The discovery of OAEs by David Kemp in the late 1970s was motivated by several puzzling and seemingly related psychoacoustic phenomena [9]. These included threshold microstructure and correlated fluctuations in loudness judgements, as well as the perception of tonal interactions like beating or distortion when presented with stimuli at frequencies near those of threshold minima [2, 10, 11].

As originally proposed by Kemp [2] and since re-iterated [8, 12,13,14], these phenomena can all be explained as the result of multiple intracochlear reflections of the waves that give rise to OAEs in the ear canal—specifically, what are now classified as reflection-source OAEs. The basic process (schematized in highly simplified form in Fig. 2) is initiated when forward-traveling waves are scattered by micromechanical impedance irregularities. Such irregularities, which are often referred to as “roughness” [12], presumably arise from variations in the structure or function of the organ of Corti along its length. For example, impedance irregularities could result from variations in the number, mechanical properties, and arrangement of the hair cells and supporting cells, or perhaps in the strength of any locally generated amplifying forces [8, 13]. Regardless of their precise origins, the resulting reverse-traveling waves propagate toward the stapes, where they are then re-reflected due to the impedance mismatch at the middle-ear boundary. This launches additional forward-traveling waves that can interfere constructively or destructively with the initial, stimulus-driven wave and contribute to further rounds of reflection.

Fig. 2

Schematic illustrating intracochlear reflections. Stimulation with a tone initiates a forward-traveling wave that is subject to scattering by micromechanical impedance irregularities, resulting in the generation of reverse-propagating waves (red, basally pointed arrow). Upon reaching the stapes, these waves vibrate the middle ear, producing an SFOAE in the ear canal. However, they are also partially re-reflected at the middle-ear boundary, initiating secondary forward-traveling waves (red, apically pointed arrow). These interfere with the stimulus-driven wave and contribute to additional rounds of reflection. Note that forward-traveling waves are reflected by spatially distributed irregularities and not a single point source. As detailed in [13], the traveling-wave reflectance, R, can be defined as the ratio of the net reverse-traveling wave to the forward-traveling wave, as evaluated at the stapes. Likewise, the reflectance at the middle-ear boundary, Rstapes, can be defined as the ratio of reflected to incident waves at the stapes. The summed influence of multiple intracochlear reflections on the initial forward-traveling wave converges to a factor of 1/(1 – RRstapes)

Maximal constructive interference theoretically occurs at frequencies where the round-trip phase accumulation associated with wave travel from the stapes to the more apical reflection sites and back is a whole number of cycles [13]. Such frequencies should therefore be associated with local enhancements in sensitivity (i.e., threshold minima). Additionally, provided sufficient amplification of forward- and reverse-traveling waves, the reflections can become self-sustaining. The presence of such spontaneous oscillations would naturally explain the tonal interactions experienced at frequencies near threshold minima. Of course, the existence of reverse-traveling waves and spontaneous oscillations was demonstrated by Kemp’s measurements of both evoked and spontaneous OAEs (SOAEs) in the ear canal [15, 16], which provided firm support for this framework.

Correlations Among Microstructures in Thresholds and Reflection-Source OAEs

Many, including Long and colleagues, have since replicated Kemp’s initial findings and provided further evidence that a common mechanism is responsible for the microstructure observed in thresholds and OAEs. As illustrated in Fig. 3, there is a close correspondence between frequencies of threshold minima, maxima in OAE amplitudes evoked by low-level tones and transient stimuli, and SOAEs [3, 5, 17, 18]. In fact, the precise morphology of the microstructure observed in hearing thresholds is nearly identical to that in SFOAEs elicited by low-level tones [4]. Furthermore, the minimum frequency spacing of adjacent threshold minima, evoked OAE maxima, and SOAEs also corresponds to the interval over which SFOAE phases accumulate one cycle [4, 13, 19, 20]. Again, this is consistent with maximal constructive interference occurring when the round-trip phase accumulated by waves traveling to the more apical, OAE-generating reflection sites and back to the stapes is an integer number of cycles. Note, however, that threshold minima and OAE amplitude maxima are not always accompanied by measurable SOAEs, since the latter occur only when the round-trip gain is sufficiently high. The detection of SOAEs is also potentially limited by middle-ear transmission and the noise floor of the measurement.

Fig. 3

Relationships among hearing thresholds, SFOAEs, and SOAEs. Pure-tone thresholds (top), SFOAE amplitudes (middle), and SFOAE phases (bottom) measured in 1/100th-octave steps from 1 to 2 kHz are shown for an individual human subject. SFOAE responses were obtained for 0–20 dB SPL stimuli in 5 dB steps (thicker/darker lines indicate responses at higher stimulus levels). Thin, dashed portions of the SFOAE curves indicate responses with amplitudes less than 6 dB above the measurement noise floor. A representative ear-canal spectrum in quiet is also shown along with the SFOAE amplitudes, revealing multiple SOAEs. Threshold minima align well with SFOAE amplitude maxima and SOAEs, and roughly one cycle of SFOAE phase is accumulated for each microstructure period. Data were adapted from supplementary material provided in [4], and thresholds were previously published in [6]. See the aforementioned references for more detailed methods

Dependence on Cochlear Amplification

The strength of the OAE-generating reflectance depends on the amplification of traveling waves by outer hair cells (OHCs). The microstructure common to reflection-source OAEs and behavioral thresholds should therefore be highly sensitive to OHC activity and cochlear status. Through painstaking measurements, Long and colleagues provided evidence that this is indeed the case. First, Long showed that threshold microstructure is largest under conditions where OHC-mediated amplification is presumably strongest and/or unsuppressed [5]. Specifically, threshold microstructure is reduced in the presence of tonal or wideband masking stimuli and is mostly absent once masked thresholds are elevated above ~ 40 dB SPL. Such effects are consistent with the notion that amplification and intracochlear reflections are both suppressed by the masking stimuli. Intracochlear reflections are also expected to saturate at moderate stimulus levels, reducing their overall influence on any stimulus-driven waves.

Most compellingly, Long and Tubis [21, 22] demonstrated that threshold microstructure could be reversibly reduced after several days of consuming aspirin, a salicylate known to inhibit cochlear amplification through its effects on OHC electromotility [23]. Reductions in microstructure depth were closely associated with decreases in the amplitudes of tone- and click-evoked OAEs. Additionally, the frequencies of threshold minima, evoked OAE maxima, and SOAEs were shifted downward in frequency during aspirin consumption. Similar reversible changes were observed by Furst et al. following intense noise exposure [24]. The effects of aspirin and noise could result from a reduction in the magnitude and change in the phase of the OAE-generating reflectance, presumably due to reduced OHC-mediated amplification. For instance, a downward frequency shift in the microstructure pattern would theoretically result from increasing phase lags in the cochlear vibrations near the reflection sites. Phase lags are, in fact, observed after loss of amplification due to acoustic overstimulation or death [25].

The dependence of microstructure on OHC function has also been explored through activation of the medial olivocochlear (MOC) efferent pathway, which reduces OHC-mediated amplification [26]. Activating the MOC efferents with contralateral sound produces upward shifts in the frequencies of SOAEs, threshold minima, and SFOAE amplitude maxima [6, 20, 27, 28]. It also reduces microstructure in SFOAEs [20] and in hearing thresholds, at least when thresholds are measured at frequencies distant from SOAEs [6]. These effects are consistent with MOC activation causing a reduction in the magnitude of the OAE-generating reflectance as well as a phase lead. Indeed, MOC activation produces phase leads in both evoked OAEs [20] and basilar membrane motion [29].

Threshold microstructure and SOAEs are also primarily found in ears with normal hearing, indicating their general dependence on cochlear function [30, 31]. However, some threshold variations and SOAEs have instead been associated with pathology [32, 33]. A possible explanation for this is that damage can induce impedance irregularities that actually increase the OAE-generating reflectance, with the largest irregularities likely being near the boundary between areas with normal and impaired OHC function.

Dependence on Middle-Ear Transmission

Microstructure is also sensitive to changes in the impedance at the middle-ear boundary. For instance, microstructure patterns and SOAEs are typically shifted upward in frequency by a few percent following changes in static ear-canal pressure or postural changes that influence the hydrostatic pressure at the oval window [2, 34,35,36,37]. Reductions in threshold microstructure depth and inversion of minima and maxima have also been observed under certain conditions [17]. Though somewhat variable, such effects could all theoretically be attributed to changes in the magnitude and phase of the stapes reflectance.

Note that middle-ear pathologies or certain experimental manipulations could both increase the reflectance at the stapes and reduce reverse middle-ear transmission. For instance, near-total reflection of reverse-propagating waves at the stapes could lead to strong intracochlear reflections while severely limiting the transmission of OAEs to the ear canal. Thus, one could potentially observe pronounced threshold microstructure in the absence of any measurable OAEs. Even under normal conditions, variability in middle-ear transmission across ears or across frequency likely complicates the relationship between microstructure depth and the overall level of SOAEs and evoked OAEs [4].

Microstructure in Ear-Canal Pressure

For SFOAEs, it is important to distinguish between the microstructure observed in the actual OAE response versus the rippling observed in the total pressure measured in the ear canal, prior to extracting the SFOAE. This rippling is due to interference between the stimulus tone and evoked SFOAE and is reduced under conditions where the relative magnitude of the SFOAE is anticipated to become small (e.g., at high stimulus levels, or in the presence of a second, suppressor tone [3, 5, 38]). It is of course this latter fact that allows estimation of the stimulus pressure so that it can be subtracted from the total measured pressure, yielding the SFOAE.

Like threshold and OAE microstructure, ripples in ear-canal pressure have a periodicity that is linked to the phase-vs.-frequency gradient of the SFOAE, with adjacent peaks being associated with ~ 1 cycle of SFOAE phase accumulation. However, as noted by Long, individual peaks and valleys in ear-canal pressure are not necessarily aligned well with those observed in behavioral and OAE microstructure [5] (see also [17, 39]). This is demonstrated in Fig. 4, which compares the frequencies of SFOAE amplitude maxima observed in Fig. 3 with the ear-canal pressure measured for this subject during the presentation of the 15 dB SPL stimulus. The lack of alignment is likely due to differences in the phases of the waves that are reflected at the stapes compared to the phases of those that reach the ear canal. The latter include additional delays associated with middle-ear transmission and propagation to the probe microphone [4]. Note that ripples in the ear-canal pressure would occur even if SFOAE amplitudes (and behavioral thresholds) were completely flat across frequency, exhibiting no microstructure. Such ripples only require that the SFOAE phase rotates relative to stimulus phase.

Fig. 4

Relationship between ripples in the total ear-canal pressure and SFOAE amplitude maxima. The ear-canal pressure in response to a 15 dB SPL swept-tone stimulus (thin black curve) is compared with the response to a 60 dB SPL stimulus that has been scaled down by 45 dB (thick gray curve). Measurements are from the subject whose extracted SFOAEs are shown in Fig. 3. Ripples are present at the lower stimulus level due to interference between the stimulus and the evoked SFOAEs. However, amplitude maxima in the extracted SFOAEs (indicated by vertical dashed lines) do not align well with maxima in the ear-canal pressure. Note that the variations in stimulus pressure not attributable to interference between the stimulus and SFOAEs result from the calibration method, which attempted to achieve a constant SPL at the eardrum, rather than at the probe microphone. Relevant methods are described in [4, 6]

Distinguishing Between SFOAE Micro- and Macrostructure

It is also important to distinguish between microstructure and the peaks, valleys, and notches that instead constitute SFOAE “macrostructure” [40]. While perhaps arbitrarily defined, macrostructure is the background structure upon which the microstructure is superimposed. This background structure often appears to have a broad quasiperiodic spacing and is generally preserved across stimulus levels. While it is easier to visualize at higher stimulus levels, where the influence of microstructure is less pronounced, it can also be observed at low levels, as shown in Fig. 5. In this example, the macrostructure was approximated by smoothing the raw SFOAE amplitudes. However, since the notches that form part of the macrostructure morphology can be quite sharp, simply smoothing SFOAE spectra can remove both microstructure and macrostructure and does not always clearly separate the two.

Fig. 5

SFOAE macrostructure. Amplitudes and phases of SFOAE responses to discrete tones presented at 18 dB forward pressure level (FPL) are shown for an individual human subject. Smoothing the raw SFOAE amplitudes highlights the macrostructure, which is also quasiperiodic. The SFOAE phase accumulates roughly three cycles between adjacent macrostructure peaks. The ear-canal spectrum measured in quiet is also shown with the SFOAE amplitudes and reveals several low-level SOAEs. Thin dashed portions of the SFOAE response curves indicate responses with amplitudes less than 6 dB above the noise floor. SFOAEs were measured using the suppression method as described in [4]

SFOAE macrostructure was recently characterized in detail by applying a more sophisticated approach to remove the long-latency SFOAE components that are associated with multiple intracochlear reflections [40]. After effectively removing the microstructure, the distance between macrostructure peaks was found to be ~ 20–30% of the stimulus frequency. This spacing corresponds to the span over which SFOAE phase accumulates roughly three cycles (as observed in Fig. 5) and is therefore approximately three times wider than the microstructure periodicity. Though the macrostructure periodicity is apparently predicted by the coherent-reflection theory of SFOAE generation, which links this periodicity to the spatial width of the dominant reflection region (i.e., the traveling wave peak) [8, 40], its importance and interpretation remain to be more fully explored.

Influence of SOAEs on Perception and Threshold Microstructure Morphology

SOAEs emerge from the same processes as threshold microstructure and can also directly impact microstructure morphology via interactions with the stimulus tone. Following several earlier reports [2, 3, 34, 41], Long and colleagues explored how external tones can interact with the SOAE to produce beats and sensations of “roughness” (not to be confused with the micromechanical impedance irregularities discussed earlier), or else entrain or suppress the SOAE [21, 42]. The type and strength of any interactions depend on the relative frequencies and levels of the external tone and SOAE, with suppression or entrainment occurring when the external tone becomes increasingly dominant. The boundary between perceiving a pure tone versus roughness also aligns well with that for the acoustic entrainment of an SOAE.

While SOAEs need not be associated with all threshold minima, Long demonstrated that microstructure-like patterns can be induced by presenting a low-level tone to mimic the presence of an SOAE in a frequency region lacking natural microstructure [21]. Threshold fluctuations caused by the presence of an SOAE-mimicking tone disappear when thresholds are measured using narrowband noise stimuli, which effectively eliminate perceptual cues like beating and distortion. In contrast, naturally occurring microstructure is largely preserved when using such narrowband noise stimuli. Thus, perceptual cues arising from the presence of SOAEs may “color” the threshold curve but are not required to produce pronounced microstructure.

Interestingly, the presence of an SOAE appears to result in an overall elevation of hearing thresholds at nearby frequencies. This was convincingly shown by Long and Tubis, who found that reductions in SOAE amplitudes during the early stages of aspirin consumption actually improved thresholds [22]. Threshold improvements coincident with reductions in SOAE amplitude have also been observed after noise exposure [24] and MOC activation [6]. Such improvements are probably due to a release from the neural masking that is caused by spontaneous cochlear oscillations [43]. A reduction in SOAE amplitude also changes how external stimuli interact with the SOAE [44], though whether this aids signal detection is unclear.

Reducing the Influence of Multiple Intracochlear Reflections

Prominent microstructure can be observed in OAE responses for stimulus levels up to ~ 20 dB SPL or more, indicating that multiple intracochlear reflections significantly influence cochlear mechanics at supra-threshold levels. While studying the microstructure itself may provide insight into the mechanisms underlying intracochlear reflections, it may otherwise be desirable to minimize the influence of such reflections. Doing so provides a more straightforward window onto the form of the underlying OAE-generating reflectance and may reduce the variability in responses across ears. While this can be accomplished by simply smoothing the responses (as in Fig.