JARO 2/1: 031–040 (2001)

DOI: 10.1007/s101620010066

Adaptation of Distortion Product Otoacoustic Emission in

Humans
D.O. KIM,1 P.A. DORN,2 S.T. NEELY,2 AND M.P. GORGA2
1
Department of Neuroscience, Division of Otolaryngology, Department of Surgery, University of Connecticut Health Center,
Farmington, CT 06030, USA
2
Boys Town National Research Hospital Omaha, NE 68131, USA
Received: 26 July 2000; Accepted: 9 October 2000; Online publication: 20 March 2001

ABSTRACT results should help advance the basic knowledge of

human cochlear mechanics operating under the con-
Previous studies of animals observed a phenomenon of trol of the MOC feedback system and contribute to
adaptation of distortion product otoacoustic emission the development of practical applications such as iden-
(DPOAE) and found that the phenomenon was medi- tifying people at high risk of acoustical injury and a
ated to a large extent by the medial olivocochlear clinical test of the functional status of the MOC system.
(MOC) reflex. The present study investigated DPOAE Keywords: olivocochlear, efferent, reflex, feedback,
adaptation in humans. The following stimuli were cochlear mechanics, auditory, hearing
used: f2/f 1 ⫽ 1.2; f2 ⫽ 2, 4, or 5.65 kHz; L2 ⫽ 50–65
dB SPL re 20 ␮Pa rms, L1 ⫺ L2 ⫽ 0–15 dB, where L1 and
L2 represent levels of the f1 and f2 tones, respectively;
duration of two-tone burst ⫽ 5.5 s; interburst gap ⫽
INTRODUCTION
20 or 30 s; number of repetitions ⫽ 40 or 64. We
analyzed the 2f1 ⫺ f2 DPOAE as a function of time
Medial olivocochlear (MOC) neurons in the brain
using a method of heterodyne envelope detection. The
stem make descending projections onto outer hair
subjects were 20 humans aged from 15 to 54 years
cells (OHCs) of the cochlea, forming a feedback or
(median ⫽ 21 years) with normal hearing. We
reflex system (e.g., Warr and Guinan 1979). The MOC
observed that (1) humans exhibited DPOAE adapta-
tion phenomenon; (2) the time course of DPOAE system is believed to influence the mechanics of OHCs
level was characterized by a 2-exponential function; mechanically because activation of the efferent system
was found to produce changes in distortion product
(3) distributions of the fast and slow time constants
were well separated with their median values being 69 otoacoustic emissions (DPOAEs), a mechanical signal
ms and 1.51 s, respectively; (4) distributions of the (Mountain 1980; Siegel and Kim 1982). This notion
magnitudes of the fast and slow adaptation compo- is consistent with the subsequent finding that OHCs
nents were largely overlapped with their median values are electromotile (Brownell et al. 1985; Zenner et al.
1985) and thus potentially able to influence the
being 0.65 and 0.40 dB, respectively; and (5) the com-
bined magnitude of the adaptation ranged from 0.4 mechanics of the cochlea, as has been postulated theo-
to 3.0 dB with a median of 1.10 dB. To our knowledge, retically in “cochlear amplifier” models (e.g., Davis
1983; Neely and Kim 1983).
the present study is the first published article to
Effects of the MOC acoustic reflex have often been
describe adaptation of DPOAE in humans. These
demonstrated with changes in various measures of
cochlear responses produced by contralateral stimula-
tion (e.g., Buño 1978; Warren and Liberman 1989;
Correspondence to: Dr. D.O. Kim • Department of Neuroscience • Folsom and Owsley 1987; Collet et al. 1990), including
University of Connecticut Health Center • Farmington, CT 06030-
3401. Telephone: (860) 679-3690; fax: (860) 679-8766; email: a change in DPOAE by contralateral stimulation (Puel
[email protected] and Rebillard 1990; Kujawa et al 1993; Moulin et al.
31
32 KIM ET AL.: Human DPOAE Adaptation

1993; Williams and Brown 1995; Puria et al. 1996). Stimulus

The contralateral influences were eliminated when the
olivocochlear bundle was transected (Warren and Lib- The following two-tone stimuli were used: f2 ⫽ 2, 4,
erman 1989; Puel and Rebillard 1990; Liberman or 5.65 kHz; f2/f1 ⫽ 1.2; L2 ⫽ 50–65 dB SPL re 20 ␮Pa
1990). rms, L1 - L2 ⫽ 0–15 dB, where L1 and L2 represent
Recently, a method has been introduced to observe levels of the f1 and f2 tones, respectively; duration of
an ipsilateral effect of the MOC-system reflex on two-tone burst ⫽ 5.5 s; interburst gap ⫽ 20 or 30 s;
DPOAE. Liberman et al. (1996) observed a phenome- number of repetitions ⫽ 40 or 64; sample rate ⫽ 48
non of adaptation of DPOAE in cats where the 2f1 ⫺ kHz. The stimulus waveform was smoothly gated on
f2 DPOAE level changed gradually over several seconds and off over 10-ms rise–fall times by multiplying the
after the onset of the primary stimulus tones. This stimulus waveform with a raised-cosine function.
phenomenon was interpreted to be mediated mainly
by the MOC system (further discussed in the Discus- Analysis of DPOAE level and phase versus time
sion section). The ipsilateral effect of the MOC reflex
is an inherent, inseparable part of the response of the A time-domain averaging of the ear-canal pressure
cochlea–brain-stem feedback system which comprises waveform (duration ⫽ 5.5 s) was performed with a
the cochlea, cochlear nucleus, and MOC neurons. sample rate of 48 kHz. Initially, we performed a fast
Whenever the cochlea–brain-stem system of an animal Fourier transform (FFT) of the complete pressure
(or a human) is acoustically stimulated, the ipsilateral waveform, 262144 points representing 5.5 s. We
reflex of the MOC system is potentially activated. obtained the magnitude and phase of the DPOAE
The phenomenon of DPOAE adaptation has subse- component as a function of time as follows. We selected
quently been observed in other animal species, i.e., a band of (positive) frequencies surrounding a particu-
guinea pigs, mice, and gerbils (Kujawa and Liberman lar DPOAE frequency and shifted these frequency
1998; Sun and Kim 1999; McGee et al. 2000). Investiga- components down such that the DPOAE frequency
tion of this phenomenon in humans should be helpful was positioned at the zero frequency. A Blackman-
in (1) advancing basic knowledge of the dynamics of shaped low-pass filter extending ⫾120 Hz around the
the human cochlea–brain-stem feedback system; (2) zero frequency was applied to the shifted spectrum.
contributing to the development of methods of practi- Then an inverse FFT was performed. The result was
cal applications such as prediction of vulnerability to a complex-valued signal spanning the duration of the
acoustical injury and a noninvasive clinical test of the original waveform. This signal represents the magni-
functional status of MOC neurons in the human brain tude and phase of the DPOAE component as a func-
stem. The goals of the present study were (1) to deter- tion of time. The magnitude-versus-time function, in
mine whether humans exhibit the phenomenon of this case, corresponds to the temporal “envelope” of
DPOAE adaptation and (2) to determine time con- the DPOAE component (e.g., Oppenheim and Shafer
stants and magnitudes of DPOAE adaptation in 1975, p. 361), sometimes called a “Hilbert” envelope.
humans. Early results of this study were reported by The magnitude and phase of the DPOAE compo-
Neely et al. (1999) and Kim et al. (2000). nent obtained by the present method represent an
analytic signal. An analytic signal has the property that
its Fourier transform is zero for “negative” frequencies
(e.g., Oppenheim and Shafer 1975, p. 358). Instead
of computing the analytic signal directly by zeroing
METHODS
the negative frequency components, we heterodyned
a band of spectral components around the DPOAE as
described above. Thus, we refer to the present method
Subjects as a “heterodyne” method. The advantages of the het-
erodyne method over direct computation of the ana-
Responses of 24 ears of 20 normal-hearing human lytic signal are that (1) the inverse FFT size is much
subjects were examined in this study. All subjects had smaller than the size of the original waveform buffer
auditory thresholds less than or equal to 20 dB HL (4096 vs. 262144), and (2) the phase deviation is
for octave frequencies from 500–800 Hz. The subjects smaller because the phase of the center frequency has
did not have a history of exposure to noise, ototoxic effectively been subtracted out. Talmadge et al. (1999)
medications, middle ear pathology, or ear surgery. For used other, similar signal processing methods to obtain
10 subjects, middle-ear function was determined to be analytic-signal representations of DPOAEs. Relkin et
normal by standard tympanometry on the day of the al. (1999) used a phase-locked amplifier, the analog
DPOAE recording. The ages of the subjects were from equivalent of the heterodyne method, to obtain
15 to 54 years (median ⫽ 21 years). DPOAE envelopes.
KIM ET AL.: Human DPOAE Adaptation 33

2-exponential fit: L(t) ⫽ Lss ⫹ mf ⭈ exp(⫺t/␶f)

⫹ ms ⭈ exp(⫺t/␶s) (2)
where mf and ms are the magnitudes of the fast and
slow components of adaptation (in dB) and ␶f and ␶s
are the time constants of the fast and slow components
of adaptation (in ms).
mcom ⫽ 앚mf앚 ⫹ 앚ms앚 (3)
where mcom is the combined adaptation magnitude
representing the sum of the absolute values of mf
and ms.
For the single-exponential fit, the initial estimate
for the time constant was always 0.1 s. The initial esti-
mate for the two-exponential fit was based on the
results of a single-exponential fit, so that it would
always be an improvement over the single-exponential
fit. The “variance accounted for” (VAF) is a metric for
the goodness of fit, defined as:
VAF ⫽ 1 ⫺ r (4)
where r is the ratio of (1) the mean-squared difference
between a fitting function and the DPOAE envelope
FIG. 1. Level (a) and phase (b) of distortion product otoacoustic
level and (2) the mean-squared deviation of the time-
emission (DPOAE) vs. time for results from normal human subject varying DPOAE envelope level from its time-averaged
No. 1. value. The value of VAF ranges from 0 (a poor fit) to
1 (a perfect fit).
Figure 2 illustrates exponential fits for results from
The DPOAE envelope magnitude (in Pa) was then subject No. 1. A 1-exponential fit (Fig. 2a) provided a
converted into DPOAE envelope level (in dB SPL). fit for the response over most of the time period except
DPOAE envelope level and phase were displayed as for the peak at the beginning. In contrast, a 2-exponen-
functions of time as illustrated in Figure 1. In an analo- tial fit (Fig. 2b) provided an improved fit for both the
gous way, the f1 and f2 primary components in the ear- initial peak and the later part of the response. The
canal pressure were also analyzed as functions of time. value of VAF increased from 0.58 with the 1-exponen-
tial fit to 0.60 with the 2-exponential fit. Figure 2c
1- and 2-exponential fits displays the response, together with the 2-exponential
fit, over the full duration of stimulus (5.5 s), whereas
A fitting function was applied to a function of DPOAE
Figure 2a, b display the response over the initial 2s.
envelope level (in dB SPL) versus time as follows. A
nonlinear (simplex) fitting procedure was performed
to obtain the least-squared difference between the
RESULTS
DPOAE envelope level and an exponential fitting func-
tion. For this step, a short offset period (typically 20–40
This article describes responses of 24 ears of 20 nor-
ms) of the DPOAE envelope level at the beginning
mal-hearing human subjects to three different f2 fre-
and the same period at the end of the stimulus dura-
quencies and several different L2/L1 values as stated
tion were excluded to remove transients at the two
in the Methods section. The time courses of DPOAE
ends.
levels were variable among subjects, with several repre-
One of the following fitting functions was applied:
sentative examples illustrated in Figure 3. Each case
1-exponential fit: L(t) ⫽ Lss ⫹ m ⭈ exp(⫺t/␶ ) in Figure 3 is shown with a 2-exponential fit. Some
exhibited a slower time course (panel a), some faster
(1)
time courses (panels b and c). Most of the cases exhib-
where L(t) is the fit for DPOAE envelope level (in dB ited a decreasing time course (panels a–c), while some
SPL) as a function of time t (in ms), Lss is the steady- exhibited a biphasic time course (panel d), where
state level of DPOAE (in dB SPL), m is the magnitude DPOAE level showed an initial decrease followed by
of adaptation (in dB), and ␶ is the time constant of an increase.
adaptation (in ms). Distributions of the fast and slow time constants of
34 KIM ET AL.: Human DPOAE Adaptation

FIG. 2. DPOAE level vs. time with a 1-exponential fit (a) or 2-

exponential fit (b, c). The abscissa in c covers a longer range than
in a and b.

DPOAE adaptation (␶f and ␶s, respectively) are shown

in Figure 4. Note that a logarithmic scale of the time FIG. 3. A variety of time courses of DPOAE level together with 2-
constants was chosen for the abscissa in this figure in exponential fits. The subject # is shown for each panel. The stimulus
order to show both ␶f and ␶s in the same graph. The conditions were (a) f2 ⫽ 5.7 kHz, L2/L1 ⫽ 60/68 dB SPL; (b) f2 ⫽ 5.7
kHz, L2/L1 ⫽ 65/73 dB SPL, (c) f2 ⫽ 4.0 kHz, L2/L1 ⫽ 60/68 dB SPL;
data of this figure demonstrate that the distributions (d ) f2 ⫽ 4.0 kHz, L2/L1 ⫽ 60/75 dB SPL.
of the two time constants are clearly separated with
little overlap between them. This observation of a sepa-
ration of the two distributions supports the concept
of two temporally distinct components of DPOAE Distributions of the magnitudes of the fast and slow
adaptation. The median values of ␶f and ␶s were 69 ms components of DPOAE adaptation (mf and ms respec-
and 1.51 s, respectively. The fact that there were many tively) are shown in Figure 5. The two distributions
cases with ␶s ⫽ 5.5 s (i.e., the rightmost bar in Fig. 4) have a large overlap. The values of mf and ms ranged
suggests that the 5.5-s stimulus duration used in this from ⫺0.3 to 2.2 dB. Their median values were 0.65
study probably was not long enough to determine ␶s and 0.40 with the fast component being larger. All
accurately. That is, data with a longer stimulus duration of the mf and most of the ms values were positive,
(perhaps 10 or 20 s) are expected to be necessary to corresponding to decreases of DPOAE level with time.
determine accurately the distribution of ␶s. In two cases, ms values were negative, corresponding
KIM ET AL.: Human DPOAE Adaptation 35

FIG. 4. Distributions of the fast and slow time constants of DPOAE

adaptation, i.e., ␶f and ␶s respectively.

FIG. 6. Top: Magnitude of the fast or slow adaptation component

(mf or ms) vs. the corresponding time constant (␶f or ␶s). Filled and
open symbols represent the fast and slow components, respectively.
Bottom: mf or ms vs. ␶f or ␶s for a representative subset of the data.
Each pair of data symbols for an individual record is connected by
a straight line. The data for the bottom panel represents every fifth
FIG. 5. Distributions of the magnitudes of the fast and slow compo- record in the data set where the data records were rank-ordered with
nents of DPOAE adaptation, i.e., mf and ms respectively. respect to ␶s/␶f.

to increases of DPOAE level with time (as shown in in the figure caption. Here, each pair of data symbols
Fig. 3d). for an individual record is connectedby a straight line.
The relationships between the magnitudes and time The graph shows that, in individual records, (1) the
constants of the adaptation are examined in Figure 6. lines connecting each pair of symbols are long and
In the top panel, filled and open circles represent (2) the lines tend to have negative slopes. These char-
the fast and slow components, respectively. The graph acteristics which are visible in the individual records
shows that larger/smaller magnitudes tend to be asso- are consistent with those visible in the scatter plot of
ciated with shorter/longer time constants, both within the data indicated above.
each group and across the two groups combined. The The distribution of the combined magnitude of
two groups of symbols are separated such that the DPOAE adaptation is shown in Figure 7. It ranged
filled circles are concentrated in an upper-left area from 0.4 to 3.0 dB with a median of 1.10 dB. A typical
and the open circles in a lower-right area. One measure time course of DPOAE adaptation in humans, as
of separation between the fast and slow components derived from a 2-exponential function using the
in individual records is ␶s/␶f. The median value of this median values of the parameters indicated above, is
measure among the data set was 24, indicating a clear shown in Figure 8. It exhibits an initial fast decline of
separation between the two time constants. The 0.65 dB followed by a slow decline of 0.4 dB, reaching
median value of mf ⫺ ms was 0.25 dB indicating that a steady-state level by about 4.5 s.
the fast component tended to have larger magnitudes. For each record of DPOAE, the present analysis
Figure 6 (bottom panel) illustrates the relationship yielded both phase and level of DPOAE as functions
between the fast and slow components in individual of time as indicated in the Methods section. Changes
records for a representative subset of the data. (Only of DPOAE phase were recognizable in some cases, and
a subset of the data are shown in this panel to make the polarity of a phase change was either in the leading
the individual lines more visible. The method of selec- or in the lagging Figure 9 illustrates an example where
tion of the data in Figure 6 bottom panel is explained the phase of DPOAE gradually changed in the leading
36 KIM ET AL.: Human DPOAE Adaptation

FIG. 7. Distribution of the combined magnitude of DPOAE

adaptation.

FIG. 9. Level (a) and phase (b) of DPOAE vs. time for results from
normal human subject No. 4, illustrating a gradual change of phase
in the leading direction.

FIG. 8. A typical time course expected for DPOAE adaptation in a function of time, i.e., DPOAE adaptation, in humans.
humans as derived from a 2-exponential function using the median
The time course of the human DPOAE adaptation was
values of the parameters shown in the inset.
well characterized by a 2-exponential function. The
fast component typically had a 69-ms time constant
direction by about 4.5⬚ as the level of DPOAE with a 0.65-dB decreasing magnitude. The slow compo-
decreased by about 1 dB. In the example of Figure 1, nent typically had a 1.51-s time constant with a 0.4-dB
the phase of DPOAE underwent a change of about decreasing magnitude. Thus, the total DPOAE adapta-
1.3⬚ where the time course of phase was suggestive of tion in humans was typically 1.1 dB.
a biphasic type, i.e., an initial lag followed by a lead. Agrama et al. (1998) described preliminary results
In most of the cases, the observed phase change of of a study of DPOAE adaptation in humans. Their obser-
DPOAE was less than 7.2⬚. However, in some cases vations are consistent with our results. A limitation of
much larger phase changes (70⬚–80⬚) were also their method, however, is that it did not make a continu-
observed. Those large phase changes were associated ous measurement of the DPOAE level versus time.
with sharp notches of DPOAE level at the onset, analo- Rather, they sampled the DPOAE level at certain desig-
gous to the observation of Talmadge et al. (1999, Fig. nated 46-ms-long time windows, e.g., a window at the
8) (see the Discussion section for further discussion.) beginning andanother at the end of the stimulus-burst
Systematic analysis of time-varying DPOAE phase char- duration, 1.7 s. They observed that the DPOAE level of
acteristics, e.g., distributions of the amount and polar- humans sampled in the 1.7-s window was 0.5–1.0 dB
ity of DPOAE phase change, have not been performed lower than that sampled in the initial window. Another
yet. Such results will be reported in the future. limitation is that the 1.7-s stimulus-burst duration would
be too short to reveal the slow component of DPOAE
adaptation in view of the present observations.
DISCUSSION
The fast and slow time constants of DPOAE adapta-
tion in animals (Liberman et al. 1996; Sun and Kim
Main findings and comparison with other 1999; McGee et al. 2000) are comparable to those of
studies humans (Figs. 3, 4). The main difference between
To our knowledge, the present study is the first pub- humans and animals regarding DPOAE adaptation is
lished article to describe changes of DPOAE level as that the adaptation magnitudes observed in humans
KIM ET AL.: Human DPOAE Adaptation 37

are smaller than those observed in animals. Although of two tones as stimuli. The equipment used in those
the reason for this difference is not clear, it may be studies could not provide stimulation, recording, and
related to the fact that the absolute level of DPOAE DPOAE analysis continuously over several seconds. In
of humans is much lower than that of nonprimate contrast, the system used in the present study [the
animals. Whether the fact that ear-canal cavities of early results of which were reported by Neely et al.
humans are larger than those of animals fully accounts (1999) and Kim et al. (2000) did provide these fea-
for lower levels of DPOAE in humans than in animals tures. Since a study of adaptation normally employs a
is not known. The observation that humans exhibit steady and continuous stimulus with a correspondingly
more prominent, transiently evoked OAEs than ani- continuous analysis of the response, the present system
mals suggest that, beyond differences in ear-canal cavi- represents an improvement over those used in previ-
ties, there may be differences in OAE mechanisms ous studies.
between humans and animals. It is desirable that a
theory of mammalian cochlear mechanics should
Middle-ear muscle and olivocochlear systems
explain both of the differences, i.e., those in the abso-
lute level and in the adaptation magnitude of DPOAE The middle-ear muscle reflex may mediate DPOAE
between humans and nonprimate animals. adaptation. This reflex, however, does not appear to
Another difference between humans and animals be a major source of DPOAE adaptation, e.g., DPOAE
regarding DPOAE adaptation is that animals show adaptation in cats with completely severed middle-ear
increasing (Liberman et al. 1996; Sun and Kim 1999) muscles was not substantially different from DPOAE
and biphasic types of time courses (McGee et al. 2000) adaptation in cats with intact middle-ear muscles (e.g.,
more prominently than humans. The biphasic-type Fig. 9 of Liberman et al. 1996). The observation that
adaptation (Fig. 3d) was rare and the increasing-type DPOAE adaptation was mostly abolished by olivococh-
adaptation was absent in the present data from humans lear transection (Liberman et al. 1996) supports the
(see below). view that DPOAE adaptation is mainly mediated by
In a study of DPOAE in guinea pigs, adaptation the olivocochlear system.
magnitude and polarity were much affected by stimu- Liberman et al. (1996) also observed that DPOAE
lus parameters, particularly L1 ⫺ L2 (Kujawa and Liber- adaptation included a component that had a slower
man 1998). It is possible that the stimulus parameters time constant and a smaller magnitude and that
of the present study differed from those that produce remained after olivocochlear transection. This obser-
maximal magnitudes of DPOAE adaptation in vation suggests that the slower component of DPOAE
humans. The recording time was much longer in adaptation, in part or in whole, is mediated by a local
humans (25–30 min per record) because of a lower cochlear process, possibly associated with generation
signal/noise ratio and a need to do more averaging of the distortion product involving OHC motility. In
compared with the recording in animals (typically the present study of humans, it was not possible to
2.5–8 min). Consequently, we have not had the oppor- ascertain how much of the DPOAE adaptation is medi-
tunity to explore the stimulus-parameter space system- ated by the olivocochlear system and how much by
atically. The information that was available in the a local cochlear process. However, we interpret the
present data regarding the effect of L1 ⫺ L2 on adapta- DPOAE adaptation of humans described in this study,
tion magnitude was not sufficient to reveal a general at least the fast component, to be mediated by the
trend. A systematic exploration of the stimulus-param- olivocochlear system. Our reason is that DPOAE adap-
eter space in future studiesof humans should help tations of humans and cats are qualitatively similar.
determine whether larger magnitudes of DPOAE Caution is needed here because the slow-adaptation
adaptation can be observed in humans under certain component was also affected by olivocochlear transec-
stimulus conditions. In another study of guinea pigs, tion in some cases (in Liberman et al. 1996). There-
adaptation magnitude was observed to be larger in an fore, the slow-adaptation component should not be
awake condition than when anesthetized (Boyev et al. automatically interpreted to be mediated entirely by
2000). An effect of anesthesia cannot be a possible a local cochlear process.
explanation for the smaller magnitude of human
DPOAE adaptation because the human subjects of the
MOC vs. LOC
present study were not anesthetized.
The olivocochlear system consists of the medial and
lateral olivocochlear (MOC and LOC) neurons, which
Different methods of observing DPOAE
project to the outer hair cells (OHCs) and to afferent
adaptation
fibers beneath the inner hear cells (IHCs), respectively
Earlier studies of DPOAE adaptation (Liberman et al. (e.g., Warr 1992; Guinan 1996). The MOC subsystem,
1996; Sun and Kim 1999) used trains of 64-ms bursts rather than the LOC subsystem, is believed to underlie
38 KIM ET AL.: Human DPOAE Adaptation

DPOAE adaptation for the following reasons. Liber- parameters of the above type of model should be use-
man et al. (1996) found that sectioning only the ful. In this regard, combined information about phase
crossed olivocochlear bundle (COCB) at the midline and level of DPOAE as functions of time should be
was as effective in disrupting the ipsilaterally evoked more useful in inferring the model parameters than
DPOAE adaptation as sectioning the complete olivo- information about DPOAE level alone. The pattern
cochlear bundle. This finding suggests that LOC neu- of a notch in the level and a large abrupt change in
rons that are spared by the midline section do not the phase of DPOAE at the stimulus onset, which was
contribute to the adaptation. In addition, the present observed by Talmadge et al. (1999, Fig. 8) and in the
view is consistent with the observations of OHC electro- present study (data not shown), has been interpreted
motility and of the influence on DPOAEs by electrical by Talmadge et al. to correspond to a destructive inter-
activation of the COCB cited in Introduction. action between a smaller contribution from the source
at the primary-frequency location and a larger and
delayed contribution from a source at the distortion-
Implications of the biphasic- and increasing-
frequency location. Compared with the studies of Tal-
type adaptation of DPOAE
madge et al. (1999), our study extends information
In some cases of the present data, the time course of about changes of DPOAE phase and level over longer
DPOAE adaptation was biphasic (Fig. 3d). In animals, periods of time, i.e., over several seconds as opposed
both biphasic-type (McGee et al. 2000) and increasing- to 200 ms.
type adaptations of DPOAE (Liberman et al. 1996; Sun In view of the above theoretical consideration, the
and Kim 1999) have often been observed. Such time fact that nonprimate animals show increasing- and
courses may arise as a result of an interaction between biphasic-type adaptations of DPOAE more promi-
contributions from two sources of DPOAE. It is nently than humans implies a large difference between
believed that two sources underlie the DPOAE signal humans and animals regarding the relative DPOAE
measured in the ear canal, i.e., one at the primary- contributions from the two sources. From the above
frequency location and the other at the distortion- discussion, the following emerges as a viable hypothe-
frequency location, as originally suggested by Kim sis: All of the several differences mentioned above
(1980, p. 312) and supported by many other studies between humans and nonprimate animals, i.e., those
(e.g., Kemp and Brown 1983; Kummer et al. 1995; regarding (1) the absolute level of DPOAE, (2) the
Brown et al. 1996; Heitmann et al. 1998; Talmadge et magnitude of DPOAE adaptation, and (3) the preva-
al. 1998; Shera and Guinan 1999; Stover et al. 1999; lence of biphasic and increasing time courses of
Knight and Kemp 2000 Moulin 2000; Parham et al. DPOAE adaptation, have a common origin, i.e., a large
2001). difference in the relative contributions to DPOAE
To explain various types of DPOAE time courses, from the two sources of DPOAE (at the primary-fre-
one may use a model of the type described by Tal- quency and distortion-frequency locations). Efforts to
madge et al. (1998, 1999) and Shera and Guinan evaluate this hypothesis should be worthwhile.
(1999). The distortion product (DP) at the 2f1 ⫺ f2
frequency is initially generated at the primary-fre-
quency location and propagated basally, constituting Prediction about basilar membrane motion
one component of the DPOAE. The DP signal gener-
ated at the primary location is also propagated apically A combined hypothesis about OHCs and MOC neu-
and reflected at the 2f1 ⫺ f2 distortion-frequency loca- rons (e.g., Kim 1984; Kim et al. 1998) postulates that,
tion, constituting the second component of DPOAE. when the ear is subjected to a steady-state stimulus
(To be accurate, one also needs to include a contribu- lasting several seconds, the cochlear amplifier gain is
tion from multiple reflections involving both a reflec- controlled by reflex activities of MOC neurons. Under
tion at the cochlear base and a reflection at the such a condition, the cochlear mechanical response
distortion-frequency location.) One can represent the is postulated to undergo a gradual change reflecting
DPOAE signal as a sum of two vectors corresponding the activities of MOC neurons. DPOAE adaptation pro-
to the two main components of DPOAE. vides strong support for this hypothesis. A prediction
Even if the effect of the MOC reflex on the DPOAE of the hypothesis is that the amplitude of basilar mem-
component contributed by the primary location is brane motion (e.g., in response to one or two tones)
always a decrease of magnitude over time, one can get should undergo an adaptation over a few seconds,
a decreasing, increasing, or biphasic time course of analogous to the DPOAE adaptation. To our knowl-
DPOAE adaptation depending upon the phase rela- edge, there are no reports of such adaptation in basilar
tionship between the two DPOAE components. Future membrane motion. A reliable measurement should
efforts to convert the experimental observations of reveal such an adaptation in basilar membrane
DPOAE adaptation in humans and animals into motion.
KIM ET AL.: Human DPOAE Adaptation 39

Practical applications of DPOAE adaptation in recording time of 25–30 min for one particular stimu-
humans lus. Efforts that attempt to obtain adequate informa-
tion more rapidly by reducing the number of
Maison and Liberman (2000) observed that, in guinea repetitions, the silent gap, or both should be helpful.
pigs, DPOAE adaptation magnitude typically under- Another obstacle to practical applications will be that
went a pattern of change that included a transition the human adaptation magnitude is small and variable
from a negative value (i.e., DPOAE level increasing among a group of subjects. Stimulus conditions that
with time) to a positive value (i.e., DPOAE level yield larger DPOAE adaptations should help reduce
decreasing with time) when the stimulus-level condi- the required measurement time.
tion changed from L1 ⫺ L2 ⫽ 15 dB to 0 dB while L1 was
fixed. They interpreted the full range of the DPOAE
adaptation magnitude (i.e., the sum of the absolute
ACKNOWLEDGMENTS
values of the most negative and the most positive mag-
nitudes of the adaptation) to be an indication of the
strength of the MOC reflex. They found that this mea- This study was supported in part by grants No. DC00360
(to Kim), No. DC02251 (to Gorga), and No. DC00982 (to
sure of the reflex strength was negatively correlated
BTNRH) from NIDCD, NIH. We thank J. Joseph, M. English,
with the degree of hearing loss produced by exposure and D. Machado for help in recording the data; R. Bir-
to an intense sound. Accordingly, a measure derived mingham, R. Pietrzak, and D. Machado for providing com-
from DPOAE adaptation magnitude may serve as a ments on the paper; and the subjects who volunteered their
useful noninvasive predictor of vulnerability to acousti- time for this study.
cal injury.
Applying the above concept to humans is desirable
since the variability in noise-induced hearing loss
among humans makes it difficult to predict such loss. REFERENCES
One practical approach may be to screen normal-hear-
ing people with a protocol of DPOAE adaptation and AGRAMA MT, WAXMAN GM, STAGNER BB, MARTIN GK, LONSBURY –
MARTIN BL. Effects of efferent activation on distortion-product
to identify individuals with a particularly small measure otoacoustic emissions in normal humans using ipsilateral acoustic
of DPOAE adaptation (i.e., weak MOC reflex) to be stimulation. Assoc. Res. Otolaryngol. Meet. Abstr. 21:152, 1998.
at high risk of acoustical injury. Such individuals can BOYEV KP, LIBERMAN MC, BROWN MC. Effects of anesthesia on olivo-
then be warned to avoid intense sounds more restric- cochlear reflex strength in guinea pigs. Assoc. Res. Otolaryngol.
Meet. Abstr. 23:288, 2000.
tively than the population at large. Results from the
BROWN AM, HARRIS FP, BEVERIDGE HA. Two sources of acoustic
present study should be helpful in applying the above distortion products from the human cochlea. J. Acoust Soc. Am.
approach to humans by demonstrating that humans 100:3260–3267, 1996.
do indeed exhibit DPOAE adaptation and by indicat- BROWNELL WE, BADER CR, BERTRAND D, de RIBAUPIERRE Y. Evoked
ing the range of adaptation magnitudes among nor- mechanical responses of isolated cochlear outer hair cells. Science
227:194–196, 1985.
mal-hearing people. BUÑO WJ. Auditory nerve fiber activity influenced by contralateral
Another potential practical application of DPOAE ear sound stimulation. Exp. Neurol. 59:62–74, 1978.
adaptation is the development of a clinical test of the COLLET L, KEMP DT, VEUILLET E, DUCLAUX R, MOULIN A, MORGON
functional status of the MOC system. For example, A. Effect of contralateral auditory stimuli on active cochlear micro-
mechanical properties in human subjects. Hear. Res. 43:251–
patients who are suspected of having dysfunction of 262, 1990.
the brain stem (e.g., Musiek and Lamb 1994) may be DAVIS H. An active process in cochlear mechanics. Hear. Res. 9:79–
subjected to a test of DPOAE adaptation. An abnor- 90, 1983.
mally small magnitude of adaptation would suggest FOLSOM RC, OWSLEY RM. N1 action potentials in humans. Influence
of simultaneous contralateral stimulation. Acta Otolaryngol
dysfunction of the MOC system. Measurements in sub-
(Stockh.) 103:262–265, 1987.
jects known to lack MOC efferents, e.g., patients who GUINAN JJ. Physiology of olivocochlear efferents. DALLOS P, POPPER
have undergone a vestibular neurectomy, would be AN, FAY RR. The Cochlea. Springer-Verlag New York, 435–502,
helpful in establishing the expected reduction in 1996.
DPOAE adaptation. Since there is no existing clinical HEITMANN J, WALDMAN B, SCHNITZLER HU, PLINKERT PK, ZENNER
HP. Suppression of distortion product otoacoustic emissions near
test, to our knowledge, that is specifically designed to 2f1 ⫺ f2 removes DP-gram fine structure—Evidence for a second-
assess the MOC system, such a test would be useful. ary generator. J. Acoust. Soc. Am. 103:1527–1531, 1998.
One major obstacle to implementing practical KEMP DT, BROWN AM. An integrated view of cochlear mechanical
applications such as the above examples will be the nonlinearities observable from the ear canal. de Boer E Viergever
MA Mechanisms of Hearing. Martinus Nijhoff Publishers Hague
requirement of a long recording time. The present 1983, 75–82.
study employed a stimulus paradigm with a long silent KIM DO. Cochlear mechanics: implications of electrophysiological
gap (20 or 30 s) and 40 or 60 repetitions for a total and acoustical observations. Hear. Res. 2:297–317, 1980.
40 KIM ET AL.: Human DPOAE Adaptation

KIM DO. Functional roles of the inner- and outer-hair-cell subsystems of adaptation of distortion product otoacoustic emission. Assoc.
in the cochlea and brainstem. BERLIN CI. Hearing Science. Col- Res. Otolaryngol. Meet. Abstr. 22:97, 1999.
lege-Hill Press San Diego, CA 1984, 241–262. OPPENHEIM AV, SCHAFER RW. Digital Signal Processing. Prentice
KIM DO, GHOSHAL S, YE Y. Integration of ascending and descending Hall Englewood Cliffs, NJ 1975.
signals representing stimulus intensity in the marginal shell of PARHAM K, SUN XM, KIM DO. Noninvasive assessment of anditory
the anteroventral cochlear nucleus. PALMER AR et al. Psychophysi- function in mice: auditory brainstem response and distortion
cal and Physiological Advances in Hearing. Whurr Publishers Lon- product otoacoustic emission. WILLOTT JAMES F. Handbook of
don 1998, 195–203. Mouse Auditory Research: From Behavior to Molecular Biology.
KIM DO, DORN PA, NEELY ST, GORGA MP. Adaptation of distortion CRC Press New York Chap. 4 2001.
product otoacoustic emission in humans. Assoc. Res. Otolaryngol. PUEL JL, REBILLARD G. Effects of contralateral sound stimulation
Meet. Abstr. 23:283, 2000. on the distortion product 2f1 ⫺ f2: Evidence that the medial
efferent system is involved. J. Acoust. Soc. Am. 87:1630–1635,
KNIGHT RD, KEMP DT. Indications of different distortion product
1990.
otoacoustic emission mechanisms from a detailed f1, f2, area study.
PURIA S, GUINAN JJ, LIBERMAN MC. Efferent-mediated effects of
J. Acoust. Soc. Am. 107:457–473, 2000.
contralateral sound: Suppression of CAP versus ear-canal distor-
KUJAWA SG, GLATTKE TJ, FALLON M, BOBBIN RP. Contralateral sound
tion products. J. Acoust. Soc. Am. 99:500—507, 1996.
suppression distortion product otoacoustic emissions through RELKIN EM, PRIEVE BA, STERNS A. The temporal envelope of the
cholinergic mechanisms. Hear. Res. 68:97–106, 1993. 2f1 ⫺ f2 distortion product otoacoustic emission. Assoc. Res. Otola-
KUJAWA SG, LIBERMAN MC. Olivocochlear reflex effects on DPOAE ryngol. Meet. Abstr. 22:101, 1999.
amplitude in anesthetized guinea pig. Assoc. Res. Otolaryngol. SHERA CA, GUINAN JJ. Evoked otoacoustic emissions arise by two
Meet. Abstr. 21:139, 1998. fundamentally different mechanisms: A taxonomy of mammalian
KUMMER P, JANSSEN T, ARNOLD W. Suppression tuning characteristics OAEs. J. Acoust Soc. Am. 105:782–798, 1999.
of the 2f1 ⫺ f2 distortion-product otoacoustic emission in humans. SIEGEL JH, KIM DO. Efferent neural control of cochlear mechanics?
J. Acoust. Soc. Am. 98:197–210, 1995. Olivocochlear bundle stimulation affects cochlear biomechanical
LIBERMAN MC. Effects of chronic de-efferentation of auditory nerve nonlinearity. Hear. Res. 6:171–182, 1982.
response. Hear. Res. 49:209–224, 1990. STOVER LJ, NEELY ST, GORGA MP. Cochlear generation of intermodu-
LIBERMAN MC, PURIA S, GUINAN JJ. The ipsilaterally evoked olivo- lation revealed by DPOAE frequency functions in normal and
cochlear reflex causes rapid adaptation of the 2f1 ⫺ f2 distortion impaired ears. J. Acoust. Soc. Am. 106:2669–2678, 1999.
product otoacoustic emission. J. Acoust. Soc. Am. 99:3572– SUN XM, KIM DO. Adaptation of distortion product otoacoustic
3584, 1996. emission in young-adult and old CBA and C57 mice. J. Acoust.
MAISON SF, LIBERMAN MC. Predicting vulnerability to acoustic injury Sco. Am. 101:3399–3409, 1999.
with a noninvasive assay of olivocochlear reflex strength. J. Neu- TALMADGE CL, TUBIS A, LONG GR, PISKORSKI P. Modeling otoacoustic
rosci. 20:4701–4707, 2000. emission and hearing threshold fine structure. J. Acoust. Soc. Am.
MCGEE J, WALSH E, NEELY ST, JOSEPH J, KIM DO. Adaptation of 104:1517–1543, 1998.
distortion product otoacoustic emission in gerbils and mice. Assoc. TALMADGE CL, LONG GR, TUBIS A, DHAR S. Experimental confirma-
Res. Otolaryngol. Meet. Abstr. 23:283, 2000. tion of the two-source interference model for the fine structure
MOULIN A, COLLET L, DUCLAUX R. Contralateral auditory stimula- of distortion product otoacoustic emissions. J. Acoust. Soc. Am.
105:275–292, 1999.
tion alters acoustic distortion products in humans. Hear. Res.
WARR WB. Organization of olivocochlear efferent systems in mam-
65:193–210, 1993.
mals. WEBSTER DB, POPPER AN, FAY RR. The Mammalian Auditory
MOULIN A. Influence of primary frequencies ratio on distortion
Pathway: Neuroanatomy. Springer-Verlag New York 410–448,
product otoacoustic emission amplitude. II. Interrelations
1992.
between multicomponent DPOAEs, tone-burst-evoked OAEs, and
WARR WB, GUINAN JJ. Efferent innervation of the organ of Corti:
spontaneous OAFs. J. Acoust. Soc. Am. 107:1471–1486, 2000. two separate systems. Brain Res. 173:152–155, 1979.
MOUNTAIN DC. Changes in endolymphatic potential and crossed WARREN EH, LIBERMAN MC. Effects of contralateral sound on audi-
olivocochlear bundle stimulation alters cochlear mechanics. Sci- tory-nerve response. I: Contributions of cochlear efferents. Hear.
ence 210:71–72, 1980. Res. 37:89–104, 1989.
MUSIEK FE, LAMB L. Central auditory assessment: an overview. Katz WILLIAMS DM, BROWN AM. Contralateral and ipsilateral suppression
J Handbook of Clinical Audiology. Williams & Wilkins Baltimore of the 2f1 ⫺ f2 distortion product in human subjects. J. Acoust.
1994, 197–211. Soc. Am. 97:1130–1140, 1995.
NEELY ST, KIM DO. An active cochlear model showing sharp tuning ZENNER HP, ZIMMERMANN U, SCHMITT U. Reversible contraction of
and high sensitivity. Hear. Res. 9:123–130, 1983. isolated mammalian cochlear hair cells. Hear. Res. 18:127–133,
NEELY ST, JOSEPH J, KIM DO. Method for continuous measurement 1985.