Frequency Specificity of the Ipsilateral, Contralateral and Binaural Medial Efferent
Reflexes in Humans
by
Watjana Lilaonitkul
B.S., Electrical Engineering (1999)
Molecular Biology (1999)
Microbiology (1999)
Purdue University
Submitted to the Department of Electrical Engineering and Computer Science
at the
Massachusetts Institute of Technology
February 2002
@ 2002 Massachusetts Institute of Technology.
All rights reserved
Signature of Author .
D o
/t1f
I -Engineering and Computer Science
January 10, 2002
C ertified by.........
.........
........
....................
..... .. ......
John J. Guinan Jr.
Associate Professor of Health Sciences and Technology
Thesis Supervisor
A ccepted by..............................
....
.....
Arthur Smith
MASSACHUSETTS N
OF TECHNOLOGY
APR 1
Chair of Committee on Graduate Students
2
LIBRARIEQ
1 BARKER
Frequency Specificity of the
Ipsilateral, Contralateral and Binaural
Medial Efferent Reflexes in Humans
by
Watjana Lilaonitkul
Submitted to the Department of Electrical Engineering and Computer Science
On January 10, 2002 in partial fulfillment of the
requirements for the Degree of Master of Science in
Electrical Engineering
ABSTRACT
A variety of evidence indicates that the brain controls the gain of the cochlea in a
frequency specific manner through the medial olivocochlear efferent pathway but the
degree of frequency specificity in humans is poorly understood. To study the frequency
specificity of the ipsilateral, contralateral and binaural medial efferent responses in
humans, changes due to the presence of bands of noise in ear canal sound pressure of
stimulus frequency otoacoustic emissions (SFOAEs) were recorded. Changes in the
SFOAEs produced by 40 dB SPL test tones with frequencies near 1 kHz or at 2.6 kHz
were monitored with a sensitive microphone in the ear canal. In the first paradigm,
efferent activity was elicited by a _ octave band of noise at 60 dB SPL with the noise
center frequency varied relative to the test tone frequency. The results show that the
maximum efferent effect was for noise bands centered near, but not always at, the test
frequency, and that in some cases noise bands centered as far as 2.5 octaves from the test
frequency elicited significant response. Equal-response tuning curves at a test frequency
of 0.77 kHz gave Q20 values which suggest that the width of efferent tuning is greater
than that of afferents in humans. In a second paradigm, the bandwidth of the elicitor noise
was varied from 1/8 to 4 octaves, keeping the energy level constant at 60 dB SPL and its
center frequency (on a log scale) fixed at the test-tone frequency. The efferent response
generally increases with increasing bandwidth, up to bandwidths of 2-4 octaves. Overall,
the results show that the medial efferent acoustic reflexes in humans exhibit some
frequency specificity, but also that they integrate acoustic energy over a wide range of
frequencies that are much broader than has previously been thought.
Thesis Supervisor: John J. Guinan, Jr.
Title: Associate Professor in Health Science and Technology
2
Content
Page
Topics
C o ver P ag e ..........................................................................................
A b stract.........................................................................................
1
.. 2
C o nte nt...............................................................................................
B ackground ....................................................................................
- Medial Olivocochlear Anatomy
- MOC Reflex and the Cochlear Amplifier
- Effects on Otoacoustic Emissions
- Experimental Issues
- Group Delay: Middle Ear Muscle effect vs. Efferent Effect
.. 4
- Ipsilateral and Binaural Measurements
- State of Alertness and the Effects on MOC Response
Introdu ction ....................................................................................
. ... 9
M easurem ent M ethods.............................................................................11
- Subjects
- Sound Stimuli
- Artifact Rejection
- Spontaneous Otoacoustic Emissions
Analysis
M eth od s.........................................................................................
- Spontaneous Otoacoustic Emissions
- SFE Envelop Extraction
- Baseline Subtraction
- Group Delay Calculation
- Data Analysis
- Noise analysis
R esu lts............................................................................................
. . 13
. . 17
- Heterodyned waveform from ear-canal pressure measurement
- Group delay measurement
- Response Curves
- Normalized Response Curves
- Variability of efferent response within a subject
- Contralateral and binaural efferent tuning curve
- Ipsilateral, contralateral and binaural responses as a function of elicitor bandwidth
D iscu ssio n ............................................................................................
B ibliography ....................................................................................
3
38
.. 41
Background
A) Medial Olivocochlear Anatomy
The Medial Olivocochlear (MOC) efferents originate in the medial part of the
superior olivary complex and terminate in the cochlea (Rasmussen, 1946, 1960). There,
the MOC fibers synapse directly on outer hair cells (OH) and on the radial auditory nerve
fibers beneath inner hair cells (Smith, 1961). The MOC pathway is binaural - meaning
that MOC fibers can be activated by elicitor sounds presented to the same ear or the
opposite ear (see figure 1). With regards to the strength of activation by ipsilateral versus
contralateral sound simulation, studies done on anesthetized cats and guinea pigs show
that roughly 2/3 of the MOC fibers are driven best by ipsilateral sounds while the other
1/3 are driven best contralaterally (Robertson and Gummer, 1985, Liberman and Brown,
1986, Brown 1989). Curiously enough, anatomically, it is known that in most species,
MOC fibers project predominantly to the contralateral cochlea (Guinan, 1996). These
fibers receive input from the contralateral cochlea and therefore produce the ipsilateral
reflex.
HIGHER CENTERS
RIGHT COCHLEA
LEFT COCHLEA
OCB
NEURONS
CONTRA
EEI n
RELAY
NEURONS
In
BRAINSTEM
Figure B 1: Schematic illustration of the basic plan of the MOC pathways to the right ear. MOC neurons
are effectors in acoustic reflexes driven from either the ipsilateral (black lines) or contralateral (gray lines)
ears. Higher centers may act to modulate the strength of this reflex, or may activate MOC neurons directly,
without changing the acoustic reflex.
MOC fibers are myelinated (Guinan, Warr, and Norris 1983) and almost all of the
information on efferent physiology, obtained from studies done on anesthetized animals,
is attributed to the MOC efferents rather than the lateral olivocochlear (LOC) efferents.
4
This is because the LOC efferents are unmyelinated. This makes their stimulation with
extracellular currents significantly harder and almost impossible for recording to be done
with high impedance pipet electrodes (Guinan, Warr, and Norris 1983).
From anesthetized animal studies, which combined electrophysiology and tracer
labeling, MOC efferents have been shown in to have tonotopic maps corresponding at
least roughly to those of afferents (Robertson and Gummer, 1985; Liberman and Brown,
1986; Brown, 1989). Hence, it seems probable that efferents tuned to some particular
frequency can receive signals from afferents tuned to similar frequencies - making the
feedback efferent response frequency specific.
B) MOC reflex and the cochlear amplifier
Efferent stimulation has been shown to depress basilar membrane motion at low sound
levels in guinea pigs with maximum depression around the center frequency of up to 2022 dB in the region of 18-20 Hz (Dolan and Nuttall, 1994). In response to sound the
basilar membrane motion causes the reticular lamina to shear radially relative to the
tectotial membrane and bends the OHC stereocilia in the process. Ion channels open from
this bending which results in an ionic flux and a change in membrane potential. OHCs are
motile and can shorten and elongate as a result of a change in its membrane potential
(Santos-Sacchi and Dilger, 1988). The OHC motility is somehow coupled back into the
basilar membrane motion such that an in-phase coupling would amplify the forward
traveling wave and vice versa. This process of OHC-induced amplification and
suppression of the basilar membrane motion is also known as the 'cochlear amplifier' and
is illustrated in figure B2 (adapted from Guinan, 1996). MOCs richly innervate and are
thought to change the properties of OHCs. The exact mechanism of this efferent control
is unknown.
C) Effects on Otoacoustic Emissions
Otoacoustic emissions (OAEs) are sounds that are produced within the cochlea,
spontaneously or as a result of some stimulation, that can be recorded in the ear canal
(Kemp, 1978). The phenomenon of OAEs is attractive for the studies of efferent
responses because they can be measured non-invasively. An efferent-induced depression
of the forward traveling wave amplitude along the basilar membrane results in a
reduction in the energy reflected back in the backward traveling wave (Guinan, 1996).
Hence, most of the existing literature on humans quantifies the amount of efferent
activation as the suppression of some kind of evoked otoacoustic emission in the
presence of an efferent elicitor.
5
Total Basilar Membrane Motion
/
A
No Medial Efferent Activity
Motion due to cochlear
amplifier incircled region
(amplitude exaggerate
--
-+
-- +
.
B
Medial Efferents Active
(OHCs tumed down)
Total Basilar Membrane Motion
C
Motion due to cochlear
amplifier in circled region
(amplitude exaggerated)
-.
0- .- O......
D
Figure B2. (Adapted from Guinan, 1996) Schematic illustrating the presumed action of the cochlear
amplifier at one place along the basilar membrane, and medial efferent inhibition of the cochlear amplifier.
Trace A is a snapshot of normal basilar membrane motion in response to a tone. The cochlear amplifier at
the circled place moves the basilar membrane at that place and creates waves that travel away in both
directions (Trace B). The forward traveling wave (TraeB ->) adds to the sound-driven wave and amplifies
it. Note that the amplitude scale is exaggerated in traces B and D relative to the traces A and C. For traces
C and D, medial efferents are activated and reduce the gain of the cochlear amplifier so that the motion
created by it is smaller (the amplitude is smaller in trace D than in trace B). Since there is less amplification
at many places along the basilar membrane, the resulting traveling wave is less ( the amplitude is smaller in
tace C than in trace A).
D) Experimental Issues:
Stimulus Frequency Otoacoustic Emissions (SFEs) are used to assay the frequency
resolution of the MOC reflexes applied to the contralateral ear, ipsilateral ear or both.
This type of OAE corresponds to the generation of additional acoustic energy from the
cochlea at the frequency of a low-level, constant tonal stimulus (Kemp and Chum, 1980a;
Probst, et al. 1990).Quantitatively, at low to moderate sound levels, SFEs can be
explained as arising from coherent scattering of the cochlear traveling waves off small
irregularities or random perturbations in the cochlear mechanics (Zweig et al. 1995). Its
magnitude and phase can be recorded non-invasively from the ear canal (Kemp and
Chum, 1980a). One advantage of using SFE to monitor efferent activation is that the test
tone itself elicits little or no efferent activity (Maison et al. 1999). Some existing
literature uses the suppression of the 2fl-f2 distortion product otoacoustic emission
(DPOAEs) as a measure of efferent activity. An advantage of using SFEs over DPOAEs
is that less energy in the SFE tonal stimulus is needed as compared to the primary levels
needed to produce a distortion product of the same magnitude. Interpretation of efferent
6
effects on SFEs is also than with DPOAEs because the effect of efferent-induced change
at the primaries and the distortion product is not well understood.
D-1) Group Delay: Middle Ear Muscle effect vs. MOC effect
In studying MOC efferents using acoustic reflex assays, it is important to be sure
that the response recorded was not a result of stapedius muscle contraction due to loud
sounds. The effects of MOC efferent versus the middle ear muscles (MEMs) can be
distinguished by their group delays. The MOC-induced effect is observed as a change in
the SFE, which is generated within the cochlea. Hence, it has a long delay - on the order
of many ms (~10 ms has been recorded for a 1kHz test tone) due to the time needed for
energy to propagate along the cochlea partition to the CF place as well as the delay
associated with the action of the cochlear amplifier (Yates, 1995, Patuzzi, 1996). MEM
contractions change the impedance of the middle ear and the travel time for sound to
reach the middle ear and return to the ear canal is significantly shorter than the travel time
into the cochlea and back (Backus et al. 1999).
So how can we measure this time delay associated with responses that originated
deep within the cochlea? In a linear dispersive medium, where the wave velocity is a
function of frequency, group delays (-d$/df) is the appropriate measure of energy
transport (Elmore and Heald, 1969). $ is the phase delay in cycles and f is frequency in
cycles per second. The group delay is the time delay that would give the same rate of
change of phase with increase in stimulus frequency.
In order for any amplitude-modulated wave to propagate through a medium
unchanged, the group velocity must be independent of frequency (Elmore and Herald,
1969), which we know is not true in the highly nonlinear cochlea. To deal with this, we
need to make sure that df is small enough so that the group velocity within that region
varies slowly enough to be assumed constant. Then the measure of the group delay is
useful to us since a fixed time delay between the stapes vibration and the basilar
membrane vibration at a particular cochlear location would produce a linear relationship
between phase and frequency (Patuzzi, 1996).
D-2) Ipsilateral and Binaural measurements
In the case of a contralateral elicitor, the resultant trace of the change in SFE vs.
time directly reveals the time course of the MOC effects. The situation is not as simple in
the case of the ipsilateral or binaural reflex. This is because the elicitor could have as one
of its frequency components, the test tone frequency which could directly interfere with
the measurements. However, this problem could be solved by reversing the polarity of
the noise on alternate bursts within the run - which upon pair-wise averaging, most of the
interference effect would cancel out (Backus et al, 1999)
Another problem is that the elicitor can suppress the test tone via the two-tone
suppression mechanism, which arises as a consequence of the cochlea's nonlinearity
(Probst et al., 1990). We need to make sure that we do not mistake this measure as MOC
7
efferent response. The way to circumvent this problem is by exploiting the fact that
suppression effects disappear almost instantaneously after the offset of suppression but
the MOC-induced change in SFE decays on a much slower time scale. Hence, an elicitor
consisting of long noise bursts with a 50 ms wait period after the offset and an analysis
window restricted to the period 50 - 150 ms after the elicitor is turned off can be used for
the measured MOC-induced effects.
D-3) State of alertness and the effects on MOC response
Another important issue to keep in mind is that selective attention is known to modify
the active micromechanical properties of the cochlea, presumably by modulating efferent
activity (Puel et al., 1988). There is also evidence for a reduction in OAEs when the
subject is asleep (Morlet et al., 1994) as well as when the subject is asked to perform a
competing visual task during the recordings (Puel et al., 1988). To help control alertness,
or at least prevent sleep, we allow subjects to take frequent breaks during intervals
between experimental runs. Another possibility, which we have yet to implement, is to
design some task for the subjects to perform in between the recordings so as to keep the
subject alert and thereby maximize the efferent effect on SFE measurements.
8
Introduction
The Medial Olivocochlear (MOC) efferent fibers provide feedback control of the
cochlear operation. They can be excited by both ipsilateral and contralateral sound
stimuli, thus forming a descending binaural-reflex pathway (Fex 1962). This efferent
bundle originates in the medial part of the superior olivary complex and synapses directly
on the numerous outer hair cells (OHCs) which are embedded in the organ of Corti (OC).
In the early 1980's, Brownell et al. reported motile behavior of OHCs when they were
stimulated in vitro with extracellular current (Brownell et al. 1985). In vivo, it is believed
that OHCs form the active and motile elements that provide the mechanical gain for the
nonlinear and highly tuned basilar membrane (BM) motion (Davis 1983, Dallos and
Evans 1995). This phenomenon is also known as the cochlear amplifier and can increase
the vibration of the BM and OC 1000-fold or more, allowing the cochlear to detect
fluctuations in acoustic sound pressure as low as 1/2,000,000,000 of normal atmospheric
pressure (20 gPa rms), (Fay 1988). So by altering the state of the OHCs, the MOC fibers
provide direct control of the sensitivity and frequency specificity of the cochlear
amplifier.
From studies done on anesthetized animals, anatomical and physiological findings
of the MOC system appear to support the idea that it can be activated in a highly
frequency-selective manner. Single fiber studies in cats showed that each MOC fiber
branches within the cochlea to innervate OHCs up to a span of one octave in cochlear
frequency and that the MOC efferents respond to ipsilateral and contralateral sound with
sharp tuning (Liberman and Brown, 1986). Intracellular labeling done on cats and guinea
pigs also revealed that there is at least a rough match between the afferent and efferent
tonotopic maps (Liberman and Brown, 1986, Brown, 1989).
In contrast to the substantial amount of published animal studies, significantly less
has been established in humans. Basic anatomical information like the spread of single
MOC neurons' peripheral projections in humans cannot be obtained directly with existing
techniques used in animals. However, a more indirect way, which is non-invasive, is to
employ acoustic reflex assays to learn more of the extent of the efferent control along the
frequency axis.
Over the years, various non-invasive techniques have been developed that allow
for a way to indirectly infer aspects of the frequency specificity of the medial efferent
reflexes in humans. In 1978, Kemp reported a phenomenon where sounds are emitted by
the cochlea, either spontaneously or in response to an acoustic stimulation. These are now
known as Otoacoustic emissions and they can be recorded in the ear canal. Later on, it
was reported that the electrical stimulation of the MOC efferents altered the cochlear
mechanics in some way, producing a change in the distortion tones measured acoustically
from the ear canal (Mountain, 1980). In accordance with current thoughts that the OHCs
have an active role in generating OAEs (Mountain and Hubbard 1989), control of
cochlear responsiveness via the MOC efferents can be inferred non-invasively through
the measurements of OAEs.
In past human studies, the magnitude of OAE suppression was usually expressed
as the dB change in emission magnitude produced by the presence of an elicitor of
efferent activity, usually with the elicitor delivered to the contralateral ear. The effects
9
reported in existing literature are usually on the order of 1-3 dB (e.g. Collet et al., 1990 a,
b; Veuillet et al., 1991, Maison et al. 2000).
Among the very few human data on the frequency specificity of the efferentinduced effect, only two show plots of the frequency specificity and it is only for the
contralateral reflex extending from 1-2 kHz (Ryan et al. 1991, Chery-Croze et al., 1993,
Norman and Thorton 1993, Maison et al. 2000). There are no published data on
ipsilateral nor binaural reflex frequency selectivity in humans or animals.
In this present study our aim is to determine the frequency resolution of the MOC
reflex in humans. We record the changes in SFEs at a fixed frequency due to a
contralateral, ipsilateral or binaural narrow-band noise elicitor at a range of frequencies.
The scope of this study shall cover a range of frequency regions from 0.77 to 2.6 kHz. To
gain some insight on the span of efferent control along the frequency axis, two paradigms
were used:
1) For a fixed test tone, the center frequency of a Narrowband Noise (NBN) MOCelicitor was systematically varied. The total energy within the noise band was kept
constant. The effects from ipsilateral, contralateral or binaural elicitors were
compared. At each test tone frequency, MOC tuning functions with evoking bursts at
several sound levels (that do not evoke MEM effects) were studied.
2) We accessed the limit of bandwidth for the MOC eliciting activity at the cochlear
region of interest by varying the bandwidth of the NBN elicitor stimuli, whose center
frequency is placed at the test tone frequency. The range of bandwidths will be from
1/16 octave to 4 octaves.
With so little known, a better understanding of human MOC functional capacity in terms
of its frequency specificity is necessary and would help bring us one small step closer to
understanding the MOC functional significance.
10
Measurement Methods
Subjects
In each of the four subjects, measurements were done from both ears. All subjects
had no history of auditory pathology and had hearing levels no higher than 20 dB SPL
between 500 Hz and 4000 Hz at octave intervals on a 1/3 octave narrowband noise
audiogram. On a test using broadband noise, hearing levels were also less than 20 dB
SPL. The subjects were seated in a comfortable reclining chair in a soundproofed room
isolated from vibrations (Ver et al. 1975).
Sound Stimuli
Stimulus waveforms were generated and the responses were recorded and
averaged digitally using a custom-built data-acquisition system implemented in
LabVIEW6 and Matlab. The sampling frequency was typically set to 20kHz. Acoustic
signals were transduced using 2 Etymotic Research ERI Oc acoustic assemblies, one in
each ear and each assembly consisting of 1 ER10b low-noise microphone with ER-10c
foam tips to keep the probe in place within the ear canal. The sound stimuli were mixed
acoustically using 2 separate sound sources to prevent intermodulation distortion in the
electrical to acoustic transduction of the two signals. In-ear calibrations were done at
regular intervals during all measurement sessions to ensure that the stimulus tones and
the components of the noise band elicitors delivered had constant level and constant
starting phase in the ear canal at all frequencies.
To induce SFEs, a pure tone stimulus was presented continuously to both ears at
40 dB SPL. We used either narrow or broadband noise to elicit efferent response in the
ipsilateral ear, contralateral ear or in both ears. The spectral components of the noise band
had equal level (i.e. were 'flat' across frequencies) and random starting phases which
were uniformly distributed between -n and n. The noise level was set to 55 or 60 dB SPL
during the experiment and the phase polarity of its spectral components was flipped in
alternate runs so that noise elicitor will cancel in the averaging of measurements.
In the time domain, stimulus presentation is done as follow:
1) The 'Baseline window' begins at 0 ms and ends at 500ms. During this time, only the
continuous tone that produces the SFE, without the elicitor was presented.
2) The elicitor noise burst began at 500 ms and lasted for 2500ms. The waveform of
interest was recorded in a 100 ms window placed 50 ms after from the falling edge
of the noise-burst. This will be referred to as the 'Response window'.
3) Finally, a rest period of 1500 ms intervened between the elicitor and the next
stimulus baseline.
11
Test tone
only
500 ms
Test tone + Noise Elicitor
Test tone
only
1500 ms
2500 ms
Baseline Window
Response Window
Figure MI. Schematic of the stimulus presentation
Artifact rejection
Real-time artifact rejection was implemented in the time domain by taking the
difference between successive data buffers to ensure that the each of the following values
falls below a set criterion.
1) the maximum absolute difference between the 2 buffers
2) the DC component or the average value in the 'Baseline window'
3) the average rms. value from the mean in both the 'Baseline window' and the
'Response window'.
Spontaneous Otoacoustic Emissions
The presence of spontaneous otoacoustic emissions (SOAEs) in each subject was
determined by recording the ear-canal pressure with the absence of any stimulus tone or
elicitor. If any SOAEs were detected, the stimulus tone for that particular subject was
chosen to be at least 50 Hz away from adjacent SOAEs to prevent entrainment - a
phenomenon where the emissions are synchronized by external tones (Long et al. 1988,
Long and Tubis, 1988a, Long et al. 1990, Tubis et al., 1989, van Dijk and Wit, 1988).
12
Analysis Methods
Spontaneous Otoacoustic Emission
The recorded ear-canal signal, without the presence of any sound stimuli, was
processed by taking its Fourier transform (FFT). To reduce the amplitude of the random
peaks in the spectra due to random noise, the FFTs were averaged in frequency (Probst,
1990).
SFE Envelop Extraction
We performed a fast Fourier transform (FFT) on the raw pressure waveform. A
band of positive spectral components surrounding the stimulus test tone frequency was
selected. These frequency components were then shifted down so as to center the
stimulus frequency at zero Hz. A 1O* order recursive exponential filtering window was
then used to remove higher frequency components of noise arising from the equipment or
the subject. This filter was defined by Shera and Zweig (1993a) as,
T
CUt
is the cutoff frequency and the function ]F(T) is defined recursively as:
1+
(f)
eF
(r)-1,
with F1 (r) = e.
The window S, (f; fu, ) has a maximum value of 1 at f=0.
The scale factor An is chosen such that the window falls to the value of 1/e at f =f
An= FY , where y,
1
= ln('Y +1)
with y=1.
Sn (f; fu, ) has a much sharper cutoff than standard window functions (e.g., the
Hamming, Blackman, etc.) and does not have any side-lobes. On the other hand, since
this function has no poles, it produces considerably less 'ringing' than the boxcar
window, which provides an infinitely sharp cutoff.
The larger the number of iterations, n, the sharper the roll-off. In this present
study, we chose n = 10 and fu, = 90 Hz. The values were chosen during preliminary
studies done on a small cavity and these parameters appeared to give the best
compromise between reduction of the background noise spectrum and edge-smearing
effects in the time domain, suitable to the time scale of the effect we are studying.
The result from this spectral component extraction is a complex-valued signal,
which represents the magnitude and phase of the SFE envelope with time. No phase
distortion is added. The process also automatically downsamples the waveform buffer.
13
Baseline Subtraction
To measure the change in SFE produced by MOC activity, we can calculate the
vector difference between the ear canal sound pressure produced by the continuous test
tone with or without the presence of the MOC activity evoked by another sound stimulus
(Guinan, 1986). Once the processed wave is transformed back into the time-domain, the
complex mean of the signal in the 'Baseline window' is subtracted from the whole
waveform. The complex average in the response window now contains the magnitude
and phase of the sound pressure due to the MOC-induced change in SFE, PASFE.
PASFE
Sound Pressure
(Imaginary Part)
PRESPONSE
Z<ASFE
1.
4.
2.
3.
Sound Pressure
(Real Part)
Figure M2. Vector diagram illustrating method of baseline subtraction
Labels:
1: PSFE with elicitor
2 : PBASELINE
3 : P TEST TONE
4 : PSFE without elicitor
It can be seen from the diagram that,
PASFE
PSFE w/ elicitor - PSFE w/o
elicitor = PRESPONSE - PBASELINE
$ASFE, is the leading phase referenced to the baseline. This makes it independent from
phase gitters in the equipment.
Group Delay Calculations
The group delay test can be applied, using BBN as the elicitor, at one frequency
to determine the highest sound level that evokes efferent effects but without evoking
significant MEM effects. Several studies have shown that the stepedius muscles are
excited significantly more with BBN than NBN, with the energy level kept constant
14
(Margolis et al., 1980). Based on this, the maximum BBN level will also be used for the
NBN.
Data Analysis
Two analysis time windows are of interest to us. The first window is from 30503150 ms from the start. This window basically starts 50 ms after the elicitor is turned off
and is used to analyze the ipsilateral, contralateral and binaural response. The second
window is from 2450-2950 ms - during which the elicitor is on. This 500 ms window is
used to analyze the contralateral response - for analysis that don't compare to ipsilateral
or binaural responses.
After the raw wave is heterodyned and the complex baseline average value is
subtracted from it, the processed waveforms are synchronously averaged point by point
with corresponding processed waveforms from different sessions. Points in the analysis
window of the final averaged heterodyned waveform are then averaged to give a complex
number which contains the magnitude and phase information of the response.
Noise Analysis
Due to lack of data to sufficiently characterize the statistics of the signal, standard
statistical detection and estimation techniques could not be employed. So for the
following set of results and subsequent discussion, signals that are at least 3 standard
deviations of the equipment noise above the mean of the noise floor are considered
significant.
The statistics of the noise were calculated from an estimate probability mass function
where the sample space consisted of JAPsFEJ obtained from non-overlapping time windows
from all runs without an elicitor within the series. In order to pool the noise data across
experiment sessions, we made the assumption that equipment noise was the main noise
contributor at frequency ranges above 400 Hz and that its statistical characteristics are
independent of the subject's physiology. Below 400 Hz, physiological noise, which
varies considerably from subject to subject, becomes important. The statistics of the noise
was remarkably consistent across subjects which seems to be in agreement with this
assumption. Preliminary noise data also seems to suggest that the phase is approximately
distributed uniformly between -180 and 180 degrees. Hence, phase information was
ignored when estimating the statistics of the noise.
The number of points used to estimate the probability mass function was 242 points
when the analysis window for the signal was placed before the elicitor was turned off and
220 points when placed after the elicitor was turned off. To calculate IAPSFE noise values
within a run, the start of the baseline window was shifted in time steps equal to the
analysis window length while keeping the duration between the baseline window and the
analysis window constant. The duration between the baseline and analysis window was
kept constant to avoid time dependent 'drifts' as seen in a random walk model for noise.
15
From a right-hand probability distribution function of the noise, the probability of the
noise taking on a value 3 standard deviations above the mean is approximately 0.5% in
all subjects.
16
Results
Heterodyned waveform from ear-canal pressure measurement
Figure 1 shows a typical time course of the magnitude and phase of the change in
ear canal pressure due to a contralateral BBN elicitor. After the onset of the BBN elicitor
at 500 ms, the magnitude of the pressure change, IAPSFEI, rose gradually over hundreds of
milliseconds before reaching saturation. After the BBN elicitor was turned off at 3000
Ms, JAPSFEI decays over several hundreds of milliseconds before falling into the noise
floor. When IAPSFE emerged substantially above the noise floor, the phase lag _psFE, was
relatively constant.
20
-20-
-0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0.5
..
1
1.5
2
2.5
3
3.5
-..-
4
4.5
5
400
0
200-
-
5
-
time (s)
Figure 1. Magnitude and phase of efferent-induced change in sound canal pressure, APSFE, as a function of
time. Test tone used was 1.03 kHz at 40 dB SPL. The contralateral elicitor was BBN at 60 dB SPL.
A different time course is observed when an ipsilateral or binaural BBN elicitor is
used. Figure 2 shows the magnitude and phase of the change in ear canal pressure due to
a BBN elicitor in both ears. At the onset of the elicitor at 500 ins, we observe a sharp rise
in |APSFEj over a time period on the order of 10 ins. This rapid rise is primarily a
consequence of the ipsilateral BBN elicitor having frequency components within the
vicinity of the test tone, which results in two-tone suppression. Hence, when consecutive
measurements from runs with elicitors of flipped polarity are averaged, we get an good
17
cancellation of the BBN elicitor but suppression dominates the response. This manifests
as the change in SFE seen at the elicitor onset. Efferent-induced SFE change occurs
simultaneously but at a much slower rate, and in this case, on a smaller magnitude scale
than the change induced by suppression. At 3000 ms, when the elicitor is turned off, we
see a rapid decrease in IAPSFEJ over tens of ms followed by a slower decrease over
hundreds of ms. This rapid decrease corresponds to the decay of the two-tone suppression
effect, after which, the slower decay of the efferent effect dominates. The _APSFEagain
seems relatively constant when the IAPSFEI is above the noise floor. It is interesting to
note, however, that the value of -APSFE changes when the efferent-induced APSFE
dominates in magnitude. This trend of phase change is seen in some subjects and not
others, with no obvious dependence on the elicitor laterality or test tone frequency.
a 1
'I)
-20-
0
0.5
1
15
2
25
3
35
4
45
5
0.5
1
1.5
2
2.5
time (s)
3
3.5
4
4.5
5
400
300
200
0
Figure 2. Magnitude and phase of efferent-induced change in sound canal pressure, APSFE, as a funCtion of
timne. Test tone used was 1.03 kHz at 40 dB SPL. The contralateral elicitor was a BBN at 60 dB SPL. 12
runs were averaged. (Subject 68)
Group delay measurement
It is important to check that the change in SFE observed is due to efferent effects
within the cochlea and not MEM contraction in response to loud sounds. This can be
done by measuring the group delay of the response. Figure 3 shows the magnitude and
phase of the change in ear canal pressure due to a BBN elicitor in the ipsilateral ear,
contralateral car, and both ears as a function of test tone frequency. The noise floor is
18
derived from measurements with no elicitor within the same run. The error bars mark the
level equivalent to 3 standard deviations above the mean of the noise floor.
In the example shown in figure 3, using a least square error linear fit on the phase
plot, we find that the group delay is approximately 15 ms for all 3 elicitors which agree
with existing literature on human SFE group delay for the test tone frequencies within the
range used (Dreisbach et al. 1998).
19
20
15
-
Figure 3. JAPSFEI (dB SPL) and _APSFE
(degrees) vs. test tone frequency (kHz).
Ipsilateral (d )contralateral (o) and binaural
(inverted triangle) elicitors were BBN at 60 dB
SPL. The test tone level was 40 dB SPL.
(.) represents runs without any elicitor which
gives an estimate of the noise floor. The error
bars correspond to the 3 std above the mean
noise floor.
10
-5
...1
0.8
0.78
0.76
test frequency (kHz)
CR
e
s
0
0.76
0.78
0 .8
test frequency (kHz)
Response Curves
Figure 4 summarizes the ipsilateral, contralateral and binaural response curves at
test tone frequencies 0.77 kHz and 1.03 kHz. The elicitor used was a 0.5 octave NBN
with a constant energy level but different center frequencies. Data points that are at least
3 standard deviations above the mean noise floor were considered significant. The signal
analysis window used was 100 ms in duration and started 50 ms after the elicitor was
turned off.
In the set of data where the test tone was 0.77 kHz, the binaural and contralateral
response curves in the left ear appears to be asymmetric with the maximum IAPSFE
occurring at elicitor center frequencies below the test tone. The ipsilateral response was
too weak under the chosen criterion to make any meaningful observations. The binaural
stimulation appeared to elicit changes in SFE even when its center frequency was 2
octaves below or 3 octaves above the test tone. The range was less with the contralateral
stimulation, where changes in SFE were observed with elicitor center frequencies 0.5
octave above and 2 octaves below the test tone. The phase of the binaural and
contralateral responses appeared relatively constant and remarkably similar in value for a
given elicitor center frequency that induced significant efferent response.
In the right ear, the binaural and ipsilateral responses were also more effective
towards the lower frequency end. The most effective elicitor in both cases was when the
center frequency was at the test tone. Changes in SFE were observed with elicitor center
frequencies 0.5 octave above and 2 octaves below the test tone. In this range of elicitor
center frequency, it is interesting to note that the phase change of the binaural and
20
ipsilateral response appears to have a negative slope instead of being approximately
constant.
In the data set where the test tone was 1.03 kHz, more effective responses were
achieved with noise bands centered above the test frequency in the right ear. This trend is
less pronounced in the left ear. There appears to be subtle peaks and valleys in the shape
of the response curves although more data is needed to determine whether or not they are
significant. Significant binaural response in the left ear was observed with noise bands
centered 2 octaves below and 2.5 octaves above the test frequency. The right ear binaural
response was induced with noise bands centered 1 octave below and 2.5 octaves above
the test frequency.
At both 0.77 and 1.03 kHz test tone, binaural stimulation was almost always more
effective in eliciting efferent activity for the elicitor center frequencies used. Noise bands
centered near the test tone also appeared more effective than those further away.
Figure 5, shows only contralateral response curves at the same test frequencies as
figure 4 but with the analysis window between 2450 and 2950 ms. Recall that the elicitor
is gated on from 500 to 3000 ms, and so no ipsilateral or binaural efferent response curve
could be reported using this analysis window since it is impossible to separate out the
changes in SFE due to suppression.
With the 0.77 kHz test tone, efferent-induced changes in SFE were seen at all
elicitor center frequencies used, which ranged from 2 octaves below to 2.5 octaves above
the test frequency. The tuning again appears asymmetric about the test frequency with the
maximum response occurring with the elicitor centered below the test frequency. In the
right ear, response was seen from 2 octaves below to 2.5 octaves above the test
frequency. However, the peak response is when the noise band is centered around the test
tone. More pronounced peak and valleys, with lAPSFE difference of approximately 10
dB SPL, are also seen at different elicitor center frequencies.
With the 1.03 kHz test tone, efferent-induced change in SFE was seen when the
elicitor was centered at frequencies above the test frequency in both the left and the right
ear - up to 2.5 octaves above. Noise bands centered at 1 to 1.5 octaves below the test tone
induced almost no SFE change. The response was greatest when the noise band is
centered above the test tone. More pronounced peak and valleys are also seen at different
elicitor center frequencies in the left ear.
Figure 6 shows the ipsilateral, contralateral and binaural response curves obtained
from both the left and right ear of Subject 61 with the 100 ms analysis window starting at
50 ms post elicitor-off. A 2.6 kHz test tone with 0.5 octave NBN elicitor of various center
frequencies was used. The left ear response was significantly weaker than that of the right
ear. The efferent response observed in both ears appeared to be stronger when the noise
bands used were centered above the test tone frequency. In the right ear, binaural and
contralateral responses were observed with noise bands centered at frequencies up to 1.5
octaves above the test tone and roughly 1 octave below. The maximum APSFE occurred
with the binaural noise band centered at 1.5 octaves above the test tone. The ipsilateral
response was weaker and only showed response values above the 3std criteria at 0.5
octave above and below the test frequency. The maximum JAPSFE in the ipsilateral and
contralateral curve was when the elicitor was centered at the test frequency. The phase
change appears relatively constant when a significant magnitude of efferent response was
observed.
21
10
10
0
0
:a
CL
-o
00
7,
-10
-10
-20
-20
1
CTT=0.77kHz
10
C
ca
400
300,
0
100
TT=0.77kHz
TT=0.77kHz
-~
0
300
0:
V
7
0
200
200
6-'
/
V
_0
100
1001
TT=0.77kHz
0
100
1i
0
10
10
0
0U)
-10
-10
"D
-20
-20
TT=1.03kHz
-30
-30
100
400
400
U)
a)
a)
0)
a)
~0
300
300
a)
Cu
-c
2001
200
aa-
Cu
TT=1.03kHz
100(
100
13
100
TT=1.03kHz
0
0
100
Center Frequency of 0.5 octave NBN
100
Center Frequency of 0.5 octave NBN
Figure 4. Response curves at test tone frequencies 0.77 kHz (Subject 61) and 1.03 kHz (Subject 68). The
magnitude and phase of the change in SFE is plotted as a function of the center frequency of the 0.5 octave
NBN elicitor (kHz). The graphs on the left and right correspond to results from the left and right ear
respectively. The test tone level and elicitor level were at 40 and 60 dB SPL respectively. 8 averages were
used to obtain the data at 0.77 kHz test tone. 12 averages were used to obtain the data at 1.03 kHz test tone.
The analysis window was of length 100 ms taken 50 ms after the elicitor was turned off. The vertical bar in
each graph marks he test tone frequency used.
-0- Elicitor in Left Ear
-A- Elicitor in Both Ears (change symbol to inverted triangle)
Elicitor in Right Ear
No Elicitor
The error bars are 3 std above mean.
22
10
10
TT=0.77kHz
TT=0.77kH
0
CO
-10
-20
(V
-20
10 0
10
400
400
.-1
0
(D
300
-20 I
-0200
a-200
0
200(
CL
c
100
_0
T T=0.77kH,
TT=0.77kH2
0
0
10 0
10
5
400
5
0
0
-- 5
-5
TT=1.03kH z
ci
-10
CL
CO
-15
20
-20
100
10
TT=1.03kHz
-
400
400
U300
300
~200
200
C
CL
a-1 00
0--
-
TT=1.03k
0
-
100
z
TT=1 .3kH;
0
0
10
Center Frequency of 0.5 octave NBN
10 0
Center Frequency of 0.5 octave NBN
Figure 5. Response curves at test tone frequencies 0.77 kHz (Subject 61) and 1.03 kHz (Subject 68). The
magnitude and phase of the change in SFE is plotted as a function of the center frequency of the 0.5 octave
NBN elicitor (kHz). The analysis window was of length 100 ms. taken 50 ms before the elicitor was turned
off. The graphs on the left and right correspond to results from the left and right ear respectively. The test
tone level and elicitor level were at 40 and 60 dB SPL respectively. 8 averages were used to obtain the data
at 0.77 kHz test tone. 12 averages were used to obtain the data at 1.03 kHz test tone.
Elicitor in Left Ear
-0Elicitor in Right Ear
No Elicitor
The error bars are 3 standard deviations above the mean noise floor.
-*-
23
Figure 7 shows only the contralateral response curves with the same experimental
parameters as that described in figure 6. However, the analysis window used is from
2450-2950 ms, during which time the elicitor is gated on. The response from the left ear
is still too weak to make any credible observation. On the right ear, the contralateral
response from 2 octaves down to 1.5 octaves above the test tone appears significant.
However, it does not have a bandpass shape within the elicitor center frequency range
used. In fact, it appears to correspond more to a highpass band. The phase change appears
relatively constant when the magnitude of efferent response was significant.
Figure 8 summarizes the ipsilateral, contralateral and binaural response for a 1.02
kHz test tone in Subject 67 and a 1.2 kHz test tone in Subject 73. We used a 100 ms
analysis window starting at 50 ms after the elicitor was off. The left ear in both subjects
gave very weak efferent response for the set criteria. In the case where the test tone was
at 1.01 kHz, the binaural and contralateral response from the right ear was only seen as
far as 1-1.5 octaves below the test tone. No significant response was seen above the test
tone. At 1.2 kHz test tone in another subject, binaural efferent response was observed as
far as 1 octave above and below the test frequency
Figure 9 summarizes the contralateral response curves from the left and right ear
of Subject 67 and 73 at 1.01 kHz test tone and 1.2 kHz test tone respectively. The
analysis window used is the same as that used to obtain figure 7. In both subjects, the left
ear contralateral response is also very weak and will not be discussed. In the data set
where the test tone was at 1.01 kHz, SFE change was observed with elicitors centered as
far as 1.5 octaves below and 2 octaves above the test frequency. There also appears to be
2 peaks in the response, one between 0.5 and 1.5 octaves below the test frequency and
one between 0.5 and 2 octaves above the test tone. The phase change of the response
appears relatively constant.
With the test frequency of 1.2 kHz, significant response was observed as far as 2
octaves below and 1 octave above the test frequency. The maximum response occurred
within the vicinity of 1.5 octaves below the test tone. Another smaller peak was seen
around 1 octave above the test tone. The phase seemed roughly constant where efferent
response was observed.
The validity of the peaks and valleys seen in all figures would need more data and
statistical analysis given the variability of the efferent within the same subject with the
same experimental parameters. This issue of intra-subject variability shall be discussed in
subsequent sections.
24
0
0
-10
a
U) -O
-20
TI=2.6kHz
-30
10 0
10
400
4( 0,
TT=2.6kHz
TT=2.6kHz
300
CO,
200
a.) 1300,
1001
10F
o -
Center Frequency of
0.5 octave
0
10
Center Frequency of 0.5 octave NBN
NBN
Figure 6. Ipsilateral, contralateral and binaural response curve at 2.6 kHz (Subject 61) test tone from the
elicitor-off analysis window. All other parameters and symbols are the same as Figure 4.
0
0
-2
TT=2.6kHz
co-I 0
CO
-10
-0
_!g-20
-20
00
100
10"0
400
TT=2.6kHz
TT=2.6kHz
00
200
0
CO)
200
0
C
100
u
5
Center Frequency of 0.5 octave NBN
Center Frequency of 0.5 octave NBN
Figure 7. Contralateral response curve at 2.6 kHz (Subject 61) test tone from the elicitor-on analysis
window. All other parameters and symbols are the same as Figure 5.
25
B. Subject 67: Right Ear
A. Subject 67: Left Ear
0
C
-10
a -1
0
7
-20
-i
-7
TT=1.SkHz
_-3
-30
100
TT=1.01kHz
D. Subject 67: Right Ear
10
C. Subject 67: Left Ear
40C
-
400
TT=1 .01kHz
TT=1.OlkHz
(D
300
m, 300(D
200
V
200
VV
7
7
a100 F
CZ
100
r
0
0
100
10
E. Subject 73: Left Ear
F. Subject 73: Right Ear
0
0
-.
-10
-10
-20
-20
2. -30
Ca
-30
0-
TT=1.2kHz
100
H. Subject 73: Right Ear
100
G. Subject 73: Left Ear
(D
V 400
400
TT=1.2kHz
300
a) 3M
200
200
C
P-
_.._
100
a- 100
(D
TT=1.2kHz
r
0
0
0
10
100
Center Frequency of 0.5 octave NBN
Center Frequency of 0.5 octave NBN
Figure 8. Ipsilateral, contralateral and binaural response curve at 1.01 kHz (Subject 61) and 1.2 kHz
(Subject 73) test tone from the elicitor-off analysis window. All other parameters and symbols are the
same as Figure 4.
26
B. Subject 67: Right Ear
A. Subject 67: Left Ear
-5
-5
-10
-0
-15
U
-20
0
TT=1.O1kHz
TT=1.01kH
-25
-25
ca
100
D. Subject 67: Right Ear
C. Subject 67: Left Ear
400
400
TT=1.01kH;
TT=1.01kHz
300
&300
200
(n20
)
100
(D
0
0
0
10"
E. Subject 73 Left Ear
10
F. Subject 73: Right Ear
0
0
-10
-10
U,
CO
-20
TT=1 2kHz
-30
-30
10 0
2
G. Subject 73: LeMI}. kHz
10
H. Subject 73: Right Ear
400
400
(D
(D
4000
300
a200
aM1 0
200
100
0
TT=1.2kHz
TT=1.2kHz
(-
0
0
100
10
Center Frequency of
Center Frequency of 0.5 octave NBN
0.5 octave
NBN
Figure 9. Contralateral response curves at 1.01 kHz (Subject 67) and 1.2 kHz (Subject 73) test tone from
the elicitor-off analysis window. All other parameters and symbols are the same as Figure 5.
27
Normalized Response curves
In order to compare the shapes of the response curves for the different test
frequencies, we normalized the response curves' IAPSFEJ values and scale the center
frequencies of the elicitor so that the curves can be superimposed. Figure 10 and 11 are
normalized response curves using the analysis windows 3050-3150ms and 2450-2950 ms
respectively. The former window lies in the elicitor-off time, while the latter lies in the
elicitor-on time. Normalized IAPSFE is defined as the difference between the IAPSFE in dB
SPL and the maximum value of IAPSFE| in dB SPL for the given test tone. The center
frequency of the noise band was also normalized by dividing it with the test frequency so
that the noise band centered around the test frequency takes on the value of 1. The
normalized response that satisfies the 3 standard deviation above the noise criteria are
marked by their respective symbols while those that do not are not marked. We notice
that the peak values from the normalized response curves do not line up. The maximum
efferent response can occur with noise bands centered below, at or above the test
frequency. There is no obvious dependence of the position of the peak response relative
to the test tone on the subject or frequency of test tone.
Variability of efferent response within a subject
Three response curves with relatively large responses were chosen to demonstrate
the variability in efferent response under the same stimulus conditions. Figure 12 shows
the contralateral response curves analyzed within the 500 ms window which ends 50 ms
before the elicitor is turned off. The first, second and third columns correspond to
response curves with test frequencies 0.77, 1.03 and 2.6 kHz respectively. The first two
rows are the SFE magnitude and phase change obtained from 12 averages. The response
values that are at least 3 standard deviations above the mean noise floor when no elicitor
is present are marked with a symbol. The bottom row of graphs show the SFE magnitude
and phase change of the three 4-average runs which combine to give the 12-average
response curves shown above. The points that are marked here correspond to those in the
12-average response curves. The horizontal line represents the 3 standard deviation above
the mean noise floor from all runs with 4 averages.
In the 4-average curves, the phase change appears very close in value to that of
the 12-average run. The spread of the response values that lie above the 3 standard
deviation above the mean noise floor criterion are as large as 4.5, 7.5 and 8.5 dB SPL
when the test frequency was 0.77 kHz, 1.03 kHz and 2.6 kHz respectively.
28
B. MOO Effect with Binaural Elicitor (Right Ear)
A.
0
A. MOO Effect with Binaural Elicitor (Left Ear)
0
,7 ~
/
-5
*
-5
/
V .
\-*
:1
'.
/
-10
/
/
N
/
-
/
7
*
*
-10
/
~
N
/
-15
-15
-20
-20
-25
-25
100
100
C. MOO Effect with
A.
0
Ipsilateral
D. MOC Effect with
Elicitor (Left Ear)
0
*
(Right
Ipsilateral Elicitor
A
Ear)
4
4.
/ 'N
*
,/
i*
/
*
/
I~
-10
-5
/
N
N'
-15
/
N.
N
/
N
/
-20
-20
-
NI
-25
-25
100
100
F. MO
COEffect with Contralateral Elicitor (Right Ear)
0
*
r
E. MOO Effect with Contralateral Elicitor (Left Ear)
0
X 1 ,.
/
/
/
N~
/
10
N
/
-10
/
-15
15
-20
-20
-25
J
N'
1
-25'
10
Normalized Center Frequency of 0.5 octave NBN
100
Normalized Center Frequency of 0.5 octave N BN
Figure 10. Normalized delta P as a function of normalized center frequency of 0.5 octave NBN elicitor.
Normalized delta P is the dB SPL difference between delta P of the raw data and the maximum delta P
within an experiment with a fixed text tone. The normalized center frequency of an elicitor is obtained by
dividing that center frequency with the test tone frequency. Plots A and B were obtained with binaural
elicitors. Plots C and D were obtained with ipsilateral elicitors. Plots E and F were obtained with
contralateral elicitors. Plots on the left panel correspond to results from the left ear while the plots on the
right panel correspond to the results from right ear. Analysis window used was 100 ms starting 50 ms from
the time the elicitor was turned off.
Subject 61: Test tone frequency = 0.77 kHz
-0Subject 67: Test tone frequency = 1.01 kHz
-A-*-
Subject 68: Test tone frequency = 1.03 kHz
-A
Subject 73: Test tone frequency = 1.20 kHz
-0-
Subject 61: Test tone frequency = 2.60 kHz
29
0
0
-2
*
i;.,
,/\
-0-
/
/
II
Il
Il
I
-6
- I
/
/
*
14
Al
-2
1'
-4
;---~
/
-
~
A
-4
I
S/-*
j
-6
III
C1,
I-
-8
-8
-
C3
10-
I,
-10
II-*
12-
-12
14 -
-14
-16
-16
10m
Normalized Elicitor Center Frequency
100
Normalized Elicitor Center Frequency
Figure 11. Normalized contralateral response curves as a function of normalized center frequency of 0.5
octave NBN elicitor. Analysis window was from 2450-2950 ms, during which the elicitor was on. All
other parameters and symbols used are the same as figure 10. Note that on the left panel which represents
data from the left ear, no data from Subject 73 was plotted as the response was too weak.
Contralateral and binaural efferent tuning curve
Figure 13 shows response curves and the resulting tuning curves obtained with
0.5 octave NBN elicitors of levels of 40, 50, 55, 60 dB SPL at a test frequency of 0.77
kHz. Only the left and right contralateral responses and the right binaural response are
presented because the response from the other elicitors did not have a good enough signal
to noise..
In many physiological studies, the sharpness of tuning curves is measured with a
quantity called Q20. Q20 is defined as the center frequency (CF) divided by the
bandwidth 20 dB above threshold.
(BW with 60 dB SPL)/threshold frequency
= (1.396;- 0.3287)/0.5445
= 1.9605
Q20 of the right contralateral tuning curve = (1.2353 - 0.3226)/0.5445
= 1.6762
= (1.2366 - 0.2226)/0.5445
Q20 of right binaural tuning curve
= 1.8623
Q20 of the left contralateral tuning curve
=
30
10
10
CO
CO
CO
Subject 61; TT=2.6 kHz
Subject 68; TT=1.03 kHz
Subject 61; TT=0.77 kHz
5
5
0
0
-5
-5
-10
-10
-15
-15
4001
400
4001
300
300.
300
200
200
200
100
100F
1001
10
100
10
10.
0
C2
0)
0
10
10r
0
0
0
-10
-10
*0'-20
100
10
10
" '10
0
*-0
0
0
10
-20
100
10,
-20
400
400k
400
300
3001
300
200
200
100
100
0
100
0
10C
100
0
10
Center Frequency of 0.5 octave NBN Elicitor
0
I1CP
Figure 12. Magnitude and phase of the change in SFE obtained with 12 or 4 averages. The top two rows are
the SFE magnitude and phase change averaged from 12 sets of-data. The errorbars in the top row represent
the standard deviation from the 3 runs, each an average of 4 sets of data, which were then averaged to give
the resulting response curve. The response values that are at least 3 standard deviations above the mean
noisefloor are marked with a circle (o). In the phase plot, (o) and (.) are change in phase with and without
an elicitor respectively. The bottom 2 rows of graphs are the three 4-average runs. The marked points
correspond to those in the top graphs. The horizontal line shown indicates the 3 standard deviation mark
above the mean noisefloor. Hence points above that line satisfy the criteria of significance.
31
A. Contralateral Effect in Left Ear B. Contralateral Effect in Right Ear
10
TT=0.77
kHz
TT=0.77 kHz
C. Binaural Effect from Right Ear
10
TT=0.77 kH2
5
5
0 -
010
-5
'
CO)
-
\
-
-5 /
-
S-10
-10-
-101
-15
-15
-20 -
-20-
f I1
/
-0-15
-
-/-/
-20
/
-
-
-
-25
---
-30
-25-
-25
-30
i-'
-
- - -
-30 -
-
--
10 0
Center Frequency of
100
0,5 octave
NBI
D. MOC Efferent Tuning Curve
60
E. MOC Efferent Tuning Curve
60
F. MOC Efferent Tuning Curve
60
58-
58
58
56-
56
aO52 --
-
54
54
52 -
52
-
50-
50 -
56
50
~48
48-
46
46-
44-
44
44
42-
42
42
48
-
46
40
40
100
40
100
1
Center Frequency of 0.5 octave NBN
Figure 13. Graph A shows response curves obtained from the left ear by eliciting MOC efferents with 0.5
octave contralateral NBN elicitor at different levels. Graphs B and C, are response curves obtained from the
right ear with contralateral NBN and binaural NBN respectively. The test tone was at 0.77 kHz in all
graphs. The NBN levels were 40 dB SPL (-0-), 50 dB SPL (-A-), 55 dB SPL (-*-) and 60 dB SPL (-0-).
Graphs D, E, and F are the MOC efferent tuning curves obtained from Graphs A, B, and C respectively.
The response curves obtained with contralateral NBN were analyzed in the 2450-2950 ms window. The
binaural response curve was analyzed in the 3050-3150 ms window.
32
Ipsilateral, contralateral and binaural responses as a function of elicitor
bandwidth
Figure 14 shows the ipsilateral, contralateral and binaural efferent response with
elicitors of different bandwidths. The test frequencies studied are 0.77 kHz, 1.03 kHz and
2.6 kHz. For points significantly above the noise floor, we see that the efferent response
generally increases with the bandwidth of the noise band elicitor and in some cases
reaching saturation. It appears that in the response that saturated, the point of saturation
occurs between bandwidths of 2 and 4 octaves. In the case when the test tone was 2.6
kHz, the right binaural and contralateral response levels appear to be higher with a 4
octave band than a BBN. Only the right ipsilateral response with test frequency of 2.6
kHz decreased with increasing elicitor bandwidths. The strength of the binaural response
appears to be largest at almost all elicitor bandwidths used. With the limited amount of
data presented, the strength of the ipsilateral and contralateral response appears to be
somewhat comparable except for the results with the test frequency of 2.6 kHz.
Figure 15 is a redisplay of the results from figure 14 such that the responses
obtained with the same elicitor orientation, ipsilateral, contralateral or binaural, are
plotted together.
Figure 16 is the normalized data from figure 14. The normalization is done by
subtracting the magnitude change in SFE ear canal pressure (in dB SPL) with the
magnitude change obtained with a 2 octave noise band. Figure 17 is the same as figure 16
but the normalized data are regrouped into ipsilateral, contralateral and binaural
responses. In both figures we note that plots of significant responses have very similar
shape and rate of growth. The ipsilateral response in one subject appears to be different; it
decreases as bandwidth increases.
33
15
20
TT=0.77 kHz
TT=0.77 kHz
....
*7
7
10
V
5
7
V
0
0-
ii'
-10-
-5
li/Il
-10
/
~
\
-20
/
I
\
\
-30
10
10
.7
-15
/
.
-41
-1
IA
-2U
1
0
I A
TT=1 .03 kHz
11
a10
Xi10
102
02
10
TT=l.03 kHz
0-
5
-I
0
-10-
X7
-------
-5
-20-10
-25[ v
-15
100
0
-5
V
TT=2.6kHz
100
10
kV
5
TT=2.6 kHzV
0
-10
-5
-15-20-
-10
-25
-15
-30
-20
-35 10~
0
10
101
-25
10
102
-1
-----
---100
10
102
Noise Elicitor Bandwidth (octave)
Noise Elicitor Bandwidth (octave)
Figure 14. Change in magnitude of SFE in the presence of efferent stimulation as a function of elicitor
bandwidth centered at the test frequency. Test tone was at 40 dBSPL. Noise bandwidths were 1/8, _, 1/2, 1,
2 octaves for test fequency 1.03 kHz and _, 1/2, 1, 2, 4 octaves for test frequencies 0.77 and 2.60 kHz. For
the plots with test tones 0.77 kHz and 2.6 kHz, the very last points on the right correspond to efferent
response to BBN elicitor which spans from 100 to 10 kHz. Elicitor level was 60 dBSPL. Analsis window
used was 100 ms in duration starting 50 ms after the elicitor is turned off.
Symbols:
Inverted triangle = Elicitor in both ears,
* = Elicitor in right ear,
o = Elicitor in left ear,
- = No elicitor
34
B. Binaural Efferent Effect in RIGHT ear
15 r
1
A. Binaural Efferent Effect in LEFT ear
20
10 0
10
*
-4T____
510 F
-30 '101
10 0
-5
10
10
0
Ca
.10
CL
U3
-20 F
M)
-40 '1
101
CO
*-
10l -
020
-30 F
- . - - -- ---
100
- - - - - - --
-30 10-1
10
E. Contralateral Efferent Effect in LEFT ear
In,
101
100
-0
(I
C')
--
D. Ipsilateral Efferent Effect in RIGHT ear
10
C. Ipsilateral Efferent Effect in LEFT ear
10
-3
*-
0-
-20 F
I
10 0
101
F. Contralateral Efferent Effect in RIGHT ear
10.
1
101-
5
0
0
-5
-10-
-10
F
-20-15
- - - - - - --- - - - - ---30
10
100
101
Noise Elicitor Bandwidth (octave)
-20'0
101
10
101
Noise Elicitor Bandwidth (octave)
Figure 15. Same as figure 14 but with the binaural, ipsilateral and contralateral responses plotted
separately.
= 0.77 kHz test frequency
*
1.03 kHz test frequency
0= 2.60 kHz test frequency
Data points marked with a symbol were found to be at least 3 standard deviations above the mean noise
floor.
35
B. Subject 61 Right Ear
A. Subject 61 Left Ear
10
20
r
10
5
0
0
V>
.
-10.
.
-5
-20-
-10
-30
15=
-401
101
100
IC -1
101
C. Subject 68 Left Ear
UI)
-2
-5
-V
-4
-10
CL
CO
-o
U)i
N
0
10'
D. Subject 68 Right Ear
0
0
100
-6
-15
-8
-20
-10
-25
-12
-30'100.
10
-14
10
10
-0.9
10 -
100.3
F. Subject 61 Right Ear
E. Subject 61 Left Ear
5
15
0
10
5
-5
0
0
-10
-5
-15
-10
-20
.0
-15
2"i-
-25
10
10
1 0Ni
Noise Elicitor Bandwidth (octave)
10
100
10-1
Noise Elicitor Bandwidth (octave)
Figure 16. Change in magnitude of SFE in the presence of efferent stimulation normalized by the response
level obtained with a 2 octave noise band elicitor as a function of elicitor bandwidth centered at the test
frequency. Test tone was at 40 dBSPL. Elicitor level was 60 dBSPL. Analsis window used was 100 ms in
duration starting 50 ms after the elicitor is turned off.
Inverted triangle = Elicitor in both ears,
Symbols:
* = Elicitor in right ear,
o
=
Elicitor in left ear,
-
=
No elicitor
Plot (A, B) = test frequency of 0.77 kHz
Plot (C, D) = test frequency of 1.03 kHz
Plot (E, F) = test frequency of 2.60 kHz
For the plots with test tones 0.77 kHz and 2.60 kHz, the very last points on the right correspond to efferent
response to BBN elicitor which spans from 100 to 10 kHz.
36
Binaural Efferent Effect in Right Ear
Binaural Efferent Effect in Left Ear
10
20
0
10
0
-10
-20
-10
-20
> -30
-U
10
o
01
0
10
lpsilateral Efferenl Effect in Right Ear
Ipsilateral EfferenPEffect in Left Ear
20
N 20
0)
101
0o
V0
0
-20
-10
E
Ipsiea
f2e0ueny.
0k
101
100
11
101
10
2epaa0e
Contralaterl Efferent Effect in Right Ear
Entc
,
f
6-40- 0...
10
.........
-20
/
0.
-20*-0/
Fiur 1 Cntralagter E
1'100
ffe~ Effc inLeft Ear 10 nc
101
10efrn tmlto
10i
octve)Noise
Elcitr
Nois
Bndwith
oraie
ytersos
0
1o
101
Elicitor Bandwidth (octave)
Figure 17.Change in magnitude of SFE in the presence of efferent stimulation normalized by the response
level obtained with a 2 octave noise band elicitor as a function of elicitor bandwidth centered at the test
frequency. Separated into Ipsilateral, contralateral and binaural plots.
=0.77 kF~z test frequency
*=1.03 k~lz test frequency
=2.60 kHz test frequency
Data points marked with a symbol were found to be at least 3 standard deviations above the mean noise
floor-
37
Discussion
The response curves presented in this study clearly demonstrate a frequency
specificity of the efferent-induced change in SFEs but also a surprising ability of distant
frequencies to activate the efferents. Response strength and tuning pattern also appeared
to vary between the 2 ears within a subject. The greatest change in IAPSFEJ were observed
when the elicitor noise bands were centered near the test frequency, but not necessarily at
the test frequency. Noise bands were also more effective when centered near the test tone
rather than further away. We also saw that in some cases, elicitors centered as far as two
and a half octaves above and/or below the test tone elicited significant changes in SFE.
This is perhaps the most significant finding in this study. The tuning observed is thus
broader than the tuning reported in other existing literature on Humans.
For example, Veuillet and colleagues (1991) pointed out a frequency specificity
of contralateral NBN stimulation on evoked otoacoustic emission (EOAE) amplitude as
seen from data averaged across subjects. The ipsilateral tone pip was limited to 1 and 2
kHz. Significant decrease in the 0.9-1.17 kHz EOAE spectrum band, induced with a 1
kHz tone pip in the ipsilateral ear, was observed with 50 dB contralateral NBN centered
at 0.75, 1.0 and 1.5 kHz. With the 2 kHz tone pip, the 1.9-2.1 kHz EOAE spectrum
decreased significantly with 1.0, 1.5 and 2.0 kHz-centered NBN elicitors.
In 1993, Chery-Croze et al. reported evidence of MOC frequency selectivity from
studies on the suppressive effects of NBN contralateral acoustic stimulation, with an
ascending and descending slope of 24 dB/octave at 55 dB SPL, on the distortion product
2f1-f2 of 1 and 2 kHz. The reported data was an average across subjects and the
significance in suppression was analyzed using the Student's t-test. They observed
greatest decreases when the contralateral NBN was centered close to the DPOAE. When
the 2f1-f2 was fixed at 2 kHz, NBN centered at 1 kHz below the DPOAE induced
significant suppression.
Lastly, Maison et al. (2000) reported an experimentally measured plot of EOAE
attenuation as a function of contralateral stimulus bandwidths centered at various
frequencies away from the 1 kHz tone pip. Again, the data was averaged across subjects.
Significant suppression was observed with NBN elicitors centered less than half an
octave above and below the nominal center frequency tested.
In the present study, we have clearly demonstrated efferent inhibition from NBN
elicitors at much further distant from the test frequency (2 and a half octaves). In
accordance with the current belief that the MOCB control on cochlear response involves
the inhibition of OHCs onto which they synapse (Guinan, 1986, Liberman and Brown
1986), the dispersion of effective elicitor center frequencies may be explained by the
spanning of each MOC efferent fiber on several OHCs (Brown, 1989) as well as the
extent to which OHCs at a specific cochlear place affects the cochlear response over a
range of frequencies (Guinan, 1996).
The normalized response curves, referenced to the maximum response magnitude
in dB SPL, exhibit no obvious similarity in shape. Even with close test frequencies like
1.01 and 1.03 kHz, the contralateral response curves obtained from the different subjects
using the 2450-2950 ms analysis window show different tuning patterns for the range of
38
elicitor used - the former appearing more bandpass while the latter more highpass. The
normalized responses were also generally asymmetric on a semi-logarithmic frequency
scale and the maximum response usually did not line up with the test frequency. These
differences suggest that we cannot assume the existence of a generic response across
subjects or even within a subject at different test frequencies.
At this point, it is worth noting that our observations of the detailed tuning
pattern are by no means conclusive. Evidence of significant intra-subject variability in the
efferent response induced under the same experimental conditions disables us from
making any convincing observations on the shape of the response curves. More repeated
measurements are needed before any statistically sound estimates can be made. In
existing literature, response variability within a subject is usually ignored in data analysis.
Perhaps this variability within a subject as well as the variability across subjects under the
same experimental parameters contributed to the narrower frequency selectivity observed
in the 3 examples (Vieullet et al., Chery-Croze et al., Maison et al.) where data was
averaged across subjects.
The MOC tuning curves obtained at a test frequency 0.77 kHz for the
contralateral response in the right and left ear as well as the binaural response in the right
ear gave Q20 values that are similar in value. The values indicate that efferent tuning is
wider than afferent tuning in humans (Shera et al. 2002).
The set of experiments in which we studied the efferent effect as a function of the
bandwidth of the elicitor centered at the test frequency also showed that efferents
integrate sounds over a wide range of frequencies. The noise bands in this study include
0.125, 0.25, 0.50, 1.0, 2.0, 4.0 octaves NBN and a BBN. The effect of broad band noise,
whose spectrum was constant in energy level from 100 Hz to 10 kHz, was also included.
The results showed that in almost all cases there is an increase in efferent response with
increasing noise bandwidth. The binaural response appears to saturate somewhere
between 2-4 octaves. This trend of increasing response with elicitor bandwidth is in
agreement with those of Maison et al (2000) where they report increasing efferentinduced suppression in EOAE with increasing bandwidth of the contralateral NBN
elicitor. However, their study only included bandwidths of 0.125, 0.25, 0.5, 1.0 and 2.0
octaves and only contralateral sound. This trend may be explained by modeling the
peripheral auditory system as an array of overlapping bandpass filters or channels and
that the larger MOC activation with wider band stimuli results from a summation of
inputs from more channels to the Cochlear Nucleus which ultimately relays the signal to
the medial part of the SOC where the cell bodies of the MOCB neurons are located.
Maison et al. (2000) also normalized their data with the number of equivalent
rectangular bandwidths (ERBs) that fall between the upper and lower cutoff frequency of
the contralateral NBN elicitor. However, a more recent study of otoacoustic and
behavioral measurements demonstrated that tuning in the human cochlea at low sound
levels is more than twice as sharp as tuning perceived with standard estimates from
traditional behavioral measurements and casts doubt on the analysis (Shera et al. 2002).
With regards to the relative strength of the responses, the binaural stimulation is
almost always the largest. In the experiment where the center frequency of a 0.5 octave
NBN was varied, the relative strength between the ipsilateral and contralateral response
appeared to vary with test frequency, elicitor band used and the subject. In the experiment
where the elicitor bandwidth was varied while keeping its center frequency at the test
39
frequency, there is too little data to show whether the ipsilateral or contralateral response
is larger. As a first approximation, however, their relative strength appears to be
comparable.
Future Studies
It would be very interesting and useful to quantify in some way the statistics of
intra-subject efferent response variability. A better understanding of this variability
appears to be an important prerequisite before the functional capacity of Human MOCB
can be clearly understood. If some tuning patterns can be well-estimated statistically, we
can then begin to probe the functionality of the system.
The response curves here are limited to test tones within mid-frequency ranges.
We intend to extend the study to higher and lower frequency regions. The difficulty of
studying low frequencies involves the extraction of the efferent response which is small
compared to the level of physiological noise floor. At higher frequencies, we are then
limited by the transfer function of the middle ear which is effective at around 1 kHz in
humans. On a different note, to capture finer details of the tuning pattern, noise bands
smaller than 0.5 octave would be needed. However, narrower noise bands may be less
able to elicit efferent activity large enough to be captured experimentally.
40
Bibliography
Backus, B, Arhonson, V., Liberman, M.C., Guinan, J.J Jr. (1999). 'Human medialefferent acoustic reflex: Ipsilateral, Contralateral and binaural reflex strengths compared
using SFE.' 1999 ARO poster session
Brown, M.C. (1989). 'Morphology and and response properties of single olivocochlear
fibers in the guinea pig,' Hear. Res. 40, 93-110.
Brownell, W.E., Bader, C.R., Bertrand, D., de Ribaupierre, Y., (1985). 'Evoked
mechanical response of isolated cochlear outer hair cells.' Science 277, 194-196.
Chery-Croze, S., Moulin, A., Collet, L., (1993) 'Effect of contralateral sound stimulation
on the distortion product 2fl -f2 in humans: Evidence of a frequency specificity.' Hear.
Res.. 68,53-58.
Collet, L., Kemp, D.T., Veuillet, E., Duclaux, R., Moulin, A., and Morgan, A., (1990a).
'Effect of contralateral auditory stimuli on active
cochlear micro-mechanical properties in human subjects.' Hear. Res. 43,
251-262.
Collet, L., Veuillet, E., Duclaux, R., and Morgan, A. (1990b).
'Influence of contralateral auditory stimulation on Evoked otoacoustic
emissions.' in Cochlear Mechanisms and Otoacoustic Emissions. Advances in
Audiology, edited by F.Grandori, G. Cianfrone, and D. Kemp (Karger, Basel),
Vol. 7, pp. 164-170.
Davis, H., (1983). 'An active process in cochlear mechanics.' Hear Res. 9, 79-90.
Dallos, P., Evans, B.N., (1995). 'High Frequency Motility of outer hair cells and the
cochlear amplifier.' Science 267, 2006-2009.
Dolan, D.F., Nuttall, A.L., (1994). 'Basilar membrane movement evoked by sound is
altered by electrical stimulation of the crossed olivocochlear bundle.' J. Acoust. Soc. Am.
83, 1081-1086.
Fay, R. (1988). 'Hearing in Vertebrates: A Psychophysics Databook'. Winnetka, IL: HillFay Associates.
Fex, J., (1962). 'Auditory activity in centrifugal and centripetal cochlear fibers in cat'.
Acta Physiol. Scand. 55, 2-68.
Guinan, J.J. Jr. (1986). 'Effect of efferent neuroactivity on cochlear mechanics.' Scand.
Audiol. Suppl. 25, 53-62.
Guinan, J.J. Jr. (1996). 'The Physiology of Olivocochlear Efferents,' in
41
The Cochlea, edited by P.J. Dallos, A.N. Popper and R.R. Fay
(Springer-Verlag, New York), pp. 435-502.
Guinan, J.J. Jr., Warr, W.B., Norris, B.E., (1983). 'Differential olivocochlear projections
from lateral vs. medial zones of the superior olivary complex,' J. Comp. Neural. 221,
358-370.
Kawase, T., Liberman, M.C. (1993). 'Antimasking effects of the olivocochlear reflex. I.
Enhancement of compound action potentials to masked tones,' J. Neurophysiol. 70,
2519-2532.
Kemp, D.T., (1978). 'Stimulated acoustic emissions from within the human auditory
system.' J. Acoust. Soc. Am. 64, 495-496
Kemp, D.T., and Chum, R.A. (1980). 'Properties of the generator of
stimulus frequency acoustic emissions.' Hear. Res. 2, 213-232.
Liberman M.C., Brown, M.C. (1986). 'Physiology and anatomy of single olivocochlear
neurones in the cat,' Hear. Res. 24, 17-36.
Long, G.R., Tubis, A., Jones, K.L., Sivaramakrishnan, S. (1988). 'Modification of
external-tone synchronization and statistical properties of spontaneous otoacoustic
emissons by aspirin consumption,' in Basic Issues in Hearing,edited by H. Duifuis, J.W.
Horst, and H.P. Wit (Academic, London), pp. 93-100.
Long, G.R., Tubis, A. (1988a). 'Investigation into the nature of the association between
threshold microstructure and otoacoustic emissions,' Hear Res. 36, 125-138.
Maison, S., Micheyl, C., Chays, A., Collet, L. (1997b). 'Medial olivocochlear system
stabilizes active cochlear micromechanical properties,' Hear Res. 113, 89-98.
Maison, S., Micheyl, C., and Collet, L. (1997). 'Medial Olivocochlear
efferent system in humans studied with amplitude modulated tones.' J.
Neurophysiol. 77, 1759-1768.
Maison, S., Micheyl, C., and Collet, L., (1998). 'Contralateral
frequency modulated tones suppress transient-evoked otoacoustic emissions in
humans,' Hear. Res. 117, 114-118.
Margolis, R.H., Dubno, J.R., and Wilson, R.H. (1980). 'Acoustic-reflex thresholds for
noise stimuli,' J. Acoust. Soc. Am. 68, 892-895.
Micheyl, C., Maison, S., Carlyon, R.P., Andeol, G., and Collet, L.
(1999). 'Contralateral suppression of transiently-evoked otoacoustic
emissions by harmonic complex tones in humans.' J. Acoust. Soc. Am. 105,
293-305.
42
Morlet, T., Ferber, C., Duclaux, R., Challel, M.J., and Collet, L.
(1994). 'Effect of sleep stages on transiently evoked otoacoustic emissions
in infants.' Brain & Develop. 16, 115-120.
Mountain, D.C., (1980). 'Changes in the endolymphatic potential and crossed
olivocochlear bundle stimulation alter cochlear mechanics.' Science 210, 71-72.
Mountain, D.C., Hubbard, A.E., (1989). 'Rapid for production in the
cochlea.' Hear. Res. 42, 195-202.
Oatman, L.C., Anderson, B.W., (1977). 'Effects of visual attention on tone burst evoked
auditory potentials.' Exp Neurol. 57, 200-211.
Patuzzi, R. (1996). 'Cochlear Micromechanics and Macromechanics,' in
The Cochlea, edited by P.J. Dallos, A.N. Popper and R.R. Fay
(Springer-Verlag, New York), pp. 186-257.
Probst, R., Lonsbury-Martin, B.L., Martin, G.K., (1990). 'A review of
otoacoustic emissions.' J. Acoust. Soc. Am. 89, 2027-2067.
Puel, J.L., Bonfils, P., and pujol, R. (1988). 'Selective attention
modifies the active micromechanical properties of the cochlea.' Brain Res.
447, 380-383.
Rajan, R., (1988). 'Effect of electrical stimulation of the crossed olivocochlear bundle on
temporary threshold shifts in auditory sensitivity. I. Dependence on electrical stimulation
parameters.' J. Neurophysiol. 60, 549-568.
Rajan, R., Johnstone, B.M., (1983). 'Crossed cochlear influences on monaural temporary
threshold shifts.' Hear Res. 9, 279-294.
Rasmussen, G.L. (1946). 'The olivary peduncle and other fiber projections of the superior
olivary complex,' J. Comp. Neurol. 84, 141-219.
Rasmussen, G.L. (1960). 'Efferent fibers of cochlear nerve and cochlear nucleus,' in
Neural Mechanisms of the Auditory and Vestibular Systems. Springfield, IL: Thomas.
pp. 105-115.
Reiter, E.R., Liberman, M.C., (1995). 'Efferent mediated protection from acoustic
overexposure: relation to 'slow effects' of olivocochlear stimulation.' J. Neurophysiol.
73, 506-514.
Robertson, D., Gummer, M. (1985). 'Physiological and morphological characterization of
efferent neurons in the guinea pig cochlea,' Hear. Res. 20, 63-77.
43
Santos-Sacchi, J., Dilger, J.P., (1988) 'Whole cell currents and mechanical responses of
isolated outer hair cells.' Hear. Res. 35, 143-150.
Shera, C.A., Zweig, G. (1993a). 'Noninvasive measurement of the cochlear traveling
wave ratio,' J. Acoust. Soc. Am. 93, 3333-3352.
Shera, C.A., Guinan, J.J., Oxenham, A.J. (2002). 'Revised Estimates of Human Cochlear
Tuning from Otoacoustic and Behavioral Measurements,' PNAS, 1-6
Smith, C.A. (1961). 'Innervation pattern of the cochlea,' Ann Oto Rhinol Laryngol. 70,
504-527.
Tubis, A., Long, G.R., Sivaramakrishnan, S., Jones, K.L. (1989). 'Tracking and
interpretive models of the active-nonlinear cochlear response during reversible changes
induced by aspirin consumption,' in Cochlear Mechanisms, Structure, Function, and
Models, edited by J.P. Wilson and D.T. Kemp (Plenum, London), pp. 323-330.
Van Dijk, Wit, H.P. (1988). 'Phase-lock of spontaneous otoacoustic emissions to a cubic
difference tone,' in Basic Issues in Hearing, edited by H. Duifuis, J.W. Horst, and H.P
Wit (Academic, London), pp. 101-105.
Ver, I.L., Brown, R.M., Kiang, N.Y., (1975) 'Low Noise Chambers for auditory
research,' J. Acoust. Soc. Am. 58, 392-398.
Veuillet, E., Collet, L. and Duclaux, R., (1991). 'Effect of
Contralateral Acoustic Stimulation on active cochlear micromechanical properties
in human subjects: Dependence on stimulus variables. 'Journal of
Neurophysiology, Vol. 65, No. 3, 724-735
Warren, E.H. III, and Liberman, M.C. (1988a). 'Effects of Contralateral
Sound on auditory-nerve response. I Contributions of cochlear
efferents.' Hear. Res. 37, 89-104
Warren, E.H. III, and Liberman, M.C. (1988a). 'Effects of Contralateral
Sound on auditory-nerve response. II Dependence on stimulus variables.'
Hear. Res. 37, 105-122
Wiederhold, M.L. (1986) 'Physiology of the olivocochlear system.' In R.A. Altschuler,
D.W. Hoffman and R.P. Bobbin (Eds.), Neurobiology of Hearing: The Cochlea. Roven
Press, New York 19, 349-370.
Winslow, R.L., Sachs, M.B., (1988). 'Single-tone intensity discrimination based on
auditory nerve rate responses in backgrounds of quiet, noise, and with stimulation of the
crossed olivocochlear bundle.' Hear res. 35, 165-190.
44
Zweig, G., Shera, C.A., 1995. 'The origin of periodicity in the spectrum of evoked
otoacoustic emissions.' J. Acoust. Soc. Am. 98, 2018-2047.
45