Olivocochlear Efferents: Their action, effects, measurement and uses, and the impact of the new conception of cochlear mechanical responses

Abstract

The anatomy and physiology of olivocochlear (OC) efferents are reviewed. To help interpret these, recent advances in cochlear mechanics are also reviewed. Lateral OC (LOC) efferents innervate primary auditory-nerve (AN) fiber dendrites. The most important LOC function may be to reduce auditory neuropathy. Medial OC (MOC) efferents innervate the outer hair cells (OHCs) and act to turn down the gain of cochlear amplification. Cochlear amplification had been thought to act only through basilar membrane (BM) motion, but recent reports show that motion near the reticular lamina (RL) is amplified more than BM motion, and that RL-motion amplification extends to several octaves below the local characteristic frequency. Data on efferent effects on AN-fiber responses, otoacoustic emissions (OAEs) and human psychophysics are reviewed and reinterpreted in the light of the new cochlear-mechanical data. The possible origin of OAEs in RL motion is considered. MOC-effect measuring methods and MOC-induced changes in human responses are also reviewed, including that ipsilateral and contralateral sound can produce MOC effects with different patterns across frequency. MOC efferents help to reduce damage due to acoustic trauma. Many, but not all, reports show that subjects with stronger contralaterally-evoked MOC effects have better ability to detect signals (e.g. speech) in noise, and that MOC effects can be modulated by attention.

1. Introduction

The olivocochlear (OC) efferents are part of brainstem-to-cochlea reflex pathways that allow stimulus-related control of the cochlea, and provide a way for the central nervous system to affect hearing at the most peripheral neural level. There have been many reviews of OC efferents (e.g. Guinan 1996; 2006; 2010; 2011; 2012). Here we concentrate on new efferent work, new interpretations based on recent cochlear-mechanical data, controversial issues, and areas that need more study. We briefly review basic, long-established efferent anatomy and physiology. Since space is limited, for citations to older literature we refer readers to the reviews listed above, and for recent work we do not cite all relevant references.

2. Efferent anatomy and action in the cochlea

2.1 Efferent anatomy

Olivocochlear (OC) neurons receive sound-driven inputs from the cochlear nucleus and from brainstem-level acoustic reflexes. OC neurons receive descending inputs that allow higher neural centers to modulate the OC reflexes. In almost all mammals there are two groups of olivocochlear efferents: medial OC (MOC) and lateral OC (LOC) efferents.

MOC neurons are located in the superior olivary complex medial, ventral and extending slightly anterior to the medial superior olivary nucleus. They have myelinated axons that synapse on outer hair cells (OHCs). In most experimental animals, about 2/3 of MOC axons cross the midline and innervate the opposite cochlea, 1/3 innervate the cochlea on the same side, and a small fraction innervate both cochleae. The proportion of crossed vs. uncrossed axons in humans is unknown. Most MOC neurons receive sound-driven inputs from both cochlear nuclei with the strongest input from the contralateral side (The recent suggestion that the MOC reflex is driven primarily by Type-II auditory-nerve (AN) fibers is almost certainly wrong – Froud et al., 2015; Maison et el, 2016). MOC neurons with crossed axons are driven mostly by sound in the opposite ear and form the ipsilateral (double crossed) MOC reflex. MOC neurons with uncrossed axons form the contralateral reflex (the signal crosses in the inputs to the MOC neurons). There are no known differences in the OHC synapses of the ipsilateral vs. contralateral MOC reflexes, but their distributions along the cochlea are different. Compared to ipsilateral-reflex axons, contralateral-reflex axons innervate OHCs over a wider span along the cochlea that extends more apically (Brown, 2014).

One aspect of MOC innervation that is not well understood is the distribution of MOC innervation around the characteristic frequency (CF) of the MOC fiber. MOC innervation of the cochlea is tonotopic (i.e. each MOC fiber innervates a cochlear region tuned to the MOC fiber’s CF). An individual MOC fiber innervates OHCs located both apically and basally around the cochlear CF region of the MOC fiber (Brown, 2014). Considerable evidence indicates that cochlear amplification takes place only basal of CF, so the MOC innervation that is apical of CF doesn’t help in reducing the gain at the MOC fiber’s CF. The implication is that MOC feedback is not anatomically arranged to be frequency specific and instead also affects nearby cochlear frequency regions.

LOC cell bodies are located in and near the lateral superior olivary (LSO) nucleus. LOC neurons have unmyelinated axons that synapse on the dendrites of AN fibers under inner hair cells (IHCs). LOC axons go predominantly to the cochlea on the same side as the axon’s cell body. Since LOC axons are thin and unmyelinated, they are difficult to record from or electrically stimulate. Consequently, little is known about when they are activated or what they do to AN firing. Unmyelinated axons conduct action potentials slowly, and the effects attributed to LOC synapses are slow, i.e. they take place over minutes.

Efferent fibers have several roles in the development of normal auditory anatomy (e.g. Yin et al., 2014; Clause et al. 2017; reviewed by Nouvian et al., 2015).

2.2 Efferent action in the cochlea

MOC synapses release acetylcholine (ACh) onto specialized (α9/α10) ACh receptors (ACHRs) on the baso-lateral walls of OHCs (reviewed by Wersinger and Fuchs, 2011). Activated α9/α10 ACHRs allow Ca⁺⁺ ions to enter the OHCs and these Ca⁺⁺ ions activate nearby channels that allow K⁺ ions to flow into and hyperpolarize the OHCs. These Ca⁺⁺ ions can also cause the release of Ca⁺⁺ from local Ca⁺⁺ stores, which increases the synaptic effect. The multistep increase in K⁺ flow into OHCs takes time and is the main determiner of the time course of MOC effects (~100 ms build up and decay times).

Activation of MOC synapses does two major things: it increases the OHC’s basolateral conductance and it hyperpolarizes the OHCs. Increasing OHC basolateral conductance shunts OHC-stereocilia mechano-electric-transduction (MET) receptor current from flowing through the normal OHC basolateral conductance, thereby reducing the OHC voltage change and the resultant somatic motility and cochlear amplification. Hyperpolarizing OHCs is also thought to reduce somatic motility and cochlear amplification by moving the OHC voltage-to-length-change function to a less favorable operating point.

LOC synapses release ACh and several other neurotransmitters and neuromodulators, notably dopamine. There is evidence for two subgroups of LOC synapses: ACh synapses and dopamine synapses (Darrow et al., 2006b). Indirectly stimulating LOC axons increased or decreased AN-fiber firing, depending on where the neural stimulation was done (Groff and Liberman, 2003). It is tempting to think that ACh produces one of these effects and dopamine produces the other, but this has not been established. In addition to LOC effects on AN coding, an important function (perhaps the main function) of LOC efferents is to reduce damage to AN fibers from excessive activation by traumatic sounds (Darrow et al., 2007; Fuente, 2015). With the recent demonstration that AN neuropathy is a common consequence of even mild acoustic overstimulation (Kujawa and Liberman, 2009), the function of LOC efferents in protecting from AN neuropathy is particularly important.

2.3 Efferent responses to sound

Most MOC neurons respond to sound in one ear and are facilitated by sound in the other ear (i.e. inputs from the other ear excite but not enough to reach threshold). MOC neurons have regular, repetitive firing. They have V-shaped, tone-response tuning curves that are slightly wider than those of AN fibers. The addition of noise in the opposite ear can greatly increase MOC tone response areas, but there are few data on MOC-neuron responses to noise, despite noise being the most potent and preferred MOC stimulus in humans. One interesting MOC response phenomenon is that high-level acoustic stimulation leads to enhanced MOC responses to later sounds. The MOC properties just described were obtained from single-fiber recordings from lightly anesthetized animals. More recent animal work shows that anesthesia greatly depresses MOC responses (Guitton et al., 2004; Chambers et al., 2012; Aedo et al., 2015).

LOC neuron responses to sound have never been measured, but it is assumed that they respond to sound, in part because their cell bodies are embedded in an auditory nucleus, the LSO. LSO neurons aid in sound localization by increasing their rate in response to the ipsi/contra sound-level difference. Presuming that LOC neurons have sound-response properties similar to LSO neurons and that LOC activity inhibits ipsilateral AN responses, it has been hypothesized that LOC neurons provide feedback to maintain a balance between activity in right vs. left ANs (Guinan, 1996). There is evidence for and against this hypothesis (Darrow et al., 2006a; Larsen and Liberman, 2010). More definitive evidence is needed.

3. Recent advances in cochlear mechanics and their impact for understanding MOC action

Over the last few years the understanding of cochlear mechanical motions has changed greatly. Basilar membrane (BM) responses to sound were known, but now there are also good measurements of responses to sound of other structures in the organ of Corti (OoC). In addition, it had been thought that IHC stereocilia are deflected only by shear between the reticular lamina (RL) and the tectorial membrane (TM) but this is now known to be an oversimplification. MOC efferents act by changing cochlear mechanics so these advances affect the interpretation of efferent mechanisms and actions.

3.1 Organ of Corti motion in response to sound

The new measurements have not changed the picture of basal-turn BM motion. Basal-turn BM motion is sharply tuned and highly sensitive at low sound levels near the local CF. BM motion near CF grows compressively so that at high sound levels its sensitivity is much less (Robles and Ruggero, 2001). The high sensitivity of BM motion at low sound levels is due to amplification of BM motion by prestin-based somatic motility in OHCs (Ashmore et al., 2010). This amplification goes away after sacrificing the animal and the response becomes “passive”. An octave or more away from CF there is no cochlear amplification and BM motion in response to sound is the same before and after death (i.e. is passive).

In high-frequency regions, knowledge of non-BM OoC motion has been greatly increased by new measurement techniques. For low-level tones near CF, the structures in the upper part of the OoC (e.g. near the RL) have been found to move much more than the BM (e.g. >6 times more), and to have slightly different tuning (Fig. 1, left and center) (Ren et al, 2016; Recio-Spinoso and Oghalai, 2017). As sound level is increased, the RL/BM amplitude difference decreases so that at 80 dB SPL, RL and BM amplitudes are nearly the same (e.g. Fig. 1D and E). An important new finding is that for frequencies an octave or more below CF (tail frequencies), where BM motion is linear and passive, RL motion is much greater than BM motion (up to 20 dB greater) and opposite in phase. Further, after sacrificing the animal, the tail-frequency RL motion decreases and becomes passive like BM motion. This indicates that in live animals, RL motion is increased above BM motion by active processes, presumably from OHC motility. Note that OHC motility has a broad frequency response and is evoked whenever OHC stereocilia are deflected, at CF or below CF.

Fig. 1.

A conception of frequency (top) and level (bottom) functions for responses to tones of the basilar membrane, the area near the reticular lamina, and auditory-nerve fibers, without and with MOC stimulation, in the basal cochlea. Based on: Cooper and Guinan (2006) for panels A & D; Ren et al., (2016) plus Recio-Spinoso and Oghalai, (2017) for no-MOC plots in panels B and E; Stankovic and Guinan (1999) plus Guinan and Stankovic (1996) for the (speculative) estimated change due to MOC stimuation in panels B and E; Guinan and Gifford (1988) for panel C; Guinan and Stankovic (1996) for panel F. SR=spontaneous rate.

In the low-frequency cochlear apex, there have been no well-controlled measurements of sound-driven motion within the OoC in sensitive cochleae, until recently. Measurements had been made in low-frequency regions, but almost all viewed the OoC through a hole in the cochlea that was not sealed adequately to prevent fast-wave artifacts from obscuring the measurements (Cooper and Rhode, 1996). Furthermore, these measurements were made without a monitor of local cochlear health, such as the compound action potential (CAP) thresholds that provide a metric of cochlear sensitivity at high frequencies, so these measurements may have been in animals with lowered apical sensitivity. These older measurements had found low-frequency OoC motion that had broad, band-pass tuning with only a small degree (~10 dB) of nonlinearity. Dong and Cooper (2006) showed that when the apical cochlear hole was sealed, the low-frequency OoC motion response became almost flat so that the response was “low-pass-like,” not band pass. However, they still found little nonlinearity. The breakthrough came by using optical-coherence tomography (OCT) which allows OoC motion to be viewed through the bony cochlear wall without making a hole. In addition to avoiding the fast-wave artifact produced by a hole, this technique also preserved cochlear sensitivity with the result that much larger nonlinearities, up to 50 dB, were seen in OoC motion at the upper-frequency edge of the low-pass-like motion response to low-frequency tones (Recio-Spinosa and Oghalai, 2017). Note that the motion response pattern in the apex depends on the frequency that the apex responds to in a given species. For instance, the mouse hears only at high frequencies (>4 kHz) and its apical motion responses have sharp tuning with high gain at CF, similar to the pattern in the base (Lee et al., 2016). However, in guinea pigs and gerbils, where apical CFs are ~<4 kHz, low-pass-like tuning has been found when no hole is made in the cochlea (Recio-Spinosa and Oghalai, 2017; Dong et al. 2017). These apical measurements viewed the cochlea through scala vestibuli so the measurements were from structures at the top of the OoC (e.g. near the RL). There are limited OCT data on BM motion in the apex, but these data indicate that BM motion is an order of magnitude less than motion at the top of the OoC (Warren et al., 2015), which is similar to the findings in the cochlear base.

Another area of cochlear mechanics about which knowledge has recently changed is the vibration pattern of the TM. In the 1980s it was hypothesized that TM radial motion had a resonance frequency below CF, and that at CF it was opposite in phase to RL radial motion in a way that would produce cochlear amplification. In contrast, recent measurements of excised TMs from mice and humans show that the TM has viscoelastic material properties that do not support a sharp resonance (Gu et al., 2008; Farrahi et al., 2016). Furthermore, the radial motions of both the RL and the TM show tuning which peaks above the CF of the BM (Lee et al. 2016). Measurements in excised TMs show that the TM supports longitudinal traveling waves of radial shear motion (Ghaffari et al., 2007). In mice with a certain mutation that affects the TM, cochlear tuning is sharper than normal, and TM radial-motion traveling waves decay faster than normal by about the same amount as the tuning sharpening (Russell et al., 2007; Ghaffari, et al. 2010). These sharpening patterns are consistent with the hypothesis that the TM radial traveling wave is strongly involved is producing the sharpness of cochlear tuning. More detailed measurements and models are needed before we understand TM and RL motions, their involvement in cochlear tuning, and how they are affected by MOC activity.

3.2 The drive to IHC stereocilia

IHC stereocilia are freestanding (their tops are not attached to the TM) so their deflection is driven by fluid forces. In the classic view, differential radial motion (shear) between the RL and TM, along with fluid-viscosity drag, produce the fluid forces that deflect IHC stereocilia. It was also thought that the gap between the RL and the TM was held constant by the stiffness of the IHC stereocilia and the inertia of the fluid in the RL-TM gap (Chadwick et al., 1996). However, Nowotny and Gummer (2006; 2011) found that sinusoidal electrical stimulation of OHC somatic motility produced oscillatory RL-TM gap changes, particularly in the region of the IHC stereocilia. Additionally, OHC stereocilia were found to change their length (Fridberger, et al., 2006; Hakizimana, et al., 2012), thus the RL-TM gap can change over both IHCs and OHCs. When the RL-TM gap changes, fluid must flow into the gap and this flow can drive IHC stereocilia (Nowotny and Gummer, 2006; Guinan, 2012). Another possible IHC drive is tilting of the RL at the top of the IHC (Steele and Puria, 2005). These multiple oscillatory drives to the IHC stereocilia are not all in phase. A change in domination by one drive to domination by another with an opposite phase seems likely to be the origin of reversals of AN responses such as Nelson’s notch (Guinan, 2012). Overall, the drive to IHC stereocilia is much more complex than classic RL-TM shear, but there is little understanding of the various drives or their relative importance.

3.3 Interpreting efferent effects in light of the new view of cochlear mechanics

In the 1990’s view of cochlear mechanics, all OoC motion was determined by, and followed, BM motion, and this motion received the most amplification at the lowest sound levels. MOC efferents reduce cochlear amplifier gain, so with the 1990’s view MOC efferents should produce the greatest inhibition at low sound levels. MOC inhibition of BM motion, AN CAP potentials and otoacoustic emissions (OAEs) are greatest at low sound levels, which fits with this view. In contrast, MOC inhibition of single AN fiber responses to tones at CF, measured as a level shift or “effective attenuation” (the amount sound has to be increased with efferent stimulation to produce the response rate obtained without efferent stimulation), is greatest at moderate-to-high sound levels (45–75 dB SPL) (Guinan and Stankovic, 1995). Measurements with a self-mixing laser pointed at the BM (the system optics had a wide depth of field so the self-mixing laser could be sensitive to motion deep in the OoC – Lukashkin et al., 2005) found the largest level shifts at 50–75 dB SPL, similar to the AN single-fiber level shifts (Russell and Murugasu, 1997). These self-mixing laser results conflict with MOC effects on BM motion from laser-Doppler measurements (Cooper and Guinan, 2006). In light of the new knowledge that RL motion is much greater than BM motion, these conflicting data can be made to fit together by the following hypothesis: The motion near the top of the OoC is larger than BM motion and dominated the self-mixing laser signal, so that the self-mixing laser measurements showed MOC inhibition of motion near the top of the OoC, not BM motion. Since motion near the top of the OoC is closer, and more directly related to the drive to IHC stereocilia than is BM motion, the MOC inhibition measured with the self-mixing laser matched the MOC inhibition of AN fibers better than the MOC inhibition of BM motion. This hypothesis fits the data and resolves the conflicting reports on MOC mechanical effects, but needs to be tested, e.g. by OCT measurements of OoC motions with and without MOC stimulation.

Another area where there is a difference between the MOC effects on BM motion and on AN single-fiber responses is at tail frequencies (frequencies an octave or more below CF). At tail frequencies BM motion is passive and there is little or no effect of MOC stimulation (Murugasu and Russell, 1996; Cooper and Guinan, 2006). In contrast, in AN single-fibers, MOC stimulation inhibits responses over a wide range of tail frequencies (Fig. 1C) (Stankovic and Guinan, 1999; 2000). It was a mystery how MOC stimulation could inhibit AN fiber responses without inhibiting BM responses until the recent finding that OoC motion near the RL receives amplification at tail frequencies. We hypothesize that RL motion is amplified by the same mechanisms as cochlear amplification at CF (i.e. by MET-current controlled OHC somatic motility), and that MOC stimulation turns down the gain for both BM and RL motion whenever they are amplified, which for RL motion includes tail frequencies. RL motion is closer to the drive to IHC stereocilia than is BM motion, so MOC inhibition of AN responses resembles the presumed MOC effect on RL motion (a reduction of amplified RL motion) and does not resemble the MOC lack-of-effect on BM motion (Fig. 1A, B). Again, this hypothesis fits the data but needs testing.

The tail-frequency responses described above sometimes show an abrupt change from MOC inhibition near threshold to no inhibition at higher levels. At frequencies that show a sharp dip in AN firing (sometimes called Nelson’s notch) and a reversal of response polarity, there is a similar pattern of MOC inhibition below the dip and no inhibition above the dip (Gifford and Guinan, 1983). These responses are consistent with the hypothesis that the drive at low levels is amplified and therefore can be MOC inhibited, whereas the drive at high levels is passive and not affected by MOC stimulation. What OoC motions correspond to these drives is not known.

One MOC effect on AN responses that still has no clear mechanical correlate is the MOC inhibition of the AN initial peak (ANIP) response to clicks that is found in low-CF fibers (Guinan et al., 2005; Guinan and Cooper, 2008). Although the AN first peak is MOC inhibited, there is no corresponding inhibition of the first peak of BM motion (measured in the basal turn). A perhaps related finding is that in low-frequency cochlear regions, the motion near the top of the OoC shows a high-gain motion response (termed a “bulge”) at the upper edge of the low-pass-like frequency response (Recio-Spinoso and Oghalai, 2017; Dong et al., 2017). The bulge response has the shortest group delay at the location of the cochlear measurement, which suggests that it may be a tone-response manifestation of a motion that has the same origin as the short-latency ANIP click response. The origin of the ANIP response is unknown, but we have hypothesized (Guinan et al., 2005) that it may be from OHC somatic motility rhythmically squeezing the OoC and producing longitudinal fluid flow within the OoC (Karavitaki and Mountain, 2007a). The ANIP and bulge responses may have particularly short latencies for the cochlear place where they are observed because they are produced by OoC squeezing by the traveling-wave response at a more basal place. Alternately, they may be from a locally produced vibration with a fast decay and therefore a shorter group delay. If the ANIP and bulge responses are produced by OoC oscillatory squeezing by OHCs, then perhaps they show a particularly high gain (or for ANIP responses, complete inhibition by MOC stimulation) because these responses don’t exist in a passive cochlea. Again, these hypotheses fit the data but need testing.

The above hypotheses indicate that RL motion is important for understanding the drive to IHC stereocilia, but TM motion is also important and may have a different pattern than RL motion. When near-RL and TM motions were both measured in the same preparation (Nowotny and Gummer, 2006; 2011; Lee, et al., 2016), their transverse motions were different, which indicates that the RL-TM gap cyclically varies and produces fluid flow in the gap that can deflect IHC stereocilia.

The most common way to measure MOC effects is to measure changes in OAEs. It has been assumed that OAEs originate from BM motion, but RL motion is bigger than BM motion so RL motion must be considered as a possible source of OAEs. Because traveling-wave energy is carried by the fluid in the scalae, what is important for the generation of OAEs is the fluid displaced by OoC movement. From measurements in electrically-stimulated, excised cochleae, it was concluded that as the RL moves up, the side of the OoC (e.g. at the Hensen cells) move inward (Karavitaki and Mountain, 2007b). This means that the large RL movements are offset, at least partly, by the movement of other OoC structures above the BM. If there were no lengthwise fluid flow within the OoC, then the fluid displaced at the top and sides of the OoC (and therefore into scala media and scala vestibule) would be exactly equal and opposite to the fluid displaced by the BM in scala tympani (this has been the normal way that OAE generation has been thought about). However, if there is longitudinal oscillatory fluid flow along the tunnel of Corti in an intact, normally working cochlea, the fluid displaced at the top and bottom of the OoC will be different. How large fluid flow within the OoC is relative to the fluid displaced by the BM is unknown. If the oscillatory fluid flow along the tunnel of Corti is only a small fraction of the fluid displaced by the BM, then the difference between the fluid displacements at the top and bottom of the OoC will be small. Nonetheless, even small differences would add an additional complexity to the production of OAEs. More mechanical data are needed before we can assess the possible impact on OAE generation of differences between RL and BM motions.

4. Measuring efferent effects: techniques and issues

4.1 The basic technique

Efferent effects are commonly assessed by the difference between some physiologic variable measured with and without efferent stimulation. The concept is simple, but it is often poorly done, most often because of inadequate signal-to-noise ratios (SNRs). Taking the difference between two quantities adds the errors from both measurements and these are the errors for the measurement of a much smaller quantity (the difference) than the original measurements, thus the SNR of the difference is much higher than the SNR of either measured quantity. While a 6 dB SNR may be adequate for an OAE measurement, attempting to measure an efferent effect with two OAE measurements that have 6 dB SNRs is grossly inadequate (Guinan, 2006; 2010). Figure 8 of Goodman et al. (2013) shows the SNR needed for detecting MOC effects of various amplitudes, e.g. to detect a 1 dB MOC change, the SNR should be >22 dB. Measurements with poor SNRs have been successfully used for group statistics, but clinically-relevant measurements require adequate SNRs on individual ears. A second issue is measurement drift (i.e. changes over time). To avoid drift from having a significant effect, multiple alternations of MOC-on, MOC-off should be done, and which condition is first should be randomized. A third issue is avoiding activation of middle-ear muscles (MEMs). Even a weak MEM contraction can have a big effect on a MOC measurement. MEM contractions can be found at levels 10–15 dB below the MEM threshold measured with clinical instruments and they can extend to frequencies of many kHz (Feeney et al., 2003). Further, MEM activation is due to a combination of the sound excitation in both ears so the threshold in one ear is lowered by sound in the other ear. The ideal MEM test would use the same stimuli as the OAE test. With click-evoked OAE (CEOAE) measurements, an efficient way to determine if there were MEM contractions is to determine if the acoustic waveform of the click itself is changed by the MOC-evoking stimulus (Guinan, 2006; Lichtenhan et al., 2016). Finally, MOC effects are typically evoked by sound in the opposite ear, but it should not be assumed that sound in the ipsilateral ear does the same thing (e.g. Fig. 2).

Fig. 2.

MOC-induced changes in SFOAEs averaged across human subjects, expressed as ΔSFOAE magnitude (top), SFOAEmoc magnitude (middle) and SFOAEmoc phase (bottom) as functions of noise-elicitor bandwidth (left) and as functions of elicitor frequency re. the probe frequency (right). SFOAEmoc is the SFOAE measured during MOC stimulation. Note the different scales at left and right. Left: 60 dB SPL elicitors centered at the 1 kHz probe frequency. Right: 60 dB SPL, half-octave-wide noise elicitors. All used 40 dB SPL, 1 kHz probe tones. Inset: Vector diagram showing the relationships of ΔSFOAE magnitude, SFOAEmoc magnitude and SFOAEmoc phase (ϕ). Data at left from Lilaonitkul and Guinan (2009b) and at right from Lilaonitkul and Guinan (2012).

Although a common way to measure MOC effects has been to measure changes in distortion-product OAEs (DPOAEs) or stimulus-frequency OAEs (SFOAEs) from fixed tones that are stepped in frequency across measurements, the use of swept tones offers several advantages (Long et al. 2008; Bennett and Ozdamar, 2010; Kalluri and Shera, 2013). Swept-tone methods are fast, they help in separating the different delay components in measurements of DPOAEs and SFOAEs, and they remove synchronized spontaneous OAEs. Swept-tone measurements have confirmed that the SFOAE delays measured from group delays correspond to actual physical delays of SFOAE components (Kalluri and Shera, 2013).

An alternate method for measuring OAE effects is the rapid-adaptation method (Liberman et al., 1996). This is a good method in species where the efferents can be cut to confirm that the adaptation is from efferents (and not from MEM contractions). It is not a suitable method for use in humans (Guinan, 2006).

4.2 MOC measurements using auditory-nerve responses

MOC effects on AN responses were first measured by the reduction of CAP responses to clicks in animals with an electrode near the round window. This is an accurate method for measuring MOC effects at low, near-threshold sound levels, because CAPs from low-level sounds are dominated by the responses of low-threshold AN fibers with high spontaneous rates (SRs). However, as noted earlier, the largest effects of MOC stimulation on AN single-fiber responses are at 45–75 dB SPL in higher-threshold, low- and medium-SR fibers. At sound levels of 45–75 dB SPL, the more-numerous high-SR fibers are saturated and show little MOC-induced change in firing rate, so CAP responses to high-level sounds don’t reflect the MOC effects on the low- and medium-SR fibers that show the largest MOC inhibitions.

MOC effects on AN responses can be measured in humans using CAP responses, but such measurements are difficult and require a lot of averaging (Lichtenhan et al., 2016; Najem et al., 2016; Smith et al., 2017; Verschooten et al. 2017). The results of human CAP studies are consistent with the conclusion that MOC effects on CAPs are similar in humans and experimental animals. An important result of these studies is that MOC inhibition is much greater when measured with CAPs than when measured with OAEs, so OAE data underestimate MOC effects. This is consistent with a similar finding in animals (Puria, et al., 1996).

4.3 MOC measurements using OAEs

MOC effects on OAEs are of particular interest because OAEs provide the easiest way to measure efferent effects in humans. However, the previous paragraph indicates the need for caution in interpreting measurements of MOC effects on OAEs. OAEs are generated by two mechanisms, coherent reflection and distortion (Shera and Guinan, 1999). Reflection emissions can be produced by a single tone (SFOAEs) or by a click (CEOAEs). Reflection emissions originate from the reflection of traveling-wave energy by minute irregularities along the cochlea, with the major reflection contribution arising near the peak of the traveling wave. Distortion emissions are typically generated using two tones (the primary tones) at frequencies f₁ and f₂ (f₂>f₁) which yield distortion products at a series of sum and difference frequencies, the most prominent being 2f₁-f₂. Distortion emissions are produced by cochlear nonlinearity (mostly in the OHC MET function) which creates a distortion product in OHC voltage, and by OHC somatic motility which injects the distortion product into OoC motion. This OoC motion creates distortion-product traveling waves going both apically (forward waves) and basally (reverse waves). The 2f₁-f₂ forward-traveling distortion product travels apically to the 2f₁-f₂ place in the cochlea where it is coherently reflected. This reflected 2f₁-f₂ wave combines with the direct-backward distortion-product wave which results in the 2f₁-f₂ DPOAE measured in the ear canal. The forward-then-reflected wave has a longer delay than the direct-backward wave, so their relative phases vary and interfere, which produces a semi-regular pattern of peaks and valleys called DPOAE “fine structure”.

For measuring MOC effects, CEOAEs are the easiest to use because the stimulus (the click) and the OAE are separated in time, and the CEOAE can be obtained by time-windowing the response. Measurements of CEOAEs with and without MOC stimulation readily yield the change produced by MOC activation. Some degree of frequency specificity can be obtained by filtering the CEOAE responses (e.g. Francis and Guinan, 2010). Also, CEOAEs provide an easy test for MEM activation, as noted earlier.

DPOAEs are also relatively easy to measure because the DPOAE is separated in frequency from the primary tones. However, measurements of MOC effects on DPOAEs must take into consideration the two-source nature of DPOAEs. The most accurate measurements are done by separating the two components. Separation can be done by a variety of methods most of which require multiple measurements (reviewed by Vetesnik et al., 2009). If MOC measurements are made without separating DPOAE components (e.g. by measuring at a single DPOAE frequency with and without MOC stimulation), the worst-case scenario is that the two DPOAE components are out of phase and cancel. Then, since the forward-then-reflected component is typically inhibited more than the direct-reverse component, the cancellation would be reduced by MOC stimulation which would result in an increased DPOAE amplitude (even though both DPOAE components were decreased). A simple way to avoid this is to measure at a peak of the DPOAE fine structure (Sun, 2008; Abdala et al., 2009).

SFOAEs are the most difficult OAE to measure because the SFOAE overlaps the evoking tone both in time and frequency. MOC effects on SFOAEs have been measured in several ways. The simplest is to measure the ear-canal sound pressure with and without MOC stimulation and take the vector difference. This cancels the evoking tone and yields the SFOAE change (ΔSFOAE) (Guinan, 2003). However, this method has the disadvantage that the ΔSFOAE metric varies with the SFOAE amplitude. To get a metric that shows the MOC effect independent of the SFOAE amplitude, the ΔSFOAE is normalized by dividing the ΔSFOAE by the magnitude of the SFOAE, which requires measuring the SFOAE itself. SFOAEs are most often measured by the suppression method (e.g. see Guinan et al., 2003), but they can also be measured by the scaling method (compression), spectral smoothing, and the double evoked method (Kemp and Chum, 1980; Kalluri and Shera, 2007; Keefe, 1998). If the overall SFOAE (or just the stimulus tone with the SFOAE suppressed) is measured by one of these methods, and measurements are made with and without MOC stimulation, then the SFOAE during MOC stimulation (SFOAEmoc) can be measured (instead of just ΔSFOAE) and the efferent effect can be expressed as a change in SFOAE amplitude and phase (e.g. as in Lilaonitkul and Guinan, 2009b; 2012). MOC effects expressed as changes in the amplitude and phase of the SFOAE (e.g. Fig. 2) are comparable to the typical method of expressing MOC-induced DPOAE changes. In contrast, ΔSFOAE combines amplitude and phase changes, which is more like typical measurements of MOC effects using CEOAEs.

One controversial issue has been whether SFOAEs originate primarily near the peak region of the cochlear response, or whether there are significant SFOAE sources far basal of the peak region. Far-basal SFOAE sources were suggested from finding large residuals at the SFOAE frequency when adding high-frequency “suppressor” tones (e.g. Guinan, 1990; Charaziak and Siegel, 2015). An alternate interpretation is that these residuals are new SFOAE-frequency sources produced by the suppressor tones driving far-basal OHCs into their nonlinear range (i.e. these residuals are not present when the suppressor tone is not present). As the suppressor frequency is changed, the far-basal-residual phases vary over several cycles, which doesn’t fit the suppressed-source hypothesis because the phase of SFOAE-frequency cochlear motion varies little in the far-basal region. While significant far-basal SFOAE sources evoked by a single tone seem unlikely, there may be significant slightly-basal sources (Moleti and Sisto, 2016).

SFOAEs and TEOAEs are produced by reflections from cochlear irregularities (Zweig and Shera, 1995). The pattern of these irregularities may be altered by efferent stimulation, which would affect the accuracy of the measurement of MOC effects with these emissions. Measurements of SFOAEs in guinea pigs with MOC activity produced by brainstem shocks show MOC-induced increases in SFOAE amplitudes that cannot be attributed to the removal of a cancellation, but are explained well by an increase in irregularities (Berezina-Greene and Guinan, 2017). Shocks produce a very irregular pattern of efferent inhibition along the cochlea. Sound-evoked MOC activity is expected to produce more even inhibition and a smaller increase in irregularities. Nonetheless, there is evidence for an MOC-induced change in the pattern of irregularities from sound-induced MOC activity in humans (Backus and Guinan, 2007). MOC-induced changes in cochlear irregularities would reduce the accuracy of measuring efferent effects using reflection emissions. It is not clear how important this effect is. More data are needed to assess this.

5. Efferent effects

5.1 MOC basic properties

If MOC-elicitor sound pressure levels are kept equal, greater MOC effects are evoked by noise than by tones, and the wider the bandwidth of the noise, the greater the MOC effect (Fig. 2, left) (Lilaonitkul and Guinan, 2009b). With frequency-specific elicitors (tones or narrow-band-noise), the MOC reflex shows some frequency-specific effects on OAEs, but reflex tuning is broad and the largest effect is not always centered at the frequency of the MOC-elicitor (Fig. 2, right) (Lilaonitkul and Guinan, 2009a; 2012; Zhao and Dhar, 2012). The reflex-tuning pattern differs when measured by SFOAE amplitude, SFOAE phase, or ΔSFOAE (Fig. 2, right). In producing ΔSFOAE, the SFOAE phase change played a bigger role than the amplitude change (Fig. 2). MOC-induced phase changes have often been found (e.g. Zhao et al., 2015) but are poorly understood. MOC-reflex tuning varies with elicitor laterality (Fig. 2, right). Wide MOC reflex tuning has also been found with psychophysical methods (Drga et al., 2016).

MOC activity produced a small broadening of cochlear tuning as assessed by MOC-induced changes in OAE delays (Francis and Guinan, 2010; Mishra and Dinger, 2016) or psychophysical tuning curves (Jennings and Strickland, 2012; Lopez-Poveda et al., 2013). However, Bhagat and Kilgore (2014) did not find a significant MOC-induced change in OAE delays.

Animal studies found twice as many MOC fibers in the ipsilateral MOC reflex as in the contralateral MOC reflex. A similar ipsi/contra ratio has been expected for humans. For noise elicitors centered on the probe tone, the ipsilateral and contralateral reflexes had the same amplitude for broad-band elicitors, but for narrow-band elicitors the ipsilateral reflex was stronger (Fig. 2 left). For narrow-band-noise elicitors at different frequencies re the probe tone, the ΔSFOAE ipsilateral/contralateral ratio was ~2, but the contralateral reflex produced almost no SFOAE amplitude change and the phase changes were complex (Fig. 2, right). Bilateral sounds appear to approximately sum the effects of ipsilateral and contralateral sounds (Fig. 2). Some differences in reflex effects across elicitor lateralities and bandwidths must originate from MOC activation patterns in the brainstem pathways, but some may also be from different ipsi/contra reflex actions in the cochlea.

The MOC reduction in cochlear amplifier gain has been estimated using psychophysical methods. Ipsilateral MOC-elicitors produced 20 dB gain reductions (Drga et al., 2016) and contralateral MOC-elicitors produced 4.4 dB gain reductions (Fletcher et al., 2016). This gain-change difference may reflect ipsi/contra reflex differences in humans, but differences in methods may also play a role. MOC-induced gain reductions are implicated in producing psychophysical “overshoot” and the “mid-level hump” in Weber fractions (Jennings et al., 2012; Roverud and Strickland, 2015). Psychophysical studies typically compare their results to BM motion, but a comparison with OoC motion near the RL is more relevant (see Section 3).

Most MOC effects in animals and humans have rise and fall times in the hundred ms range (James et al., 2005; Backus and Guinan, 2006). In guinea pigs, there is also a MOC effect that changes on a time scale of tens seconds, the MOC “slow effect”. The MOC slow effect is produced by a different OHC mechanism than the two mechanisms employed by normal MOC inhibition (Cooper and Guinan, 2003). Attempts to measure MOC slow effects in humans found only very small effects (Zhao and Dhar, 2011).

5.2 MOC and hearing in noise

The expectation that MOC efferents aid hearing in noise had been explored by many papers (Guinan 2006). Recent work conclusively shows this MOC benefit (Abdala et al., 2014; Mishra and Lutman, 2014; Bidelman and Bhagat, 2015; Maruthy et al. 2017). However, other recent work failed to detect a benefit (Wagner et al., 2008; Garinis et al., 2011) which indicates that there is still an incomplete understanding of the conditions under which MOC activity aids hearing in noise.

5.3 Efferent effects and acoustic trauma

A large body of data indicates that a major role of both LOC and MOC efferents is to reduce the damage from traumatic sounds. For reviews see Maison et al. (2013), Fuente (2015) and Liberman and Kujawa (2017). MOC inhibition decreases with age (Yilmaz et al, 2007; Abdala et al., 2014). This may be a contributing factor to age-related hearing loss and synaptopathy (Chumak et al., 2015).

Many studies have inquired whether abnormal MOC reflexes have a role in pathologic conditions. Increased contralaterally-elicited MOC-inhibition of DPOAEs has been found in subjects with tinnitus and hyperacusis (Knudson et al., 2014 – see this for a review) and increased inhibition of TEOAEs has been found in children with autism spectrum disorder (Wilson et al. 2017). Stronger MOC reflexes may be a result, rather than a cause, of these conditions.

6. Descending control of efferents and attention

Olivocochlear efferents can be influenced by cortical activity (Dragicevic et al., 2015; Jager and Kossl, 2016). One manifestation is that selective attention to auditory or visual stimuli can change OAE amplitudes and MOC-reflex strength (Srinivasan et al., 2014; Wittekindt et al., 2014; Mishra and Lutman, 2014; Walsh et al., 2015; Smith and Cone, 2015; Terreros et al., 2016). Various papers have reported increases, decreases and no change in MOC activity. Several organizing principles have been proposed (e.g. non-attended frequencies are inhibited; MOC activity changes when this confers a task benefit) but none fits all of the data. Descending efferent control may be important in auditory plasticity (Kumar et al., 2010).

7. Conclusions

New data on cochlear mechanical motion has brought about a re-interpretation of how MOC effects are produced, and have resolved some previous data conflicts. MOC tuning is broad and there are differences between the ipsilateral and contralateral MOC reflexes. Recent experiments show that efferents aid hearing and are influenced by attention, but the data do not fit into easily-described conceptual frameworks. Efferent prevention of damage is significant.

Highlights.

• The anatomy and physiology of olivocochlear efferents are reviewed.

• To interpret olivocochlear effects, new cochlear mechanical data are reviewed.

• Auditory-nerve MOC effects follow reticular-lamina, not basilar-membrane, motion.

• Efferents help in detecting signals in noise, in attention, and in reducing trauma.