference between [da] and [ga] is in the direction of the transition in the third formant, F3, with F3 sweeping downward in [da] and upward in [ga]. The formants of the vowel portion of the syllables ([a]) were constant across the continuum, with the first to fourth formants (F1–F4) at ∼750, ∼1,050, ∼2,250, and ∼3,500 Hz, respectively.

Fig. 1.

Fig. 1.

Open in a new tab

Spectrograms of the endpoints of the [da]/[ga] continuum used in the consonant-vowel (CV) discrimination-in-noise task. The white arrowheads mark the third formant transitions (F3), which are the most salient difference between the 2 syllables.

The speech sounds were presented in a background of broadband (0–22 kHz) Gaussian noise, presented continuously from a 60-s circular buffer, generated with Adobe Audition (Adobe Systems, San Jose, CA), at an average sensation level (SL) of 40 dB. The speech sounds were presented at a RMS level of 10 dB above that of the noise (SNR = 10 dB).

Syllable discrimination thresholds were measured using a three-interval, two-alternative AXB task, where X was equally likely to be identical to either A or B. The participants' task was to indicate whether X was identical to A or B by pressing one of two response buttons. Discrimination threshold was estimated using a three-down, one-up adaptive staircase procedure, which tracks the distance between A and B that yields 79% correct responses. In each trial, the A and B sounds were taken from symmetrically opposite points along the 96-step continuum between [da] and [ga], with A equally likely to be closer to either endpoint. In the first trial of each adaptive track, the distance between A and B was maximal (i.e., A and B were at the continuum endpoints). After each three consecutive correct responses, the A–B distance was then decreased by 10 steps (i.e., both A and B were moved by five steps toward the continuum center) or increased by the same amount after each incorrect response. The step size of the staircase increments and decrements was reduced to six after six reversals in the A–B distance, and the track was terminated after 12 reversals, or 60 trials, whichever occurred earlier. Each threshold estimate was taken as the arithmetic mean of the A–B distance at the last five reversals within each track and expressed as a percentage of the maximum possible distance (i.e., 94 steps between the continuum endpoints). Each participant performed three consecutive blocks of three adaptive tracks, with short breaks in between blocks, during which the masking noise was paused.

The experiment was controlled using an in-house psychoacoustic software package (STAR), written in Visual Basic. The interval among the stimuli in each trial (A, X, and B) was 500 ms, and the intertrial interval was 2 s. Visual feedback was given at the end of each trial. The speech sounds and noise were mixed digitally and digital-to-analog converted with a 16-bit amplitude resolution and a 44.1-kHz sampling rate using an M-Audio Audiophile 2496 soundcard (Avid Technology, Burlington, MA). The soundcard output was fed through a Hatfield attenuator (Type 2125) and presented to the participants' right ear via a general-purpose OAE probe transducer (Otodynamics, Hatfield, UK).

OAE measurements.

Click-evoked OAEs were recorded using an in-house system for recording auditory-evoked responses (MLS2001). The system consists of a digital signal processing board and is controlled by custom-written (Visual Basic) software. The stimulus was a 100-μs click, generated at a 30-kHz sampling rate, and presented to the participants' right ear via the same OAE probe transducer as used in the CV discrimination-in-noise task. The clicks were presented at a rate of 20/s and two different peak-equivalent (pe) levels of 60 and 70 dB sound pressure level (SPL). OAEs were recorded using the OAE probe microphone, digitized at a 30-kHz sampling rate using an 18-bit ∑Δ analog-to-digital (AD) converter, and averaged online over 1,500 trials, which were rejected if the response amplitude exceeded 5 mPa within the period from 6 to 16 ms after click presentation.

To assess MOC activity, continuous broadband Gaussian noise was presented to the contralateral (left) ear during one-half of the trials. The noise was generated in the same way as for the CV discrimination-in-noise task and presented at 40 dB SL via a second OAE probe. A total of four OAE recordings was made for each stimulus level, two with and two without contralateral noise in alternating order. This yielded two OAEs, or “replicates”, for each condition (i.e., each combination of click level and contralateral noise condition). Each recording lasted ∼2 min. Consecutive recordings were separated by ∼30 s of silence. The order of presentation of stimulus levels was counterbalanced across participants.

Offline processing of OAEs was performed in MATBLAB (MathWorks, Natick, MA). First, the OAEs were bandpass filtered from 250 to 6,000 Hz with zero-phase delay by applying a second-order Butterworth filter in both forward and reverse time direction (yielding a −12-dB/octave filter rolloff). To avoid contamination by the stimulus artefact, present within the earlier period of the response (<∼6 ms), each OAE was multiplied with a 10-ms window, applied between 6 and 16 ms after the click. The window edges were rounded according to a 0.5-ms quarter-cosine function. The OAE amplitude for each condition was taken as the real part of the cross-spectrum between the respective replicates. This procedure avoids noise bias (Wong and Bickford 1980). OAEs were accepted as valid only if the correlation between the two replicates for each condition, referred to as reproducibility, was >0.5 (Lutman 1993). Two participants failed to meet this criterion, and so, their OAEs were excluded from further analysis. One participant only just passed the criterion with a reproducibility of 0.5 for one of the conditions. The remaining participants had a reproducibility of 0.7 or greater for all conditions. The reproducibility for all included OAEs averaged 0.95 (±0.008).

The OAE amplitude evoked by clicks is known to increase with increasing stimulus level (Kemp 1978). In normal ears, and without contralateral noise stimulation, the function relating OAE amplitude (in logarithmic units) to stimulus level, which is often referred to as input/output (I/O) function, tends to be compressive (i.e., has a slope of less than unity). Thus a 10-dB increase in stimulus level produces an increase in OAE amplitude of <10 dB [typically between 3 and 5 dB (Uppenkamp and Kollmeier 1994; Veuillet et al. 1996)]. This is due to the cochlear gain decreasing with increasing stimulus level and thus producing compressive growth of the cochlear response (Robles and Ruggero 2001). Reduction in cochlear gain (e.g., through hearing loss) reduces cochlear compression and thus increases the slope of the I/O function (Veuillet et al. 1996). This means that the I/O function slope can be taken as a measure of cochlear gain (with larger slopes indicating less gain). In the current study, the I/O function slope was estimated by subtracting the OAE amplitude for the 60-dB SPL clicks from that for the 70-dB SPL clicks and dividing the difference by the difference in click level (10 dB). An increase in I/O slope by the contralateral noise, referred to as I/O suppression, would be assumed to reflect a reduction in cochlear gain through noise-evoked MOC activation. In addition to the I/O suppression, we analyzed the suppression in OAE amplitude at each click level separately (difference between OAE amplitudes without and with contralateral noise). OAE amplitude suppression is the most common measure of MOC activity.

To control for activation of the MEM reflex by the contralateral noise, which could also cause a reduction in OAE amplitude, the contralateral MEM reflex threshold for broadband noise (0.125–4 kHz) was measured for each participant using a GSI-33 tympanometer (Grason-Stadler, Eden Prairie, MN) and an ER-2 insert earphone (Etymotic Research, Elk Grove Village, IL) for presenting the contralateral noise. The MEM reflex threshold ranged from 55 to 80 dB normal hearing level (nHL), with an average of 65 dB nHL, and thus exceeded the noise level used for the OAE measurements (40 dB SL) by ∼25 dB. This means that it is unlikely that the MEM reflex had a significant contribution to OAE suppression.

ABR measurements.

ABRs were acquired using the same recording system as used for the OAE measurements (MLS2001). In this case, the stimulus was either a 100-μs click or a 40-ms synthetic [da] syllable, generated as in Johnson et al. (2005) using the Klatt synthesizer (Klatt 1980). The syllable comprised a 10-ms onset burst frication at the third–fifth formant frequencies (F3–F5), followed by a 30-ms formant transition period, during which F0 was fixed at 110 Hz, F1 increased from 220 to 720 Hz, F2 decreased from 1,700 to 1,240 Hz, and F3 decreased from 2,580 to 2,500 Hz. F4 and F5 remained constant at 3,600 and 4,500 Hz, respectively, throughout the voiced portion of the syllable.

The click and syllable were presented at rates of 20/s and 10/s, respectively, and a pe level of 80 dB SPL. For the syllable, this corresponds to a RMS level of 51 dB SPL. The stimuli were generated at a 30-kHz sampling rate and presented to the participants' right ear via the same OAE probe as used in the other two experiments. Sound-level calibration was performed using the OAE probe microphone and a 1-cm3 test cavity (Otodynamics). Stimuli were presented with opposite polarity on alternate trials (alternating polarity stimulation) to minimize contamination by the stimulus artefact or cochlear microphonics (CMs). Both the stimulus artefact and CMs preserve stimulus polarity and will thus cancel out in the average response when using alternating-polarity stimulation.

The click was presented only in quiet. The syllable was presented in quiet and in two levels of continuous noise. The noise was generated in the same way as for the other two experiments and presented at a RMS level of either 31 or 41 dB SPL, yielding a SNR of 20 or 10 dB, respectively. As in the CV discrimination-in-noise task, the noise was presented to the same ear as the syllable (right ear) using the second transducer in the OAE probe.

ABRs were recorded with three electrodes, placed at the vertex (noninverting), ipsilateral (right) earlobe (inverting), and forehead (ground). Electrode impedances were below 5 kΩ, and interelectrode impedance differences were no greater than 2 kΩ. The electrode signals were amplified by a factor of 50,000, lowpass filtered at 3,000 Hz (−24 dB/octave rolloff), and highpass filtered at 30 Hz (−12 dB/octave rolloff) using an ICP511 alternating current-difference amplifier (Grass Telefactor, West Warwick, RI) located inside the recording booth. The responses were digitized at a 15-kHz sampling rate using the same 18-bit ∑Δ AD converter as for the OAE measurements and averaged over 3,000 trials. The click ABR was recorded first in all participants and then the responses to the [da] stimulus, with the order of the in-quiet condition and the two noise-masked conditions (20 and 10 dB), counterbalanced, according to a three-by-three Latin square design.

As for the OAEs, offline processing of ABRs was performed in MATLAB. To reduce the remaining noise contamination, the digitized responses were further bandpass filtered between 100 and 2,000 Hz by applying a second-order Butterworth filter in both forward and reverse time direction (−12-dB/octave rolloff). All individual click ABRs exhibited a clear, positive peak at a latency of ∼6 ms, which is the ABR wave V (see Fig. 5A). In the [da]-evoked ABR (speech ABR), five positive peaks could be identified clearly in the average response across participants (grand average response), with peak latencies similar to those observed by Johnson et al. (2005). For each participant, the latencies of wave V in the click ABR and the five peaks in each of the three speech ABRs (in-quiet and two noise-masked conditions) were measured by finding the largest positive peak within the 3-ms time window around the respective grand-average peak latency. To avoid mistaking peaks in the recording noise for genuine ABR peaks, Peirce's criterion for outliers (Ross 2003) was used to discard peaks with outlier latencies. Five participants failed to show reliable latencies for all five peaks in all three speech ABRs and were thus excluded from further analysis.

Fig. 5.

Fig. 5.

Open in a new tab

Auditory brain stem responses (ABRs) evoked by the click and the [da] stimulus presented in quiet (A and B) or in noise (C). A: click-evoked (black lines) and [da]-evoked (red lines) ABRs for 3 representative participants (S17, S35, and S14; top 3 traces) and averaged over 24 participants (bottom trace). The blue arrowheads mark the onset responses (OnRs) to the click and [da] stimulus, and the green arrowheads mark the 4 additional peaks observed in the [da]-evoked response. B: aligning the [da]-evoked ABR (red line) with the stimulus waveform (purple line) shows that the first 3 of the additional peaks in this response [labeled glottal pulse (GP)1–GP3] were elicited by the glottal pulses in the vowel portion of the stimulus (marked by green arrowheads), and the 4th peak [off-response (OffR)] was elicited by the stimulus offset (cyan arrowhead). C: grand-average [da]-evoked ABRs in quiet (red line) and in 2 noise levels, corresponding to signal-to-noise ratios (SNRs) of 20 (green line) and 10 dB (blue line).

Statistics.

All statistical analyses were performed in SPSS (IBM, Armonk, NY). Repeated-measures (RM) ANOVA and analysis of covariance (ANCOVA) were performed using a type-III general linear model. F-tests were Huynh-Feldt corrected when equality of variance (sphericity) did not apply, and Bonferroni correction was used for multiple comparisons in post hoc tests. Correlations among the measures obtained from the three experiments (CV discrimination-in-noise task, OAE suppression, and speech ABR measurements) were calculated using Pearson's product-moment correlation coefficient, r. Only participants who had not been excluded based on their OAE or ABR results were used for the correlation analyses (18 participants).

Participants.

A total of 24 participants (12 male, mean age 22.6 yr, SD 3.3 yr) took part in this study after having given written, informed consent. Prior to testing, all participants were screened for normal hearing thresholds at audiometric frequencies (0.5, 1, 2, 4, and 8 kHz) in both ears using a Kamplex KC50 audiometer. Participants were also screened for normal middle-ear pressure (between −100 and 50 daPa) and normal middle-ear compliance (between 0.2 and 2.5 ml) using a GSI-33 middle-ear analyzer (Grason-Stadler). No bone-conduction testing was performed. Table 1 shows that most participants had a hearing threshold of 15 dB HL or better in both ears. One participant had a hearing threshold of 25 dB HL at 8 kHz in the left ear. Whereas 25 dB HL borders on mild hearing loss, we decided to nevertheless include this participant, because the loss was at a relatively remote frequency and in the ear opposite to the ear to which the speech stimuli were presented. All participants spoke English as their first language and were right-handed, according to the Edinburgh inventory (Oldfield 1971). The procedures used complied with the Code of Ethics of the World Medical Association (Declaration of Helsinki) and were approved by the Southampton and South West Hampshire Research Ethics Committee.

Table 1.

Mean audiometric threshold (in dB hearing level) and threshold range across 24 participants for each audiometric frequency separately and averaged across frequencies for the left and right ears

Ear Freq. (kHz) 0.5 1 2 4 8 Pure tone average
Right Mean 2.9 3.5 2.7 0.2 2.1 2.3
SD 5.7 5.4 6.1 5.0 8.1 4.2
Min. −5 −5 −10 −10 −10 −4
Max. 10 15 10 10 15 10
Left Mean 4.8 3.1 2.7 1.5 2.5 2.9
SD 5.4 5.7 6.3 6.3 8.1 4.8
Min. −5 −5 −10 −10 −10 −5
Max. 15 15 15 10 25 16
Open in a new tab

RESULTS

CV discrimination in noise.

Discrimination of the [da]/[ga] syllable contrast in noise (SNR = 10 dB) proved to be a challenging task; the average discrimination threshold across 24 participants corresponded to 76% (±1.5%) of the [da]/[ga] continuum (i.e., sounds along a linear continuum between [da] and [ga] were just discriminable when the acoustic difference between them amounted to 76% of the acoustic difference between [da] and [ga]; see methods). Performance was relatively stable across the nine threshold estimates obtained for each participant (Fig. 2A; thresholds were measured in three blocks of three adaptive tracks; see methods); a RM ANOVA of the threshold estimates from individual tracks, with block and track number as within-participants factors and sex as between-participants factor, revealed no significant main effect of block or track number. There was, however, some improvement in syllable discrimination performance during the first block of tracks and some deterioration during the second and third blocks, giving rise to a marginally significant interaction between block and track number [F(4.0,87.3) = 2.3, P = 0.063]. The performance improvement during the first block may reflect rapid perceptual learning (Hawkey et al. 2004) or familiarization with the task (procedural learning) (Wright and Fitzgerald 2001), and the performance deterioration during the second and third blocks was probably due to fatigue or attentional drift. Apart from these small, within-block threshold variations, individual performance was reasonably consistent over time, as shown by a significant correlation between the average thresholds for the first and third blocks of tracks (r = 0.5, P = 0.01). This implies a reliable ranking of participants in terms of CV discrimination-in-noise performance. The ANOVA showed no significant main effect of, or interactions with, sex. This means that the observed ranking of participants in terms of CV discrimination-in-noise performance was not due to a difference between male and female participants.

Fig. 2.

Fig. 2.

Open in a new tab

CV discrimination-in-noise thresholds, expressed as a percentage of the maximal possible difference within the CV continuum. A: average CV discrimination-in-noise thresholds over 24 participants as a function of the block and track number. Error bars denote the SE. B: correlation across participants between CV discrimination-in-noise thresholds, averaged over tracks (ordinate), and audiometric thresholds, averaged over frequencies (abscissa). Open circles indicate female (F; n = 12) and closed circles male (M; n = 12) participants. The dash-dotted line is the regression line. The correlation parameters [Pearson's product-moment correlation coefficient (r)] are shown in the top right corner. nHL, normal hearing level.

Table 1 (see methods) indicates that the audiometric thresholds differed by as much as 15–25 dB among participants. This interindividual variability in hearing sensitivity might be expected to explain at least some of the observed variability in CV discrimination-in-noise performance. This expectation is supported by Fig. 2B, which shows that there was a tendency for participants with better average hearing thresholds across audiometric frequencies (0.5–8 kHz) to show better CV discrimination-in-noise performance. This correlation between the average audiometric and CV discrimination-in-noise thresholds, however, was only marginally significant (r = 0.38, P = 0.067).

Effect of contralateral noise on click-evoked OAEs.

The amplitude of the click-evoked OAEs was generally smaller (suppressed) when the contralateral noise was present (Fig. 3A) than when it was absent. This was true for both click levels used (60 and 70 dB peSPL). The mean suppression amounted to 1.1 dB and was confirmed to be significant [F(1,20) = 47.8, P < 0.001] by a RM ANOVA of the individual OAE amplitudes, with click level (60 and 70 dB peSPL) and contralateral noise condition (on/off) as within-participants factors and sex as between-participants factor. The ANOVA also showed a highly significant interaction between click level and contralateral noise condition [F(1,20) = 16.3, P = 0.001], indicating that the contralateral suppression effect was stronger at the lower (60 dB peSPL) rather than higher (70 dB peSPL) click level (1.3 vs. 0.9 dB). This means that the slope of the function relating OAE amplitude to click level (I/O function) was steeper (less compressive) when the contralateral noise was present than when it was absent. As explained in methods, an increase in the I/O function slope by contralateral noise stimulation (I/O suppression) indicates a reduction in cochlear amplification gain through MOC activation. Finally, there was also a marginally significant, three-way interaction among click level, contralateral noise condition, and sex [F(1,20) = 4.1, P = 0.056]. As can be seen from Fig. 3B, this interaction came about as a result of the female participants showing a greater effect of click level on OAE suppression (i.e., difference in suppression at 60 vs. 70 dB peSPL) than the male participants (0.73 vs. 0.24 dB in females and males, respectively). This meant that the I/O suppression was stronger in the female than male participants. The fact that this difference was only marginally significant (P = 0.056) was due to a relatively large within-sex variability in OAE suppression.

Fig. 3.

Fig. 3.

Open in a new tab

Suppression of click-evoked otoacoustic emissions (OAEs) by contralateral (contra) noise. A: OAE amplitudes at 2 click levels [60 and 70 dB peak-equivalent sound-pressure level (peSPL)] with contralateral noise on (white bars) or off (black bars). B: OAE amplitude suppression at 60 (dark gray bars) and 70 (light gray bars) dB peSPL, averaged over female (n = 11) and male (n = 11) participants separately. Error bars denote the SE. Significant differences are *P < 0.05; **P < 0.01.

As for the CV discrimination-in-noise thresholds, the variability in OAE suppression might be expected to be at least partly due to variability in hearing sensitivity (e.g., Keppler et al. 2010). Whereas there were no significant simple correlations between the average audiometric thresholds and any of the OAE suppression measures, there was a partial negative correlation between the audiometric thresholds and OAE amplitude suppression for the 70-dB peSPL click level when using sex as a controlling variable (Fig. 4B; r = −0.46, P = 0.032). This indicates that participants with better hearing tended to show stronger OAE amplitude suppression at the higher click level. The correlation was nonsignificant at 60 dB peSPL (Fig. 4A). For I/O suppression, there was a tendency for better hearing to be associated with lesser suppression, but this tendency was nonsignificant (Fig. 4C). These results suggest that, as a measure of MOC activity, I/O suppression may be relatively more robust to variability in hearing sensitivity than OAE amplitude suppression.

Fig. 4.

Fig. 4.

Open in a new tab

Partial correlations across participants between OAE suppression measures (ordinates) and average audiometric thresholds (abscissae), with sex as a controlling variable (i.e., each data point was means adjusted to remove sex-related differences). A: OAE amplitude suppression at 60 dB peSPL click level. B: OAE amplitude suppression at 70 dB peSPL click level. C: input/output (I/O) suppression. As before, open circles indicate female (n = 11) and closed circles male (n = 11) participants. The dotted lines are the regression lines. The correlation parameters are shown in the top right corner of each panel.

ABRs to the click and speech stimuli in quiet.

The ABRs to both the click and the [da] stimulus in quiet showed a robust onset response (OnR), which was highly consistent across participants (Fig. 5A). The OnR to the click stimuli contained a prominent, positive deflection at an average latency of 6 ms (±0.05 ms), which is consistent with previous findings for wave V of the click-evoked ABR (Burkard and Don 2007). Apart from a slightly longer latency (7.1 ± 0.06 ms), the OnR to the [da] stimulus had a very similar morphology as the click-evoked OnR. Moreover, the latencies of the click- and [da]-evoked OnRs were highly correlated across participants (r = 0.8, P < 0.001), supporting previous suggestions that click- and speech-evoked ABRs have a similar neural origin (Song et al. 2006). The [da]-evoked ABR exhibited four additional peaks not observed in the click-evoked ABR. These additional peaks can be seen particularly clearly in the grand-average response in Fig. 5A. Temporally aligning the [da]-evoked ABR with the stimulus waveform (by advancing the ABR by the latency of its OnR; Fig. 5B) shows that the first three of the four additional peaks were time locked to the onsets of the glottal pulses (GPs) in the vowel portion of the stimulus. For that reason, these peaks will henceforth be labeled GP1–GP3. They have been referred to as “frequency-following response” in previous studies (Hornickel et al. 2008). In contrast, the fourth, additional peak was time locked to the syllable offset, suggesting that it represents an off-response (OffR).

Effect of noise on the speech ABR.

In line with previous findings, noise masking caused a reduction in the amplitudes and an increase in the latencies of the [da]-evoked ABR peaks (Fig. 5C). Only the latency effect will be considered further here, as it has been shown to represent a more reliable measure of noise masking on ABRs than the amplitude effect (Burkard and Hecox 1983; Kramer and Teas 1982). The individual peak latencies were submitted to a RM ANOVA, with peak label (OnR, GP1–GP3, OffR) and noise condition (in quiet, SNR = 20 and 10 dB) as within-participants factors. Sex, which is known to influence the latency of wave V in the click-evoked ABR (Durrant et al. 1990), was included as a between-participants factor. As expected based on Fig. 5C and previous findings, the ANOVA yielded a highly significant main effect of noise condition [F(1.8,30.6) = 174.8, P < 0.001], brought about by an increase in ABR peak latencies in the two noise-masked conditions (SNR = 20 and 10 dB) compared with the in-quiet condition. Post hoc comparisons showed that the latency increase, averaged over all peaks, was significant at both SNRs (20 and 10 dB, both P < 0.001) and was significantly larger for the lower (10 dB) than the higher (20 dB) SNR (P < 0.001). There was no significant main effect of, or interaction with, sex on the ABR latencies. The ANOVA also revealed a significant interaction between peak label and noise condition [F(6.2,105.9) = 7.2, P < 0.001]. Post hoc comparisons indicated that this interaction arose as a result of the OnR peak exhibiting a larger noise-induced latency shift (i.e., latency difference between each of the two noise-masked conditions and the in-quiet condition) than any of the later peaks (GP1–GP3 and OffR; see Fig. 5C). This was true for both noise levels (P < 0.02; Fig. 6). There were no significant differences in latency shift among any of the later peaks (Fig. 6).

Fig. 6.

Fig. 6.

Open in a new tab

Average noise-induced latency shifts (n = 19) for each of the 5 peaks within the [da]-evoked ABR (OnR, GP1–GP3, OffR) at SNR = 20 dB (light gray bars) and SNR = 10 dB (dark gray bars). Error bars denote the SE.

To test whether the degree of noise degradation of the speech ABR depended on hearing sensitivity, a RM ANCOVA was performed on the noise-induced ABR peak-latency shifts, with peak label (OnR, GP1–GP3, OffR) and SNR (20 and 10 dB) as within-participants factors, sex as between-participants factor, and average audiometric threshold as covariate. This analysis revealed no significant main effect of average audiometric threshold [F(1,16) = 0.6, P = 0.461] and no interaction between audiometric threshold and SNR [F(1,16) = 2.4, P = 0.141]. There was, however, a marginally significant interaction between audiometric threshold and peak label [F(2.9,46.5) = 2.8, P = 0.053], indicating that the effect of hearing sensitivity on the ABR latency shifts varied among peaks. To investigate this interaction further, we averaged the ABR latency shifts across SNRs and calculated their correlations with the audiometric thresholds. Only the correlation for the OffR peak was significant, with larger latency shifts (i.e., greater effects of noise masking) for higher audiometric thresholds (r = 0.48, P = 0.037).

As for the absolute latencies, the main effect of sex on the latency shifts was nonsignificant [F(1,16) = 1.8, P = 0.193], but there was a significant interaction between sex and peak label [F(2.9,46.5) = 3.5, P = 0.023]. Independent t-tests showed that this interaction was due to the GP3 peak, but not any of the other peaks, showing a larger latency shift (averaged over both SNRs) in male compared with female participants (difference = 0.38 ms, P = 0.21).

Relationship between perceptual and ABR measures of speech-in-noise processing.

Based on previous findings (Cunningham et al. 2001; Parbery-Clark et al. 2009), participants with poorer performance in the CV discrimination-in-noise task were expected to exhibit a greater noise effect on the speech ABR (i.e., larger noise-induced latency shifts). To test this expectation, the ABR latency shifts were submitted to an RM ANCOVA, with CV discrimination-in-noise threshold as covariate. As before, the analysis also included peak label (OnR, GP1–GP3, OffR) and SNR (20 and 10 dB) as within-participants factors, sex as a between-participants factor, and average audiometric threshold as another covariate. It revealed a significant main effect of CV discrimination-in-noise threshold [F(1,15) = 6.2, P = 0.025], as well as a significant interaction with SNR [F(1,15) = 6.9, P = 0.019]. These effects were independent of peak label [three-way interaction among CV discrimination-in-noise thresholds, SNR, and peak label was nonsignificant; F(4,60) = 0.8, P = 0.533]. Figure 7, A and B, shows the relationship between the CV discrimination-in-noise thresholds and the ABR latency shifts (averaged over all peaks) for each SNR separately. The figure shows that there was a highly significant, positive correlation between CV discrimination-in-noise performance and the size of the ABR latency shifts at the lower SNR of 10 dB (Fig. 7B; r = 0.63, P = 0.004) but not at the higher SNR of 20 dB (Fig. 7A; r = 0.09, P = 0.721). The correlation implies that participants with greater noise effects on the speech ABR tended to show poorer CV discrimination-in-noise performance. The absence of any correlation at the higher SNR was probably due to the noise effect being too small at this SNR. Importantly, both the main effect of audiometric threshold [F(1,15) = 0.02, P = 0.887] and the interaction between audiometric threshold and SNR [F(1,15) = 1.0, P = 0.326] were nonsignificant. This means that the relationship between the CV discrimination-in-noise thresholds and the ABR latency shifts was not caused by a shared dependence on hearing sensitivity.

Fig. 7.

Fig. 7.

Open in a new tab

Correlations across participants between noise-induced ABR latency shifts, averaged over ABR peaks (ordinates), and CV discrimination-in-noise thresholds (abscissae), calculated for each SNR separately. A: latency shifts at SNR = 20 dB. B: latency shifts at SNR = 10 dB. Open circles indicate female (n = 9) and closed circles male (n = 10) participants. The dash-dotted lines are the regression lines. The correlation parameters are shown in the top right corner of each panel.

Relationship between OAE suppression and speech-in-noise processing.

According to the antimasking hypothesis, the MOC system enhances speech-in-noise perception by reflexively turning down cochlear gain. Participants with stronger OAE suppression (indicative of greater MOC activity) were thus expected to show better performance (lower thresholds) in the CV discrimination-in-noise task and a lesser effect of noise (smaller noise-induced latency shifts) on the speech ABR. To test this expectation, we performed two separate RM ANCOVAs on the ABR latency shifts. In the first analysis, the OAE amplitude suppression at both click levels (60 and 70 dB peSPL) was included as covariates. As before, the analysis also included peak label (OnR, GP1–GP3, OffR) and SNR (20 and 10 dB) as within-participants factors, sex as between-participants factor, and average audiometric threshold as another covariate. The results revealed a highly significant main effect of OAE amplitude suppression at the lower [60 dB peSPL; F(1,13) = 15.0, P = 0.002], but not the higher [70 dB peSPL; F(1,13) = 2.7, P = 0.126], click level. There were no significant interactions. Figure 8A illustrates that the main effect of OAE amplitude suppression at 60 dB peSPL was due to a significant positive, partial correlation (using sex as the controlling variable) between the OAE suppression and the ABR latency shifts averaged over both SNRs and all peaks (i.e., stronger OAE suppression was associated with larger ABR latency shifts; r = 0.59, P = 0.013). No such correlation was found at the 70-dB peSPL click level (Fig. 8B). Unlike the previous analyses, the current ANCOVA also revealed a significant main effect of sex [F(1,13) = 10.1, P = 0.007], with larger ABR latency shifts in the male than female participants (difference between marginal means = 0.2 ms). The reason why the previous analyses did not show this effect was because they did not account for the sex differences in OAE suppression.

Fig. 8.

Fig. 8.

Open in a new tab

Partial correlations across participants between OAE suppression measures (ordinates) and noise-induced ABR latency shifts, averaged over ABR peaks and SNRs (abscissae), with sex as a controlling variable (i.e., each data point was means adjusted to remove sex-related differences). A: OAE amplitude suppression at 60 dB peSPL click level. B: OAE amplitude suppression at 70 dB peSPL click level. C: I/O suppression. The correlations were calculated for n = 9 female (open symbols) and n = 9 male (closed symbols) participants. The dotted lines are the regression lines (see the top right corner of each panel for correlation parameters).

The second analysis was identical to the first, except that I/O suppression, rather than OAE amplitude suppression, was used as covariate. This analysis revealed a highly significant main effect of I/O suppression [F(1,14) = 17.6, P = 0.001]. Figure 8C shows that as for the OAE amplitude suppression at 60 dB peSPL, the main effect of I/O suppression was due to a significant positive, partial correlation between the I/O suppression and the ABR latency shifts (r = 0.74, P = 0.001). As for the first analysis, there was a significant main effect of sex [F(1,14) = 18.7, P = 0.001]. None of the interactions was significant. Neither analysis showed any significant main effect of, or interaction with, audiometric threshold. When the partial correlations in Fig. 8, A–C, were recalculated, while also controlling for audiometric thresholds, the correlations became more significant for the OAE amplitude suppression at 60 dB peSPL (r = 0.7, P = 0.001), remained nonsignificant for the OAE amplitude suppression at 70 dB peSPL (r = 0.23, P = 0.371), and were essentially unchanged for the I/O suppression (r = 0.61, P = 0.009). This indicates that the observed correlations between the OAE suppression measures and the ABR latency shifts were not caused by a shared dependence on hearing sensitivity.

The positive correlation between the OAE suppression measures and the ABR latency shifts on the one hand, together with the positive correlation between the ABR latency shifts and the CV discrimination-in-noise thresholds on the other hand, suggested that contrary to the antimasking hypothesis, stronger OAE suppression would be associated with poorer, rather than better, CV discrimination-in-noise performance. Figure 9C shows that there was indeed a highly significant positive correlation between the I/O suppression and the CV discrimination-in-noise thresholds (r = 0.68, P < 0.001). This correlation was practically unaffected when audiometric sensitivity was controlled for. The correlation with the OAE amplitude suppression at 60 dB peSPL was only marginal (r = 0.36, P = 0.097) but became significant when audiometric sensitivity was controlled for (r = 0.48, P = 0.028). The correlation with the OAE amplitude suppression at 70 dB peSPL was nonsignificant (r = −0.07, P = 0.775) and remained nonsignificant when controlling for audiometric thresholds (r = 0.13, P = 0.566).

Fig. 9.

Fig. 9.

Open in a new tab

Partial correlations across participants between OAE suppression measures (ordinates) and CV discrimination-in-noise thresholds (abscissae), with sex as a controlling variable (i.e., each data point was means adjusted to remove sex-related differences). A: OAE amplitude suppression at 60 dB peSPL click level. B: OAE amplitude suppression at 70 dB peSPL click level. C: I/O suppression. The correlations were calculated with n = 11 female (open symbols) and n = 11 male (closed symbols) participants. The dotted lines are the regression lines (see the top right corner of each panel for correlation parameters).

DISCUSSION

This study set out to test the antimasking hypothesis of MOC function, which posits that MOC activation enhances speech-in-noise perception by a reflexive turning-down of cochlear amplification. The data showed a significant correlation between OAE suppression (which would be presumed to index the general responsiveness of the MOC system in a given individual) and performance in a CV ([da]/[ga]) discrimination-in-noise task. However, the direction of this correlation was opposite to that predicted by the MOC antimasking hypothesis, in that participants with stronger OAE suppression showed poorer, rather than better, CV discrimination-in-noise performance. Importantly, a similar correlation was found between OAE suppression and noise-induced latency shifts of CV ([da])-evoked ABRs, with larger ABR latency shifts being associated with stronger OAE suppression. Noise-induced ABR latency shifts are thought to be indicative of the degree of noise masking (Burkard and Hecox 1983). In accordance with this assumption, we also found larger latency shifts of the CV-evoked ABR to be associated with poorer CV discrimination-in-noise performance. The ABR latency shifts and the CV discrimination-in-noise performance were measured under vastly different task conditions: the ABR latency shifts were measured under passive listening conditions (while participants were watching a subtitled movie), whereas the CV discrimination-in-noise performance was measured in a complex and cognitively demanding active listening task. The fact that both measures nevertheless showed a similar correlation with OAE suppression indicates that this correlation is unlikely to have been mediated by task-related effects on MOC activity. Rather, the correlation probably reflects interindividual differences in the general responsiveness of the MOC system to reflexive sound stimulation and thus the size of its modulatory effect on cochlear gain.

Note also that the correlation between OAE suppression and CV-in-noise processing (CV discrimination-in-noise performance and noise-induced ABR latency shifts) cannot have been mediated by a shared dependence on hearing sensitivity. Whereas both the CV discrimination-in-noise thresholds and the OAE amplitude suppression at the higher click level (70 dB peSPL) correlated with the audiometric thresholds, the two correlations were of the same direction (i.e., higher audiometric thresholds were associated with poorer CV discrimination-in-noise performance and weaker OAE amplitude suppression). Thus their combination could only have created a positive, but not the observed negative, correlation between OAE suppression and CV-in-noise processing.

On average, the female participants showed stronger OAE suppression and smaller ABR latency shifts than the male participants. However, these sex-related differences were of opposite direction to the overall correlation between the OAE suppression and ABR latency shifts. This indicates that the correlation was not caused by the sex differences. It also suggests that the sex differences in the OAE suppression and ABR latency shifts are functionally unrelated. It is likely that the sex difference in the OAE suppression is related to the previously reported sex difference in overall OAE amplitudes (McFadden 2009) and similarly, that the difference in the ABR latency shifts is related to the difference in absolute ABR latencies (Durrant et al. 1990).

The relationship between OAE suppression and CV-in-noise processing was dependent on the click level used to elicit the OAEs: only the OAE amplitude suppression at the lower click level (60 dB peSPL) and the I/O suppression, but not the OAE amplitude suppression at the higher click level (70 dB peSPL), showed a significant correlation with CV-in-noise processing. The absence of any significant correlation at the higher click level is likely to be related to the decrease in cochlear gain with increasing sound level (Robles and Ruggero 2001): lesser gain at higher levels would be expected to mean lesser sensitivity to MOC-induced reduction in gain.

Burkard and Hecox (1983) showed that the effect of noise on the click-evoked ABR is similar to the effect of adaptation. Like noise, prior stimulation with a train of adapting clicks reduced the amplitude and increased the latency of the ABR wave V. Importantly, the adaptation- and noise-induced effects were mutually occlusive, in that adapted responses showed lesser noise effects than unadapted responses and vice versa. This suggests that adaptation and noise effects are caused by common physiological mechanisms. This might explain why the OnR peak in the speech ABR measured in the current study showed a larger noise-induced latency shift than the later peaks (GP1–GP3, OffR); the later peaks were presumably adapted by the preceding ones. Adaptation is thought to be caused by synaptic depletion and somatic afterhyperpolarization (Kramer and Teas 1982; Smith 1979). These processes are affected by auditory neuropathy, which is also known to create speech-in-noise deficits (Rance 2005). Thus it is possible that subclinical levels of this pathology may also have contributed to the interindividual variability in CV-in-noise processing observed in the current study.

In the current study, the audiometric thresholds were a poor predictor of CV-in-noise processing. Nevertheless, it is possible that some of the observed variability in CV-in-noise processing was due to interindividual differences in the overall cochlear gain, which, in a given individual, is determined by the integrity of the outer hair cells, of which audiometric thresholds are known to be a poor measure (Moore 2007). Behavioral estimates of cochlear nonlinearity, which is indicative of cochlear gain (for a review, see Robles and Ruggero 2001), have been shown to exhibit considerable variability even among people with normal hearing and have also been found to be predictive of speech-in-noise performance (Dubno et al. 2007; Horwitz et al. 2007; Sommers and Gehr 2010).

Finally, some of the variability in CV-in-noise processing may have been due to interindividual differences in higher-level auditory function and cognitive ability (for a review, see Chandrasekaran and Kraus 2009).

The current findings suggest a detrimental effect of MOC-induced reduction in cochlear gain on speech-in-noise processing and thus conflict with some previous studies that have found a beneficial effect (Giraud et al. 1997; Kumar and Vanaja 2004). A particularly pertinent comparison is with the study by de Boer and Thornton (2008). Despite using similar equipment and identical procedures to the current study, they found a positive, rather than negative, correlation between OAE suppression and CV discrimination-in-noise performance. Apart from the participants being different, the only other material difference was that de Boer and Thornton (2008) used the [bi:]/[di:] rather than [da]/[ga] contrast to measure CV discrimination-in-noise performance. Both the consonant and the vowel context of a CV syllable determine which of the available acoustic cues (release burst vs. formant transitions; Kewley-Port 1982) are most relevant for the identification of the syllable's consonant (Smits et al. 1996). The vowel context also determines the spectral region within which the formant transitions occur. Findings by Micheyl and colleagues (Micheyl and Collet 1996; Micheyl et al. 1995) suggest that the discrepancy between the current and de Boer and Thornton's (2008) studies is a result of these acoustic differences. They indicate that the direction of the correlation between OAE suppression and signal-in-noise performance depends crucially on the acoustic properties of the signal. Using pure tones as signals, Micheyl and Collet (1996) found the correlation to be positive, whereas Micheyl et al. (1995) found a negative correlation using a complex-tone signal. The differences in correlation direction in both our and Micheyl's studies (1995, 1996) may have been caused by differences in the mechanisms by which masking was mediated. Perceptual models of cochlear amplification suggest that masking produced by suppression or on-frequency excitation would be ameliorated by MOC-induced reduction in cochlear gain (Strickland 2001, 2008), whereas masking produced by an upward spread of excitation would be aggravated (Jennings et al. 2009; Krull and Strickland 2008). Furthermore, the effect of MOC-induced reduction in cochlear gain would also be expected to depend on the spectral region within which masking occurs (Guinan and Gifford 1988; Vinay and Moore 2008). Taken together, the current and previous results indicate that the masking of complex signals, such as speech, by noise can be affected both beneficially and detrimentally by MOC activation. It is thus unlikely that reflexive, and thus unspecific, reduction in cochlear gain by the MOC system would be able to confer a beneficial overall effect on speech-in-noise processing.

However, results by Maison et al. (2001) and de Boer and Thornton (2007) indicate that MOC activation is not entirely reflexive but is amenable to modulation by attention. They measured contralateral suppression of OAEs and found that when attention was directed to the ear containing the OAE-evoking stimuli, suppression was reduced (de Boer and Thornton 2007), whereas when attention was directed to the ear containing the contralateral suppressor, suppression was increased (Maison et al. 2001). The latter effect was found to be frequency specific, in that it was only observed when the attended frequency corresponded to the frequency of the OAE-evoking stimulus. These results suggest that MOC activity can be “tuned” to exert weaker suppression on the attended parts of the incoming sound mixture and stronger suppression on the unattended parts. The MOC system might thus enhance speech-in-noise processing by selectively increasing the response to the speech over that to the noise. Attentional information would be assumed to originate from the cortex and be transmitted through the corticofugal pathways to the periolivary regions (for a review, see Winer 2006).

There is some evidence that attention-dependent MOC effects might be malleable to experience (de Boer and Thornton 2008; Veuillet et al. 2007). For instance, in addition to measuring the correlation between OAE suppression and CV discrimination-in-noise performance, de Boer and Thornton (2008) conducted a short-term training on the CV discrimination-in-noise task. Interestingly, they only found a significant correlation between OAE suppression and CV discrimination-in-noise performance before the training. After training, the correlation had disappeared due to changes in those participants who had performed poorly before training. This suggests that the correlation before training was mediated by reflexive, or unspecific, MOC activation. Its disappearance through training might indicate a change to a more selective, attention-driven mode of operation of the MOC system.

The MOC system's dependence on attentional and experience-related factors may have contributed to the disparities among previous studies on MOC function. Taking these factors into account is likely to yield more conclusive results about the role of the MOC system in signal-in-noise perception.

GRANTS

This research was supported by the Medical Research Council (UK).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: J.d.B. and A.R.D.T. conception and design of research; J.d.B. performed experiments; J.d.B. analyzed data; J.d.B. and K.K. interpreted results of experiments; J.d.B. prepared figures; J.d.B. and K.K. drafted manuscript; J.d.B., A.R.D.T., and K.K. edited and revised manuscript; J.d.B., A.R.D.T., and K.K. approved final version of manuscript.

REFERENCES

  1. Burkard R, Don M. The auditory brainstem response. In: Auditory Evoked Potentials: Basic Principles and Clinical Application, edited by Burkard RE, Don M. Philadelphia, PA: Lippincott Williams & Wilkins, 2007 [Google Scholar]
  2. Burkard R, Hecox K. The effect of broadband noise on the human brainstem auditory evoked response. I. Rate and intensity effects. J Acoust Soc Am 74: 1204–1213, 1983 [DOI] [PubMed] [Google Scholar]
  3. Chandrasekaran B, Kraus N. The scalp-recorded brainstem response to speech: neural origin and plasticity. Psychophysiology 47: 1–11, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cooper NP, Guinan JJ. Efferent-mediated control of basilar membrane motion. J Physiol 576: 49–54, 2006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cunningham J, Nicol T, Zecker SG, Bradlow A, Kraus N. Neurobiologic responses to speech in noise in children with learning problems: deficits and strategies for improvement. Clin Neurophysiol 112: 758–767, 2001 [DOI] [PubMed] [Google Scholar]
  6. de Boer J, Thornton AR. Effect of subject task on contralateral suppression of click evoked otoacoustic emissions. Hear Res 233: 117–123, 2007 [DOI] [PubMed] [Google Scholar]
  7. de Boer J, Thornton AR. Neural correlates of perceptual learning in the auditory brainstem: efferent activity predicts and reflects improvement at a speech-in-noise discrimination task. J Neurosci 28: 4929–4937, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Dubno JR, Horwitz AR, Ahlstrom JB. Estimates of basilar-membrane nonlinearity effects on masking of tones and speech. Ear Hear 28: 2–17, 2007 [DOI] [PubMed] [Google Scholar]
  9. Durrant JD, Sabo DL, Hyre RJ. Gender, head size, and ABRs examined in large clinical sample. Ear Hear 11: 210–214, 1990 [DOI] [PubMed] [Google Scholar]
  10. Giraud AL, Garnier S, Micheyl C, Lina G, Chays A, Chery-Croze S. Auditory efferents involved in speech-in-noise intelligibility. Neuroreport 8: 1779–1783, 1997 [DOI] [PubMed] [Google Scholar]
  11. Guinan JJ. Olivocochlear efferents: anatomy, physiology, function, and the measurement of efferent effects in humans. Ear Hear 27: 589–607, 2006 [DOI] [PubMed] [Google Scholar]
  12. Guinan JJ. Physiology of olivocochlear efferents. In: The Cochlea, edited by Dallos PP, Fay RR. New York: Springer-Verlag, 1996 [Google Scholar]
  13. Guinan JJ, Gifford ML. Effects of electrical stimulation of efferent olivocochlear neurons on cat auditory-nerve fibers. I. Rate-level functions. Hear Res 33: 97–113, 1988 [DOI] [PubMed] [Google Scholar]
  14. Hawkey DJ, Amitay S, Moore DR. Early and rapid perceptual learning. Nat Neurosci 7: 1055–1056, 2004 [DOI] [PubMed] [Google Scholar]
  15. Hornickel J, Skoe E, Kraus N. Subcortical laterality of speech encoding. Audiol Neurootol 14: 198–207, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Horwitz AR, Ahlstrom JB, Dubno JR. Speech recognition in noise: estimating effects of compressive nonlinearities in the basilar-membrane response. Ear Hear 28: 682–693, 2007 [DOI] [PubMed] [Google Scholar]
  17. Jennings SG, Strickland EA, Heinz MG. Precursor effects on behavioral estimates of frequency selectivity and gain in forward masking. J Acoust Soc Am 125: 2172–2181, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Johnson KL, Nicol TG, Kraus N. Brain stem response to speech: a biological marker of auditory processing. Ear Hear 26: 424–434, 2005 [DOI] [PubMed] [Google Scholar]
  19. Kawase T, Delgutte B, Liberman MC. Antimasking effects of the olivocochlear reflex. II. Enhancement of auditory-nerve response to masked tones. J Neurophysiol 70: 2533–2549, 1993 [DOI] [PubMed] [Google Scholar]
  20. Kawase T, Liberman MC. Antimasking effects of the olivocochlear reflex. I. Enhancement of compound action potentials to masked tones. J Neurophysiol 70: 2519–2532, 1993 [DOI] [PubMed] [Google Scholar]
  21. Kemp DT. Stimulated acoustic emissions from within the human auditory system. J Acoust Soc Am 64: 1386–1391, 1978 [DOI] [PubMed] [Google Scholar]
  22. Keppler H, Dhooge I, Corthals P, Maes L, D'haenens W, Bockstael A, Philips B, Swinnen F, Vinck B. The effects of aging on evoked otoacoustic emissions and efferent suppression of transient evoked otoacoustic emissions. Clin Neurophysiol 121: 359–365, 2010 [DOI] [PubMed] [Google Scholar]
  23. Kewley-Port D. Measurement of formant transitions in naturally produced stop consonant-vowel syllables. J Acoust Soc Am 72: 379–389, 1982 [DOI] [PubMed] [Google Scholar]
  24. Klatt DH. Software for a cascade-parallel formant synthesizer. J Acoust Soc Am 67: 971–995, 1980 [Google Scholar]
  25. Kramer SJ, Teas DC. Forward masking of auditory nerve (N1) and brainstem (wave V) responses in humans. J Acoust Soc Am 72: 795–803, 1982 [DOI] [PubMed] [Google Scholar]
  26. Krull V, Strickland EA. The effect of a precursor on growth of forward masking. J Acoust Soc Am 123: 4352–4357, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kumar UA, Vanaja CS. Functioning of olivocochlear bundle and speech perception in noise. Ear Hear 25: 142–146, 2004 [DOI] [PubMed] [Google Scholar]
  28. Liberman MC, Guinan JJ. Feedback control of the auditory periphery: anti-masking effects of middle ear muscles vs. olivocochlear efferents. J Commun Disord 31: 471–483, 1998 [DOI] [PubMed] [Google Scholar]
  29. Lutman ME. Reliable identification of click-evoked otoacoustic emissions using signal-processing techniques. Br J Audiol 27: 103–108, 1993 [DOI] [PubMed] [Google Scholar]
  30. Maison S, Liberman MC. Predicting vulnerability to acoustic injury with a noninvasive assay of olivocochlear reflex strength. J Neurosci 20: 4701–4707, 2000 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Maison S, Micheyl C, Collet L. Influence of focused auditory attention on cochlear activity in humans. Psychophysiology 38: 35–40, 2001 [PubMed] [Google Scholar]
  32. McFadden D. Masculinization of the mammalian cochlea. Hear Res 252: 37–48, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Micheyl C, Collet L. Involvement of the olivocochlear bundle in the detection of tones in noise. J Acoust Soc Am 99: 1604–1610, 1996 [DOI] [PubMed] [Google Scholar]
  34. Micheyl C, Morlet T, Giraud AL, Collet L, Morgon A. Contralateral suppression of evoked otoacoustic emissions and detection of a multitone complex in noise. Acta Otolaryngol 115: 178–182, 1995 [DOI] [PubMed] [Google Scholar]
  35. Micheyl C, Perrot X, Collet L. Relationship between auditory intensity discrimination in noise and olivocochlear efferent system activity in humans. Behav Neurosci 111: 801–807, 1997 [DOI] [PubMed] [Google Scholar]
  36. Moore BC. Cochlear Hearing Loss. Chichester, UK: Wiley-Blackwell, 2007 [Google Scholar]
  37. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9: 97–113, 1971 [DOI] [PubMed] [Google Scholar]
  38. Parbery-Clark A, Skoe E, Kraus N. Musical experience limits the degradative effects of background noise on the neural processing of sound. J Neurosci 29: 14100–14107, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Rajan R. Centrifugal pathways protect hearing sensitivity at the cochlea in noisy environments that exacerbate the damage induced by loud sound. J Neurosci 20: 6684–6693, 2000 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Rance G. Auditory neuropathy/dys-synchrony and its perceptual consequences. Trends Amplif 9: 1–43, 2005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Robles L, Ruggero MA. Mechanics of the mammalian cochlea. Physiol Rev 81: 1305–1352, 2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ross SM. Peirce's criterion for the elimination of suspect experimental data. J Eng Technol 20: 38–41, 2003 [Google Scholar]
  43. Smith RL. Adaptation, saturation, and physiological masking in single auditory-nerve fibers. J Acoust Soc Am 65, 166–178, 1979 [DOI] [PubMed] [Google Scholar]
  44. Smits R, ten Bosch L, Collier R. Evaluation of various sets of acoustic cues for the perception of prevocalic stop consonants. I. Perception experiment. J Acoust Soc Am 100: 3852–3864, 1996 [DOI] [PubMed] [Google Scholar]
  45. Sommers MS, Gehr SE. Two-tone auditory suppression in younger and older normal-hearing adults and its relationship to speech perception in noise. Hear Res 264: 56–62, 2010 [DOI] [PubMed] [Google Scholar]
  46. Song JH, Banai K, Russo NM, Kraus N. On the relationship between speech- and nonspeech-evoked auditory brainstem responses. Audiol Neurootol 11: 233–241, 2006 [DOI] [PubMed] [Google Scholar]
  47. Strickland EA. The relationship between frequency selectivity and overshoot. J Acoust Soc Am 109: 2062–2073, 2001 [DOI] [PubMed] [Google Scholar]
  48. Strickland EA. The relationship between precursor level and the temporal effect. J Acoust Soc Am 123: 946–954, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Uppenkamp S, Kollmeier B. Narrowband stimulation and synchronization of otoacoustic emissions. Hear Res 78, 210–220, 1994 [DOI] [PubMed] [Google Scholar]
  50. Veuillet E, Duverdy-Bertholon F, Collet L. Effect of contralateral acoustic stimulation on the growth of click-evoked otoacoustic emissions in humans. Hear Res 93: 128–135, 1996 [DOI] [PubMed] [Google Scholar]
  51. Veuillet E, Magnan A, Ecalle J, Thai-Van H, Collet L. Auditory processing disorder in children with reading disabilities: effect of audiovisual training. Brain 130: 2915–2928, 2007 [DOI] [PubMed] [Google Scholar]
  52. Vinay Moore BC. Effects of activation of the efferent system on psychophysical tuning curves as a function of signal frequency. Hear Res 240, 93–101, 2008 [DOI] [PubMed] [Google Scholar]
  53. Winer JA. Decoding the auditory corticofugal systems. Hear Res 212: 1–8, 2006 [DOI] [PubMed] [Google Scholar]
  54. Wong PK, Bickford RG. Brain stem auditory evoked potentials: the use of noise estimate. Electroencephalogr Clin Neurophysiol 50: 25–34, 1980 [DOI] [PubMed] [Google Scholar]
  55. Wright BA, Fitzgerald MB. Different patterns of human discrimination learning for two interaural cues to sound-source location. Proc Natl Acad Sci USA 98: 12307–12312, 2001 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Neurophysiology are provided here courtesy of American Physiological Society

RESOURCES