Despite research spanning more than two decades, the term APD is poorly defined and refers to a collection of different functional impairments (Jerger and Musiek, 2000; American Speech-Lanuage-Hearing Association, 2005; American Academy of Audiology, 2010). Children with APD typically have difficulties in complex listening situations, such as understanding speech in background noise or understanding rapid or degraded speech. However, the mechanisms of APD are unclear.

The conventional hypothesis proposes that APD results from impaired “bottom-up” sensory processing and abnormal neural representation of complex acoustic signals, e.g., speech. This may involve lesions in the central auditory nervous system or functional impairment of basic auditory processing (Jerger and Musiek, 2000; Musiek et al., 2005; Richard, 2011); the central auditory system begins in the cochlear nucleus of the medulla and ends with the auditory cortex. However, some researchers argue that the problem may be entirely cognitive, primarily affecting attention, memory or language processing, and exerting a non-specific effect on perception (Moore, 2006, 2012; Moore et al., 2010, 2011), while others claim that APD and attentional deficits are not always related (Sharma et al., 2009). Further, there is a continuing discussion on the relationship between APD and other neurodevelopmental disorders such as specific language impairment and dyslexia (King et al., 2003; Sharma et al., 2006, 2009; Dawes et al., 2008; Ferguson et al., 2011; Moore and Hunter, 2013).

Recently, Ludwig et al. (2014) demonstrated that children with (suspected) APD with normal peripheral function [i.e., normal audiograms, stapedial reflexes, and otoacoustic emissions (OAEs)] but no associated language/reading impairments have, in fact, impaired sensory processing, manifested as a discrimination deficit in spectral and temporal domains. Their results clearly imply that APD, as a distinct clinical entity, cannot be entirely dismissed. They suggested that dichotic listening deficits in these children could be influenced by the subcortical efferent auditory system, specifically the medial olivocochlear (MOC) efferents. Accordingly, it is plausible that processing deficits expressed in the ascending auditory system could reflect the “top-down” influences from cortical centers that may exert their influence on the auditory cortex, and be conveyed to auditory periphery via the efferent pathways (Moore and Hunter, 2013).

Another potential mechanism of speech-in-noise deficits despite a normal audiogram is hypothesized to be reduced temporal encoding precision of supra-threshold sound – termed as cochlear neuropathy (Bharadwaj et al., 2014). This can manifest both behaviorally, for example, frequency modulation discrimination, and in subcortical steady state responses in humans. Whether this is the same as APD is currently not known. Such lack of consensus in APD has affected diagnosis and management of this poorly understood disorder. For instance, one of the major issues in clinical assessment of APD is that most diagnostic tests lack clear scientific underpinnings and empirical evidence.

Regardless of whichever hypothesis of the pathophysiology of APD may turn out to be true, the descending system plays a significant role in hearing and auditory learning (Suga et al., 2000, 2002; Xiao and Suga, 2002; Suga and Ma, 2003; Bajo et al., 2010). The auditory system has an elaborate system of descending, efferent neural pathways that extend to the cochlea (Suga et al., 2000; Xiao and Suga, 2002). Efferent effects shape or modulate the bottom-up afferent processing. This review focuses on the medial efferents (Fex, 1962; Guinan, 2006). Auditory nerve fibers innervate reflex interneurons in the contralateral posteroventral cochlear nucleus. Axons of these interneurons cross the brainstem ventrally and innervate the medial efferent neurons on the ipsilateral side. These neurons project to the ipsilateral cochlea through the uncrossed olivocochlear bundle and modulate outer hair cell (OHC) activity (Guinan, 2006). The OHC functioning can be measured by OAEs. OAEs are sounds of cochlear origin, which can be recorded by a small microphone fitted into the ear canal. They are generated by the motion of the OHCs as they respond to auditory stimulation (Kemp, 1978).

In humans, acoustic stimulation of the MOC system results in altered cochlear amplification that is often described in terms of changes in amplitude and phase of OAEs, and is typically termed OAE suppression or MOC reflex (Guinan, 2006, 2010). In this report, the term MOC reflex is used. Animal work suggest a “MOC unmasking” hypothesis in which stimulation of the MOC efferents reduces cochlear responses to continuous noise, allowing greater responsiveness to transient signals embedded in noise (Winslow and Sachs, 1988; Kawase and Liberman, 1993; Kawase et al., 1993). While a complete understating of the role of the human MOC system in hearing is emerging (Guinan, 2006, 2010), one of the putative functions of this system is to aid in listening-in-noise (Andéol et al., 2011; de Boer et al., 2012; Abdala et al., 2014; Mishra and Lutman, 2014).

Since children with APD clinically present listening deficits in noise, a potential involvement of the MOC efferents has been investigated by some researchers as one of the underlying mechanisms of the APD (Muchnik et al., 2004; Clarke et al., 2006; Sanches and Carvallo, 2006; Garinis et al., 2008; Butler et al., 2011). The benefits of measuring MOC inhibition in children with APD are twofold. It could be useful to better define APD and identify the potential mechanisms of this disorder. Additionally, such work may elucidate the functional significance of MOC efferents in listening in complex environments during the developmental period. The potential role of the MOC system in APD pathophysiology, should it be confirmed, would be of significant clinical interest because current APD clinical test batteries lack mechanism-based physiologic tools that are relatively quick (Emanuel, 2002; Emanuel et al., 2011). In addition, it could offer rehabilitation options for the management of APD or its subgroup through auditory training since MOC reflex strength has been shown to improve with training (Veuillet et al., 2007; de Boer and Thornton, 2008).

The objective of this review was to critically evaluate the current evidence to determine whether the MOC system is altered in individuals with APD. The literature was examined in the context of current understanding of the generation mechanism of OAEs (Shera and Guinan, 1999) and recent advancements in human MOC work (Guinan, 2006, 2010). Specifically, the included studies were analyzed for study population characteristics (how the APD was defined) and the rigor of the OAE-based MOC tests, for instance, the type of OAEs used, the quality of OAE measurements in terms of signal-to-noise ratio (SNR), techniques used to control middle-ear muscle reflexes (MEMRs), and method of computation of OAE differences.

Considering the controversies that surround APD, it is important to clarify that the intent of this review was not to precisely define APD or to agree/disagree with the concept of APD per se.

One of the major concerns in OAE-based MOC studies is the possibility that the middle-ear muscle reflex contributes to the MOC reflex (or change in OAE) magnitude by influencing the stimulus and/or emissions as they transmit through the middle-ear (Guinan et al., 2003; Guinan, 2006). Depending on the level, MEMR can be evoked by contralateral broad-band noise (BBN) and/or by OAE-evoking stimulus. Included studies either did not use any MEMR test, did not report (Veuillet et al., 1999, 2007; Sanches and Carvallo, 2006) or used clinical acoustic reflex procedures (Muchnik et al., 2004; Clarke et al., 2006; Burguetti and Carvallo, 2008; Garinis et al., 2008; Yalçinkaya et al., 2010; Butler et al., 2011); see Table 1 under MEMR control. Clinical 226-Hz-probe tone procedures have been shown to yield on average 12–14 dB higher acoustic reflex thresholds, for 1000 and 2000 Hz activators, compared to wideband acoustic reflectance methods (Feeney et al., 2003). It is likely that MOC reflex estimation was compromised due to potential MEMR influences in studies that used high click levels to evoke OAEs (Muchnik et al., 2004; Sanches and Carvallo, 2006; Yalçinkaya et al., 2010) and/or high BBN levels to stimulate the MOC pathway (Muchnik et al., 2004; Yalçinkaya et al., 2010).

The noise floor associated with the OAE recording is an important variable for MOC reflex quantification because shifts in OAE level due to noise can be confused with true physiological inhibition. Likewise, lack of contralateral acoustic stimulation (CAS)-induced OAE shifts could be due to lack of inhibition or to low SNR, making the results difficult to interpret. The magnitude of the MOC effect has been reported to be dependent on the quality or SNR of the OAE data (Goodman et al., 2013). While a high SNR is desirable to detect robust MOC effects, a minimum 6 dB SNR has been shown to produce repeatable results (Francis and Guinan, 2010; Mishra and Lutman, 2013). Unfortunately, other than one study (Butler et al., 2011), most studies had low (3 dB) or unspecified SNR (see Table 1 under OAE protocol). Additionally, these studies did not examine potential SNR differences between the study and control groups.

Two studies (Muchnik et al., 2004; Yalçinkaya et al., 2010) used the non-linear CEOAE recording method (Kemp et al., 1986) to assay MOC inhibition. Although the non-linear click method of recording CE/TEOAEs eliminates stimulus ringing artifacts, it cancels the linear portion of the emission, potentially removing inhibition information. In contrast, the linear click method captures the entire emission and retains complete MOC effects.

The DPOAE is a vector combination of at least two components, termed as distortion (generated at the overlap of primary tones f₁ and f₂) and reflection, produced at the f_dp (2f₁ − f₂) place (Kim, 1980; Shera and Guinan, 1999; Talmadge et al., 1999; Knight and Kemp, 2000). The interaction of these two components in the ear canal produces oscillations in DPOAE level and phase – fine-structure. The phase interference between distortion and reflection components can confound DPOAE-based measures of the MOC reflex, occasionally creating enhancements (Abdala et al., 2009; Deeter et al., 2009). Enhancements are more common at fine-structure dips or minima where the components are out of phase. The MOC effect differentially influences the two components, with greater changes in the reflection component (Abdala et al., 2009, 2013; Deeter et al., 2009). The effect is not merely reduction in DPOAE level, but shifting of the fine-structure pattern mainly through the reflection component (Henin et al., 2011). Without due consideration of these physiologic mechanisms, the MOC-induced DPOAE changes may be inconclusive. For APD work, only one study used DPOAEs (Butler et al., 2011), but without any consideration to fine-structure or component separation.

Conventionally, the MOC reflex has been quantified by computing “raw dB differences” between OAEs with and without CAS. Most APD studies used raw dB metric to show MOC effects (Muchnik et al., 2004; Clarke et al., 2006; Sanches and Carvallo, 2006; Yalçinkaya et al., 2010; Butler et al., 2011). Such computation may not be mathematically accurate, particularly when the OAE levels between the study and control groups are different (Veuillet et al., 1999; Muchnik et al., 2004; Garinis et al., 2008; Yalçinkaya et al., 2010). Recent studies have used a variety of novel methods to compute MOC reflex (Abdala et al., 1999; Backus and Guinan, 2007; Deeter et al., 2009; Henin et al., 2011; Mishra and Lutman, 2013, 2014), for instance, a normalized index that considers the baseline OAE level or a vector index that takes into account the phase information. Three studies have specifically shown that use of normalized index provided different results or study outcomes compared to the raw dB index (Backus and Guinan, 2007; Garinis et al., 2011; Mishra and Lutman, 2014).

Other OAE- and CAS-related factors, such as slower OAE probe-drifts (Henin et al., 2011; Goodman et al., 2013) and BBN bandwidths (Maison et al., 2000; Lilaonitkul and Guinan, 2009), respectively, could influence MOC reflex estimates. Lack of these specifications in published reports also makes comparison across studies difficult.

Auditory attention, a subject-related variable, has been shown to influence MOC reflex magnitude (Maison et al., 2001; de Boer and Thornton, 2007). Controlling attention by including task-dependent measures might reduce variability. This is highly relevant for comparing APD and control groups, since a well-designed, population-based study demonstrated auditory inattention as one of the underlying mechanisms of APD (Moore et al., 2010). It is plausible that the effect of auditory attention may be different in children with APD. This prediction, however, requires experimental verification.

An obvious challenge is the lack of consensus on what constitutes an APD. Consequently, studies have used a bewildering array of tests to recruit clinical participants that lack clear theoretical import. However, this is a generic APD research problem. Until APD is precisely defined, a working ad hoc approach could be to define the experimental group as listeners with impaired speech/listening-in-noise performance, with normal peripheral function and with typical speech-language and learning abilities. It would also be informative to test whether a relationship exists between MOC impairment and the behavioral test outcome in these children (Veuillet et al., 2007).