You are on page 1of 157

University of Iowa

Iowa Research Online

Theses and Dissertations

Fall 2013

The origin of short-latency transient-evoked otoacoustic


emissions
James Douglas Lewis
University of Iowa

Follow this and additional works at: https://ir.uiowa.edu/etd

Part of the Speech and Hearing Science Commons

Copyright 2013 James Lewis

This dissertation is available at Iowa Research Online: https://ir.uiowa.edu/etd/5013

Recommended Citation
Lewis, James Douglas. "The origin of short-latency transient-evoked otoacoustic emissions." PhD (Doctor
of Philosophy) thesis, University of Iowa, 2013.
https://doi.org/10.17077/etd.gv4ccdxa

Follow this and additional works at: https://ir.uiowa.edu/etd


Part of the Speech and Hearing Science Commons
THE ORIGIN OF SHORT-LATENCY
TRANSIENT-EVOKED OTOACOUSTIC EMISSIONS

by
James Douglas Lewis

A thesis submitted in partial fulfillment of the


requirements for the Doctor of Philosophy degree
in Speech and Hearing Science in the Graduate College of
The University of Iowa

December 2013

Thesis Supervisor: Assistant Professor Shawn S. Goodman


Graduate College
The University of Iowa
Iowa City, Iowa

CERTIFICATE OF APPROVAL

PH.D. THESIS

This is to certify that the Ph. D. thesis of

James Douglas Lewis

has been approved by the Examining Committee


for the thesis requirement for the Doctor of
Philosophy degree in Speech and Hearing
Science at the December 2013 graduation.

Thesis Committee:
Shawn S. Goodman, Thesis Supervisor

Paul J. Abbas

Ruth A. Bentler

Lenore A. Holte

Christopher W. Turner
To Kelli & Emelia.

  ii  
Listen to this, O Job,
Stand and consider the wonders of God.
Do you know how God establishes them,
And makes the lightning of His cloud to shine?
Do you know about the layers of the thick clouds,
The wonders of one perfect in knowledge,
You whose garments are hot,
When the land is still because of the south wind?

-Job 37:14-17

  iii  
ACKNOWLEDGEMENTS

Thanks to Shawn Goodman who first welcomed me into his lab back in 2006
when we were both just beginning in the department. I am now happy to call Shawn both
a mentor and friend. Shawn taught me how to program, think critically, seek answers, and
ask questions. He has been an example of one who cares about his work and wants to
make a meaningful contribution to the field of hearing science.
Thanks to Paul Abbas, Ruth Bentler, Lenore Holte, and Chris Turner who, along
with Shawn, served on my dissertation committee. Each of these individuals has made
significant contributions to my development both as a researcher and clinical audiologist.
I want to extend my gratitude to Rachel Stanziola and Brittany James who worked
with me to collect the data presented in this thesis. Over 80 hours of data were collected
during the final month of the semester thanks to these two. I’d also like to thank the
Executive Council for Graduate and Profession Students at the University of Iowa who
generously provided the funding to complete the experimental work described in the
following pages.
Thanks also to my parents, Doug and Christine. Two things stand out to me as I
think of them. First, they are the hardest working people I know. Second, everything they
touch is made better. If I possess even the slightest measure of these qualities it is
because of my dad and mom.
My wife Kelli deserves greatest thanks. Her devotion has never been contingent
on my success but, rather, something that is always present and can be relied upon even
in the midst of failure or disappointment. I can’t really imagine going through the Ph.D.
program without my wife. She has brought me so much joy and happiness, without which
I’m not sure how long I would have lasted. Kelli has never yielded her support or been
lacking in encouragement for me over these past few years. I’m glad to be moving-on
with her by my side.

  iv  
ABSTRACT

Bandpass filtered transient-evoked otoacoustic emission (TEOAE) waveforms are


composed of short-latency (SL) and long-latency (LL) portions. The LL portion has
latency consistent with generation through linear coherent reflection at the tonotopic
place on the basilar membrane. The short-latency (SL) portion occurs earlier in time and
exhibits less compressive growth. Several mechanisms have been hypothesized to explain
generation of the SL portion, including 2 f1 ! f2 intermodulation distortion and coherent
reflection basal to the tonotopic place. Two experiments were designed to examine the
generation mechanism and generation location of the SL portion. Experiment 1 tests the
hypothesis that the SL portion results from low-side, cubic intermodulation distortion.
Experiment 2 determines the region along the basilar membrane at which the SL portion
of the TEOAE is generated.
The null hypothesis that the SL portion of the TEOAE is generated through low-
side, cubic intermodulation distortion requires stimuli with broad frequency content.
Stimulus energy at different frequencies ( f1 and f2 ) is presumed to interact
simultaneously across the cochlear partition, generating a distortion-source OAE. To test
this hypothesis, OAEs were evoked using 2 kHz tone-bursts with durations spanning the
time-frequency continuum between a click and a pure tone. As tone-burst duration
increases, stimulus energy at the primary frequencies ( f1 and f2 ) decreases and the input
to any nonlinear distortion source is reduced. Accordingly, if generated through 2 f1 ! f2
distortion, the magnitude of the SL portion of the TEOAE was expected to decrease as
tone-burst duration increased. Results were inconsistent with generation of the SL portion
through intermodulation distortion. As tone-burst duration increased, the SL portion
remained present in the TEOAE. The presence of the SL portion influenced the level-
dependency of TEOAE latency and magnitude to the same extent across all tone-burst
durations.

  v  
The region of generation along the cochlear partition of the SL portion has
implications for the mechanism through which it is generated. Generation through low-
side, cubic intermodulation distortion ( 2 f1 ! f2 ) would occur near the f2 tonotopic place.
If generation is through coherent reflection, a generation region basal to that of the
tonotopic place is hypothesized. To determine the cochlear region where the SL portion is
generated, TEOAEs were evoked by 2 kHz tone-bursts in the presence of different
suppressor stimuli. The amount of suppression induced by each suppressor on the OAE
was measured, and the suppressor frequency causing greatest suppression of a given
portion of the TEOAE was interpreted as corresponding to that portion’s generation place
along the basilar membrane. For analysis purposes, the SL portion was divided into two
SL time-windows (SL1 and SL2). The LL portion of the TEOAE was maximally
suppressed by a 2.07 kHz tone, consistent with generation at the tonotopic place. Both SL
components were generated basal to the tonotopic place. The later-occurring SL portion
of the TEOAE (SL1) was generated between 1/4-1/3-octave basal to the tonotopic place
while the earlier-occurring SL portion (SL2) was generated 3/5-octave basal to the
tonotopic place. The generation region of the SL1 portion of the TEOAE was too apical
to be consistent with generation through 2 f1 ! f2 distortion. Although the generation
region of the SL2 portion was what would be expected for a 2 f1 ! f2 distortion-source

OAE, the latency was too early. Generation of both SL portions may be explained
through basal linear coherent reflection. Per this mode of generation, the SL1 and SL2
portions of the TEOAE each likely mirror the underlying mechanics at different regions
along the cochlear partition.

  vi  
TABLE OF CONTENTS

LIST OF TABLES .............................................................................................................ix

LIST OF FIGURES ............................................................................................................x

LIST OF ABBREVIATIONS ...........................................................................................xii

INTRODUCTION ...............................................................................................................1

CHAPTER

I. MECHANICS OF THE COCHLEA ...................................................................5

1.1 Passive Cochlear Mechanics .................................................................5


1.2 The Cochlear Amplifier .........................................................................7
1.3 Active Cochlear Mechanics ...................................................................8

II. OTOACOUSTIC EMISSIONS.........................................................................11

2.1 OAE Classification...............................................................................11


2.2 OAEs and Cochlear Mechanics............................................................13
2.3 Generation Mechanisms.......................................................................16
2.3.1 Underlying Theories..............................................................16
2.3.2 Generation of DPOAEs.........................................................19
2.3.3 Generation of SFOAEs..........................................................21
2.3.4 Generation of TEOAEs .........................................................23
2.3.4.1 The Long-Latency TEOAE....................................24
2.3.4.2 The Short-Latency TEOAE ...................................25

III. EXPERIMENT 1: THE EFFECT OF TONE-BURST


DURATION ON THE SHORT-LATENCY TEOAE .....................................28

3.1 Introduction ..........................................................................................28


3.2 Methods................................................................................................31
3.2.1 Subjects .................................................................................31
3.2.2 Signal Generation and Data Acquisition...............................32
3.2.3 Measurement and Analysis of TEOAEs ...............................32
3.2.3.1 Stimuli ....................................................................32
3.2.3.2 Calibration..............................................................35
3.2.3.3 Analysis..................................................................36
3.3 Results ..................................................................................................41
3.3.1 TEOAE Envelopes ................................................................41
3.3.2 TEOAE Latency-Intensity Functions....................................43

  vii  
3.3.3 TEOAE Input-Output Functions ...........................................49
3.4 Discussion ............................................................................................51
3.4.1 Generation Mechanism of the SL TEOAE ...........................51
3.4.2 Origin of TEOAE envelope morphology ..............................57

IV. EXPERIMENT 2: TWO-TONE SUPPRESSION


OF SL AND LL PORTIONS OF THE TEOAE..............................................74

4.1 Introduction ..........................................................................................74


4.2 Methods................................................................................................77
4.2.1 Subjects .................................................................................77
4.2.2 Signal Generation and Data Acquisition...............................78
4.2.3 Measurement and Analysis of TEOAEs ...............................78
4.2.3.1 Stimuli ....................................................................78
4.2.3.2 Calibration..............................................................80
4.2.3.3 Analysis..................................................................83
4.3 Results ..................................................................................................87
4.3.1 Tone-Burst Suppressor Condition.........................................87
4.3.2 Pure Tone and Noise-Burst Suppressor Conditions .............90
4.4 Discussion ............................................................................................95
4.4.1 SL and LL Portions of the TEOAE.......................................95
4.4.2 Generation Regions of the TEOAE .....................................96
4.4.3 Generation Mechanisms of the TEOAE ..............................98
4.4.4 Relation of TEOAEs to BM Mechanics .............................100
4.4.4.1 Cochlear Compression ........................................100
4.4.4.2 Basal Shift in the Peak of the Traveling Wave ....102
4.4.5 Implications for Ears with SNHL .......................................103

V. SUMMARY AND CONCLUSIONS .............................................................117

5.1 SL and LL Portions of the TEOAE....................................................117


5.1.1 Growth and Latency Behavior ............................................117
5.1.2 Proposed Mechanisms of Generation .................................118
5.2 Experimental Work to Determine the Origin of the SL TEOAE.......120
5.2.1 Experiment 1 .......................................................................120
5.2.1.1 Design and Analysis.............................................120
5.2.1.2 Results ..................................................................122
5.2.1.3 Discussion ............................................................122
5.2.2 Experiment 2 .......................................................................123
5.2.2.1 Design and Analysis.............................................123
5.2.2.2 Results ..................................................................124
5.2.2.3 Discussion ............................................................124
5.3 Conclusions and Future Directions ....................................................125

REFERENCES ................................................................................................................128

  viii  
LIST OF TABLES

Table 3.1. Mean latencies (ms) ± 1 standard deviation (ms) for each
tone-burst duration and level. 62

Table 3.2. Mean magnitudes (dB SPL) ± 1 standard deviation (dB) for
each tone-burst duration and level. 63

  ix  
LIST OF FIGURES

Figure 3.1. Time- and frequency-domain representations of the 2 kHz


tone-burst stimuli. 64

Figure 3.2 Empirical distribution functions used to determine SNR


criterion. 65

Figure 3.3 TEOAE envelopes for three subjects across different levels
and tone-burst durations. 66

Figure 3.4 TEOAE envelopes for three subjects across different levels
and tone-burst durations. 67

Figure 3.5 Latency-intensity functions for each tone-burst stimulus for


all 12 subjects. 68

Figure 3.6 The effect of temporal interaction between the SL and LL


components on the 24-cycle TEOAE latency calculation. 69

Figure 3.7 Group latency-intensity functions for each tone-burst


stimulus. 70

Figure 3.8 Input-output functions for each tone-burst stimulus for all
12 subjects. 71

Figure 3.9 Group input-output functions for each tone-burst stimulus. 72

Figure 3.10 Simulation results for the multiple component model to


explain TEOAE envelope morphology. 73

Figure 4.1 Probe- and suppressor-stimulus paradigm used to evoke the


suppressed TEOAE. 106

Figure 4.2 Illustration of the technique used to specify the time-


windows corresponding to different-temporal portions of
the TEOAE. 107

Figure 4.3 TEOAE envelopes measured for different level tone-bursts


in the tone-burst suppressor condition. 108

Figure 4.4 TEOAE envelopes measured for different level tone-bursts


in the tone-burst suppressor condition. 109

  x  
Figure 4.5 Level-series illustration of the different-temporal portions
of the TEOAE for the tone-burst suppressor condition,
collapsed across all subjects. 110

Figure 4.6 Mean input-output, slope-intensity, and latency-intensity


functions for the different-temporal portions of the TEOAE
and the total TEOAE. 111

Figure 4.7 Contour plots showing the effect of different frequency


suppressors on the TEOAE. 112

Figure 4.8 Contour plots showing the effect of different frequency


suppressors on the TEOAE. 113

Figure 4.9 Mean iso-level suppression curves for the different-latency


portions of the TEOAE and the total TEOAE. 114

Figure 4.10 Slope-intensity functions for the SL2, SL1, and LL portions
of the TEOAE compared to those derived from
measurements at proportional locations on the BM. 115

Figure 4.11 Input-output functions for the LL, SL1, and SL2 portions of
the TEOAE compared to those at proportional locations on
the basilar membrane. 116

  xi  
LIST OF ABBREVIATIONS

BLN – band-limited noise

BM – basilar membrane

CEOAE – click evoked otoacoustic emission

CF – characteristic frequency

CP – cochlear partition

cSPL – bandwidth compensated sound pressure level

dB – decibel

DPOAE – distortion product otoacoustic emission

EEOAE – electrically evoked otoacoustic emission

FFT – Fast Fourier transform

FIR – finite impulse response

HL – hearing level

Hz – hertz

IHC – inner hair cell

IO – input-output

kHz – kilohertz

LCR – linear coherent reflection

LL – long-latency

m – meters

mm – millimeters

ms – millisecond

  xii  
NLD – nonlinear distortion

OAE – otoacoustic emission

OHC – outer hair cell

PC – personal computer

pSPL – peak sound pressure level

PSR – probe-to-suppressor ratio

RMS – root-mean-square

SFOAE – stimulus frequency otoacoustic emission

SL – short-latency

SL1 – short-latency (occurring later-in-time)

SL2 – short-latency (occurring earlier-in-time)

SL3 – short-latency (occurring earliest-in-time)

SNHL – sensorineural hearing loss

SNR – signal-to-noise ratio

SOAE – spontaneous otoacoustic emission

SPL – sound pressure level

SSOAE – synchronous spontaneous otoacoustic emission

TBOAE – tone-burst evoked otoacoustic emission

TEOAE – transient evoked otoacoustic emission

  xiii  
  1  

INTRODUCTION

Ototacoustic emissions (OAEs) are sounds generated within the cochlea and
transmitted through the middle ear to the ear canal. This thesis is concerned with the
generation of one specific type, transient-evoked otoacoustic emissions (TEOAEs).
Specifically, the generation mechanism and the generation location within the cochlea are
examined. Chapter I provides a brief introduction to the passive and active mechanics of
the cochlea. Chapter II examines the relationship between OAEs and cochlear mechanics,
with emphasis on the generation mechanisms of the most commonly measured OAEs:
distortion product (DPOAEs), stimulus frequency (SFOAEs), and transient-evoked
(TEOAEs). Within the discussion of TEOAE generation mechanisms, attention is
directed toward those mechanisms that may account for the short-latency portion of the
emission. Two potential generation mechanisms are introduced, intermodulation
distortion and basal reflection. Chapter III describes the experimental work carried out
for this thesis to determine whether intermodulation distortion or basal reflection is the
more likely generator of the short-latency portion of the TEOAE. Related to the
generation mechanism is the generation location along the cochlear partition. Chapter IV
describes the experimental work performed to identify the generation location of the

short-latency portion of the TEOAE. A summary of the findings of this thesis and ideas
for future research are presented in Chapter V.
When an acoustic stimulus is presented to the healthy cochlea, soft sounds, known
as OAEs, can be measured in the ear canal. OAEs are generated within the cochlea and
provide insight into the mechanics of the cochlea. The generation of OAEs within the
cochlea is dependent on the cochlear amplifier; a biological process by which cochlear
vibration is enhanced, resulting in increased behavioral auditory sensitivity and frequency
discrimination. OAEs arise as a byproduct of the amplification process; therefore the
measurement of OAES provides an indirect and noninvasive means to examine cochlear

 
  2  

mechanics and cochlear health in humans. However, realization of the full utility of
OAEs likely depends on an accurate understanding of how and where they are generated.
There is still much to be learned about the various types of OAEs. This thesis focuses on
understanding how and where TEOAEs are generated.
As the name implies, TEOAEs are evoked by temporally short stimuli including
acoustic impulses, or clicks, and tone-bursts. Given their transient nature, the spectral
bandwidths of these stimuli are broad. Upon presentation to the cochlea, TEOAEs are
evoked across an equally broad frequency range. In other words, the TEOAE contains
multiple frequency components. When evoked by a low-level stimulus, each frequency
component is generated near its tonotopic place along the cochlear partition. Evidence for
generation at the different tonotopic places is revealed by the time delay, or latency, of
the different frequency components in the TEOAE. As the frequency of the TEOAE
decreases, the associated latency exponentially increases. The latency of a given
frequency in the TEOAE approximates the round-trip travel time required for
propagation of the cochlear traveling wave to its characteristic frequency location and
back to the middle ear. In this thesis, TEOAE energy with this delay behavior is referred
to as the long-latency (LL) portion of the TEOAE. When evoked by a moderate- to high-
level transient stimulus, each frequency in the TEOAE also contains energy at much

shorter latencies. In this thesis, this earlier-occurring energy is referred to as the short-
latency (SL) portion of the TEOAE.
The growth rates of the LL and SL portions are different. The magnitude of the
LL portion of the TEOAE grows very compressively as stimulus level increases. In
contrast, the SL portion grows less compressively, with growth rates approaching unity
(1 dB/dB). Both the mechanism and place of generation for the SL portion are poorly
understood. Given its short latency and the broad excitation of the cochlear partition
caused by transient stimuli, intermodulation distortion has been proposed as the likely

 
  3  

generation mechanism. An alternative mode of generation may be through coherent


reflection from a location basal to the tonotopic place.
Experiment 1 was designed to test the hypothesis that the SL portion of the
TEOAE is generated through intermodulation distortion. This was done by exploiting the
presumed dependency of the SL portion on stimulus energy at two specific frequencies,

f1 and f2 . In this experiment, TEOAEs were evoked by 2 kHz tone-bursts of different

lengths, from very short, click-like bursts to relatively long bursts that begin to
approximate pure tones. As tone-burst duration increases, the bandwidth of the stimulus
narrows and the spatial excitation of the cochlear partition induced by the stimulus at
frequencies f1 and f2 decreases. Accordingly, if the SL portion of the TEOAE is
generated through intermodulation distortion, its magnitude should become smaller as the
tone-burst increases in duration. The alternative hypothesis, generation through basal
reflection, predicts contrasting behavior. If generated through basal reflection, the SL
portion depends primarily on the level of stimulus energy at the tone-burst center
frequency (2 kHz, in this case) and should therefore remain unchanged in amplitude as
the tone-burst duration increases and bandwidth becomes increasingly narrow around 2
kHz.
Experiment 2 was designed to identify the region along the basilar membrane

where the SL portion of the TEOAE is generated. Two-tone suppression was utilized to
localize the regions along the basilar membrane contributing to generation of the different
latency portions of the TEOAE. TEOAEs were evoked by 2 kHz tone-bursts in the
presence of different suppressor stimuli that spanned a range of frequencies both higher
and lower than 2 kHz. Although the main goal of this experiment was not to determine
the generation mechanism but rather the generation place, the results were anticipated to
provide either support for or against generation through intermodulation distortion.
As a final note about the format of this thesis, Experiment 1 (Chapter III) and
Experiment 2 (Chapter IV) were each written with an eye towards publication as stand-

 
  4  

alone papers in peer reviewed journals. Therefore, in the context of this thesis, these two
chapters contain some overlapping and redundant content, particularly with regard to
portions of the background and methodology.

 
  5  

CHAPTER I
MECHANICS OF THE COCHLEA

Natural auditory perception of sound depends on the functions of the external,


middle, and inner ears. The external ear (pinna and ear canal) collects, amplifies, and
directs acoustic energy to the tympanic membrane where it is transduced into mechanical
vibrations. Within the middle ear these vibrations are amplified by the ossicles as they are
directed toward the inner ear where they will become fluid pressure waves. The
amplification provided by the ossicles is necessary to minimize the impedance mismatch
between air and the cochlear fluids. Stapes vibration against the oval window of the
cochlea completes the impedance matching process and gives rise to the cochlear
traveling wave across the cochlear partition (CP). The vibration of the CP will eventually
result in deflection of the stereocilia of the inner hair cells (IHCs) and be transduced to
electrical impulses that are interpreted by the central auditory system. Facilitating the
transduction process are the outer hair cells (OHCs). The OHCs work as biological
amplifiers and increase the response of the CP to an acoustic stimulus. The action of the
OHCs is restricted to the region along the CP corresponding to the frequency of the
acoustic stimulus. Accordingly, only the forces acting upon a subset of the sensory IHCs

are amplified. At low stimulus levels, the contribution of the OHCs to the cochlear
response is significant and the mechanics are said to be active. In contrast, at high
stimulus levels, or in the presence of sensorineural hearing loss (SNHL), the contribution
from the OHCs to the cochlear response is diminished and the cochlea is characteristic of
a passive system.
1.1 Passive Cochlear Mechanics
Békésy’s measurements of CP displacement were the first to demonstrate the
concept of a cochlear traveling wave (von Békésy, 1947; 1949; 1953). In response to the
motion of the stapes against the oval window, a pressure gradient is induced across the

 
  6  

perilymphatic scalae. As a result of the pressure differential between scalae, the CP


(basilar membrane and organ of corti) becomes displaced and vibrates at the frequency of
stapes motion. The amplitude and phase lag (relative to stapes motion) of CP
displacement is not constant across the partition’s length. Amplitude gradually increases
until a maximum is reached at a given spatial location on the CP, the characteristic
frequency (CF) location; beyond this region displacement rapidly diminishes. Phase lag
similarly increases until CP displacement diminishes. It is this longitudinal pattern of
increasing CP displacement and phase lag that describes the cochlear traveling wave.
The location along the CP where the traveling wave’s displacement is greatest
depends on the frequency of stapes vibration. Békésy demonstrated that as the
stimulating frequency increases, the location of maximum displacement along the CP
shifts basally. This observation has resulted in the formulation of frequency-place maps
describing the tonotopic arrangement of the CP (Greenwood, 1961; 1990).
The tonotopic organization of the CP is a direct consequence of the partition’s
graded impedance. Compared to its apex, the base of the basilar membrane is relatively
thick and narrow resulting in very low compliance and impedance that is stiffness
dominated. As the distance from the stapes increases, the basilar membrane decreases in
thickness but increases in width. This effectively increases the compliance of the cochlear

partition while maintaining nearly constant mass. As a result, impedance becomes mass
dominated. It is this change in impedance from the cochlear base to apex that causes each
location along the partition to have a unique resonant frequency.
Békésy’s traveling waves were characteristic of a highly damped response in that
the peak of the wave was relatively broad and poorly localized along the CP. These data
conflicted with neural tuning curves and behavioral frequency discrimination (Shower &
Biddulph, 1931; Evans, 1972; 1975), both of which demonstrated high degrees of tuning.
It has since been demonstrated that although Bekesy’s measurements were accurate, they
were not representative of the traveling wave induced by low-level tones in a living

 
  7  

cochlea (Rhode, 1971; Sellick et al., 1982; Johnstone et al., 1986; Rhode, 2007).
Békésy’s measurements, which were made in dead (or physiologically compromised)
cochleae using high-level tones (130 dB SPL), characterized the passive mechanics of the
cochlear partition.
1.2 The Cochlear Amplifier
The traveling wave in the healthy cochlea is shaped by the action of the cochlear
amplifier, a coin termed by Davis (1983) to refer to the mechanism through which the
damping forces of the cochlear fluids are overcome (Neely & Kim, 1983; for review see
de Boer, 1996 and Neely & Kim, 2008). The existence of a similar mechanism was
originally proposed by Gold (Gold & Pumphrey, 1948; Gold, 1948) during the same time
as Békésy’s original measurements. Gold suggested that it was inaccurate to think of the
cochlea as simply a passive detector of sound energy. Shortly after Kemp’s initial
discovery of evoked otoacoustic emissions (OAEs; Kemp, 1978), Kemp (1979a) and
Wilson (1980) both reported that some OAEs occurred spontaneously and could not be
explained by a passive cochlea. It is now well established that the displacement of the CP
is a result of both passive and active mechanisms, the latter of which are associated with
the cochlear amplifier.
The OHC is the likely structure through which the cochlear amplifier enhances

the CP response. For instance, loss of the OHCs is known to reduce behavioral sensitivity
and frequency selectivity while at the same time causing abnormal loudness growth, all
of which are mediated through the action of the cochlear amplifier (discussed below).
Moreover, the OHC is electromotile presumably through the membrane protein prestin
(Zheng et al., 2000; Liberman et al., 2002; Cheatham et al., 2004; Dallos et al., 2008;
Santos-Sacchi et al., 2006), such that depolarization shortens the cell while
hyperpolarization lengthens the cell (Brownell et al., 1985). The forces generated by
these length changes have been shown to be sufficient to overcome the damping forces of

 
  8  

the cochlear fluids and amplify the CP response (Iwasa and Chadwick, 1992).
Accordingly, the OHC is widely considered to be the location of the cochlear amplifier.
The action of the cochlear amplifier occurs within a feedback system whereby
cochlear partition displacement activates the amplifier and is adjusted by the amplifier
(Patuzzi & Robertson, 1988; Hubbard 1993; Geisler, 1998). In short, lateral displacement
of the OHC stereocilia by the motion of the CP induces current flow within the OHC
(either positive or negative, depending on the direction of ciliary deflection). The current
generates a voltage potential across the basolateral membrane of the OHC, which then
causes conformational changes to the motor protein prestin, either lengthening or
shortening the OHC. The forces generated by the OHC length changes act upon the
cochlear partition to reduce the damping of the cochlear fluids and amplify displacement.
The current generated through ciliary deflection has been modeled by saturating
nonlinearities including the hyperbolic tangent and Boltzmann functions (Weiss &
Leong, 1985; Kros & Richardson, 1992; Lukashin & Russell, 1998; Bian et al., 2002).
The receptor potential generated by this current is therefore also nonlinear. As a result of
this nonlinearity, the forces generated through OHC length changes saturate at higher-
levels of stimulation (Santos-Sacchi, 1992; Evans and Dallos, 1993; for review see
Patuzzi, 1996). Furthermore, as the stimulation level increases, the forces generated by

the OHCs become less effective in overcoming the damping forces of the cochlear fluids.
In other words, the gain provided by the cochlear amplifier decreases as input level
increases.
1.3 Active Cochlear Mechanics
Measurements of CP displacement in living cochlea establish the role of the
cochlear amplifier in auditory perception. In the healthy cochlea, displacement of the CP
is amplified by 50 – 80 dB, relative to stapes displacement, for low-level inputs (Robles
et al., 1986; Gorga et al., 2003; Rhode, 2007). The gain provided by the cochlear
amplifier can thus account for behavioral detection of very soft sounds. Moreover, the

 
  9  

low-level gain provided by the amplifier is restricted to a narrow region at and


immediately basal to the CF location. The result is partition displacement that is highly
localized to a specific spatial location that depends on the frequency of stimulation. The
tuning afforded by this localized amplification is similar to that of auditory neurons
(Robles et al., 1986; Khanna & Leonard, 1986; Ruggero et al., 1997; Ren & Nuttall,
2001) and is likely the origin of behavioral frequency discrimination.
The level and frequency dependent gain provided by the cochlear amplifier results
in compressive growth of CP displacement (Rhode, 1971; Rhode, 1978; Sellick et al.,
1982; Johnstone et al., 1986; Ruggero et al., 1997; Rhode, 2007). Accordingly, the
healthy cochlea is nonlinear. In addition to compression, some of the other nonlinear
behaviors exhibited by the cochlea include biasing (Sellick et al., 1982; Patuzzi et al.,
1984; Bian et al., 2002), two-tone suppression (Geisler et al., 1990; Cooper & Rhode,
1992), and generation of distortion (harmonic or intermodulation; Robles et al., 1991;
Patuzzi, 1996; Rhode, 2007).
Cochlear nonlinearity is most pronounced at and near the CF location where the
contribution from the cochlear amplifier is greatest (de Boer, 1983a; 1983b).
Compression at the CF place has been shown to occur for input levels as low as 0 dB SPL
(Rhode, 2007). Other data is more consistent with near linear growth at the CF place for

input levels up to ~ 40 dB SPL (Ruggero et al., 1997). As the input level increases
beyond 40 dB SPL, compression becomes increasingly pronounced with growth rates
between 0.1 – 0.3 dB/dB (Ruggero et al., 1997; Rhode, 2007). Compression also occurs
at spatial locations basal to the CF place; however, the amount of compression is reduced.
By approximately 1/2-octave basal, basilar membrane growth is nearly linear (Ruggero et
al., 1997; Robles & Ruggero, 2001). As stimulus level increases, the forces generated by
the OHCs are insufficient to overcome the energy loss associated with the traveling wave
approaching its resonance place. In other words, the mechanics shift from active to
increasingly passive. As a result, the peak of the traveling wave shifts basally (Ruggero et

 
  10  

al., 1997; Recio et al., 1998) to regions just beyond the resonance place. The amplitude of
the peak continues to grow but at less compressive rates than those for lower stimulus
levels. Accordingly, the interplay between mechanics presumably accounts for the broad
dynamic range over which perception occurs (~120 dB; for review see Oxenham &
Bacon, 2003).
In summary, the sound induced vibrations of the cochlear partition are a
combination of passive and active mechanics. The active mechanics are attributable to
the action of the cochlear amplifier in the OHCs and result in cochlear nonlinearity. This
nonlinearity is beneficial to audition and allows for detection of very soft sounds,
discrimination between similar sounds, and perception of sound across a broad range of
levels. Although the active mechanics of the living cochlea in human are unobservable
due to the invasiveness of the measurements, a byproduct of the active mechanism,
OAEs, can be used to provide “far-field” insight into human cochlear processing.

 
  11  

CHAPTER II

OTOACOUSTIC EMISSIONS

In 1978 David Kemp demonstrated that the cochlea generates an “echo” when
presented with an acoustic stimulus (Kemp, 1978). In his original work, Kemp presented
acoustic impulses to the ear and observed that the decay of the ear canal pressure
response was not monotonic. Rather, several milliseconds after stimulus cessation a small
pressure peak was evident that was not present in either a coupler or ears with middle ear
pathology. He hypothesized that the echo, or otoacoustic emission (OAE), was a
byproduct of Gold’s active mechanism (Gold, 1948) within the cochlea and that the OAE
could provide insight into cochlear mechanics. The discovery of OAEs measured in the
absence of external acoustic stimulation, spontaneous otoacoustic emissions (SOAEs),
provided further support of an active cochlea (Kemp, 1979a; Wilson, 1980).
Since the earliest discoveries by Kemp and later, Wilson, OAEs have garnered
considerable interest. OAEs are now viewed as being byproducts of the cochlear
amplifier. The relation between OAEs and active cochlear mechanics (reviewed below)
has been an area of special attention since measures including single-nerve fiber
recordings and interferometer measures of basilar membrane vibration are invasive and
unattainable in humans. If OAEs, which are non-invasive, objective measurements and
easily measured in humans, provide equivalent information, then greater knowledge of
the underlying mechanisms responsible for human audition may be realized. Such
knowledge would certainly be beneficial in the diagnosis of and differentiation between
hearing pathologies. Prior to further discussion of OAEs and cochlear mechanics it is
worthwhile to review the different types of OAEs.

2.1 OAE Classification


The current naming conventions used to classify OAEs is somewhat ambiguous.
At the broadest level, OAEs can be either evoked by external stimuli (evoked OAEs) or

 
  12  

can arise spontaneously (Kemp, 1979a; Wilson, 1980). There are two types of
spontaneously generated OAEs including SOAEs and synchronous-spontaneous OAEs
(SSOAEs). The former were defined above and are found in approximately 60 - 80% of
all healthy ears (Martin et al., 1990; Talmadge et al., 1993; Penner & Zhang, 1997).
SSOAEs are emissions that are elicited by acoustic stimulation but then may persist up to
hundreds of milliseconds after stimulus cessation before decaying. Both SOAEs and
SSOAEs may originate from instability in the active feedback mechanism (Gold, 1948;
Camalet, 2000; Sisto et al., 2001; Martin et al., 2001) and/or the creation of standing
waves within the cochlear cavity (Kemp, 1979a; 1979b; Talmadge & Tubis, 1993; Shera,
2003).
Evoked OAEs can be elicited by either electrical stimuli (EEOAEs; Hubbard &
Mountain, 1983; Ren & Nuttall, 1996) or acoustic stimuli. Acoustically-evoked OAEs
have traditionally been classified according to characteristics of the stimulus. The most
commonly used stimuli are short transients or long pure tones. Emissions evoked by
relatively short-duration stimuli may be referred to generally as transient-evoked OAEs
(TEOAEs). They may also be given more specific names, such as clicked-evoked OAEs
(CEOAEs) when the stimuli are acoustic impulses (clicks) or tone-burst OAEs
(TBOAEs) when the stimuli are short, ramped tone bursts. Emissions evoked by

relatively long-duration sinusoidal stimuli are referred to as stimulus frequency OAEs


(SFOAEs) when the stimulus is a single tone or distortion-product OAEs (DPOAEs)
when the stimuli are pairs of tones.
Of course, acoustically-evoked OAEs can be elicited by any acoustic stimulus,
including those that fall somewhere in between short transients and long pure tones. For
example, frequency glides have been explored as stimuli (Neumann et al., 1994; Kalluri
& Shera, 2013), as have long bursts of white noise (Maat et al., 2000). In this thesis,
some of the stimuli used to elicit OAEs are of medium length, falling in between the

 
  13  

space between short transients and long tones. In all of these cases, the traditional naming
classification system based on stimulus type is not particularly useful.
More recently, an attempt was made to reclassify OAEs based on their generation
mechanisms (Shera & Guinan, 1999). According to this classification scheme, SFOAEs
and TEOAEs are considered reflection-source emissions since they are thought to arise
through reflection of the forward-moving traveling wave off impedance irregularities
along the basilar membrane (discussed further in subsequent sections). In contrast to SF
and TEOAEs, DPOAEs are classified both as a reflection-source and a distortion-source
OAE. DPOAEs contain two components (discussed further in subsequent sections); the
distortion-source component is generated through nonlinear distortion in the cochlear
mechanics and the reflection-source is generated through the same mechanism as SF and
TEOAEs. Unfortunately, this classification scheme has proved less useful than initially
hoped, because some OAEs appear be generated by more than one mechanism,
depending on stimulus level. Further, referring to all OAEs as being one of two types
does not give enough information about how they were elicited. At this time, it appears
that the practice of classifying OAEs by the eliciting stimuli will continue, with the
addition of specifying generation mechanism when that information is deemed useful.
2.2 OAEs and Cochlear Mechanics

The association between OAEs and the cochlear amplifier and, therefore, active
cochlear mechanics, is attested to through the observation that OAEs are either
diminished or absent in ears with sensorineural hearing loss (SNHL; Probst et al., 1987;
Collet et al., 1989; Smurzynski et al., 1990). Correlation between the behavioral
audiogram and OAE magnitude has also been demonstrated and implicates OAEs as
frequency specific indicators of cochlear amplifier integrity (Collet et al., 1991; Boege &
Janssen, 2001; Mertes & Goodman, 2013). Unfortunately, attempts to predict threshold
from OAEs have shown limited success (Avan et al., 1991; Hussain et al., 1998; Dorn et
al., 1999; Gorga et al., 2003). The limited utility of OAEs in predicting threshold may

 
  14  

result from several factors. First, there is evidence that the OAE of a particular frequency
(TEOAE, SFOAE, or DPOAE) is generated across a broad region of the basilar
membrane (Siegel et al., 2005; Johnson et al., 2007; Withnell et al., 2008; Goodman et
al., 2011; Moleti et al., 2013; Martin et al., 2013) and; therefore, the OAE may not be as
frequency-specific of a measure compared to behavioral threshold. Second, behavioral
threshold reflects both peripheral and central auditory processing. In contrast, OAEs, as
typically measured, likely reflect only the integrity of the OHC. In light of the limitations
in predicting hearing loss from OAEs, the clinical utility of OAEs has been relegated to
the detection of hearing loss in infants, young children, and other individuals who are
unable to perform more advanced behavioral measures.
Despite the limited usefulness of evoked OAEs in predicting auditory threshold,
OAEs are able to provide insight into various aspects of active cochlear mechanics. For
instance, the tonotopic organization of the cochlear partition is demonstrated by OAE
latency (the occurrence in time of the OAE relative that of the evoking stimulus). As the
frequency of the evoking stimulus increases, the latency of the OAE decreases (Kemp,
1978; Wilson, 1980; Norton & Neely, 1987; Tognola et al., 1997; Shera & Guinan, 2003;
Sisto et al., 2007). The decrease in OAE latency is consistent with the traveling wave
propagating longer distances across the cochlear partition as frequency decreases.

OAE latency has also been used as a measure of cochlear tuning (Shera et al.,
2002; Shera & Guinan, 2003; Moleti & Sisto, 2003). If the cochlear partition (CP) is
modeled as a series of minimum-phase, bandpass filters (Zweig, 1976), the bandwidth of
each filter will be inversely proportional to its group delay. In cases of SNHL, damage to
the OHCs (or the organ of Corti, more generally speaking) results in a decrease in the
amount of gain afforded by the cochlear amplifier and the bandwidths of the cochlear
filters increase. OAEs reflect this loss of tuning in that OAE latencies in ears with SNHL
tend to be shorter than latencies in ears with normal hearing (Konrad-Martin & Keefe,
2005; Keefe, 2012; see Don et al., 1998). In addition to depending on the integrity of the

 
  15  

cochlear amplifier, cochlear gain is also level dependent such that greatest gain is
provided for low-level stimuli. As stimulus level increases, the gain provided by the
cochlear amplifier decreases and the excitation pattern along the CP broadens (Ruggero
et al., 1997; Rhode & Recio, 2000; Rhode, 2007). Not unexpectedly, the latency of OAEs
evoked by low-level stimuli is longer than that of OAEs evoked by higher-level stimuli
(Neely et al., 1988; Schairer et al., 2006; Sisto & Moleti, 2007).
OAE magnitude has been used to assay active mechanics including cochlear
amplifier gain and tuning. This has been achieved through OAE suppression experiments
and the formulation of OAE tuning curves (Brass & Kemp, 1993; Abdala et al., 1996;
Abdala, 2001; Gorga et al., 2003; Gorga et al., 2011; Keefe et al., 2008). When two tones
are presented to the cochlea the displacement at each tone’s characteristic frequency (CF)
location is suppressed by the presence of the other tone (Geisler et al., 1990; Ruggero,
1992; Ruggero et al., 1992; Geisler & Nuttall, 1997). Suppression is greatest in the region
of active mechanics – at and immediately basal to the tonotopic place (de Boer, 1983a;
1983b; Brass and Kemp, 1993; Killan et al., 2012). OAE amplitude is similarly subject to
the effects of suppression and has been used to examine the effect of SNHL on cochlear
tuning. For instance, OAE tuning curves have been measured in ears with normal hearing
and ears with hearing loss. The latter demonstrate reduced cochlear gain compared to the

former and are consistent with diminished action of the cochlear amplifier (Gorga et al.,
2003).
Yet another aspect of active mechanics that may be examined from OAEs is the
nonlinear growth of CP displacement. Similar to the CP response, OAE magnitude
increases nonlinearly with stimulus level (Grandori, 1985; Probst, et al., 1986; Dorn et
al., 2001; Schairer et al., 2003). At low stimulus levels the OAE grows nearly linearly. As
stimulus level increases, the growth of the OAE becomes increasingly compressive. This
basic pattern of linear transitioning into compressive growth resembles that of the CP at a

 
  16  

given measurement location in response to its CF (Robles, et al., 1986; Ruggero et al.,
1997; Rhode & Recio, 2000).
The potential usefulness of measuring OAEs to assay active cochlear partition
mechanics depends on an accurate understanding of how and where the OAE is
generated. A well-known example of this is illustrated by using DPOAEs versus SFOAEs
to predict hearing loss at a given frequency. Although the frequency of these OAEs might
be identical, 2 kHz for instance, they are generated at different locations along the
cochlear partition. The 2 kHz DPOAE is generated near the 3.1 kHz spatial location
while the 2 kHz SFOAE is generated near the 2 kHz spatial location. Accordingly, it
would be inaccurate to try to predict the 2 kHz behavioral threshold using a 2 kHz
DPOAE.
Despite the large body of research concerning the generation mechanisms and
locations of OAEs, knowledge gaps still exist. The following sections review what is
known regarding the generation mechanisms of OAEs and highlights inconsistencies in
the literature. Many of these inconsistencies can be traced to incomplete understanding of
the generation of TEOAEs and SFOAEs and form the basis for the experiments described
in the following chapters.
2.3 Generation Mechanisms

2.3.1 Underlying Theories


The two most common theories of OAE generation are linear coherent reflection
(LCR) and nonlinear distortion (NLD; Zweig & Shera, 1995; Shera & Guinan, 1999).
LCR depends on the presence of randomly distributed impedance irregularities across the
basilar membrane. In other words, the basilar membrane is not smooth but rough. As the
cochlear traveling wave moves along the cochlear partition, it encounters the impedance
discontinuities and portions of the wave are reflected back toward the stapes. Majority of
these backscattered wavelets interact destructively with each other due to phase
differences; however, wavelets reflected near the peak of the traveling wave interact

 
  17  

constructively and can be measured as an OAE in the ear canal (Zweig & Shera, 1995).
The peak of the traveling wave is optimal for constructive interaction between reflected
wavelets because of its large amplitude, broad spatial width, and constant wavelength (for
review see Shera & Guinan, 2008). The term place-fixed (Kemp, 1986) is frequently used
to refer to the LCR mechanism since the impedance discontinuities responsible for the
OAE are a physical characteristic of the basilar membrane.
Nonlinear distortion OAE generation is founded upon cochlear nonlinearity. The
traveling wave induces distortion at and around the spatial location of its peak where the
cochlear response is highly nonlinear. The distortion generates a reverse-traveling
wavelet of the same frequency as the evoking stimulus. If the wavelet is of sufficient
amplitude, an OAE may be measured in the ear canal. Additionally, if multiple, closely
spaced tones are simultaneously presented to the ear, intermodulation distortion occurs,
resulting in wavelets at frequencies corresponding to integer combinations of the evoking
stimulus frequencies. These wavelets can similarly be measured in the ear canal as OAEs.
The term wave-fixed (Kemp, 1986) is frequently used to refer to the NLD mechanism
since the OAE is generated by a distortion source induced by the cochlear traveling
wave(s).
LCR and NLD each predict contrasting phase behavior for the OAE across

frequency, so that the phase of the OAE can be used to determine its generation
mechanism. The phase behavior of OAEs generated through NLD depends on the
cochlea being scaling-symmetric (Rhode, 1971; Zweig, 1976). In a perfectly scaling-
symmetric cochlea, the traveling wave rotates through the same number of cycles to
reach its CF place, regardless of frequency. Consider two OAEs, POAE1 and POAE2,
evoked by non-simultaneously presented pure tones, f1 and f2 ( f2 > f1 ), respectively: The
cochlear traveling waves for f1 and f2 both rotate through the same number of cycles to
their characteristic locations on the basilar membrane (BM), x1 and x2, respectively,
despite x1 being a further distance from the cochlear base than x2. At each site, the

 
  18  

mechanics are nonlinear and a distortion source OAE is generated. Because the traveling
waves for the two tones accumulated the same amount of phase delay by the time they
reached their characteristic locations and induced distortion, the generated OAEs also
have the same phase, !POAE1 " !POAE 2 (Shera & Guinan, 1999).
NLD similarly predicts constant phase across frequency for OAEs generated by
intermodulation distortion. For two pure tones simultaneously presented to the cochlea,
intermodulation distortion occurs at the cochlear location corresponding to maximum
overlap of the two tones’ traveling waves (Hall, 1974; Gaskill & Brown, 1990). The
distortion source generates forward and reverse traveling waves. The reverse traveling
wave is measured as a distortion-source OAE. If the ratio of the two pure tones is
maintained, such that the spatial distance on the basilar membrane is held constant, and
the frequencies are swept, the different frequency distortion-source OAEs will have
nearly identical phases (Shera & Guinan, 1999; Knight & Kemp, 2001).
In contrast to NLD generation, the OAE resulting from coherent reflection does
not originate from a source induced by the wave but rather through reflection of the wave
off pre-existing irregularities in the BM mechanics. The phase of the emitted emission
thus corresponds to the time-delay imposed by the round-trip travel time of the cochlear
traveling wave from the stapes to the characteristic frequency spatial location (Shera &

Guinan, 1999). Consistent with the tonotopic arrangement of the basilar membrane, the
phase of the OAE is expected to rotate rapidly across frequency.
The phase predictions of both theories imply different OAE latencies when
latency is quantified in terms of group delay ( ! ),

""
! =! , Eq. 2.1
"#

where ! is in seconds and the quantity !"! "" is minus the first derivative of
unwrapped phase with respect to radian frequency. Since the phase of the OAE across

 
  19  

frequency is predicted to be constant for NLD generation, the expected group delay is
approximately zero milliseconds, ! NLD ! 0 milliseconds. In practice, group delay is small,
but non-zero due to the cochlea not being perfectly scaling symmetric. In contrast, LCR
theory predicts an OAE group delay approximately twice the basilar membrane group
delay, ! LCR ! 2! BM (Shera & Guinan, 1999; Shera & Guinan, 2003; see Shera et al.,
2008). Accordingly, the generating mechanism of a particular OAE can be inferred from
the phase response.
2.3.2 Generation of DPOAEs
There is general consensus that the DPOAE includes contributions from both
NLD and LCR generators (Shera & Guinan, 1999; Talmadge et al., 1999; Knight &
Kemp, 2001; Kalluri & Shera, 2001; Konrad-Martin et al., 2001; Long et al., 2008;
Vetešník et al., 2009). The observation that the simultaneous presentation of two tones, f1
and f2 ( f2 > f1 ), causes the generation of a third tone, fDP, necessarily implies a nonlinear
mechanism. Nonlinear system theory would suggest that distortion is most pronounced
when the input to the underlying nonlinearity is greatest. In the cochlea, this presumably
occurs in the region where the traveling waves of the two primary tones maximally
overlap – near the f2 tonotopic place (Hall, 1974; Gaskill & Brown, 1990; Martin et al.,
1998; Talmadge et al., 1999; Knight & Kemp, 2000; Gorga et al., 2011; but see Martin et

al., 1987; Knight & Kemp, 2001; Martin et al., 2013 for nonlinear interactions basal to
the f2 place). The distortion induced at f2 generates both forward- (toward the cochlear
apex) and reverse-traveling (toward the cochlear base) wavelets at various
intermodulation frequencies. In mammals, the dominant intermodulation distortion
product is 2 f1 ! f2 ( f2 f1 = 1.22 ). In the case of 2 f1 ! f2 distortion, the reverse-traveling
wavelet is transduced into an acoustic pressure in the ear canal and can be measured as an
OAE. This particular component of the DPOAE is commonly referred to as a distortion-
source OAE, given its nonlinear origin. Consistent with NLD generation theory and a

 
  20  

scaling-symmetric cochlea, the distortion-source OAE exhibits nearly constant phase as


the primary frequencies are swept at a fixed f2 f1 ratio (Knight & Kemp, 2001).
The linear coherent reflection component of the DPOAE is generated as a result
of the nonlinear-induced forward-traveling wave. As this wave propagates to its tonotopic
place, fDP, it encounters impedance irregularities in the cochlear mechanics and is
partially reflected toward the stapes. Majority of these reflected wavelets interact
destructively with each other, with the exception of those generated at the peak of the
traveling wave. These reflections near the peak coherently sum to form a reverse-
traveling wavelet that is eventually measured in the ear canal as an OAE. In contrast to
the distortion-source component, the phase of this reflection-source component rotates
rapidly with frequency (Shera & Guinan, 1999).
The relative contributions of the distortion and reflection components to the total
DPOAE primarily depend on the frequency ratio of the evoking tones ( f2 f1 ) and the
specific intermodulation distortion product being considered ( 2 f1 ! f2 , 2 f2 ! f1 , etc.).
When both sources contribute equally, the DPOAE spectrum exhibits spectral peaks and
nulls depending on the phase relationship between the components (Heitmann et al.,
1998; Talmadge et al., 1999; Mauermann et al., 1999). For the low-side 2 f1 ! f2 cubic
distortion product, the reflection component is dominant for primary frequency ratios

between 1.0 – 1.1. For larger ratios (1.1 – 1.35), the distortion component tends to
dominate (Knight & Kemp, 2001; Dhar et al., 2005). The contribution to the high-side

2 f2 ! f1 DPOAE is dominated by the reflection component across all frequency ratios


from 1 – 1.35 (Knight & Kemp, 2001). Although both reflection and distortion generation
mechanisms contribute to the DPOAE, the reflection component exists only because of a
distortion source induced in the region of spatial interaction between the traveling waves
of the primary frequencies. In other words, the reflection component would not be
generated without distortion first occurring.

 
  21  

2.3.3 Generation of SFOAEs


SFOAEs, when evoked by low-level tones, are generally consistent with
generation through a place-fixed mechanism at the peak of the cochlear traveling wave,
namely, LCR (see Lichtenhan, 2012). For stimuli at or below 40 – 50 dB SPL, SFOAE
phase rotates rapidly with frequency, and group delay approximates twice that of the
basilar membrane (Shera & Guinan, 2003; Goodman et al., 2003; Schairer et al., 2006;
Kalluri & Shera, 2007; see Shera et al., 2008). Similarly, two-tone suppression
experiments have demonstrated that greatest suppression of the low-level SFOAE is
achieved by introducing a second tone of similar frequency to that used to evoke the
OAE (Brass & Kemp, 1993; Keefe et al., 2008). The presence of the suppressor tone is
thought to saturate the nonlinear elements (OHCs) at and immediately basal to the
tonotopic place (de Boer, 1983a; 1983b; Brass & Kemp, 1993) which, in the absence of
the suppressor, determine the shape of the cochlear traveling wave that gives rise to OAE
generation through LCR.
Despite the evidence implicating LCR as the generation mechanism of SFOAEs,
it may not be the only mechanism. Siegel et al. (2005) reported very short SFOAE group
delay latencies (in chinchilla) that were inconsistent with generation through LCR at the
peak of the traveling wave. They proposed that the SFOAE is generated through a

distributed region of distortion basal to the peak of the cochlear traveling wave. Several
two-tone suppression experiments have seemingly supported a broader region of SFOAE
generation. Both Guinan (1990) and Siegel et al. (2004) measured suppression of the
SFOAE by frequencies much higher than that of the SFOAE. However, Shera and
colleagues (2004) countered these findings by demonstrating that introducing a much
higher frequency tone than that used to evoke the SFOAE may actually induce distortion
product OAEs and give the illusion of a suppressive effect. Accordingly, the two-tone
suppression results showing suppression by frequencies much higher than that of the
SFOAE should be viewed with caution.

 
  22  

Concerning the findings of Siegel et al. (2005) demonstrating very short SFOAE
latencies, Shera and colleagues (Shera et al., 2008) analyzed the same data and concluded
that the SFOAE is composed of two components that differ in latency. The short-latency
(SL) component exhibits a shallow-phase gradient while the long-latency (LL)
component has phase that changes rapidly with frequency. The phase response of the LL
component is well predicted by cochlear models incorporating LCR from the peak of the
traveling wave; however, this same mechanism cannot explain the existence of the SL
component.
In contrast to the theory proposed by Siegel et al., Shera and colleagues suggested
a nonlinear coherent reflection mechanism was responsible for the SL SFOAE
component. This reasoning was based on a theory proposed earlier by Talmadge and
colleagues (2000) to account for SFOAE spectral fine structure. According to this theory,
coherent reflection occurs not only through a linear mechanism but also through wave-
fixed distortion. Consistent with a nonlinear distortion mechanism (albeit coherent
reflection), the phase-gradient of the OAE is expected to be very shallow (Shera &
Guinan, 1999; Talmadge et al., 2000).
SFOAE data from humans also suggest that multiple components may contribute
to the total SFOAE. Schairer et al. (2003) observed notches in SFOAE growth functions.

Non-monotonic growth is not predicted if the emission results solely from linear
reflection at a fixed location on the basilar membrane. In contrast, DPOAE growth
functions are known to exhibit notches both as a result of the nonlinearity of the
mechanoelectric transducer function from which DPOAEs originate (Lukashkin &
Russell, 1998; 1999) and destructive interaction between the distortion-source and
reflection-source emissions (Heitmann et al., 1998; Talmadge et al., 1999; Mauermann et
al., 1999;). A qualitatively similar interference between different components may occur
for SFOAEs. Most recently, evidence for distinct SFOAE components in humans was

 
  23  

reported by Sisto et al. (2013; see Moleti et al., 2012b) who, in similar fashion to Shera
and colleagues (2008), “unmixed” the total SFOAE into SL and LL components.
Choi et al. (2008) presented what could be considered a modified LCR generation
theory that may account for short latency SFOAEs (Siegel et al., 2005), non-monotonic
SFOAE magnitude growth (Schairer et al., 2003), and SFOAE spectral fine structure
(Talmadge et al., 2000). They modeled the SFOAE as consisting of contributions from
linear coherent reflection occurring at the traveling wave peak and at regions basal to the
peak (similar to Siegel et al., 2005). If the reflected wavelets from the two regions are of
equal magnitude and anti-phase, a null in both the growth function and spectral
distribution will occur. As suggested by Siegel et al. (2005), with such a mechanism,
overall SFOAE group delay latency would depend on the component with the largest
magnitude.
2.3.4 Generation of TEOAEs
Similar to SFOAEs, TEOAEs contain both SL and LL portions (Carvalho et al.,
2003; Withnell & McKinley, 2005; Withnell et al., 2008; Goodman et al., 2009;
Goodman et al., 2011; Moleti & Sisto, 2012a). In this thesis, these portions will
sometimes be referred to as the SL TEOAE and the LL TEOAE. Both the SL and LL
TEOAEs have similar frequency content but occur at different moments in time. These

different portions of the TEOAE also exhibit different growth rates. The SL TEOAE
exhibits less compressive growth than the LL TEOAE and is generally evoked only for
higher-level stimuli (Withnell et al., 2005; 2008; Goodman et al., 2011; Moleti et al.,
2012a). The majority of studies examining TEOAEs have focused only on the LL
portion, perhaps unknowingly. This is due, in part, to traditional TEOAE analysis
techniques wherein the first 5 milliseconds of the response are discarded due to excessive
contamination from stimulus artifact (Kemp et al., 1986). The SL portion often occurs
within this timeframe and is thus excluded from analysis.

 
  24  

2.3.4.1 The Long-Latency TEOAE


The phase response and amplitude growth of the LL TEOAE are strikingly similar
to those of both the LL SFOAE component (Kalluri & Shera, 2007) and the reflection-
source DPOAE component (Knight & Kemp, 1999). This similarity implies a shared
generation mechanism – linear coherent reflection (see earlier discussion). Consistent
with LCR generation, two-tone suppression experiments and measurements of the latency
of the LL portion of the TEOAE both demonstrate that this portion of the emission
originates at its tonotopic location on the basilar membrane (Neely et al., 1988;
Tavartkiladze et al., 1994; Zettner & Folsom, 2003; Harte et al., 2009; Goodman et al.,
2009).
Data demonstrating that LL TEOAEs adhere to the principle of superposition
further supports generation through a linear mechanism, namely, LCR. Norton and Neely
(1987) reported good correlation between the spectra of the LL TEOAE and the stimulus,
indicating that no additional distortion frequencies were generated, as would happen if
nonlinear distortion were the mechanism. Xu et al. (1994) and Prieve et al. (1996)
demonstrated that the spectrum for a broadband TEOAE (including primarily the LL
portion) could be reconstructed with reasonable accuracy through summation of the
spectra from narrow-band TEOAEs (also primarily including the LL portion). Data from

Kalluri and Shera (2007) suggested that the long-latency TEOAE is simply the super-
position of multiple low-level SFOAEs.
A handful of studies have reported TEOAE behavior (presumably for the LL
portion) that is seemingly more consistent with generation through nonlinear
intermodulation distortion (Avan et al., 1995; 1997; Yates & Withnell, 1999; Withnell et
al., 2000; Carvalho et al., 2003). In guinea pig, TEOAE energy has been measured
outside of the evoking stimulus’ bandwidth (Yates & Withnell, 1999). TEOAE amplitude
and phase (human and guinea pig) have also been found to depend on the integrity of
cochlear regions basal to the TEOAE’s tonotopic place (Avan et al., 1995; 1997;

 
  25  

Withnell et al., 2000; Lucertini et al., 2002; Jedrzejczak et al., 2005). Both of these
findings are at odds with the notion that the frequency channels through which TEOAEs
are generated are independent of each other (Norton & Neely, 1987; Xu et al., 1994;
Prieve et al., 1996; Kalluri & Shera, 2007).
At first glance, the transient nature of the stimuli used to evoke TEOAEs seems to
necessitate at least some generation of the OAE through intermodulation distortion.
Presentation of an acoustic impulse to the cochlea causes broad excitation across the
basilar membrane (Recio et al., 1998; Recio & Rhode, 2000). If nearby areas were
excited simultaneously, they would necessarily induce intermodulation distortion through
the interaction of different frequency components on the nonlinear CP. Despite this, the
majority of evidence supports LCR as the generating mechanism for the LL portion of the
TEOAE. The answer to this apparent discrepancy presumably lies in the spatial-time
filtering imposed by the CP, whereby the various frequency components of a transient
stimulus do not excite the basilar membrane simultaneously (Shera, 2001).
2.3.4.2 The Short-Latency TEOAE
In contrast to the general consensus of LCR as the mechanism responsible for the
LL TEOAE, no such agreement surrounds that of the SL TEOAE. Notwithstanding the
potential limitation of cochlear spatial-time filtering, intermodulation distortion is a

plausible candidate. For instance, the phase response of the SL TEOAE rotates relatively
slowly with frequency (Withnell & McKinley, 2005; Withnell et al., 2008) and is similar
to distortion-source emissions. Moreover, the time-domain latency of the SL portion of
the TEOAE approximates cochlear model predictions for a 2 f1 ! f2 distortion-source
emission (Moleti et al., 2012a). Finally, the 2 f1 ! f2 distortion-source component grows
much more linearly than its reflection-source counterpart (Mauermann & Kollmeier,
2004), suggesting a shared generator with the more linear growing SL TEOAE (Moleti et
al., 2012a).

 
  26  

Some limited data has cast doubt on intermodulation distortion as the generator,
however. Recall that the generation place along the basilar membrane of a 2 f1 ! f2
distortion-source OAE corresponds to the f2 tonotopic location. To test whether
generation of the SL TEOAE occurs at the f2 place, Goodman et al. (2011) analyzed the
2.5 kHz TEOAE evoked in the absence and presence of a highpass noise suppressor. The
spectral bandwidth of the noise included 3.2 kHz, the f2 frequency for a 2.5 kHz 2 f1 ! f2
distortion-source OAE ( f2 f1 = 1.22 ). This noise was expected to suppress the magnitude
of the SL portion of the TEOAE if it was generated through 2 f1 ! f2 intermodulation
distortion. However, no change in the SL TEOAEs magnitude was observed when
measured in the presence of the noise suppressor, thereby suggesting generation through
a mechanism other than intermodulation distortion.
The presence of SL components in SFOAEs (Siegel et al., 2005; Schairer et al.,
2006; Shera et al., 2008; Moleti et al., 2012b; Sisto et al., 2013) may also suggest that
intermodulation is not the generating mechanism of the seemingly analogous component
in TEOAEs. For instance, to evoke the SL portion of the TEOAE, assuming it is a cubic
distortion product, stimulus energy is required at primary frequencies f1 and f2. This
condition is easily realized by using a spectrally flat, high-level, acoustic click as the
evoking stimulus. In contrast, a pure tone would not be expected to evoke a SL

intermodulation distortion component because there is minimal energy outside of the


frequency of the tone. If the SL TEOAE and SFOAE are truly analogous to each other,
intermodulation distortion cannot be the generating mechanism.
An alternative mechanism that may account for the generation of the SL portion
of the TEOAE is basal LCR (Goodman et al., 2011; Moleti et al., 2013; see Choi et al.,
2008 for SFOAEs). Several models of basal LCR have been presented in the literature,
and it is worthwhile to differentiate between them. Choi et al., modeled how linear
coherent reflection across the tail regions of the cochlear traveling wave could result in
the generation of a SL SFOAE and give rise to SFOAE spectral fine-structure.

 
  27  

Presumably, this same mechanism could also account for SL TEOAE generation. A
slightly different model of basal reflection as it applies to the SL TEOAE was discussed
by Goodman et al (2011) and later modeled by Moleti and colleagues (2013). Recall that,
as discussed earlier, linear coherent reflection describes the process through which energy
from the forward traveling cochlear wave is reflected off impedance irregularities
distributed randomly across the basilar membrane. The peak of the traveling wave
induces a region of reflection across which the reflected wavelets coherently sum and the
resulting wavelet is measured in the ear canal as a reflection-source OAE. At low
stimulus levels and for a specific frequency, the peak of the traveling wave corresponds
to the frequency’s characteristic place on the basilar membrane. However, as stimulus
level increases, the peak of the traveling wave broadens and shifts basally as the
underlying mechanics shift from active to passive. Zweig and Shera (1995) hypothesized
that this would expand the region of coherent reflection and the latency of the OAE
would decrease. Accordingly, the SL TEOAE may simply be a consequence of the
traveling wave peak shifting basally as stimulus level increases. Goodman et al. (2011)
argued that generation basal to the tonotopic place would explain both the shorter-latency
and less compressive growth of the SL portion of the TEOAE compared to the LL
portion, given the fewer phase rotations of the cochlear traveling wave and more linear

growth of the basilar membrane response, respectively (Ruggero et al., 1997; Recio et al.,
1998; Rhode, 2007).
The experiments described in the subsequent chapters further address the question
of SL TEOAE generation. Experiment 1 tests the widely accepted hypothesis that the SL
portion of the TEOAE is generated through 2 f1 ! f2 intermodulation distortion.
Experiment 2 investigates where along the cochlear partition the SL TEOAE is
generated. The findings from these experiments are expected to provide further insight
into the relationship between TEOAEs and cochlear mechanics. Such understanding may
increase the clinical usefulness of TEOAEs.

 
  28  

CHAPTER III
EXPERIMENT 1: THE EFFECT OF TONE-BURST
DURATION ON THE SHORT-LATENCY TEOAE

3.1 Introduction
Transient-evoked otoacoustic emissions (TEOAEs) are sounds generated within
the cochlea in response to external acoustic stimulation by temporally short signals such
as clicks and tone-bursts (Kemp, 1978; Wilson, 1980). Given their transient nature, these
stimuli have wide frequency bandwidths and cause excitation across a broad region of the
basilar membrane (Don & Eggermont, 1978). The resulting TEOAEs are similarly
broadband but dispersed in time, with high frequency energy occurring earlier than low
frequency energy. The relationship between TEOAE latency and frequency is such that
the time-domain latency exponentially decreases with increasing frequency, suggesting
that TEOAE frequency components are generated at or near their tonotopic locations
(Greenwood, 1961; Tognola et al., 1997; Sisto et al., 2007; Goodman et al., 2009;
Rasetshwane et al., 2013). Further evidence of the tonotopic origin of TEOAEs comes
from two-tone suppression experiments demonstrating that TEOAE frequency
components are maximally suppressed by tones of similar frequency (Tavartkiladze et al.,

1994; Zettner and Folsom, 2003).


Linear coherent reflection (LCR, Zweig and Shera, 1995; Shera and Guinan,
1999) has been proposed as the mechanism through which low-level TEOAEs are
generated. For each frequency component in the evoking stimulus, the cochlear traveling
wave propagates along the basilar membrane towards its tonotopic place, encountering
randomly-distributed impedance irregularities along the way. These irregularities result in
partial reflections of the apical-moving travelling wave. The wavelets reflected from the
peak region of the traveling wave sum coherently, and the resulting sum is measurable in
the ear canal as an OAE. Reflection from the peak region of the traveling wave accounts

 
  29  

for the time-frequency organization of TEOAEs. A hallmark sign of generation through


LCR is that the phase response of the OAE rotates rapidly with frequency (Shera and
Guinan, 1999), resulting in a group delay that decreases with increasing frequency. When
evoked by low-level stimuli, both stimulus frequency (SF) OAEs and TEOAEs show
similar patterns of decreasing group delay with frequency and are presumed to be
generated through the same LCR mechanism (Knight and Kemp, 1999; Schairer et al.,
2006; Kalluri and Shera, 2007).
Although there is general consensus for the role of LCR in TEOAE generation at
low stimulus levels, the mechanism responsible at higher-levels of stimulation is unclear.
Multiple studies have confirmed the presence of TEOAE energy at latencies much earlier
than would be expected from generation at the tonotopic location through LCR (Withnell
and McKinley, 2005; Withnell et al., 2008; Goodman et al., 2009; Goodman et al., 2011;
Moleti et al., 2012a). In these studies, bandpass filtered TEOAE waveforms exhibit two
magnitude peaks. The later occurring peak has latency consistent with generation at the
tonotopic place through LCR. This long-latency (LL) portion of the TEOAE dominates
the waveform at low stimulus levels and grows compressively with rates as low as 0.1
dB/dB (Goodman et al., 2011). As stimulus level increases, the earlier-occurring peak
emerges and grows less compressively than the LL TEOAE. This short-latency (SL)

portion of the TEOAE dominates the TEOAE waveform at high stimulus levels. The SL
and LL portions differ in latency by a factor of approximately 1.6 (Goodman et al., 2009;
Moleti et al., 2012a).
One hypothesis concerning the origin of the SL portion of the TEOAE is that it is
generated through intermodulation distortion (Withnell & McKinley, 2005; Moleti et al.,
2012a). Under this hypothesis, various frequency components of the evoking stimulus
interact within the cochlea to generate distortion-source OAEs. Intermodulation distortion
has been shown to play a role in generation of TEOAEs in the guinea pig, but it is not
clear whether this is also true in humans. For instance, TEOAE energy was measured

 
  30  

outside of the stimulus bandwidth in guinea pigs (Yates and Withnell, 1999) but not in
humans (Prieve et al., 1986; Xu et al., 1994; Konrad-Martin and Keefe, 2005). However,
nonlinear interaction between the different frequency components within transient stimuli
does occur in humans in the form of two-tone suppression (Konrad-Martin and Keefe,
2003; 2005) thereby implying that intermodulation distortion may also occur.
Furthermore, in the frequency domain, the phase response of the SL portion of the
TEOAE changes slowly with frequency, consistent with a nonlinear distortion generation
mechanism (Shera and Guinan, 1999; Knight and Kemp, 2000; Withnell and McKinley,
2005; Withnell et al., 2008).
Time-domain latency measures of the SL portion are also consistent with
generation through intermodulation distortion. For instance, latency predictions of the
cubic distortion product ( 2 f1 ! f2 , f2 f1 = 1.22 ) obtained from one cochlear model
approximate observed latencies of the SL TEOAE (Moleti et al., 2012a). Similarly,
comparing 2 f1 ! f2 DPOAE onset latency to the SL TEOAE latency at equivalent
frequencies shows relative agreement between the two (Whitehead et al., 1996; Goodman
et al., 2009). The less compressive growth of the SL portion is likewise consistent with
generation through intermodulation distortion, specifically the cubic distortion product.
Due to the compressive nonlinearity underlying basilar membrane growth, higher order

distortion products are expected to show less compressive growth than 1st-order linear
reflections (Sisto and Moleti, 2008; Sisto et al., 2013). Mechanism-specific growth is
evidenced in the differential growth rates of DPOAE components, in that the distortion-
source component grows less compressively than the reflection-source component
(Mauermann and Kollmeier, 2004; Abdala et al., 2011).
Despite the evidence cited above, other evidence has caused some to question
generation of the SL TEOAE through intermodulation distortion. First, the amplitude of
the SL TEOAE appears to be relatively insensitive to two-tone suppression effects when
the suppressor overlaps with the f2 generation site of a 2 f1 ! f2 distortion-source OAE

 
  31  

(Goodman et al., 2011). Second, SL SFOAEs have recently been measured and shown to
possess similar latencies and growth rates to the SL TEOAE (Sisto et al., 2013). The
similarity between growth rates and latencies of the SL SF and TEOAEs suggest that they
are generated through a common mechanism (see Kalluri & Shera, 2007). However,
generation of a SL SFOAE would not be expected to occur through intermodulation
distortion since stimulus energy at the requisite primary frequencies is absent in the
evoking pure tone stimulus. Accordingly, if the SL TEOAE is analogous to the SL
SFOAE, some other mechanism must be responsible for its generation.
The current study was designed to further examine the possible role of cubic
intermodulation distortion in the generation of the SL portion of the TEOAE in humans.
TEOAEs were evoked using 2 kHz tone-bursts of varying durations and levels. The
durations of the different tone-bursts were selected to span the time-frequency continuum
between a click and a pure tone to provide insight into the similarity and/or dissimilarity
between the generation mechanisms responsible for TEOAEs and SFOAEs. In order for
the 2 kHz SL TEOAE to be generated by cubic distortion, stimulus energy at primary
frequencies f1 and f2 , 2.5 kHz and 3.1 kHz, respectively, would be required. As the
duration of the tone-burst increases, the bandwidth narrows, and the stimulus energy at
these frequencies decreases. Accordingly, if generated through cubic intermodulation

distortion, the amplitude of the SL portion of the TEOAE should decrease as tone-burst
duration increases.
3.2 Methods
3.2.1 Subjects
Thirteen subjects (4 male) between the ages of 18 – 29 years old participated in
the study. All subjects had normal hearing thresholds (≤ 20 dB HL) at the octave
frequencies between 0.25 – 8 kHz in the test ear (8 right ears, 5 left ears). Data collection
was completed over the course of 1 visit lasting approximately 2 hours. To rule out

 
  32  

middle-ear pathology, 226 Hz tympanometry was performed in the subject’s test ear. The
research protocol was approved by the Internal Review Board at the University of Iowa.
3.2.2 Signal Generation and Data Acquisition
Two-channel stimuli were digitally created at a 44.1 kHz sampling rate using
custom written MATLAB (Mathworks, Inc.) software and a Windows XP (Microsoft,
Inc.) personal computer (PC). Stimuli were routed from the PC to an external 24-bit
soundcard (UltraLite-mk3 Hybrid, Mark of the Unicorn) through a USB 2.0 interface.
The resulting two-channel electrical signal was directed through a pair of earphones
(Sennheiser IE8) and transduced into acoustic pressure. The output of each earphone was
guided through silicone tubing to a receiver port of an ER10B+ probe assembly
(Etymotic Research, Inc.). The ER10B+ probe assembly was coupled to the ear canal by
an ER10-14 foam eartip (Etymotic Research, Inc.). The ear canal pressure responses to
the acoustic stimuli were transduced into an electrical signal by a miniature microphone
housed in the ER10B+ probe assembly. The microphone voltage was amplified 20 dB by
the ER10B+ preamplifier, routed to the external soundcard, and converted to its digital
equivalent at a sampling rate of 44.1 kHz. The digital signal was stored on the PC hard
drive for offline analysis. Signal presentation and data acquisition were controlled by
MATLAB using custom written software.

3.2.3 Measurement and Analysis of TEOAEs


3.2.3.1 Stimuli
TEOAEs were evoked using 2 kHz tone-bursts with durations of 3, 6, 12, and 24
cycles. Tone-bursts were generated in cosine phase and windowed using a hann-window.
Figure 3.1 shows the time- and frequency-domain representations for the different
duration tone-bursts. As tone-burst duration increases from 3 to 24 cycles, the bandwidth
of the stimulus becomes increasingly narrow. Of particular importance for this study, the
stimulus levels at 2.5 kHz and 3.1 kHz (the primary frequencies f1 and f2 for a 2 kHz
2 f1 ! f2 distortion-source OAE, f2 f1 = 1.22 ) decrease by approximately 57 dB. Across

 
  33  

the different tone-burst durations, the level ratio ( L1 − L2 , where L1 and L2 correspond to
the dB levels of f1 and f2 , respectively) remains nearly constant at 16 dB, since each
doubling of tone-burst duration decreases L1 and L2 by approximately equal amounts (19
dB). The magnitude of a 2 f1 ! f2 distortion-source OAE depends both on the level ratio
between the primary frequencies and the level of f1 (Whitehead et al., 1995a; 1995b).
Largest OAE magnitudes result when the excitation patterns across the basilar membrane
completely overlap at the f2 place (Gaskill and Brown, 1990). When L1 is high, the
largest OAE is generated when L2 = L1 . However, as L1 decreases, L2 must decrease at a
faster rate to evoke a large magnitude OAE. The reason for this dependency is due to the
amount of basilar membrane excitation at the f2 place, induced by the f1 primary,
relative to that induced by the f2 primary. For a given decrease in L1 and L2 , the
excitation caused by the f1 primary at the f2 place is more greatly reduced than the
excitation caused by the f2 primary. Accordingly, the degree to which the excitation
patterns overlap also decreases. All this is to say that the fixed level ratio across the
different tone-burst durations is not optimal for distortion-source OAE generation.
Accordingly, the likelihood of a distortion-source OAE being generated by anything
other than the 3-cycle tone-burst seems minimal. Therefore, the presence of a SL portion
in the TEOAEs evoked by the longer-duration tone-bursts would be strong evidence that

the SL portion is not generated through 2 f1 ! f2 distortion.


Tone-bursts were presented at 8 levels to examine the presence of the different-
temporal portions of the TEOAEs (SL and LL) at different stimulus levels and to
calculate latency-intensity and input-output functions. The 3-, 6-, 12-, and 24-cycle tone-
bursts were calibrated to have peak levels from 44 – 86 dB peak sound pressure level
(pSPL), 41 – 83 dB pSPL, 38 – 80 dB pSPL, and 35 – 77 dB pSPL (6 dB steps),
respectively, in a calibration cavity (see below). These levels were chosen so that the
level-per-cycle (dB re: 1 Hz bandwidth) at 2 kHz was equivalent across the different
duration stimuli. As discussed by Kalluri and Shera (2007), when comparing OAEs

 
  34  

evoked by stimuli with different durations and bandwidths it is necessary to equate the
stimulus levels between the different stimuli. In their work, a measure referred to as
bandwidth compensated SPL (cSPL) was derived and shown to account for stimulus
bandwidth differences. Specifically, the spectral characteristics (phase and magnitude) of
TEOAEs and SFOAEs were nearly identical when the different OAE types were
compared at an equivalent dB cSPL. Pilot data suggests that equating the different-
duration tone-bursts in terms of their level-per-cycle may similarly account for bandwidth
differences. Lewis and Goodman (2012) measured comparable magnitudes and latencies
for TEOAEs evoked by different-duration tone-bursts when the tone-bursts were equated
in dB level-per-cycle. Accordingly, this same technique was used in the current study.
Calculation of the scaling factors to determine the appropriate dB pSPL levels at
which the different tone-bursts would have the same 2 kHz level-per-cycle was
performed in the frequency domain. The different duration tone-bursts were scaled to
have peak amplitudes of 1 (electrical units). Tone-bursts were transformed to the
frequency domain via the Fast Fourier transform (FFT). The size of the FFT depended on
the length of the tone-burst; therefore, the number and width of the frequency bins were
different for each tone-burst stimulus. For a given tone-burst duration, the energy within
each frequency bin was divided by the bandwidth of the bins. The magnitude spectra of

the different tone-bursts were compared and scaling factors were derived relating the 2
kHz energy (dB re: 1 Hz bandwidth) of the 3-, 6-, and 12-cycle tone-bursts to that of the
24-cycle tone-burst. All tone-bursts levels, regardless of duration, are hereafter
referenced according to those of the 24-cycle tone-burst (35 – 77 dB pSPL).
Stimuli were presented using an equal-level, double-evoked, double-source
paradigm (Keefe, 1998; Keefe and Ling, 1998; Schairer et al., 2003). A stimulus buffer
was composed of three consecutive stimulus intervals ( s1 , s2 , and s1,2 ). Each interval
was 40 ms long, resulting in a total buffer duration of 120 ms. Stimulus interval s1
contained the tone-burst and was routed through channel 1 of the transducer. Stimulus

 
  35  

interval s2 contained the same tone-burst as s1 but was routed through channel 2 of the
transducer. Tone-bursts were presented simultaneously through channels 1 and 2 of the
transducer for stimulus interval s1,2 . The number of stimulus buffers presented depended
on the tone-burst level, with more presentations for lower level tone-bursts to improve the
signal-to-noise ratio (SNR) of the evoked emission. Two thousand buffers were presented
for levels between 44 – 56 dB pSPL, 1600 were presented for levels between 62 – 74 dB
pSPL, and 1200 were presented for levels between 80 – 86 dB pSPL.
3.2.3.2 Calibration
The peak amplitudes of the tone-bursts were calibrated in terms of the pressure
generated in a long, plastic tube (2 m length, 7 mm inner diameter) terminated by a steel
bearing. The long-tube calibration technique was chosen because it is insensitive to
standing-wave effects that are present in traditional in-situ dB SPL ear canal calibrations
(Sachs & Burkhard, 1972; Stinson et al., 1982; Gilman & Dirks, 1986). Additionally, the
maximum voltage drive to the transducer is strictly controlled and the possibility of
inadvertently overdriving the transducers is eliminated. This technique has been
described previously (Goodman et al., 2009). Briefly, the probe tip is placed into the long
tube and stimuli are presented and recorded repeatedly. The pressure recorded for each
stimulus presentation includes the incident pressure delivered by the probe and multiple

reflections resulting from the stimulus pressure traveling back-and-forth between the two
ends of the tube. With each round trip, the amplitude of the reflected stimulus pressure is
attenuated due to viscothermal losses. In order to isolate the incident pressure delivered
by the probe to the tube it is necessary to use a short-duration stimulus and a slow
stimulus repetition rate. If the stimulus duration is too long, the incident pressure will
overlap in time with the reflected pressure. If the stimulus repetition rate is too fast, the
incident pressure will overlap with the internal reflections of the previously presented
stimulus. The appropriate stimulus duration can be calculated from the round-trip travel
time of sound in the tube. For the 2 m long tube used in the current study, the first

 
  36  

reflection will occur at approximately 11.6 ms. The appropriate presentation rate depends
on various factors of the tube including its length, material, and how each end is
terminated (i.e., an absorptive versus reflective material). Accordingly, the presentation
rate must often be determined empirically.
One important assumption made when using the long-tube calibration is that the
acoustic surge impedance of the tube is similar to the ear canals being tested. This
impedance is determined primarily by the cross sectional area of the cavity. The average
adult ear canal is approximately 8 mm in diameter (Stinson & Lawton, 1989), close to the
7 mm diameter tube used for calibration. Another important assumption is that individual
differences in ear canal and middle ear impedances are relatively small. This assumption
appears justifiable in the present study, given that only young adults were tested and the
inclusion criteria included normal hearing and normal middle ear function.
Sixty-four repetitions of a 10 ms 2 kHz tone-burst were delivered to the
calibration tube at a rate of 1 per second and recorded. Recordings were highpass filtered
using a finite impulse response (FIR) digital filter (250 Hz cut, 256 order). Only the
initial 10 ms of each filtered recording was retained. The pressure measured in this time
window described the incident pressure delivered by the transducer to the tube and was
thus free from internal reflections occurring at the distal end of the tube and propagating

back toward the microphone. The peak amplitude of the pressure generated within the
tube was calculated. A calibration factor relating the peak electrical stimulus to that of the
acoustic pressure in the tube was calculated. The amplitudes of the electrical tone-burst
stimuli were adjusted to yield output levels (as measured in the long tube) at the dB pSPL
levels previously specified.
3.2.3.3 Analysis
The TEOAE was extracted by first dividing the ear canal pressure recordings for
each stimulus buffer into three time segments ( p1 , p2 , and p1,2 ). Each time segment

 
  37  

corresponded to each of the three stimulus intervals ( s1 , s2 , and s1,2 ). The nonlinear
differential response ( pD ) was calculated as

pD = ( p1 + p2 ) ! p1,2 . Eq. 3.1

The nonlinear differential response represents the nonlinear-growing portion of the OAE
evoked by tone-burst stimulus. Any linear growing portion of the TEOAE and stimulus is
cancelled in the calculation of pD .
The initial and final 5 ms of the nonlinear differential responses were ramped-on
and –off, respectively, using a 1/2-cycle hann window. These responses were then
highpass filtered using an FIR digital filter (250 Hz cut, 128 order). The root-mean-
square (RMS) levels of the filtered responses were calculated and subjected to an artifact
rejection algorithm (Goodman et al., 2009; see Hoaglin et al., 1983) in order to identify
recordings contaminated by high-levels of intermittent noise. The first- and third-
quartiles of the RMS levels were calculated and the interquartile range was derived.
Recordings with an RMS exceeding the third-quartile by more than 1.5 times the
interquartile range were excluded from further analysis.
Retained responses were filtered using a 1/3-octave bandpass FIR digital filter (2

kHz center frequency, 512 order). The K filtered responses were synchronously
averaged in the time domain to yield an estimate of the 2 kHz TEOAE signal ( pOAE ),

1 K
pOAE [ n ] = ! pD [ n, k ]. Eq. 3.2
K k=1

The analytic expression ( p̂OAE [ n ] ) of the TEOAE signal was calculated using the discrete
Hilbert transform (denoted by H ),

 
  38  

p̂OAE [ n ] = pOAE [ n ] + jH ( pOAE [ n ]). Eq. 3.3

From the analytic signal, the TEOAE instantaneous magnitude ( p̂OAE ), or envelope, was
calculated as

2 2
( ) (
p̂OAE [ n ] = Re p̂OAE [ n ] + Im p̂OAE [ n ] . ) Eq. 3.4

An estimate of the time-domain noise was calculated as the standard error of the
mean TEOAE instantaneous magnitude. First, the analytic signal of each nonlinear
differential response was calculated,

p̂D [ n, k ] = pD [ n, k ] + jH ( pD [ n, k ]). Eq. 3.5

The mean instantaneous magnitude ( p̂D [ n ] ) was subsequently calculated,

1 K
p̂D [ n ] = ! p̂D [ n, k ] , Eq. 3.6
K k=1

and the noise was computed as the standard error of the mean ( SEM [ n ] ),

K
1 2
(
" p̂D [n, k ] ! p̂D [n]
K !1 k=1
)
SEM [ n ] = . Eq. 3.7
K

Note that the operations used to calculate the noise are performed on the individual
buffers while the operations used to calculate the signal are performed on the mean of the
individual buffers.

 
  39  

The TEOAE signal and noise estimates were constrained to an analysis window
extending from stimulus-onset to stimulus-offset plus 10 ms. The duration of this window
was determined based on the expected latency of a low-level 2 kHz TE/SFOAE (Neely et
al., 1988; Tognola et al., 1997; Shera et al., 2002; Sisto et al., 2007). A time vector was
mapped to the analysis window with time zero corresponding to the peak of the tone-
burst stimulus. TEOAE magnitude and latency were calculated within the analysis
window for each tone-burst stimulus. Magnitude ( mOAE ) was calculated as the root-mean-
square of the windowed TEOAE envelope,

N!1
1
"( p̂ [n] ).
2
mOAE = OAE
Eq. 3.8
N n=0

This definition of magnitude is different from more common definitions performed on the
time-domain waveform of the TEOAE signal. The numeric difference between the RMS
of the TEOAE envelope (Eq. 3.8) and the RMS of the TEOAE waveform amounts to
3 dB with the former being larger than the latter. The origin of this difference is the
inclusion of the imaginary part of the analytic signal when calculating the TEOAE
envelope (see Eq. 3.4). Compared to the real part of the analytic signal, the imaginary
part has the same amplitude but differs in phase by ! 2 radians. The phase information
is discarded upon the squaring operation and intensity is effectively doubled. This
doubling exhibits itself as a 3 dB increase in the RMS of the TEOAE envelope compared
to that of the TEOAE waveform.
TEOAE latency ( ! OAE ) was also defined using the TEOAE envelope,

N!1

"( p̂ [n] t [n])


2
OAE
! OAE = n=0
N!1
, Eq. 3.9
"( p̂ [n] )
2
OAE
n=0

 
  40  

where t was the time vector for the analysis window (ms). This definition quantifies
latency as the energy-weighted mean of the time-vector and gives greatest weight to time
indices corresponding to peaks in the TEOAE envelope, where the SNR is largest. By
attributing greatest weight to high-SNR regions of the TEOAE, the variability commonly
associated with OAE latency measurements is reduced (Shera & Bergevin, 2012).
In order to determine whether a TEOAE was truly present in the analysis window,
the SNR was calculated and compared to a criterion SNR of 12 dB. The 12 dB criterion
was determined using a bootstrap technique. First, 2000, 40 ms noise frames were
generated using a normally distributed random-number generator in MATLAB. The 2000
frames were filtered using the same FIR filters applied to the ear canal pressure
recordings and the analytic signals were calculated (see Eq. 3.3). From the 2000 noise
frames, 2000 were chosen using a resample-with-replacement algorithm. From the
resampled noise frames, the mean instantaneous magnitude (see Eq. 3.4) and standard
error of the mean (see Eq. 3.7) were calculated. The former represented an estimate of the
signal and the later represented an estimate of the noise. The signal and noise estimates
were truncated to 4 different time windows corresponding to those used in the analysis of
the TEOAEs. For each time window, the SNR was calculated as the RMS of the signal

divided by the RMS of the noise (linear units). The resampling process was subsequently
repeated 5000 times, resulting in 5000 SNR calculations for each time window. The
empirical distribution function for each time window was determined and the SNR
corresponding to the 95th-percentile identified. This SNR corresponded to a 5%
probability of erroneously classifying a TEOAE as being present when it was not. Figure
3.2 shows the empirical distributions function for each of the different time windows
used in this study. The SNRs corresponding to the 95th-percenitles were similar across the
3-, 6-, and 12-cycle time windows varying between 11.7 – 12.07 dB. In contrast, the SNR
at the 95th percentile for the 24-cycle time window was 10.7 dB. Given the difference

 
  41  

between these SNRs, a 12 dB SNR criterion was initially adopted for analysis of the 3-,
6-, and 12-cycle tone-burst TEOAEs and an 11 dB SNR criterion for analysis of the 24-
cycle tone-burst TEOAEs. Analysis was subsequently repeated using a single 12 dB SNR
criterion for all tone-burst durations. Results were minimally different between the two
analyses, and only the 12 dB SNR analysis is reported in this paper.
In contrast to the recent analysis reported by Sisto et al. (2013), no attempt was
made to isolate the different-temporal portions of the TEOAEs. Rather, characteristics of
the “mixed” TEOAE, including the TEOAE envelope, latency-intensity function, and
input-output function, were measured and compared across the entire tone-burst duration.
If the dominant generation mechanism changed from the 3- to 24-cycle TEOAE then
differences in the envelopes, latency-intensity functions, and input-output functions
should be apparent. If, on the other-hand, the generation mechanism were the same across
the different tone-burst durations then similarity between the envelopes, latency-intensity
functions, and input-output functions of the different TEOAEs would be expected.
3.3 Results
3.3.1 TEOAE Envelopes
Figures 3.3 and 3.4 show the TEOAE envelopes evoked by each tone-burst
stimulus (duration and level). Data from 6 subjects are shown. These data were

characteristic of all subjects. In each plot, the magnitudes of the envelopes were
normalized to the peak magnitude of the envelope evoked by the highest-level tone-burst.
The dashed line in each plot is located at a latency of 6.03 ms and represents the earliest
expected latency (95th percentile) for a 2 kHz, low-level SFOAE/TEOAE based on data
reported by Shera et al. (2002). TEOAE energy that occurred later than 6.03 ms is
hereafter referred to as the long-latency (LL) portion. TEOAE energy that occurred
earlier than 6.03 ms is referred to as the short-latency (SL) portion.
Regardless of tone-burst duration, TEOAE envelopes included both a SL and LL
portion. The presence of a SL portion in the TEOAEs evoked by the longer-duration

 
  42  

tone-bursts suggests that its generation was not dependent on stimulus energy at the f1
(2.6 kHz) and f2 (3.1 kHz) primaries required to evoke a 2 kHz distortion-source OAE.
Recall that the energy at these frequencies had decreased by 57 dB from the 3-cycle to
the 24-cycle tone-bursts (see Fig. 3.1). The following paragraphs discuss the morphology
of the TEOAE envelopes in further detail, highlighting nuances in the envelopes of the
different duration tone-bursts and between different subjects. However, in spite of these,
the main finding remains unaltered in that the SL portion of the TEOAE does not
strongly depend on stimulus energy outside of 2 kHz. This finding conflicts with the
hypothesis that this part of the TEOAE is generated through intermodulation distortion
(Withnell et al., 2008; Goodman et al., 2009; Moleti et al., 2012a).
The data shown in Fig. 3.3 was characteristic of the majority of subjects who
participated in the study. For these subjects, the TEOAE envelopes evoked by the 3-cycle
tone-bursts exhibited multiple amplitude peaks. Each peak occurred at a different
moment in time, or latency, which did not change with stimulus level. In other words, the
latency of each peak was level-invariant (Goodman et al., 2009). Two peaks occurred
within the SL time window and a single peak occurred in the LL time window. The
envelopes for the 6-cycle tone-bursts closely resembled those for the 3-cycle tone-bursts.
The same number of peaks occurred at the same latencies as those in the envelopes

evoked by the 3-cycle tone-bursts. However, the depths of the magnitude nulls separating
the peaks was reduced and the peaks were not as clearly resolved. This was especially
true for the two peaks that were generally evident in the time window of the SL TEOAE
(the initial 6.03 ms of the response). When tone-burst duration increased to 12- and 24-
cycles, the distinction between the SL peaks was further reduced. For the TEOAE
envelopes evoked by these tone-bursts, two SL peaks were no longer apparent. In
contrast to the distinction between the two SL peaks, the distinction between the SL and
LL peaks was relatively well preserved across the different tone-burst durations.

 
  43  

The TEOAE envelopes shown in Fig. 3.4 lacked the distinct magnitude peaks and
nulls that characterized the envelopes in Fig. 3.3. Rather, the envelopes for all tone-burst
durations were nearly continuous across time. SL and LL portions of the TEOAE were
still evident, however.
For all data, the SL and LL portions of the TEOAE evoked by a given tone-burst
duration showed a similar level-dependence to the SL and LL portions of the TEOAEs
evoked by all other tone-burst durations. First, at low stimulus levels, the magnitude of
the LL TEOAE portion was larger than that of the SL portion. In many cases, the SL
portion of the TEOAE did not emerge from the noise floor until higher stimulus levels. A
good example of this is seen in the envelopes for subject 05L (Fig. 3.3, 1st column).
Second, the SL portion of the TEOAE grew less compressively than the LL portion. This
difference between the growth rates resulted in the magnitude of the TEOAE in the SL
window exceeding that of the magnitude in the LL window at the highest stimulus levels.
Goodman and colleagues (2009; 2011) as well as Moleti and colleagues (2012a; Sisto et
al., 2013) have previously reported similar behavior of the different portions of the
TEOAE.
In summary, the OAEs evoked by the different duration tone-bursts all exhibited
SL TEOAE energy. Recall that the stimulus levels at f1 and f2 decreased by 19 dB with

each doubling in tone-burst duration. Accordingly, the continued presence of the SL


TEOAE energy in the tone-bursts longer than 3-cycles suggests that it was not generated
through 2 f1 ! f2 intermodulation distortion, as had previously been hypothesized.
3.3.2 TEOAE latency-intensity functions
Latency-intensity functions were derived for each subject at all tone-burst
durations. The purpose of this analysis was to determine whether the latencies of the
TEOAEs evoked by the different tone-bursts all exhibited a similar dependency on
stimulus level. Latency was calculated as the energy-weighted mean of the time vector
(Eq. 3.9) and included both the SL and LL portions of the TEOAE. Due to the energy

 
  44  

weighting applied, latency will be biased toward the time index associated with the
largest magnitude peak in the TEOAE envelope. Recall that the levels of the tone-bursts
were equalized across the different durations by calibrating in terms of dB SPL re: 1 Hz
bandwidth (at 2 kHz) and only the energy at frequencies outside of 2 kHz decreased as
tone-burst duration increased. If generation of the SL portion of the TEOAE depends on
energy outside of 2 kHz (as would be the case if generated through cubic intermodulation
distortion), then its contribution to TEOAE latency should systematically decrease as
tone-burst duration increases beyond 3 cycles. As a result, the latency-intensity functions
for the TEOAEs evoked by the different-duration tone-bursts should be dissimilar.
Latency-intensity functions for the TEOAEs evoked by the longer-duration tone-bursts
should show relatively little change in latency with stimulus level, compared to those for
the shorter duration tone-bursts, since only the LL portion of the TEOAE would be
present – the latency of this portion of the TEOAE is level invariant. The qualitative
analysis of the TEOAE envelopes presented above suggests that this will not be the case
since the SL portion of the TEOAE was present for all tone-burst durations. However,
analyzing the TEOAE in terms of its latency provides a quantitative approach to
determining the dependence of the SL TEOAE portion on stimulus energy at the f1 and

f2 primary frequencies. If the SL portion of the TEOAE makes comparable contributions


to the TEOAE for all tone-burst durations, the latency-intensity functions should be
similar and all exhibit a strong dependency of latency on stimulus level.
Figure 3.5 plots the latency-intensity functions for the 12 subjects included in the
study. Marker type denotes the tone-burst duration used to evoke the OAE. For each
subject and across all tone-burst durations, TEOAE latency decreased as stimulus level
increased. The decrease in latency with increasing stimulus level was generally
monotonic. Despite the different duration tone-bursts used to evoke the TEOAE, the
resulting latency-intensity functions for a given subject were similar both in terms of their
shape and latency values. For instance, the different tone-burst latency-intensity functions

 
  45  

were nearly identical to each other for subjects 12R, 13R, and 19L. When discrepancies
between the different duration tone-burst latency-intensity functions were apparent, the
function for the 24-cycle tone-burst evoked OAE typically departed from the functions
for the TEOAEs evoked by the shorter-duration tone-bursts. In nearly all of these cases,
the latency of the TEOAE evoked by the 24-cycle tone-burst was slightly longer than the
latencies of the TEOAEs evoked by the shorter tone-bursts at isolated stimulus levels (see
subjects 01L, 05L, 14L, and 15R). Only a single subject exhibited latencies for the
TEOAE evoked by the 24-cycle tone-burst that were consistently longer than the
latencies for the shorter-duration tone-burst evoked TEOAEs (subject 11R).
The reason for the prolonged latencies in the 24-cycle tone-burst TEOAEs is not
clear. Figure 3.6 illustrates how these instances coincided with the transition from a
dominant magnitude peak in the LL portion of the TEOAE to a dominant peak in the SL
portion of the TEOAE. Each panel shows the 24-cycle tone-burst TEOAE envelopes for a
different subject. The envelope represented by the black tracing indicates the TEOAE
evoked for the stimulus level at which the latency was prolonged relative to the shorter-
duration tone-burst TEOAEs (59 dB pSPL for all subjects, see Fig. 3.5). Triangles are
placed on each envelope at the time index corresponding to its calculated latency. At
lower stimulus levels, the TEOAE envelope exhibited a dominant peak in the LL time

window. However, as stimulus level increased, an earlier peak emerged in the SL time
window and a notch developed in the TEOAE envelope between the SL and LL TEOAEs
(indicated by arrows). It was at these stimulus levels where the 24-cycle latency increased
and was longer than latencies at comparable stimulus levels for the shorter-duration tone-
bursts. At higher stimulus levels, the SL peak grew rapidly and dominated the TEOAE.
Latency subsequently decreased and approached the values for the shorter-duration tone-
burst OAEs.
As mentioned, it is not clear as to why the latency-intensity functions for the 24-
cycle tone-bursts exhibited this behavior. It does not seem to be indicative of a different

 
  46  

generation mechanism for the TEOAE evoked by the 24-cycle tone-burst since, aside
from these isolated instances, latency still decreases and the 24-cycle latency-intensity
function is generally similar to those for the shorter tone-bursts. One hypothesis relies on
first interpreting each TEOAE peak as representative of a distinct component. Within this
framework, the morphology of the TEOAE envelope is a result of phase interactions
between the components across the temporal regions where they overlap. As tone-burst
duration increases, the duration of each component will also increase and the temporal
overlap between adjacent components expands. The change in the amount of temporal
overlap between components will be relatively small when increasing from a 3- to 12-
cycle tone-burst as stimulus duration increases from 1.5 to 6 ms. However, the increase in
overlap can be expected to be much larger when going from a 12- to 24-cycle tone-burst
as stimulus duration increases from 6 to 12 ms. It is possible that this much more
pronounced overlap between components in the 24-cycle response somehow gives rise to
errant latency values.
To quantify the dependence of TEOAE latency on stimulus level, TEOAE
latency-intensity functions for each tone-burst were collapsed across subjects and fit with
1st-order polynomials,

! n = an Lstim + bn , Eq. 3.10

where ! was the predicted latency (ms), a was a fitting constant describing the slope of
the latency-intensity function (ms/dB), b was the y-intercept (ms), Lstim was stimulus
level (dB pSPL), and n was the tone-burst duration (3-, 6-, 12-, or 24-cycles). The
resulting models (black lines), coefficients, and individual observations (gray lines) are
shown in Fig. 3.7; each panel represents a different duration tone-burst. The latencies of
the TEOAEs evoked by the different duration tone-bursts all showed similar dependency
on stimulus level: Latency decreased with increasing stimulus level at rates between 0.10

 
  47  

– 0.12 ms/dB. The modeled dependency of latency on stimulus level is similar to that
observed by other investigators. Sisto and Moleti (2007) and Rasetshwane et al. (2013)
reported 2 kHz TEOAE latencies that decreased at a rate of 0.117 ms/dB and 0.133
ms/dB, respectively. SFOAE data from Schairer et al. (2006) demonstrated a decrease in
latency at a rate of approximately 0.15 ms/dB.
Table 3.1 provides the mean latencies and standard deviations for each tone-burst
duration – level combination shown in Fig. 3.7. At the lowest stimulus level (35 dB
pSPL), latency ranged from 7.66 – 8.42 ms across the different tone-bursts. This value is
comparable to previously reported latencies for low-level 2 kHz TE/SFOAEs (Tognola et
al., 1997; Shera et al., 2002; Sisto & Moleti, 2007). At the highest level (77 dB pSPL),
latency ranged from 3.50 – 3.78 ms, comparable to the ~4 ms latency Rasetshwane et al.
(2013) measured for 2 kHz TEOAEs at similar stimulus levels, but shorter than the 5.7
ms latency measured by Sisto and Moleti (2007). The discrepancy between the current
results and those of Sisto and Moleti (2007) may be due to several factors. First, they
measured OAEs using an ILO system (Otodynamics, Ltd.) whereby the initial 2.5 ms are
discarded and the subsequent 2.5 ms are onset-ramped. It is possible that some of the SL
portion of the TEOAE was discarded in these investigators analysis, biasing latency to
longer values. Second, it may not be entirely appropriate to compare the latency of the 77

dB pSPL TEOAEs from the current study to data from Sisto and Moleti (2007) since they
used acoustic clicks to evoke the OAE.
The variability (quantified by the standard deviation) in latency across subjects
decreased from an average of 1.11 ms at 35 dB pSPL to 0.69 ms by 77 dB pSPL.
Variability was slightly higher for the 24-cycle tone-burst latencies compared to the
shorter tone-bursts. Overall, standard deviations were approximately 3 times smaller than
those reported by Rasetshwane et al. (2013). One explanation for the smaller standard
deviations in the current study may lie in the slightly different method used to calculate
latency. Rasetshwane et al. calculated latency from the rectified TEOAE time waveform

 
  48  

while the current study calculated latency from the TEOAE time waveform envelope.
The process of converting the time waveform to its analytic equivalent and then
extracting the instantaneous magnitude attenuates high-frequency noise thereby acting as
a low-pass filter. This does not occur with full-wave rectification. Accordingly, an
improvement in SNR associated with using the instantaneous magnitude may have
yielded less variability in the estimation of TEOAE latency.
The similarity between TEOAE latency values and the dependence of TEOAE
latency on stimulus level across the different duration tone-bursts suggests that TEOAE
latency was independent of the tone-burst duration. To test this hypothesis, a two-way
analysis of variance (ANOVA) was performed with main factors tone-burst duration (

n = 4 ) and tone-burst level ( n = 8 ). Data from all subjects were included in the analysis (

n = 12 ); however, data points with SNRs less than 12 dB were excluded. Main effects of
tone-burst duration (F(3, 368) = 4.5, p = 0.004) and level (F(7, 368) = 130.47, p << 0.01)
were found. The interaction term was not significant (F(21, 368) = 0.25, p = 1). The
finding that the interaction term was not significant was consistent with stimulus level
having the same effect on TEOAE latency regardless of the tone-burst’s duration. In
other words, an increase in stimulus level caused a similar decrease in TEOAE latency
across all tone-burst durations.

The significant effect of tone-burst duration on TEOAE latency was somewhat


surprising. To investigate this further, the means of the different tone-burst TEOAE
latencies were compared using multiple two-sample t-tests with a Bonferroni correction
(p = 0.05, 6 comparisons). The 24-cycle mean latency was significantly different from
both the 6- and 12-cycle mean latencies. No other comparisons were significantly
different. In both cases, the mean latency of the 24-cycle TEOAE was longer than either
the latencies for the 6- or 12-cycle TEOAEs: The mean latency of the 24-cycle TEOAE
was 6.16 ms compared to 5.71 ms for both the 6- and 12-cycle TEOAEs (see Table 3.1).
Rasetshwane et al. (2013) also found a significant effect of stimulus duration on TEOAE

 
  49  

latency in that the latency increased as stimulus duration increased. However, their
finding showed a systematic increase in latency with tone-burst duration. This was not
replicated in the current study. The origin of Rasetshwane et al.’s findings is likely due to
referencing latency to tone-burst onset as opposed to the peak of the tone-burst. As the
rise-time of the tone-burst increases, the peak occurs at a later moment in time relative to
stimulus onset, thereby causing the peak of the TEOAE to be increasingly delayed
relative to stimulus onset. It is not immediately obvious as to the cause of the slightly
longer latency for the 24-cycle TEOAE in the current study.
3.3.3 TEOAE Input-Output Functions
Input-output (IO) functions were derived for each subject at all tone-burst
durations. The purpose of this analysis was to determine whether the magnitudes of the
TEOAEs evoked by the different tone-bursts all exhibited a similar dependency on
stimulus level. In other words, did the different TEOAEs show similar growth rates. For
this analysis, magnitude was calculated as the RMS of the TEOAE envelope (Eq. 3.8)
across the full time-window over which the TEOAE occurred. Therefore, both SL and LL
portions of the TEOAE were included in the calculation. Recall that the levels of the
tone-bursts were equalized across the different durations by calibrating in terms of dB
SPL re: 1 Hz bandwidth (at 2 kHz) and only the energy at frequencies outside of 2 kHz

decreased as tone-burst duration increased. If the generation mechanism of the SL


TEOAE depends on energy outside of 2 kHz (as would be the case if generated through
cubic intermodulation distortion), then its contribution to the total TEOAE magnitude
should systematically decrease as tone-burst duration increases beyond 3 cycles. As a
result, the IO functions for the different duration tone-burst evoked OAEs should be
dissimilar. In contrast, if contribution from the SL TEOAE is comparable across tone-
burst duration, the IO functions for the different duration tone-burst evoked OAEs should
be similar, suggesting an alternate generation mechanism. Both the qualitative analysis of
the TEOAE envelopes and the quantitative analysis of the latency-intensity functions

 
  50  

suggest that similarity will be evident in the IO functions of the different TEOAEs. This
final analysis provides another quantitative assessment of the dependence of SL TEOAE
energy on stimulus energy outside of 2 kHz.
Figure 3.8 plots the IO functions for the 12 subjects included in the study. Marker
type denotes the tone-burst duration used to evoke the OAE. For each subject, TEOAE
magnitude increased with stimulus level, for all tone-burst durations. Similar to their
latency-intensity functions, the shapes and magnitude values of each subject’s IO
functions were similar across the different tone-burst durations. In many cases, it is
difficult to differentiate between the different tone-burst evoked TEOAE IO functions as
they overlay on each other. For each subject, the differences between TEOAE
magnitudes across the tone-burst stimuli were generally less than 5 dB. Accordingly,
TEOAE magnitude seemed relatively independent of tone-burst duration while exhibiting
a strong dependence on tone-burst level.
TEOAE IO functions for each tone-burst stimulus were collapsed across subjects
and fit with 1st-order polynomials,

Ln = an Lstim + bn , Eq. 3.11

where L was the predicted magnitude (dB SPL), a was a fitting constant describing the
slope of the IO function (dB/dB), b was the y-intercept (dB SPL), Lstim was stimulus
level (dB pSPL), and n was the tone-burst duration (3-, 6-, 12-, or 24-cycles) . The
resulting models (black lines), fitting coefficients, and individual observations (gray
lines) are shown in Fig. 3.9, each panel represents a different duration tone-burst. Model
coefficients were similar across all tone-burst durations. Magnitude grew compressively
with stimulus level at a rate of 0.42 – 0.49 dB/dB. Growth rates for TEOAEs evoked by
the longer duration stimuli (12- and 24-cycles) were slightly less compressive than rates
for the TEOAEs evoked by the shorter-duration stimuli. TEOAE and SFOAE growth

 
  51  

rates have previously been measured in the range of 0.3 – 0.6 dB/dB (Kemp & Chum,
1980; Probst et al., 1986; Schairer et al., 2003). Table 3.2 provides the mean magnitudes
and standard deviations for each tone-burst duration – level combination shown in Fig.
3.9.
The similarity between both the TEOAE magnitude values and the dependence of
TEOAE magnitude on stimulus level across the different duration tone-bursts suggested
that TEOAE magnitude was independent of the tone-burst duration. To test this
hypothesis, a two-way analysis of variance (ANOVA) was performed with main factors
tone-burst duration ( n = 4 ) and tone-burst level ( n = 8 ). Data from all subjects were
included in the analysis ( n = 12 ); however, data points with SNRs less than 12 dB were
excluded. A main effect of tone-burst level (F(7, 368) = 57.57, p << 0.01) was found.
Neither tone-burst duration (F(3, 368) = 2.54, p = 0.057) nor the interaction term (F(21,
368) = 0.17, p = 1) were significant. Accordingly, TEOAE magnitude did not depend on
stimulus duration and, therefore, the compressive growth of the TEOAE was essentially
identical across the different length tone-bursts.
3.4 Discussion
3.4.1 Generation Mechanism of the SL TEOAE
The primary goal of the current study was to determine if 2 f1 ! f2 intermodulation

distortion is the generation mechanism responsible for the SL portion of the bandpass
filtered TEOAE. As discussed in the introduction and briefly reviewed here, many of the
characteristics of the SL TEOAE are consistent with generation through nonlinear
distortion. First, a short latency would be expected if the generation site was basal to the
OAE’s characteristic frequency (CF) location. The cubic distortion product ( 2 f1 ! f2 ) is
generated near the f2 region (Gaskill & Brown, 1990; Martin et al., 1998; Talmadge et
al., 1999; Knight & Kemp, 2000), basal to its CF place, and thus has a shorter latency
than a reflection-source emission of the same frequency (Konrad-Martin & Keefe, 2003;
Moleti et al., 2012a). Second, the growth of the SL TEOAE is less compressive than that

 
  52  

of LL TEOAE. Differential growth rates are evident in DPOAEs with the distortion-
source component exhibiting less compressive growth than the reflection-source
component (Mauermann and Kollmeier, 2004; Abdala et al., 2011). Third, the transient
nature of the evoking stimulus seemingly invites nonlinear interaction between the
stimulus’ various frequency components. For instance, intermodulation distortion has
been evidenced in guinea pig TEOAEs (Yates and Withnell, 1999) while two-tone
suppression has been demonstrated in human TEOAEs (Konrad-Martin and Keefe, 2003;
2005).
In light of the characteristics cited above, the current study adopted a null
hypothesis that the SL portion of the TEOAE is generated through 2 f1 ! f2
intermodulation distortion. To test this hypothesis, TEOAEs were evoked using 4 tone-
bursts of progressively longer durations (3-, 6-, 12-, and 24-cycles). As tone-burst
duration increased, the spectrum became increasingly narrow and stimulus energy at f1
and f2 decreased by approximately 19 dB with each doubling of tone-burst duration (see
Fig. 3.1). Accordingly, it was expected that if generated through cubic intermodulation
distortion, the SL TEOAE would decrease in magnitude with increasing stimulus
duration.
In contrast to this expectation, all TEOAEs, regardless of the duration of the

evoking tone-burst, included comparable contributions from a SL and LL portion. Within


the TEOAE envelopes, the SL portion of the TEOAE occurred within the initial 6.03 ms
of the response. The LL portion occurred after 6.03 ms. The two portions of the TEOAE
exhibited different growth rates. Consistent with previous reports (Goodman et al., 2009;
2011; Moleti et al., 2012a; Sisto et al., 2013), the SL portion grew less compressively
than the LL portion.
To provide a more quantitative assessment of the contribution the SL portion of
the TEOAE makes to the total TEOAE, latency-intensity and input-output functions were
derived for each tone-burst duration and compared. The dependence of TEOAE latency

 
  53  

on tone-burst level was similar across the different tone-burst durations. On an individual
subject level, TEOAE latencies for a given tone-burst stimulus were similar to latencies
for all other tone-burst stimuli. Group analysis demonstrated that TEOAE latency
decreased at a rate of approximately 0.11 ms per 1 dB increase in stimulus level, for all
tone-burst durations. By examining the TEOAE envelopes for the 3-, 6-, and 12-cycle
tone-bursts, the origin of the level-dependency of TEOAE latency may be traced to the
less compressive growth of SL portion of the TEOAE relative to the LL portion (Sisto et
al., 2007; Goodman et al., 2011; Moleti et al., 2012a). Not only was the level-dependence
of the TEOAEs evoked by different tone-bursts similar, the TEOAE latency values were
also similar. This was especially true when comparing the latencies of the 3-, 6-, and 12-
cycle TEOAEs. No significant difference was found between these latencies. The latency
of the 24-cycle TEOAE was significantly longer than the latencies of both the 6- and 12-
cycle TEOAEs (but not the 3-cycle TEOAE); however, the difference between these
latencies was less than 0.5 ms.
The input-output functions relating TEOAE magnitude to stimulus level were
nearly identical for the TEOAEs evoked by different duration tone-bursts. For each
individual subject, TEOAE magnitudes across the different tone-burst stimuli were
approximately equal to each other at comparable stimulus levels. Accordingly, the IO

functions overlaid each other and exhibited similar growth rates. Across all subjects, the
different TEOAE IO functions showed compressive growth ranging from 0.44 – 0.52
dB/dB. No significant main effect of tone-burst duration was found, nor was the
interaction between tone-burst duration and level on TEOAE magnitude found to be
significant.
Similar TEOAE latencies, magnitudes, and growth rates across the different
duration tone-bursts would be expected if the TEOAEs shared a common generation
mechanism (Kalluri and Shera, 2007; Sisto et al., 2013). In this case, “common” need not
necessarily be interpreted in the sense that the SL and LL portions of the TEOAE are

 
  54  

generated through the same mechanism. Rather, the mechanism(s) is/are the same for all
tone-burst durations. In other words, the results of this study suggest that whatever
mechanism(s) are responsible for the SL and LL portions of the TEOAE evoked by the 3-
cycle tone-burst are the same in the TEOAEs evoked by the 6-, 12-, and 24-cycle tone-
bursts. This finding agrees with several previous studies suggesting that TEOAEs
(evoked by either clicks or very short tone-bursts) and SFOAEs (evoked by pure-tones,
long tone-bursts, or frequency-glides) arise through the same mechanism(s) (Kalluri &
Shera, 2007; Sisto et al., 2013; Kalluri & Shera, 2013).
One difficulty with making comparisons between these different OAEs is that of
matching stimulus levels between stimuli that differ in their duration and frequency
content. Kalluri and Shera (2007) empirically derived a measure of stimulus level
referred to as bandwidth compensated dB SPL (or dB cSPL) to equate click and pure-
tone levels and subsequently measured nearly identical growth functions and spectral
characteristics between each OAE type. However, their analysis limited comparisons to
low-level SFOAEs and the LL portion of the TEOAE. Sisto et al. (2013) hinted at the
equivalence across the SL and LL portions of the TEOAE and SFOAE in that they
exhibited similar growth rates and latencies; however, an effort to control for differences
in stimulus level was not made and measurements were restricted to relatively low

stimulus levels. The current study matched stimulus level across the different duration
tone-bursts by calibrating them in terms of dB SPL re: 1 Hz bandwidth of the tone-burst
center frequency. By controlling for differences in stimulus level across the different-
duration tone-bursts, the results from the current study demonstrate approximate
equivalence between the generation mechanisms of the TEOAE (approximated by the 3-
cycle tone-burst) and SFOAE (approximated by the 24-cycle tone-burst) across a wide
range of stimulus levels.
The commonality between the mechanisms contributing to the 3-, 6-, 12-, and 24-
cycle tone-burst emissions disqualifies 2 f1 ! f2 distortion as the generation mechanism

 
  55  

responsible for the SL portion of the TEOAE. The basis of the argument in favor of

2 f1 ! f2 distortion is that the evoking stimulus contains energy at the requisite


frequencies to induce a 2 f1 ! f2 distortion-source OAE (Yates & Withnell, 1999; Moleti
et al., 2012a). However, the major finding of the current study was that the SL portion of
the TEOAE is also generated in the near absence of stimulus energy at these primary
stimulus frequencies. To generate a 2 f1 ! f2 , 2 kHz, distortion-source OAE, stimulus
energy at 2.5 kHz ( f1 ) and 3.2 kHz ( f2 ) is required ( f2 f1 = 1.22 ). The energy at these
frequencies in the 24-cycle tone-burst was 57 dB less than the energy in the 3-cycle tone-
burst, yet the TEOAEs exhibited similar magnitudes, latencies and changes in magnitude
and latency with stimulus level. Granted, it is theoretically possible that intermodulation
distortion is responsible for generation of the SL portion of the TEOAE when the evoking
stimulus is very short and another mechanism is responsible when the stimulus is long.
However, in order for such a system to explain the similarly between the latency-intensity
and IO functions across tone-burst durations, both mechanisms would have to generate
emissions with similar latencies and magnitudes and compensate for each other as
stimulus duration changes, which seems unlikely.
While the focus of the current study was to examine the possible role of 2 f1 ! f2
distortion in the generation of the SL TEOAE, it is also worthwhile to consider whether

this portion of the TEOAE may arise through 2 f2 ! f1 distortion. Compared to the

2 f1 ! f2 distortion product, which is most robust for a frequency ratio ( f2 f1 ) of 1.22, the
optimal frequency ratio for the 2 f2 ! f1 distortion product is approximately 1 (Fitzgerald
and Prieve, 2005). Accordingly, the 3-, 6-, 12-, and 24-cycles tone-bursts would all have
sufficient energy at the requisite primaries to generate a 2 f2 ! f1 distortion-source OAE.
However, the dominant region of generation for the 2 f2 ! f1 OAE is near the fDP place (

fDP = 2 f2 ! f1 ; Martin et al., 1998) and presumably occurs through a reflection


mechanism (Knight & Kemp, 2001). The latency of a 2 kHz, 2 f2 ! f1 OAE would
therefore be approximately the same as that of the 2 kHz LL TEOAE, generated through

 
  56  

linear coherent reflection at the tonotopic place. As such, the latency of a 2 f2 ! f1 OAE is
too long to explain the SL portion of the TEOAE.
An alternative mechanism to explain the generation of SL TEOAEs is linear
coherent reflection at basilar membrane regions slightly basal to the tonotopic place
(Goodman et al., 2011; Sisto et al, 2013; Moleti et al., 2013). Both the low-level TEOAE
and SFOAE are presumably generated at the tonotopic place through back-scattering of
the forward moving traveling wave off pre-existing impedance discontinuities in the
basilar membrane mechanics (Zweig and Shera, 1995; Shera and Guinan, 1999; Kalluri
and Shera, 2007; Lichtenhan, 2012). Although reflections of the traveling wave occur
across the length of the basilar membrane, the shape of the wave at the tonotopic place
results in constructive phase interactions between the reflected wavelets, creating a robust
reverse traveling cochlear wave which can be measured as an OAE in the ear canal. As
the input level to the cochlea increases; however, the peak of the traveling shifts basally
(Ruggero et al., 1997; Rhode, 2007) and may induce an additional region of coherent
reflection basal to the tonotopic place (Goodman et al., 2011; see Zweig and Shera,
1995). Any resulting OAE would be expected to have both shorter latency and less
compressive growth compared to the OAE generated at the tonotopic place. Moleti et al.
(2013) recently demonstrated the plausibility of basal coherent reflection in the

generation of the SL portion of the TEOAE through the use of a nonlinear cochlear
model. The investigators simulated TEOAEs while selectively turning on/off the
impedance discontinuities, or roughness, basal to the tonotopic place. When roughness
was preserved across the length of the BM, the TEOAE included both SL and LL
portions. However, when roughness was turned off up to 1/4-octave basal to the
tonotopic place, the SL portion of the TEOAEs disappeared. Accordingly, coherent
reflection basal to the tonotopic place is at least a conceivable mechanism for SL TEOAE
generation. Other mechanisms have also been suggested to explain SL OAEs, including
nonlinear coherent reflection (Talmadge et al., 2000), distributed sources of nonlinear

 
  57  

distortion (Siegel et al., 2005), and reflection from the tail portion of the cochlear
traveling wave (Guinan, 1990; Siegel et al., 2004; Choi et al., 2008).
3.4.2 Origin of TEOAE Envelope Morphology
While the TEOAE envelopes all exhibited a SL and LL portion, differences in the
morphology of the envelopes was apparent across subjects and across tone-burst
durations. Recall that sharp magnitude nulls and peaks punctuated the envelopes evoked
by the 3- and 6-cycle tone-bursts for many of the subjects (Fig. 3.3). As tone-burst
duration increased, the resolution of individual peaks was decreased, especially within
the SL time window. The envelopes for other subjects lacked magnitude peaks and nulls
(Fig. 3.4) and exhibited minimal changes in morphology at tone-burst duration increased.
These differences may be explained under several different frameworks.
The presence of multiple peaks within the TEOAE has previously been
interpreted as indicative of different-latency components within the TEOAE. The
different components occur at different latencies but all share similar frequency content
(Withnell & McKinley, 2005; Goodman et al., 2011; Moleti et al., 2012a). Per this
interpretation, the SL portion of TEOAE in the envelopes evoked by the shorter-duration
tone-bursts (3- and 6-cycles) shown in Fig. 3.3 would include two components
(evidenced through two magnitude peaks). The LL portion of the envelope would include

only a single component (evidenced through one magnitude peak). Although these
components have different latencies, they can be expected to partially overlap in time.
Depending on their relative phases, deconstructive or constructive interactions will occur
across the region of overlap. These interactions will be evidenced in the morphology of
the TEOAE envelope. If the components share common phases, distinct magnitude peaks
might be expected only for the shortest duration tone-bursts (3- and 6-cycles), assuming
the components minimally overlap in time. However, as the duration of the tone-burst
increases and the components overlap more in time, the peaks that were initially present
can be expected to merge into a single peak since constructive interactions are occurring

 
  58  

between the components across the region they overlap. On the other hand, if the
components are initially in anti-phase, distinct magnitude peaks will be evident in all
TEOAE envelopes, regardless of tone-burst duration.
A simple model to illustrate this phenomenon was developed. Three TEOAE
components, a LL component and two SL components (SL1 and SL2), were modeled by
2 kHz tone-bursts whose durations were either 3-, 6-, 12-, or 24-cycles. The LL
component occurred latest in time and the SL2 component occurred earliest in time. The
SL1 component occurred between the SL2 and LL components. Two milliseconds
separated adjacent components. The phase of each component was specified relative to
the phase of its earlier-occurring neighbor component: The phase of the SL1 component
lagged that of the SL2 component by 2! radians and the phase of the LL component
lagged that of the SL1 component by ! radians. All components had the same peak
amplitude. These phases were chosen to provide examples of both anti- and in-phase
interactions between adjacent components.
Figure 3.10 presents the results from the simulation. Data are presented in the
same format as Figs. 3.3 and 3.4. Considering first the simulated TEOAE envelope
evoked by the 3-cycle tone-burst, three distinct peaks are present corresponding to two
SL components and a single LL component. Sharp magnitude nulls separate adjacent

components. When the tone-burst duration increases to 6-cycles three peaks are still
evident; however, the peaks corresponding to the two SL components are not as clearly
resolved as they were for when the TEOAE was evoked by the 3-cycle tone-burst. Recall
that these components shared similar phases. In contrast to the two SL components, the
LL component remains clearly resolved. The phase of this component was anti-phase of
the adjacent SL1 component. As the duration of the evoking tone-burst increases further,
the two SL peaks merge into a single peak thereby giving the illusion of a single SL
component. Meanwhile, the LL component remains separated from the SL component by
a sharp magnitude null. These results are generally characteristic of the observed data and

 
  59  

increased temporal-overlap between components seems a reasonable explanation for the


change in TEOAE envelope morphology that accompanied increases in tone-burst
duration (see Fig. 3.3).
The component model may also explain the data in Fig. 3.4. As mentioned,
distinct peaks were not readily apparent in these envelopes. If the TEOAE was composed
of multiple components, these components would need to have similar phases and
overlap in time, even for the shortest-duration tone-burst. The model illustration does not
demonstrate this, however, since the SL1 and LL components were in-phase but still
resolved in the envelope. However, if the components were closer together in time, a
more continuous response would have resulted.
The different morphologies of the TEOAE envelopes and how they changed with
increases in tone-burst duration can also be explained through a framework that does not
include different latency components. Rather, the TEOAE for a specific frequency may
be only a single component whose energy is dispersed across time. This TEOAE will
interact with TEOAEs at adjacent frequencies, giving rise to peaks and nulls in the
TEOAE envelope. The resolution of the peaks and nulls will be proportional to the
bandwidth of the filtered TEOAE. Consider the 3-cycle tone-burst; this stimulus has
energy across a broad range of frequencies and, therefore, evokes OAEs across an

equally broad frequency range. As a result, the 1/3-octave bandpass filtered 2 kHz
TEOAE contains OAE energy not only at 2 kHz but also at adjacent frequencies (1900
Hz, 2010 Hz, 2031 Hz, etc.). Since these different frequency components would overlap
in time, wave-interference would be expected to occur and result in a series of amplitude
peaks and nulls in the TEOAE envelope. However, as tone-burst duration increases,
excitation of the cochlear partition is reduced to a narrower frequency region and the 2
kHz bandpass filtered TEOAE would include less energy at frequencies outside of 2 kHz.
As a result, wave-interference would be reduced and the resolution of peaks and nulls in
the envelope would decrease.

 
  60  

Wave-interference between different frequency components could also explain


the difference in the envelope morphologies in Fig. 3.3 compared to Fig. 3.4. The exact
morphology of the envelope may depend on the cochlear reflectivity – the underlying
roughness pattern along the basilar membrane that causes backscattering of the forward
traveling cochlear wave – across the region where the TEOAE is generated. If the evoked
TEOAE originates from a region with unusually large reflectivity relative to adjacent
regions (characteristic of a peak in the spectral fine structure; Zweig & Shera, 1995), the
envelope might be expected to be more continuous in nature. In contrast, if the TEOAE
originates from a region where reflectively is nearly constant across adjacent regions, the
TEOAE envelope would exhibit distinct magnitude peaks and nulls.
Presumably, both temporal-overlap between different components of the same
frequency and wave-interference between components of different frequencies can
account for TEOAE envelope morphology. There is evidence for the both SL and LL
components in the SFOAE. For instance, Schairer et al. (2003) measured notches in the
IO functions for the SFOAE. Notches would be expected when two components that are
in anti-phase temporally overlap, thereby cancelling each other. Additionally, Sisto et al.
(2013) were able to separate the SFOAE into SL and LL components based on their
different phases (see Moleti et al., 2012b for technique). Notches in the OAE envelope, as

observed in the current study, may similarly implicate SL and LL components that are in
anti-phase. DPOAE envelopes show similar behavior due to anti-phase interactions
between two components (see Talmadge et al., 1999 and Vitesník et al., 2009). However,
as was demonstrated in the wave-interference simulation, these same effects may arise
through wave-interference. More advanced modeling efforts may be able to provide
further insight as to whether one model is more appropriate than the other. Additionally,
studies examining how the TEOAE envelope changes depending on its location in the
spectral fine structure may be beneficial. While these effects deserve further

 
  61  

consideration, they do not alter the conclusions of the current study, namely, that the SL
portion of the TEOAE is not generated through cubic intermodulation distortion.

 
  62  

3 Cycles 6 Cycles 12 Cycles 24 Cycles All Cycles


35 dB pSPL 7.66 ± 0.95 7.95 ± 1.08 7.84 ± 1.15 8.42 ± 1.30 7.97 ± 1.11
41 7.46 ± 0.79 7.42 ± 0.94 7.51 ± 0.94 8.11 ± 1.19 7.60 ± 0.96
47 6.99 ± 0.79 6.90 ± 0.77 6.86 ± 0.89 7.10 ± 1.10 6.96 ± 0.88
53 6.35 ± 0.74 6.08 ± 0.77 6.02 ± 0.78 6.53 ± 1.35 6.24 ± 0.94
59 5.60 ± 1.02 5.37 ± 0.91 5.55 ± 0.84 6.05 ± 1.38 5.64 ± 1.06
65 4.85 ± 0.94 4.57 ± 0.89 4.57 ± 0.90 5.16 ± 1.21 4.79 ± 0.99
71 4.27 ± 0.79 3.93 ± 0.65 3.86 ± 0.82 4.15 ± 1.13 4.05 ± 0.86
77 3.69 ± 0.60 3.50 ± 0.54 3.50 ± 0.58 3.78 ± 0.99 3.62 ± 0.69
All Levels 5.86 ± 0.82 5.71 ± 0.82 5.71 ± 0.86 6.16 ± 1.21 5.86 ± 0.94
Table 3.1. Mean latencies (ms) ± 1 standard deviation (ms) for each tone-burst duration
and level.

 
  63  

3 Cycles 6 Cycles 12 Cycles 24 Cycles All Cycles


35 dB pSPL -5.8 ± 5.1 -5.0 ± 5.4 -3.2 ± 4.8 -6.7 ± 4.6 -5.2 ± 5.0
41 -4.0 ± 6.0 -1.9 ± 5.9 -2.4 ± 5.9 -1.5 ± 5.2 -2.6 ± 5.7
47 -0.6 ± 6.0 0.1 ± 5.8 0.6 ± 6.0 -0.0 ± 6.7 0.0 ± 6.0
53 1.0 ± 6.1 2.4 ± 5.9 2.8 ± 6.2 2.5 ± 7.1 2.2 ± 6.2
59 3.6 ± 5.6 4.8 ± 5.8 4.9 ± 7.1 5.1 ± 6.4 4.6 ± 6.1
65 6.3 ± 5.1 7.6 ± 5.5 8.6 ± 6.5 7.7 ± 6.7 7.6 ± 5.9
71 8.8 ± 4.8 10.7 ± 4.9 12.3 ± 5.6 11.8 ± 5.9 10.9 ± 5.4
77 11.4 ± 4.2 13.5 ± 4.6 15.5 ± 4.9 15.1 ± 5.4 13.9 ± 5.0
All Levels 2.6 ± 5.4 4.0 ± 5.5 4.9 ± 5.9 4.25 ± 6.0 3.9 ± 5.7
Table 3.2. Mean magnitudes (dB SPL) ± 1 standard deviation (dB) for each tone-burst
duration and level.

 
  64  

0
ï10
ï20

Magnitude (dB re: 1 Hz)


ï30
ï40
Amplitude

ï50
ï60
ï70
ï80
ï90
ï100
0 3 6 9 12 0.5 1 1.5 2 2.5 3 3.5
Time (ms) Frequency (kHz)

Figure 3.1. Time- and frequency-domain representations of the 2 kHz tone-burst stimuli.
The left panel shows the waveforms for the 3-, 6-, 12-, and 24-cycle stimuli. The right
panel shows the corresponding spectra. Line thickness is used to show increasing number
of cycles in the tone-burst. The two broken, gray, vertical lines on the right panel identify
the location of the primary frequencies (f1 and f2) corresponding to a 2 kHz cubic
distortion product OAE ( f2 f1 = 1.22 ).

 
  65  

1
3ïCycle Window
0.9 6ïCycle
12ïCycle
0.8
24ïCycle
Cummulative Probability

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 5 10 15
SNR

Figure 3.2. Empirical distribution functions used to determine SNR criterion. Line-style
denotes the length of the time-window across which the empirical distribution function
was derived.

 
  66  

1 05L 11R 19L


3 Cycles
0.8

0.6

0.4

0.2

1
6 Cycles
0.8

0.6

0.4
Magnitude (Normalized)

0.2

1
12 Cycles
0.8

0.6

0.4

0.2

1
24 Cycles
0.8

0.6

0.4

0.2

0
0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15
Latency (ms)

Figure 3.3. TEOAE envelopes for three subjects across different levels and tone-burst
durations. Each column shows responses for an individual subject, and each row shows
responses for a given tone-burst duration. Increasing line thickness shows increasing
tone-burst level. The dashed line indicates the earliest expected latency (95th percentile)
for a 2 kHz low-level TE/SFOAE from data reported by Shera et al. (2002).

 
  67  

1 01L 12R 13R


3 Cycles
0.8

0.6

0.4

0.2

1
6 Cycles
0.8

0.6

0.4
Magnitude (Normalized)

0.2

1
12 Cycles
0.8

0.6

0.4

0.2

1
24 Cycles
0.8

0.6

0.4

0.2

0
0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15
Latency (ms)

Figure 3.4. TEOAE envelopes for three across different levels and tone-burst durations.
Data is presented in identical format to Fig. 3.3.

 
  68  

10

2 01L 05L 06L

10

4
Latency (ms)

2 10R 11R 12R

10

2 13R 14L 15R

10

2 17R 18R 19L

35 45 55 65 75 35 45 55 65 75 35 45 55 65 75
Stimulus Level (dB pSPL)

Figure 3.5. Latency-intensity functions for each tone-burst stimulus for all 12 subjects.
Each panel shows the functions for a single subject. Marker style denotes tone-burst
duration (dot = 3-cycles, asterisk = 6-cycles, square = 12-cycles, triangle = 24-cycles).

 
  69  

05L 14L 18R

Magnitude (Normalized)

ï5 0 5 10 15 ï5 0 5 10 15 ï5 0 5 10 15
Latency

Figure 3.6. The effect of temporal interaction between the SL and LL peaks on the 24-
cycle TEOAE latency calculation. Each panel shows the 24-cycle TEOAE envelopes for
a specific subject. The black tracing corresponds to the 59 dB SPL tone-burst where each
subject exhibited prolonged latency compared to the latencies for the shorter-duration
tone-burst emissions (see Fig. 3.5). The arrows indicate magnitude peaks of the SL and
LL TEOAEs. The triangles correspond to the calculated latency of each envelope. See
text for further explanation.

 
  70  

12
a = ï0.10 a = ï0.11
10 b = 11.62 b = 11.97
8
6
4
2
Latency (ms)

3 Cycles 6 Cycles
0
12
a = ï0.11 a = ï0.12
10 b = 11.98 b = 12.68
8
6
4
2 12 Cycles 24 Cycles
0
35 45 55 65 75 35 45 55 65 75
Stimulus Level (dB pSPL)

Figure 3.7. Group latency-intensity functions for each tone-burst stimulus. Each gray line
represents an individual subject’s latency-intensity function. The thick black line in each
plot represents the 1st-order polynomial fit describing the data. Model parameters are
provided in the upper right-hand corner of each plot.

 
  71  

30
01L 05L 06L
20

10

ï10

ï20
30
10R 11R 12R
20

10

0
Magnitude (dB SPL)

ï10

ï20
30
13R 14L 15R
20

10

ï10

ï20
30
17R 18R 19L
20

10

ï10

ï20
35 45 55 65 75 35 45 55 65 75 35 45 55 65 75
Stimulus Level (dB pSPL)

Figure 3.8. Input-output functions for each tone-burst stimulus for all 12 subjects. Each
panel shows the functions for a single subject. Marker style denotes tone-burst duration
(dot = 3-cycles, asterisk = 6-cycles, square = 12-cycles, triangle = 24-cycles).

 
  72  

30 a = 0.42 a = 0.44
20 b = ï20.68 b = ï20.41

10
0
Magnitude (dB SPL)

ï10
3 Cycles 6 Cycles

30 a = 0.46 a = 0.49
20 b = ï21.18 b = ï23.10

10
0
ï10
12 Cycles 24 Cycles
35 45 55 65 75 35 45 55 65 75
Stimulus Level (dB pSPL)

Figure 3.9. Group input-output functions for each tone-burst stimulus. Each gray line
represents an individual subject’s input-output function. The thick black line in each plot
represents the 1st-order polynomial fit describing the data. Model parameters are provided
in the upper left-hand corner of each plot.

 
  73  

3 Cycles
1

0.5

0
6 Cycles
1
Magnitude (Normalized)

0.5

0
12 Cycles
1

0.5

0
24 Cycles
1

0.5

0
0 5 10 15
Time (ms)

Figure 3.10. Simulation results for the multiple component model to explain TEOAE
envelope morphology. The TEOAE is modeled as three different-latency components
(two SL components and 1 LL component) that partially overlap in time. The dashed line
delineates the SL portion of the TEOAE from the LL portion. See text for further model
details.

 
  74  

CHAPTER IV
EXPERIMENT 2: TWO-TONE SUPPRESSION OF
SL AND LL PORTIONS OF THE TEOAE

4.1 Introduction
Otoacoustic emissions (OAEs) generated in response to the presentation of a
temporally short acoustic signal are known as transient-evoked OAEs (TEOAEs). When
bandpass filtered, TEOAE waveforms typically exhibit multiple amplitude peaks. The
relative prominence of these different peaks depends on the level of the evoking stimulus.
At very low-levels only a single peak is present. The time delay (relative to the stimulus)
of this peak shows a frequency dependency similar to that seen for low-level SFOAEs
(Tognola et al., 1997; Shera et al., 2002; Schairer et al., 2006; Sisto et al., 2007;
Goodman et al., 2009). As the stimulus level increases, at least one other magnitude peak
emerges. The latency of this peak is shorter than that of the peak previously discussed by
a factor of approximately 1.6 (Goodman et al., 2009; Moleti et al., 2012a).
These peaks have traditionally been interpreted as being representative of distinct
packets of similar frequency energy with different latencies (Withnell & McKinley, 2005;
Goodman et al., 2011; Moleti et al., 2012a; Mertes & Goodman, 2013). The peak

occurring at the longer latency has been called the long-latency (LL) component while
the earlier occurring peak has been labeled the short-latency (SL) component. This
terminology, although useful in describing the TEOAE, may not be entirely accurate. For
instance, not all waveforms exhibit distinct peaks (Lewis & Goodman, 2013). In these
waveforms only a single peak is evident; however, as stimulus level increases, the latency
of this peak progressively decreases. Accordingly, it seems more appropriate to refer to
the TEOAE as containing a SL portion and a LL portion.
Aside from their latencies, the SL and LL portions of the TEOAE differ in how
their magnitudes grow with increases in stimulus level. The LL portion exhibits highly

 
  75  

compressive growth, with rates as low as 0.1 dB/dB (Goodman et al., 2009; 2011). The
SL portion shows less compressive growth that can approach 1 dB/dB (Goodman et al.,
2011). When a magnitude peak is present within either the SL or LL portions of the
TEOAE, the latency of the peak does not change with increases in stimulus level. In other
words, the latency of each peak is level-invariant (Carvalho et al., 2003; Goodman et al.,
2009; Goodman et al., 2011; Moleti et al., 2012a). Traditional calculations of TEOAE
latency include both the SL and LL portions of the emission (e.g. Sisto and Moleti, 2007;
Rasetshwane et al., 2013). As a result, the latency of the TEOAE decreases with
increasing stimulus level. This decrease in latency is due to the dominant magnitude
portion of the TEOAE shifting from the LL to the SL as stimulus level increases
(Goodman et al., 2011; Moleti et al., 2012a).
The contrasting latencies and growth rates of the SL and LL portions of the
TEOAE imply that the different portions are generated through different mechanisms
and/or at different places in the cochlea. It is generally accepted that the LL portion arises
through linear coherent reflection (Zweig and Shera, 1995; Shera and Guinan, 1999;
Kalluri and Shera, 2007). The generation mechanism of the SL portion is less certain
although intermodulation distortion has traditionally been hypothesized as the likely
mechanism (Withnell and McKinley, 2005; Goodman et al., 2009; Moleti et al., 2012a).

Per this theory, the SL portion of the TEOAE is a 2 f1 ! f2 distortion-source OAE (Moleti
et al., 2012a) generated near the f2 place on basilar membrane (BM; Martin et al., 1998;
Knight & Kemp, 2001) where the excitation patterns of the f1 and f2 primaries
maximally overlap (Hall, 1974; Gaskill & Brown, 1990). Although it has received less
attention, another possibility is that the SL portion arises from the same mechanism and
place as the LL portion, but travels back to the stapes via a fast compression wave (Ren,
2002). Other theories posit generation of the SL portion basal to that of the LL, either
through nonlinear distortion (Siegel et al., 2005) or coherent reflection (Choi et al., 2008;
Goodman et al., 2011; Moleti et al., 2013). Accordingly, as part of understanding the

 
  76  

generation of the SL portion of the TEOAE, it is necessary to know where along the
length of the cochlear partition it is generated.
Recently the role of intermodulation distortion in the generation of the SL portion
of the TEOAE has been called into question. For instance, using a suppression paradigm
with highpass noise suppressing the f2 location, Goodman and colleagues (2011)
measured no change in the SL magnitude. Had the SL portion of the TEOAE been
generated through 2 f1 ! f2 distortion, the highpass noise would have been expected to
reduce its magnitude. Findings from Sisto et al. (2013) similarly cast doubt on SL
generation through intermodulation distortion. These investigators measured similar
growth rates and latencies between SL portions in both the TEOAE and SFOAE. This
was despite the fact that the stimulus used to evoke the SFOAE did not have energy at
either f1 or f2 .
In light of these findings, basal reflection has been suggested as a likely
mechanism through which the SL portion of the TEOAE is generated (Goodman et al.,
2011). Per this theory, the different-latency portions of the TEOAE mirror the underlying
mechanics of the BM at different locations. Moleti and colleagues (2013) have recently
presented simulations from a nonlinear cochlear model demonstrating the plausibility of
the different-latency portions of the TEOAE arising from different cochlear locations

through coherent reflection. In their model, impedance irregularities along the cochlear
partition (referred to as cochlear roughness and required for linear coherent reflection;
Zweig and Shera, 1995) were removed at various cochlear locations and the effect on the
simulated TEOAE was examined. When roughness was preserved at locations basal to
the tonotopic place, the TEOAE included both a SL and LL portion; however, upon
removing impedance irregularities basal to the tonotopic place, the SL portion was either
reduced in magnitude or eliminated while the LL portion persisted. Their results
suggested that the greatest contribution to the SL portion occurs at BM locations basal to
the tonotopic place by at least 1/4-octave.

 
  77  

Not all data agrees with an area of TEOAE generation constrained to a basal
region close to the CF place, however. Several studies have demonstrated that cochlear
trauma restricted to regions beyond 1/2-octave basal to the CF place may affect both
TEOAE magnitude and latency (Avan et al., 1995; 1997; Withnell et al., 2000; Lucertini
et al., 2002; Jedrzejczak et al., 2005). If SFOAEs and TEOAEs are assumed to represent
the same cochlear mechanisms (Kalluri and Shera, 2007; Sisto et al., 2013) then
additional evidence seems to implicate a region of generation extending to much more
basal regions of the cochlea. For instance, both Guinan (1990) and Siegel et al. (2004)
have measured two-tone suppressive effects on the SFOAE by suppressor tones at
frequencies more than 2-octaves higher than the frequency of the SFOAE.
Previous TEOAE suppression studies have examined the effect of suppression on
the LL portion of the TEOAE only, due to either low-stimulus levels and/or analysis
techniques whereby the initial milliseconds of the TEOAE are discarded (Tavartkiladze
et al., 1994; Zettner & Folsom, 2003). Knowledge of the generation location(s) of the SL
portion is expected to provide greater insight into the mechanisms responsible for its
generation. The goal of the current study was to determine the generation locations along
the BM of the SL and LL portions of the TEOAE using a two-tone suppression paradigm.
4.2 Methods

4.2.1 Subjects
Ten subjects (4 male) between the ages of 18 – 29 years old participated in the
study. All subjects had normal hearing thresholds (≤ 20 dB HL) at the octave frequencies
between 0.25 – 8 kHz in the test ear (5 right ears, 5 left ears). Data collection was
completed over the course of 2 visits (~2 hours each) spaced within 1 month of one
another. Tympanometry (226 Hz) was performed at each visit to ensure the subject’s test
ear showed no evidence of middle ear pathology. The Internal Review Board at the
University of Iowa approved all testing.

 
  78  

4.2.2 Signal Generation and Data Acquisition


Stimuli were digitally created at a 44.1 kHz sampling rate using custom written
MATLAB (Mathworks, Inc.) software and a personal computer (PC) running the
Windows XP (Microsoft, Inc.) operating system. Stimuli were routed from the PC to an
external 24-bit soundcard (UltraLite-mk3 Hybrid, Mark of the Unicorn) through a USB
2.0 interface. The resulting two-channel electrical signal was directed through a pair of
earphones (Sennheiser IE8) and transduced into acoustic pressure. The output of each
earphone was guided through silicone tubing to a receiver port of an ER10B+ probe
assembly (Etymotic Research, Inc.). The ER10B+ probe assembly was coupled to the ear
canal by an ER10-14 foam eartip (Etymotic Research, Inc.). The ear canal pressure
responses to the stimuli were detected and transduced into an electrical signal by a
miniature microphone housed in the ER10B+ probe assembly. The microphone voltage
was amplified 20 dB by the ER10B+ preamplifier, and subsequently routed to the
external soundcard and digitized at a sampling rate of 44.1 kHz. The digital signal was
sent to the PC and stored on the hard drive for offline analysis. Stimulus delivery and
recording was accomplished using MATLAB and custom written software.
4.2.3 Measurement and Analysis of TEOAEs
4.2.3.1 Stimuli

TEOAEs were measured across three different suppressor conditions; suppressors


included: a) 2 kHz tone-bursts, b) different frequency pure tones, and c) high-frequency,
band-limited noise. For each condition, stimuli were presented to the ear using a double-
evoked, double-source paradigm (Keefe, 1998; Keefe and Ling, 1998; Keefe et al., 2008).
A stimulus buffer was composed of three consecutive stimulus intervals ( s1 , s2 , and s1,2 ).
The duration of each interval depended on the suppressor stimulus: Intervals were 40 ms
long for the tone-burst suppressor condition (total buffer duration of 120 ms) and 50 ms
long for the pure tone and noise-burst conditions (total buffer duration of 150 ms).
Stimulus interval s1 contained the TEOAE probe stimulus (a 2 kHz tone-burst) routed

 
  79  

through channel 1 of the transducer. Stimulus interval s2 contained the suppressor


stimulus routed through channel 2 of the transducer. Both stimuli were presented
simultaneously during stimulus interval s1,2 . Figure 4.1 illustrates the double-evoked
paradigm. The top panel shows the probe (a 2 kHz tone-burst). Each subsequent panel
shows a different suppressor condition, described in more detail below.
a. Tone-burst suppressor. The tone-burst suppressor condition used a 3-cycle, 2
kHz tone-burst as both the probe (Fig. 4.1, 1st panel) and suppressor (Fig 4.1, 2nd panel).
The tone-burst was generated in cosine phase and windowed with a hann window. Tone-
bursts were calibrated to have levels between 44 – 86 dB peak SPL (pSPL) in a
calibration cavity (see section below). The probe and suppressor tone-bursts were always
the same level. When presented simultaneously during stimulus interval s1,2 , the
resulting tone-burst generated in the ear canal was 6 dB higher than either tone-burst
presented alone. The number of stimulus buffers presented depended on the tone-burst
level with more presentations for lower level tone-bursts to improve the signal-to-noise
ratio (SNR) of the evoked emission – 2000 buffers were presented for levels between 44
– 56 dB pSPL, 1600 were presented for levels between 62 – 74 dB pSPL, and 1200 were
presented for levels between 80 – 86 dB pSPL.
b. Pure tone suppressor. The probe for this condition was the same 3-cycle, 2 kHz

tone-burst described above, presented at a level of 74 dB pSPL. The suppressor was a 28


ms pure tone (Fig. 4.1, 3rd panel). The first and last 5 ms were windowed by a 1/2-cycle
hann window. The frequency of the suppressor tone ranged from 2-octaves below the
probe frequency (0.5 kHz) to 1.585 octaves above the probe frequency (6 kHz). Specific
frequencies included 0.5, 0.71, 1 – 3.86 (~1/7-octave steps), 5 and 6 kHz.
Suppressor tones were presented at probe-to-suppressor level ratios (PSRs) of 0,
12, and 24 dB. Suppressor levels were calibrated in terms of the dB pSPL generated in a
calibration cavity (see below). Not all suppressor frequencies were presented at each
PSR, as pilot data suggested minimal effects of certain suppressor frequencies when

 
  80  

presented at the lower PSRs (Lewis and Goodman, 2012). At 0 dB PSR, all frequencies
were presented. At 12 dB PSR, frequencies between 1 – 6 kHz were presented. At 24 dB
PSR, frequencies between 1 – 3.86 kHz were presented. The suppressor tone (for a given
frequency) alternated between cosine and negative cosine phase across adjacent stimulus
buffers. The alternating of phase between adjacent stimulus buffers constituted a buffer
pair and was necessary to ensure that any OAEs induced by the suppressor tone would
cancel upon averaging (Shera et al., 2004). Eighteen hundred stimulus buffers (900 buffer
pairs) were presented for each suppressor tone condition.
c. Noise-burst suppressor. The probe for this condition was the 3-cycle, 2 kHz
tone-burst described above, presented at a level of a 74 dB pSPL. The suppressor was a
28 ms noise-burst, band-limited from 8 – 16 kHz (Fig. 4.1, 4th panel). The noise was
calibrated to have a flat spectrum from 8 – 16 kHz in the calibration cavity and a root-
mean-square (RMS) level of 80 dB SPL (see below). Thirty different frozen noise-bursts
were randomly generated for each subject and served as the suppressor stimuli. Each
frozen noise stimulus alternated between in-phase and out-of-phase across adjacent
stimulus buffers to form buffer pairs, similar to the pure-tone suppressor conditions. Each
frozen noise-burst pair was presented 15 times before a different frozen noise-burst pair
was presented. Altogether, 1800 stimulus buffers (900 buffer pairs) were presented.

4.2.3.2 Calibration
All probe stimuli and the suppressor stimuli for the tone-burst and pure tone
suppressor conditions were calibrated in terms of the peak pressure generated in a long,
plastic tube (2 m length, 7 mm inner diameter) terminated on one end by a steel bearing.
The noise-burst suppressor was calibrated in terms of the RMS pressure in the long-tube.
The long-tube calibration technique was chosen because it is insensitive to standing-wave
effects that are present for traditional in-situ dB SPL ear canal calibrations (Sachs &
Burkhard, 1972; Stinson et al., 1982; Gilman & Dirks, 1986). Additionally, the maximum
voltage drive to the transducer is strictly controlled and the possibility of inadvertently

 
  81  

overdriving the transducers is eliminated. This technique has been described previously
(Goodman et al., 2009). Briefly, the probe tip is placed into the long tube and stimuli are
presented and recorded repeatedly. The pressure recorded for each stimulus presentation
includes the incident pressure delivered by the probe and multiple reflections resulting
from the stimulus pressure traveling back-and-forth between the two ends of the tube.
With each round trip, the amplitude of the reflected stimulus pressure is attenuated due to
viscothermal losses. In order to isolate the incident pressure delivered by the probe to the
tube it is necessary to use a short-duration stimulus and a slow stimulus repetition rate. If
the stimulus duration is too long, the incident pressure will overlap in time with the
reflected pressure. If the stimulus repetition rate is too fast, the incident pressure will also
overlap with the internal reflections of the previously presented stimulus. The appropriate
stimulus duration can be calculated from the round-trip travel time of sound in the tube.
For the 2 m long tube used in the current study, the first reflection will occur at
approximately 11.6 ms. The appropriate presentation rate depends on various factors of
the tube including its length, material, and how each end is terminated (i.e., an absorptive
versus reflective material). Accordingly, the presentation rate must often be determined
empirically.
One important assumption made when using the long-tube calibration is that the

acoustic surge impedance of the tube is similar to the ear canals being tested. This
impedance is determined primarily by the cross sectional area of the cavity. The average
adult ear canal is approximately 8 mm in diameter (Stinson & Lawton, 1989), close to the
7 mm diameter tube used for calibration. Another important assumption is that individual
differences in ear canal and middle ear impedances are relatively small. This assumption
appears justifiable in the present study, given that only young adults were tested and the
inclusion criteria included normal hearing and normal middle ear function. Further
calibration details specific to each suppressor type are given below.

 
  82  

a. Tone-burst suppressor. Sixty-four repetitions of a 10 ms 2 kHz tone-burst were


delivered to the calibration tube at a rate of 1/s and recorded. Recordings were highpass
filtered using a finite-impulse-response (FIR) digital filter (250 Hz cutoff, 256 order).
Only the initial 10 ms of each filtered recording was retained. The pressure measured in
this time window was the incident pressure delivered by the transducer to the tube and
free from internal reflections arising from the distal end of the tube. The amplitude of the
electrical tone-burst stimulus was adjusted to yield acoustic levels (as measured in the
long tube) from 44 – 86 dB pSPL (6 dB steps).
b. Pure tone suppressor. Calibration of the pure tone suppressors was achieved
following procedures similar to that described for the tone-burst suppressors. Tone-bursts
(10 ms) at each suppressor frequency were presented and a separate calibration was made
for each suppressor frequency. The electrical stimuli were adjusted in amplitude to yield
acoustic levels (as measured in the long tube) of 74 dB pSPL (0 dB PSR), 62 dB pSPL
(12 dB PSR), and 50 dB pSPL (24 dB PSR).
c. Noise-burst suppressor. Sixty-four repetitions of a 10 ms frozen, white noise-
burst were delivered to the calibration tube at a rate of 1/s and recorded. Recordings were
high-pass filtered using a FIR digital filter (250 Hz cutoff , 256 order) and synchronously
averaged in the time-domain. The electrical drive to the transducer for the white noise-

burst ( xcal [ n ] , n denotes time in terms of sample number) and the average pressure
recorded in the long tube ( ycal [ n ] ) were transformed to the frequency domain ( Xcal [ m ]
and Ycal [ m ] , respectively, m denotes frequency bin number) using the discrete Fast
Fourier transform (FFT). The magnitude of the frequency-domain inverse transfer
function of the calibration tube was then calculated,

Xcal [ m ]
!1
H cal [m] = . Eq. 4.1
Ycal [ m ]

 
  83  

Since the noise-burst suppressor was to have a bandwidth from 8 – 16 kHz, the
magnitudes for the frequency bins corresponding to 0 – 8 kHz and 16 – 22.05 kHz were
set to 0. Phase was set to zero at all frequencies. The inverse transfer function was
converted from polar to rectangular notation and converted back to the time domain ( hcal
!1
)
via the inverse discrete FFT. Only the real part of the inverse transfer function was
retained. The time-domain inverse transfer function was shifted, truncated to 128
samples, and windowed using a hann window to generate an impulse response.
The frozen white noise-burst ( xcal ) was convolved with the impulse response to
yield a high-frequency, band-limited noise-burst ( zcal ) that would have a flat magnitude
spectrum between 8 – 16 kHz in the calibration tube. The noise-burst was corrected for
the group delay of the impulse response and bandpass filtered by an FIR digital filter (8
kHz low-cut, 16 kHz high-cut, 1024 order) to further attenuate any residual energy
outside of the 8 – 16 kHz frequency band. Sixty-four repetitions of the filtered noise-
burst were delivered to the calibration tube at a rate of 1/s and recorded. After truncating
the recorded tube pressure to the initial 10 ms and highpass filtering (see above), a
calibration factor relating the RMS of the electrical stimulus ( zcal ) to that of the acoustic
pressure measured in the calibration tube was calculated. The 30 different frozen noise-
bursts suppressors were multiplied by the calibration factor to yield RMS levels of 80 dB

SPL in the calibration tube.


4.2.3.3 Analysis
The suppressed TEOAE was extracted by first dividing the ear canal pressure
recordings for each stimulus buffer into three time segments ( p1 , p2 , and p1,2 ). Each time
segment corresponded to each of the three stimulus intervals ( s1 , s2 , and s1,2 , see Fig.
4.1). The nonlinear differential response ( pD ) was calculated by adding the ear canal
pressures for the probe alone ( p1 ) and the suppressor alone ( p2 ) and subtracting the ear
canal pressure for the probe and suppressor together ( p1,2 ). The nonlinear differential

 
  84  

response quantifies the nonlinear interaction between the probe and suppressor stimuli
within the cochlea. In the absence of nonlinearity, pD contains only noise.
The nonlinear differential responses for each suppressor condition were highpass
filtered using an FIR digital filter (250 Hz cut, 128 order). The highpass filtered
responses were then bandpass filtered using a 1/3-octave bandpass FIR digital filter (2
kHz center frequency, 512 order). The RMS levels (Pa) of the filtered responses were
calculated and subjected to an artifact rejection algorithm (Goodman et al., 2009; see
Hoaglin et al., 1983) in order to identify recordings contaminated by high-levels of
intermittent noise. The first- and third-quartiles of the RMS levels were calculated and
the interquartile range was derived. Recordings with an RMS exceeding the third quartile
by more than 1.5 times the interquartile range were excluded from further analysis.
Whenever a response was discarded, so was its pair (adjacent response where the
suppressor was of opposite phase – pure tone and noise-burst suppressor conditions).
Retained responses were divided into 2 collection buffers. Odd numbered response pairs
were stored in one buffer and even numbered pairs were stored in a second buffer. For
each collection buffer, the recordings were synchronously averaged in the time domain.
An estimate of the 2 kHz TEOAE signal was calculated as the mean of the averaged
recordings for the 2 collection buffers ( pOAE ). The analytic expression ( p̂OAE [ n ] ) of the

TEOAE signal was calculated using the discrete Hilbert transform (denoted by H ),

p̂OAE [ n ] = pOAE [ n ] + jH ( pOAE [ n ]). Eq. 4.2

From the analytic signal, the TEOAE instantaneous magnitude ( p̂OAE ), or envelope, was
calculated as

2 2
( ) ( )
p̂OAE [ n ] = Re p̂OAE [ n ] + Im p̂OAE [ n ] . Eq. 4.3

 
  85  

An estimate of the time-domain noise was calculated by subtracting the averaged


responses of the two collection buffers and dividing by 2 (Kemp et al., 1990).
In order to measure the effects of the different suppressors on the total TEOAE
(including the SL and LL portions) as well as on each individual portion, the TEOAE
signal and noise estimates were analyzed across various time windows. The first window
extended from stimulus onset to 10 ms post stimulus offset. The duration of this window
was determined based on the expected latency of a low-level 2 kHz TEOAE (Neely et al.,
1988; Tognola et al., 1997; Sisto et al., 2007). This analysis window included both the SL
and LL portions of the TEOAE and is hereafter referred to as ALL. A time vector was
mapped to the analysis window with time zero corresponding to the peak of the tone-
burst stimulus. The initial 1 ms, corresponding to time from -0.75 – 0.25 ms was set to
zero in order to remove any non-cancelling stimulus artifact from the analysis. The ALL
time window was partitioned into multiple, shorter-duration, windows corresponding to
different-temporal portions of the TEOAE. The following method, performed on a
subject-by subject basis, was used to determine the bounds of these windows: TEOAE
envelopes measured in the tone-burst suppressor condition were averaged across the
different stimulus levels to yield the mean envelope. Magnitude nulls in the mean
envelope were identified, and the time between consecutive nulls was used to define the

time window for a specific portion of the TEOAE. Figure 4.2 shows an example of this
method. Three different temporal portions of the TEOAE were evident for this particular
subject. The long-latency (LL) portion was defined between 6.01 – 10.77 ms. Two SL
portions were identified. The first SL portion (SL1) was defined from 3.4 – 6.01 ms, and
the second SL portion (SL2) between 0.25 – 3.4 ms. Although three temporal portions of
the TEOAE were defined for the subject shown in Fig. 4.2, this was not always the case.
As discussed in the introduction, the TEOAE envelope does not always exhibit well-
defined magnitude peaks and nulls. Accordingly, the number of temporal regions defined

 
  86  

varied from subject-to-subject. However, the TEOAE was always divided into at least
two temporal portions corresponding to LL and SL energy.
TEOAE magnitude and latency were calculated within each time window for
each suppressor condition. Magnitude was calculated as the RMS of the TEOAE
envelope within the analysis window. This definition of magnitude differs slightly from
more traditional measures that are computed on the TEOAE waveform. When performed
on the envelope, the RMS of the TEOAE is 3 dB higher than when performed on the
waveform. This difference occurs because the calculation of the envelope includes both
the real and imaginary parts of the analytic signal (see Eq. 4.3). The magnitudes of the
real and imaginary parts of the analytic signal are identical while their phases differ by

! 2 radians. When the respective magnitudes are squared and summed, intensity is
doubled. When the RMS of the waveform is calculated, only the real part of the analytic
signal is involved in the calculation and no doubling of intensity occurs.
Latency ( ! OAE ) was calculated as the energy-weighted mean of the time vector ( t
) for the analysis window,

N!1

"( p̂ [n] t [n])


2
OAE
! OAE = n=0
N!1
. Eq. 4.4
"( p̂ [n] )
2
OAE
n=0

Similar to the calculation for TEOAE magnitude, the latency calculation also used the
TEOAE envelope (the square of the envelope served as the weighting function). It is
important to note that latency was calculated for each analysis window and; therefore, the
value of the time vector at n = 0 depended on the particular time window under
consideration.
To determine whether a TEOAE was present in the analysis window, the SNR
was calculated by dividing the RMS of the TEOAE envelope by the RMS of the noise

 
  87  

envelope (linear units). Again, the SNR was calculated across each time window. An
SNR criterion value of 6 dB was adopted. Whenever the SNR of the time-windowed
portion of the TEOAE exceeded 6 dB, the TEAOE was defined as being present.
4.3 Results
4.3.1 Tone-Burst Suppressor Condition
Figures 4.3 and 4.4 show the TEOAE envelopes measured for the tone-burst
suppressor condition. Each panel shows the level-series for a particular subject (indicated
by the subject number and test ear). Multiple magnitude peaks characterized majority of
the TEOAE envelopes. As mentioned in the Introduction, these peaks have traditionally
been interpreted as representing different packets of similar-frequency TEOAE energy
(referred to as components). In contrast, the TEOAE envelopes for subjects 01L and 12R
lacked distinct magnitude peaks and were more consistent with a single packet of
TEOAE energy dispersed across time.
Of the 10 subjects tested, 2 temporal portions of the TEOAE were identified for 1
subject (12R), 3 portions were identified for 7 subjects, and 4 portions were identified for
2 subjects (17R and 18R). Across all subjects, several trends were apparent. At the lowest
stimulus levels, the peak magnitude of the LL portion of the TEOAE was largest of the
different temporal portions. As stimulus level increased, magnitude peaks within the SL

portions of the TEOAE emerged, growing less compressively than the LL peak. At the
highest stimulus levels, the peak magnitudes of the SL portions were often larger than
that of the LL portion. The time index associated with each of the different magnitude
peaks remained nearly constant as stimulus level increased.
Across all subjects, the mean time window for the LL portion of the TEOAE was
bounded between 7.01 (standard deviation = 0.90) – 10.38 (0.53) ms. The mean time
window for the SL1 portion extended from 3.91 (1.50) – 7.01 (0.90) ms. The mean time
window for the SL2 portion was 0.68 (0.86) – 4.31 (0.81) ms. Since an SL3 portion was
only evident in 2 subjects, it was not included in the group analysis. Figure 4.5 plots the

 
  88  

magnitudes (RMS) of the different-latency portions of the TEOAE for each stimulus
level as a function of latency (calculated by Eq. 4.4). Each vertical line illustrates the
magnitude growth for a specific portion of the TEOAE measured in a single subject. The
magnitudes for each portion of the TEOAE were normalized so that their magnitudes
were 0 dB for the lowest-level tone-burst (44 dB pSPL). Latency changed only slightly
with stimulus level (shown by most of the lines being nearly vertical). This finding is
consistent with previous reports that have noted the level-invariant nature of the different
portions of the TEOAE (Goodman et al., 2009; Goodman et al., 2011; Moleti et al.,
2012a). The mean latency of the LL TEOAE, averaged across subjects and stimulus
levels, was 8.56 ms (standard deviation = 0.64) compared to 5.49 ms (0.81) and 2.89 ms
(0.79) for the SL1 and SL2 TEOAEs, respectively. Relative to the LL TEOAE, the SL1
latency was shorter by a factor of 1.56. Both Goodman et al. (2009) and Moleti et al.
(2012a) have reported a similar difference between the LL and SL1 latencies. The latency
of the SL2 TEOAE was shorter by factors of 2.95 and 1.89 relative to the latencies of the
LL and SL1 TEOAEs, respectively. Also consistent with previous reports, the SL
portions of the TEOAE (SL1 and SL2) grew less compressively than the LL portion.
The top panel of Fig. 4.6 plots the input-output (IO) functions of the different-
temporal portions of the TEOAE (LL, SL1, and SL2), averaged across all subjects. These

functions were each fit by a 2nd-order polynomial (indicated by the solid line). The
middle panel of Fig. 4.6 plots the slopes of the IO functions. Slopes were calculated by
differentiating the polynomial fits to the IO functions. Accordingly, the slope at a given
stimulus level represents the growth rate of the TEOAE. The bottom panel of Fig. 4.6
plots the latency-intensity functions of the different-temporal portions of the TEOAE,
averaged across all subjects. Included in each plot are the measurements for the TEOAE
across the ALL time window (this time window spanned all temporal portions of the
TEOAE). Traditionally, calculations of TEOAE magnitude and latency are not performed

 
  89  

on the individual portions of the TEOAE but rather across the total TEOAE. The ALL
response thus allows for comparisons with previously published data.
The mean data of Fig. 4.6 are comparable to the individual results discussed
previously. The top panel shows the magnitudes of the different temporal portions of the
TEOAE and how the portion of the TEOAE with the largest magnitude depended on
stimulus level. For stimulus levels below 62 dB pSPL, the magnitude of the LL portion
was largest. For stimulus levels between 62 – 74 dB pSPL, the magnitude of the SL1
portion was largest. At stimulus levels beyond 74 dB pSPL, the SL2 magnitude was
largest. The middle panel illustrates the compressive growth of the different temporal
portions. All portions of the TEOAE grew compressively, suggesting that generation
occurred within the region over which the cochlear amplifier was active. Across the
different time windows, the growth of the LL TEOAE was most compressive.
Compression increased with stimulus level, evidenced by the growth rate decreasing from
0.49 dB/dB at the lowest levels to less than 0 dB/dB at the highest levels. The SL1
TEOAE also exhibited increasingly compressive growth as stimulus level increased but
compression was less than that of the LL TEOAE. Growth decreased from 0.87 dB/dB at
the lowest stimulus levels to 0.12 dB/dB at the highest levels. Compared to the SL1 and
LL portions of the TEOAE, the growth rate of the SL2 portion was less compressive.

Across the stimulus levels where the SNR was at least 6 dB, the growth rate remained
relatively stable around 0.82 dB/dB. The bottom panel shows the relationship between
the latency of each temporal portion of the TEOAE and stimulus level. As previously
mentioned, latency was nearly invariant with stimulus level.
Considering the total TEOAE (ALL) mean response (gray lines) in Fig. 4.6, the
IO function was compressive, and compression increased with stimulus level. The growth
rate decreased from 0.55 dB/dB at the lowest stimulus level to 0.40 dB/dB at the highest
stimulus level. Previous reports of TEOAE growth rates have ranged from 0.3 – 0.5
dB/dB (Kemp & Chum, 1980; Probst et al., 1986; Verhulst et al., 2011). In contrast to the

 
  90  

latency behavior of the different-temporal portions of the TEOAE, the latency of the total
TEOAE changed with stimulus level. Latency decreased from 7.80 ms at 44 dB pSPL to
3.70 ms by 86 dB pSPL. This represented a decrease on the order of 47% and was the
result of the dominant magnitude portion of the TEOAE changing from the LL at low-
levels to the SL at high-levels. TEOAE data have shown similar decreases in latency
when compared across a comparable range of stimulus levels (Neely et al., 1988; Sisto &
Moleti, 2007; Rasetshwane et al. 2013). This decrease has been attributed to the
generation region of the TE/SFOAE shifting basally as the peak of the cochlear traveling
wave moves to more basal regions at higher stimulus levels (Ruggero et al., 1997; Rhode,
2007). In order for this explanation to be valid, the SL portions of the TEOAE (SL1, SL2,
etc.) would need to be generated basal to the generation site of the LL portion. For it is
the presence of the SL portions of the TEOAE and their increasingly large magnitudes at
higher stimulus levels that give rise to the level-dependency of the total TEOAE latency.
Recall that when considered in isolation, the latency of each portion of the TEOAE is
level-invariant. The following section presents the two-tone suppression data from the
pure tone and noise-burst suppressor conditions used to determine the generation places
of the different-temporal portions of the TEOAE.
4.3.2 Pure Tone and Noise-Burst Suppressor Conditions

To determine the regions along the basilar membrane contributing to the TEOAE,
the suppression induced by different suppressor frequencies on each portion of the
TEOAE was measured. Figures 4.7 and 4.8 show suppression contour plots for the 10
subjects whose TEOAE envelopes were shown in Figs. 4.3 and 4.4, respectively. These
graphs illustrate the amount of suppression induced by each suppressor frequency, and
the band-limited noise-burst (BLN), on different-temporal portions of the 2 kHz TEOAE.
Light regions represent time-frequency combinations where suppression was minimal
and dark regions indicate time-frequency combinations where suppression was greatest.
Overlaid on each plot are the time windows corresponding to the locations of each

 
  91  

temporal portion of the TEOAE (determined from the tone-burst suppressor condition). If
the different portions of the TEOAE were generated at different locations along the
basilar membrane, then each would be most sensitive to suppression by a different range
of frequencies.
Contour plots are shown only for the 12 dB PSR condition (including the noise-
burst suppressor). Corresponding plots for the 0 dB PSR condition exhibited broad
regions of suppression making it difficult to determine whether the different portions of
the TEOAE were differentially suppressed by the suppressor frequencies. The contour
plots for the 24 dB PSR condition were similarly difficult to interpret; however, this was
due to a poor SNR. Group effects of suppressor level on the different portions of the
TEOAE will be considered in further detail following the discussion of the individual 12
dB PSR contour plots.
In very general terms, the temporal portion of the TEOAE suppressed by a given
pure tone was inversely proportional to the tone’s frequency: Lower frequency tones
suppressed the LL portion of TEOAE, mid-frequency tones suppressed the SL1 portion,
and higher frequency tones suppressed the SL2 and SL3 (when present) portions. These
results are consistent with the TEOAE being composed of contributions from different
regions along the basilar membrane and the SL portion being generated basal to the LL

portion. A very clear example of the differential effect suppressors had on the TEOAE
portions is evident in the contour plot for subject 19L (Fig. 4.7, bottom-right). The LL
portion (7 – 10.77 ms) was maximally suppressed by frequencies between 1.4 – 2.1 kHz,
the SL1 portion (3.51 – 7 ms) was maximally suppressed by frequencies between 2.1 –
2.8 kHz, and the SL2 portion (0.25 – 3.51 ms) was maximally suppressed by frequencies
between 2.5 – 5 kHz.
In some instances, the time window for a particular portion of the TEOAE
exhibited multiple frequency regions across which suppression was especially effective.
A good example of this is found in the data for subject 14L (Fig. 4.7, bottom-middle)

 
  92  

within the SL2 time-window. One region of suppression is centered at 3.1 kHz, a second
region is centered at 3.9 kHz, and a third region at 6 kHz. For this particular subject, each
of these different “hot-zones” of suppression is associated with a slightly different time
index. The 6 kHz suppressor causes suppression on an earlier portion of the TEOAE in
the SL2 window compared to the 3.9 kHz suppressor which, in-turn, suppresses an earlier
portion of the TEOAE than the 3.1 kHz suppressor. These results further support the idea
that the bandpass filtered TEOAE originates across a distributed region of the BM and SL
portions are generated basal to the LL portion.
In certain cases the most effective suppressors of one temporal portion of the
TEOAE were also the most effective suppressors of a different portion. For instance,
overlap between the most effective suppressors is especially evident across the SL1 and
SL2 portions of the TEOAE for subject 06L (Fig. 4.8, top-right panel). This could
suggest that both portions of the TEOAE were generated at the same location on the BM
but differed in how they propagated back towards the stapes. For instance, the SL2
portion may have propagated backward through a fast compression wave (Ren, 2002)
while the SL1 component propagated backward through a slower reverse-traveling wave.
Another possibility is that the excitation on the basilar membrane across the generation
region of the SL2 TEOAE was simply more sensitive to low-side suppression

(suppression by tones with lower-frequencies than that of the generation place). The BM
response to the probe stimulus (2 kHz tone-burst) was likely larger at the SL1 generation
region than at the more basal SL2 region. Accordingly, frequencies that caused greatest
suppression at the SL1 place would also be expected to be effective suppressors of the
SL2 place. This explanation seems more likely than the idea that a portion of the SL
energy returns to the stapes through a fast compression wave. For instance, although the
SL2 portion of the TEOAE for subject 06L was suppressed by some of the same
frequencies as the SL1 portion, suppression was also induced by higher frequencies that
did not suppress the SL1 portion. Therefore, it seems likely that the SL2 portion of the

 
  93  

TEOAE was primarily generated across a different region of the basilar membrane than
the SL1 portion of the TEOAE.
Minimal suppression of the TEOAE was typically induced by suppressor
frequencies above 4 – 5 kHz. This was especially true concerning the LL and SL1
portions of the TEOAE. In terms of the SL2 portion, there were several instances where
the 6 kHz suppressors and the high-frequency, band-limited noise-burst (BLN) induced
suppression. Subjects 10R and 14L (Fig. 4.7, top-right and bottom-middle, respectively)
both show a region of suppression beginning around 5 kHz and extending at least to 8
kHz (the lower frequency limit of the noise burst). However, in most cases, the upper
limit of suppression for the SL2 portion of the TEOAE was 5 kHz.
Figure 4.9 shows iso-level curves (suppressor level), averaged across subjects,
for each portion of the TEOAE (LL, SL1, and SL2) and the total TEOAE (ALL). Line
thickness denotes PSR condition (0, 12, and 24 dB) with the thickest line corresponding
to 0 dB PSR. There are several features common to the different portions of the TEOAE
(LL, SL1, and SL2). First, the greatest amount of suppression occurred when the
suppressor was the same peak level as the probe (0 dB PSR). For this PSR, the shapes of
the suppression curves generally represented lowpass filters (component SL2 was
somewhat of an exception to this). This shape was the result of greater low-side

suppression (suppressor frequencies < 2 kHz) compared to high-side suppression


(frequencies > 2 kHz). Second, as the suppressor level decreased, the shapes of the
suppression curves became increasingly narrow and changed from resembling a lowpass
filter to a bandpass filter. This effect was primarily due to low-side suppression
decreasing with suppressor level at a faster rate than high-side suppression.
Despite their similarities, the iso-level curves were not identical across the
different portions of the TEOAE. First, each portion of the TEOAE was most effectively
suppressed by a different suppressor frequency. The LL portion (1st panel) was most
sensitive to suppression by a 2.07 kHz tone, the SL1 portion (2nd panel) was most

 
  94  

sensitive to suppression by a 2.55 kHz tone, and the SL2 portion (3rd panel) was most
sensitive to suppression by a 3.14 kHz tone. These frequencies were determined from the
24 dB PSR condition and are indicated by the vertical gray lines in Fig. 4.9. A second
difference between the temporal portions of the TEOAE was the change in suppression
that accompanied a decrease in the suppressor level. As the suppressor level decreased,
the degree to which the different portions of the TEOAE were suppressed also decreased.
However, this was not constant across the different TEOAE portions. The most apparent
difference was for the SL2 portion compared to the later occurring portions (LL and
SL1). For a 24 dB decrease in the suppressor level, the amount of suppression induced by
the 3.14 kHz tone (the frequency at which the SL2 TEOAE was most sensitive to
suppression) on the SL2 portion decreased by less than 2 dB. In contrast, identical
decreases in the suppressor levels at 2.07 kHz and 2.55 kHz reduced suppression of the
LL and SL1 portions by 7.5 dB and 9.8 dB, respectively.
This difference may be explained by first interpreting the peaks of the suppression
curves as corresponding to the approximate generation regions of the different-temporal
portions of the TEOAE. Accordingly, the LL portion would be generated at
approximately the 2 kHz place (corresponding to the tonotopic region of the 2 kHz tone-
burst), the SL1 portion would be generated between 1/4 - 1/3-octaves basal, and the SL2

portion would be generated approximately 3/5-octaves basal. Suppression of TEOAEs is


presumably the result of the nonlinear BM response to the probe frequency being reduced
by the presence of the suppressor frequency (Killan et al., 2012; see Brass and Kemp,
1993 for SFOAE suppression). If the displacement caused by the suppressor tone at the
particular place on the BM is much larger than the displacement caused by the probe, the
response to the probe will be minimal. At the SL2 generation place, basal to both the SL1
and LL places, the response to the probe is likely smaller than the responses at the SL1
and the LL places. Thus, the effectiveness of a suppressor within the generation region of
the SL2 TEOAE would be greater than for suppressors within the generation regions of

 
  95  

the later portions of the TEOAE. This argument, however, is not entirely satisfying since
the SL1 component would also be expected to be more effectively suppressed than the
LL component, which was not the case.
The iso-level curves for the ALL TEOAE time window (Fig. 4.9, 4th panel) were
qualitatively very similar to those for the LL, SL1, and SL2 time windows. The most
notable difference was that the suppressor frequency at which greatest suppression was
induced increased as the suppressor level decreased. For the 0 dB PSR condition, greatest
suppression occurred for the 2.07 kHz tone. However, the 2.55 kHz and 3.14 kHz tones
caused the greatest amount of suppression when the PSR was 12 dB and 24 dB,
respectively. Since the ALL TEOAE included all portions of the TEOAE, the suppression
curves are essentially the averages of the curves for the LL, SL1, and SL2 portions. The
observed increase in the most effective suppressor frequency with decreasing suppressor
level can thus be traced to the temporal portion of the TEOAE that had the largest
suppressed magnitude. For instance, the suppressed magnitude of the SL2 portion at 3.14
kHz was the largest across all suppressor frequencies for the 24 dB PSR condition. As a
result, the 24 dB PSR suppression curve of the ALL TEOAE peaked at 3.14 kHz.
4.4 Discussion
4.4.1 SL and LL Portions of the TEOAE

The findings of this study build on previous work demonstrating that the bandpass
filtered TEOAE contains both LL and SL portions that exhibit different growth rates
(Goodman et al., 2009; 2011; Molet et al., 2012a; Sisto et al., 2013). The differential
growth rates between the different-temporal portions of the TEOAE have the effect of the
dominant portion of the TEOAE (portion with the largest magnitude) changing with
stimulus level. If the TEOAE is analyzed across a broad time-window (such as was used
for the ALL condition), the shift in the dominant portion of the TEOAE from the LL to
the SL2 is evidenced by a decrease in TEOAE latency with increasing stimulus level
(Neely et al., 1988; Sisto and Moleti, 2007; Rasetshwane et al., 2013; Moleti et al.,

 
  96  

2012a). Recall that the latencies of the LL, SL1, and SL2 portions were level-invariant.
Traditional measures of latency, whether frequency- or time-domain based, attach
greatest weight to the time-index associated with the peak in TEOAE amplitude.
Accordingly, as the peak shifts from the LL to the SL portions, a decrease in latency will
be measured. A recent report by Rasetshwane et al. (2013) illustrates this effect in that 2
kHz TEOAE latency, calculated across a broad time window, decreased from ~11 ms – 4
ms across a stimulus range spanning 20 – 90 dB SPL. This decrease in latency
approximates the difference between the SL2 and LL components measured in the
present study.
4.4.2 Generation Regions of the TEOAE
The two-tone suppression results are consistent with the different portions of the
TEOAE being generated at different places along the basilar membrane. The LL portion
was most sensitive to suppression by a 2.07 kHz tone and implies generation at the
tonotopic place (the evoking stimulus was a 2 kHz tone-burst). This finding agrees with
several earlier studies that have measured suppression tuning curves for TEOAEs evoked
by tone-burst stimuli. Tavartkiladze et al. (1994) concluded that the TEOAE was
generated at its tonotopic place after finding that the peak of the suppression tuning curve
for a 1.5 kHz and 2.5 kHz tone-burst TEOAE was roughly 1.5 kHz and 2.5 kHz,

respectively. Similarly, Zettner and Folsom (2003) reported the peak of the suppression
tuning curve to be within 1/4-octave above the tonotopic place for a 4 kHz tone-burst
TEOAE.
Although neither of the previously mentioned studies explicitly examined
suppression of only the LL portion of the TEOAE, it is likely that this dominated their
measurements. Both investigators used versions of the ILO system (Otodynamics Ltd.),
which eliminates the initial 2.5 ms of the TEOAE and then windows the subsequent 2.5
ms to reduce contamination from stimulus artifact (Kemp et al., 1986). This type of
windowing would likely eliminate the SL2 portion and reduce the magnitude of the SL1

 
  97  

portion. Zettner and Folsom (2003) also used relatively low-level tone-bursts, 40 and 50
dB SPL, likely evoking a TEOAE with dominant LL energy (see Fig. 4.6). It is difficult
to judge the dB SPL of the stimuli used by Tavartkiladze et al. (1994) as they were
calibrated in terms of dB sensation level (SL).
Both the SL1 and SL2 portions of the TEOAE appear to be generated basally to
the tonotopic place. The SL1 portion was most sensitive to suppression at 2.55 kHz
corresponding to a generation region between 1/4 - 1/3-octave basal to the CF location of
the evoking tone-burst’s center frequency (2 kHz). The SL2 portion was most sensitive to
suppression at 3.14 kHz corresponding to a generation region 3/5-octave basal to the CF
location. Negligible contributions to the SL portions of the TEOAE from regions beyond
1-octave basal to the CF location were observed. This conflicts with several studies that
have suggested a much more dispersed region of generation extending to the cochlear
base (Guinan, 1990; Siegel et al., 2005).
The influence of the SL portions of the TEOAE on the suppression pattern for the
total TEOAE (ALL) was evidenced by a shift in the peak of the pattern from 2.07 kHz to
3.14 kHz as the suppressor level decreased 24 dB (see Fig. 4.9). This represents a shift in
the region most sensitive to suppression from the CF place to 3/5-octave basal and is
consistent with the dominant generation region of the TEOAE changing. Keefe et al.

(2008) noticed a similar effect in SFOAE suppressor tuning curves when the SFOAE was
evoked by higher-level probe tones. While that finding may indicate a basal shift in the
generation region of the SFOAE, the investigators cautioned that the effect was
potentially a result of the suppressor tone inducing a region of distortion at its
characteristic frequency place, basal to that of the probe tone (see Siegel et al., 2004;
Shera et al., 2004). Accordingly, it is important to consider whether the suppressor tones
used in the current study simply induced distortion as opposed to suppressing the actual
TEOAE. As mentioned in the methods portion of the paper, each suppressor tone/noise
was paired with an anti-phase version of itself. This pairing was done so that, upon

 
  98  

averaging, any distortion-source OAEs generated by the nonlinear interaction between


the suppressor and probe would cancel. An additional step taken to limit potential
confounding effects from the suppressor-probe interaction was that suppression was only
analyzed across time windows that contained SL TEOAE energy. These time windows
were determined from the measurement condition where the probe and suppressor were
identical (both were tone-bursts). Accordingly, the results are likely valid indicators of
the suppression caused by the suppressor on the TEOAE.
4.4.3 Generation Mechanisms of the TEOAE
The LL portion of the TEOAE, originating from the tonotopic place, is consistent
with generation through linear coherent reflection (Zweig and Shera, 1995; Shera and
Guinan, 1999; Kalluri and Shera, 2007) whereby a portion of the forward traveling
cochlear wave is reflected off pre-existing impedance irregularities in the basilar
membrane mechanics. Backscattering of the traveling wave presumably occurs across the
length of the BM basal to the CF location; however, the shape of the traveling wave
induces a dominant region of coherent reflection at its peak. At low stimulus levels where
the gain of the cochlear amplifier is greatest, the peak of the traveling wave is
approximately at the CF place (Ruggero et al., 1997) and generates a TEOAE with a
long-latency.

The presence of the SL portion of the TEOAE at moderate and high stimulus
levels has traditionally been interpreted as being a byproduct of nonlinear
intermodulation distortion induced by the various frequency components in the evoking
stimulus (Yates and Withnell, 1999). Similar to a distortion-source OAE, the phase
gradient of the SL TEOAE rotates at a much slower rate across frequency compared to
that of reflection-source OAEs (Knight and Kemp, 2000; Withnell and McKinley, 2005;
Withnell et al., 2008). Furthermore, the less compressive growth of the SL portion
relative to the LL portion is similar to that observed between the distortion-source and
reflection-source DPOAE components (Mauermann & Kollmeier, 2004; Abdala et al.,

 
  99  

2011). However, linear coherent reflection may similarly account for the generation of
the SL portion of the TEOAE. As stimulus level increases, the peak of the traveling wave
shifts basally and may induce either a new region of coherent reflection or broaden the
region over which coherent reflection occurs (Zweig and Shera, 1995; Konrad-Martin
and Keefe, 2005). The forward travel time of the cochlear wave to these basal regions
would be expected to be shorter than that to the tonotopic place as would the reverse
travel time associated with any reflections. Accordingly, the latency of OAEs arising
through a basal shift in the traveling wave peak would be expected to be shorter than that
of the OAE localized to the tonotopic place. Moreover, the magnitude of the resulting
OAE might be expected to be greater than that of the OAE generated at the tonotopic
place simply because the peak response has shifted basally.
Further efforts to understand the mechanism(s) responsible for the SL portion of
the TEOAE have relied on comparisons between cochlear model predictions of latency
and empirical data. Moleti et al. (2012a) hypothesized that if the SL portion (specifically,
the SL1 portion) was a 2 f1 ! f2 distortion-source OAE (generated at the f2 place), its
latency would be ~3/5 that of the latency for the LL portion generated at the tonotopic
place. Their predictions were based on latency calculations from a linear transmission-
line cochlear model. Subsequent comparisons between the 1.5 kHz SL and LL portions of

the TEOAE supported their hypothesis: The mean latency of the LL portion was ~12 ms
compared to ~7 ms for the SL portion. The current study’s results demonstrate a
comparable ratio between the latencies of the SL1 and LL TEOAES; however, a
generation region corresponding to the f2 place was not supported. Instead of occurring
at 3.1 kHz (the f2 place for a 2 kHz distortion-source OAE; f2 f1 = 1.22 ), generation
was localized to the 2.55 kHz region.
More recent modeling efforts by Moleti et al. (2013) agree with the current
study’s empirical estimate of the generation location for the SL1 portion of the TEOAE.
Using a nonlinear cochlear model, Moleti et al. examined the generation regions and

 
  100  

mechanisms of the SL and LL portions of the TEOAE by selectively adding/removing


roughness along the cochlear partition. As discussed earlier, the roughness along the
cochlear partition causes backscattering of the forward moving cochlear wave thereby
resulting in the generation of OAEs (Zweig and Shera, 1995; Shera and Guinan, 1999).
Model simulations indicated that the SL1 portion was generated approximately 1/4-
octave basal to the tonotopic place through linear reflection.
The measured generation region of the SL2 portion of the TEOAE (~3.1 kHz) is
similar to that expected for a 2 kHz, 2 f1 ! f2 distortion-source OAE. However, the SL2
latency relative to that of the LL seems too short to be generated through intermodulation
distortion. The latency of the SL2 TEOAE was 1/3 that of the LL TEAOE as opposed to
the expected difference of 3/5 (Moleti et al., 2012a). Unfortunately, the results from
Moleti et al. (2013) do not provide direct insight into the generation mechanism
responsible for the SL2 portion of the TEOAE since they did not explicitly note the
presence of this portion in either their simulation or empirical data. The reason for the
lack of the SL2 portion in their data is unclear but could be due to different stimulus
levels and/or analysis techniques. Considering the SL1 portion of the TEOAE may be
explained through linear reflection, it is possible that this same mechanism is responsible
for the SL2 portion.

4.4.4 Relation of TEOAEs to BM Mechanics


4.4.4.1 Cochlear Compression
The growth rates of the different-temporal portions of the TEOAE may reflect the
underlying nonlinear growth of the BM at corresponding locations: The LL portion
corresponding to the CF place, the SL1 portion corresponding to 1/4 - 1/3-octave basal to
CF, and the SL2 portion corresponding to 3/5-octave basal to CF. To examine this
potential relationship, BM response slope-intensity functions from chinchilla (as reported
in Ruggero et al., 1997; Rhode & Recio, 2000; Rhode et al., 2007) were derived and
compared to those for the different portions of the TEOAE (from Fig. 4.6). Each of the

 
  101  

studies reporting BM responses measured BM IO functions to different frequency pure


tones at a fixed measurement location near the base of the cochlea (CF between 9 – 12
kHz). These data were fit using 2nd-order polynomials of identical form to those used to
fit the TEOAE IO functions in order to compare with the data from the current study. The
resulting fits were differentiated to yield slope-intensity functions.
Although each set of BM data was measured at a single, fixed location on the
BM, by assuming a nearly scaling symmetric cochlea (Rhode, 1971; Zweig, 1976), the
responses can be understood in terms of how regions at and basal to a particular CF
location respond to the CF tone. For instance, when the 10 kHz place is stimulated by a 7
kHz tone, the resulting IO or slope-intensity function is that which would be expected at
the region 1/2-octave basal to the 10 kHz place, when a 10 kHz tone is presented. By
interpreting the data in this manner, TEOAE and BM IO and slope-intensity functions
from approximately proportional places on the BM (relative to the CF place) can be
compared. Accordingly, each panel of Fig. 4.10 shows the slope-intensity functions from
a given region along the basilar membrane, relative to the CF location. The top-panel
shows responses at the CF location, the middle-panel shows responses from 1/8 – 1/3-
octave basal to the CF location, and the bottom-panel shows responses from 1/3 – 3/4-
octaves basal to the CF location. Overlaid on the BM responses in each panel is the

slope-intensity function of the TEOAE portion originating from a proportional region as


the BM data.
The bottom panel of Fig. 4.10 shows BM slope-intensity functions measured at
the CF location in response to a pure tone at the CF. Growth becomes increasingly
compressive as stimulus level increases. Depending on the particular study, growth rates
range from approximately 0.25 – 0.5 dB/dB at the lowest stimulus levels and decrease to
anywhere between 0 – 0.05 dB/dB by 90 dB SPL. Compared to these mechanical
measurements of cochlear compression, the growth of the LL TEOAE is similar;

 
  102  

however, the TEOAE exhibits greater compression than the BM data at stimulus levels
above 70 dB SPL.
Compared to cochlear compression at the CF location when stimulated by a tone
at CF, the growth of the response to lower frequency tones is more linear (Fig. 4.10,
middle and top panels). The growth rate of the SL1 portion of the TEOAE, generated
approximately 1/3-octave basal to the CF place, is within the range expected of the BM
response at similar locations (middle-panel). Similarly, the growth of the SL2 portion of
the TEOAE, generated approximately 3/5-octave basal to the CF place, approximates BM
growth at comparable regions (top-panel).
4.4.4.2 Basal Shift in the Peak of the Traveling Wave
The peak BM response for a particular frequency of stimulation shifts basally as
stimulus level increases. The total shift in the traveling wave peak has been measured to
be in the vicinity of 1/2-octave basal to the characteristic frequency place (Ruggero et al.,
1997; Ren and Nuttall, 2001). The IO functions for the different portions of the TEOAE
suggest that the basal shift of the traveling wave peak can be determined from the
TEOAE. Figure 4.11 (top-left panel) re-plots the IO functions of the different-temporal
portions of the TEOAE from Fig. 4.6, as well as BM IO functions (other panels)
measured in chinchilla. The BM data comes from the same studies considered in Fig.

4.10. Each plot of Fig. 4.11 shows the actual data points (indicated by the markers)
reported by the corresponding investigators as well as the 2nd-order polynomial fits (lines)
used to derive the slope-intensity functions of Fig. 4.10. Markers have been chosen to
indicate BM responses coming from proportional regions, relative to the CF place, as the
different portions of the TEOAE.
Considering first the TEOAE data (Fig. 4.11, top-left panel), the LL portion of the
TEOAE, generated at the CF place, is largest in magnitude when stimulus level is below
60 dB pSPL. Above 60 dB pSPL, the peak response shifts to the SL1 portion, which is
generated, between 1/4 - 1/3-octave basal to the CF place. As stimulus level increases

 
  103  

beyond 74 dB pSPL, the SL2 portion has the largest magnitude. This component is
generated approximately 3/5-octave basal to the tonotopic place. If these magnitudes
intimate the magnitudes of the BM response at different spatial locations, the traveling
wave peak shifts basally from the tonotopic place by 3/5-octaves.
The BM data show similar shifts in the location that corresponds to the largest
response. Data from the 10 kHz location, reported by Ruggero et al. (1997; top panel),
demonstrate basal shifts, relative to the CF place (10 kHz), of ~1/3-octave (10 kHz to 8
kHz) and ~1/2-octave (10 kHz to 7 kHz) at 60 dB SPL and 80 dB SPL, respectively. The
data from Rhode and Recio (2000) show the peak response shifting basally by ~1/4-
octave (relative to the CF place) at approximately 55 dB SPL. No further shift is evident
as stimulus level increases beyond 55 dB SPL; however, this could be due to the limited
upper-range of stimulus levels tested by the investigators. Had higher levels been tested,
it is probable that the peak response would have shifted even more basally. For instance,
at a stimulus level of 70 dB SPL, the response to the 8 kHz tone (representing ~3/5-
octave basal to the CF place) is nearly equal to that of the 10 kHz tone (1/4-octave basasl
to the CF place) but is growing less compressively. The data from Rhode (2007; bottom-
right panel) measured at the 9.1 kHz place also show a basal shift occurring at
approximately 45 dB SPL. In this case, the peak response shifts by ~1/4-octave (9.1 kHz

to 7.7 kHz). Since 7.7 kHz was the only lower-frequency tone presented to the 9.1 kHz
place, insight into any additional basal shift that might have occurred is not possible.
4.4.5 Implications for Ears with SNHL
Relevant to the finding of a basally generated portion of the TEOAE is a body of
work examining the effect of basal cochlear trauma on lower-frequency TEOAEs. Avan
et al. (1995) and Withnell et al. (2000) measured a decrease in TEOAE magnitude at
frequencies presumably unaffected by noise-induced cochlear trauma, in guinea pig.
Similar decreases in magnitude have been measured in humans with high-frequency
sensorineural hearing loss (SNHL) compared to normal hearing controls (Lucertini et al.,

 
  104  

2002). These findings are seemingly consistent with the TEOAE being composed of
contributions from both the tonotopic place and more basal regions. Presumably, basal
trauma would eliminate the basal contributions to the TEOAE and cause a decrease in
overall TEOAE magnitude.
Conflicting evidence surrounds the effect of hearing loss (whether restricted to
basal regions or not) on TEOAE latency. If the cochlea is imagined as a series of
minimum phase filters (Zweig, 1976) then the filter delay incurred by the cochlear
traveling wave as it approaches the tontopic location should be proportional to the
bandwidth of the underlying cochlear filter (Shera et al., 2002; Sisto et al., 2007; Shera et
al., 2010). When the gain of the cochlear amplifier is decreased due to damage to the
outer hair cells, the bandwidths of the cochlear filters effectively increase (Sellick et al.,
1982; Khanna & Leonard, 1986; Don et al., 1998; Robles & Ruggero, 2001).
Accordingly, TEOAE latency is expected to decrease in the presence of SNHL. Such a
decrease has been found (Avan et al., 1993; Konrad-Martin and Keefe, 2005; Keefe,
2012); however, studies have also shown either no change (Prieve et al., 1996) or an
increase in TEOAE latency (Lucertini et al., 2002).
The conflicting results concerning the effect of hearing loss on TEOAE latency
may be due, in part, to a differential effect of SNHL on the different portions of the

TEOAE. Typically, damage is greater at more basal cochlear regions. If the damage
across the generation region of the SL portion of the TEOAE is greater than that at the
generation region of the LL portion, the contribution to the TEOAE from the SL
generation place might be considerably reduced compared to the contribution from the
LL place. This would result in an increase in TEOAE latency (Lucertini et al., 2002).
Alternatively, if the SL portion of the TEOAE is less dependent on the integrity of the
cochlear amplifier compared to the LL portion of the TEOAE and damage to the
generation regions of these different portions of the TEOAE is similar, the contribution
from the LL portion to the TEOAE would be more greatly reduced than that of the SL

 
  105  

portion. This would result in a decrease in TEOAE latency (Avan et al., 1993; Konrad-
Martin & Keefe, 2005; Keefe, 2012). Mertes and Goodman (2013) recently examined the
predictive capacity of the different-latency portions of TEOAE for a given frequency as
they relate to behavioral threshold at the same frequency. The magnitude of the LL
portion was more strongly correlated with behavioral threshold suggesting a closer link to
the integrity of the cochlear amplified than that of the SL portion.
Finally, different analysis techniques between studies may also partially explain
inconsistent findings of the effect SNHL has on TEOAE latency (i.e., Prieve et al., 1996
and Lucertini et al., 2002). If the initial milliseconds of the TEOAE are discarded, as is
often the case in an effort to eliminate stimulus artifact, the SL portion of the TEOAE
will be lost. As a result, its contribution to TEOAE latency will be reduced and that of the
LL portion of the TEOAE will dictate the latency of the TEOAE. Since the latency of the
LL portion of the TEOAE is level-invariant, TEOAE latency may not show any
difference between ears with SNHL and ears without SNHL. Obviously, further study is
warranted concerning the effect of cochlear damage on the generation of the different
latency components and how (of if) they might beneficial in a clinical setting.

 
  106  

s1 s2 s1,2

probe Ch. 1
Amplitude (Normalized)

suppressor ï a Ch. 2

suppressor ï b Ch. 2

suppressor ï c Ch. 2

0 20 40 60 80 100 120 140


Time (ms)

Figure 4.1. Probe- and suppressor-stimulus paradigm used to evoke the suppressed
TEOAE. Each panel includes 3 stimulus intervals ( s1 , s2 , and s1,2 ). During s1 , only the
probe tone-burst stimulus is presented through channel 1 of the transducer. During s2 ,
only the suppressor stimulus is presented through channel 2. During s1,2 the probe and
suppressor stimuli are simultaneously presented through channel 1 and 2, respectively.
The same 3-cycle, 2 kHz tone-burst probe stimulus was used for all suppressor conditions
(1st panel). Three different suppressor stimuli were used: 3-cycle, 2 kHz tone-burst (2nd
panel), 28 ms pure-tone (3rd panel), and 28 ms narrow-band noise-burst (4th panel). Note
that for the tone-burst suppressor condition, each stimulus interval was 40 ms as opposed
to the 50 ms shown in the figure (see text for additional details).

 
  107  

1.2
SL2 SL1 LL
1 05L
Mag. (Normalized)

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10
Time (ms)

Figure 4.2. Illustration of the technique used to specify the time-windows corresponding
to different-temporal portions of the TEOAE. Gray lines indicate the TEOAE envelopes
for the tone-burst suppressor condition (thicker lines indicate higher stimulus levels). The
solid black line is the mean TEOAE envelope averaged across all stimulus levels. The
dashed lines delineate the time-windows for the different-latency TEOAEs.

 
  108  

SL2 SL1 LL SL2 SL1 LL SL2 SL1 LL


1 01L 05L 10R

Magnitude (Normalized) 0.8

0.6

0.4

0.2

SL2 SL1 LL SL2 SL1 LL SL2 SL1 LL


1 11R 14L 19L
Magnitude (Normalized)

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
Time (ms)

Figure 4.3. TEOAE envelopes measured for different level tone-bursts in the tone-burst
suppressor condition. Each panel shows data from an individual subject. Line thickness is
proportional to stimulus level (44 – 86 dB pSPL in 6 dB steps). The shaded gray regions
represent each subject’s noise floor. The different-temporal portions of the TEOAE are
delineated by the dashed lines and indicated along the top of each panel.

 
  109  

SL2 SL1 LL SL1 LL


1 06L 12R
Magnitude (Normalized)
0.8

0.6

0.4

0.2

SL3 SL2 SL1 LL SL3 SL2 SL1 LL


1 17R 18R
Magnitude (Normalized)

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 0 2 4 6 8 10
Time (ms)

Figure 4.4. TEOAE envelopes measured for different level tone-bursts in the tone-burst
suppressor condition. Format is the same as in Fig. 4.3.

 
  110  

40
LL
35 SL1
SL2
30
Mag. (dB, Normalized)

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Time (ms)

Figure 4.5. Level-series illustration of the different-temporal portions of the TEOAE for
the tone-burst suppressor condition, collapsed across all subjects. Each line represents the
magnitude of either the LL, SL1, or SL2 portion of the TEOAE (indicated by marker
type) across stimulus levels. In this figure, the SL1 portion of the TEOAE is represented
by an open triangle; however, a filled triangle is used in all subsequent figures.

 
  111  

10

dB SPL 0
ALL
ï10 LL
SL1
SL2
ï20

1.5
Slope (dB/dB)

0.5

10

8
Latency (ms)

0
40 50 60 70 80 90
Stim. Level (dB pSPL)

Figure 4.6. Mean input-output, slope-intensity, and latency-intensity functions for the
different-temporal portions of the TEOAE and the total TEOAE. The top panel shows the
input-output functions of the LL, SL1, and SL2 portions of the TEOAE (indicated by
marker type) when averaged across all subjects. The gray line is the input-output function
of the total TEOAE (ALL). The open symbols indicate data points where the SNR was
less than 6 dB. The middle panel shows the slopes of the input-output functions versus
stimulus level. The bottom panel shows the latency of the different TEOAEs as a
function of stimulus level.

 
  112  

BLN
6 SL2 SL1 LL SL2 SL1 LL SL2 SL1 LL
5 01L 05L 10R
4
Suppressor (kHz)
3

1
BLN
6 SL2 SL1 LL SL2 SL1 LL SL2 SL1 LL
5 11R 14L 19L
4
Suppressor (kHz)

1
0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
Time (ms)

Figure 4.7. Contour plots showing the effect of different frequency suppressors on the
TEOAE. Data is shown for the pure tone suppressor condition (12 dB PSR) and the
noise-burst suppressor condition. Each panel shows the data for an individual subject
(panels correspond with those of Fig. 4.3). The darkness of each contour is proportional
to the amount of suppression caused by the suppressor on the TEOAE. The different-
temporal portions of the TEOAE are delineated by the dashed lines and are indicated
along the top of each panel.

 
  113  

BLN
6 SL2 SL1 LL SL1 LL
5 06L 12R
Suppressor (kHz) 4

1
BLN
6 SL3 SL2 SL1 LL SL3 SL2 SL1 LL
5 17R 18R
4
Suppressor (kHz)

1
0 2 4 6 8 10 0 2 4 6 8 10
Time (ms)

Figure 4.8. Contour plots showing the effect of different frequency suppressors on the
TEOAE. Data is shown for the remaining 4 subjects (each panel corresponds to those of
Fig. 4.4).

 
  114  

5
SL2 SL1 LL ALL

Magnitude (dB SPL) 0

ï5

ï10

ï15

ï20
0.5 1 2 4 8 0.5 1 2 4 8 0.5 1 2 4 8 0.5 1 2 4 8
Suppressor Frequency (kHz)

Figure 4.9. Mean iso-level suppression curves for the different-latency portions of the
TEOAE and the total TEOAE. Each panel shows the suppression patterns, averaged
across subjects, at each PSR condition for a given temporal portion of the TEOAE
(indicated in the upper right corner). Line thickness corresponds to the level of the
suppressor tone and is thus inversely proportional to the PSR condition (i.e., thickest line
is 0 dB PSR). The vertical thin line corresponds to the peak suppressor frequency for the
24 dB PSR condition. The broken line in each plot indicates the mean noise floor
averaged across the different PSR conditions.

 
  115  

1.5

7 kHz
1 6 kHz
9 kHz
0.5 8 kHz
SL2
TEOAE: SL2
0 BM: 1/3 ï 3/4 oct. basal to CF

1.5 TEOAE: SL1


BM: <= 1/3 oct. basal to CF 9 kHz
Slope (dB/dB)

1 8 kHz
11 kHz
10 kHz
0.5 7 kHz
SL1
0

1.5 TEOAE: LL
BM: CF
1 10 kHz
12 kHz
9 kHz
0.5 LL

0
30 40 50 60 70 80 90 100
Stim. Level (dB SPL)

Figure 4.10. Slope-intensity functions for the SL2, SL1, and LL portions of the TEOAE
compared to those derived from measurements at proportional locations on the BM. Line
style denotes the source of the data: solid black – TEOAE data from current study, dash-
dash gray – BM data from Ruggero et al., (1997), dot-dash gray – BM data from Rhode
and Recio (2000), dot-dot – BM data from Rhode (2007). The bottom panel shows the
slope-intensity functions for the LL portion of the TEOAE and BM responses at different
CF locations to the CF tone. The middle panel shows the slope-intensity functions for the
SL1 portion of the TEOAE and BM responses at different CF locations to tones 1/8 –
1/3-octave lower than the CF. The top panel shows the slope-intensity functions for the
SL2 portion of the TEOAE and BM responses at different CF locations to tones 1/3 –
3/4-octave lower than the CF.

 
  116  

LL 0 10 kHz
0
SL1 ï10 8
SL2 7
dB (Normalized) ï20
ï10
ï30

ï40
ï20
ï50
TEOAE IO BM IO: CF = 10 kHz
Current Study ï60 Ruggero et al., 1997
ï30
ï70
45 55 65 75 85 30 40 50 60 70 80 90

0 12 kHz 0 9.1 kHz


10 7.7
ï10 8
ï10
dB (Normalized)

ï20

ï30 ï20

ï40
ï30
ï50 BM IO: CF = 12 kHz BM IO: CF = 9.1 kHz
Rhode & Recio, 2000 Rhode, 2007
ï60 ï40

20 30 40 50 60 70 20 30 40 50 60 70 80 90
Stimulus Level (dB SPL) Stimulus Level (dB SPL)

Figure 4.11. Input-output functions for the LL, SL1, and SL2 portions of the TEOAE
compared to those at proportional locations on the basilar membrane. The top-left panel
shows the TEOAE input-output functions (this is the same data shown in Fig. 4.6).
Marker type indicates the temporal portion of the TEOAE. The remaining panels show
input-output functions measured in chinchilla at fixed locations on the basilar membrane,
in response to different frequency pure-tones (indicated by marker and line type), as
reported in the literature. Markers were chosen to identify the BM response(s) that
approximated the generation locations of the different-latency portions of the TEOAE
(relative to the CF place). For instance, the 7 kHz response can be interpreted as the
response approximately 1/2-octave basal to the CF location; since the SL2 portion of the
TEOAE is generated in this same region, they share the similar marker types.

 
  117  

CHAPTER V
SUMMARY AND CONCLUSIONS

5.1 SL and LL Portions of the TEOAE


5.1.1 Growth and Latency Behavior
Transient evoked otoacoustic emissions (TEOAEs) contain both short-latency
(SL) and long-latency (LL) portions. The SL portion occurs earlier in time than the LL
portion by a factor of roughly 3/5 (Goodman et al., 2009; Moleti et al., 2012a). The
latencies of both portions are invariant with stimulus level (Goodman et al., 2011). The
growth rates of the two portions are also different: The SL portion grows less
compressively than the LL portion.
Traditionally, the different latency portions of the TEOAE have not been
individually studied. Rather, analysis techniques have utilized broad time windows that
span the temporal locations of both the SL and LL portions. This type of analysis results
in the measured growth of the overall TEOAE falling somewhere between the growth
rates of the SL and LL portions. For instance, typical growth rates for TEOAEs analyzed
across broad time windows fall within 0.3 – 0.5 dB/dB (Kemp and Chum, 1980; Probst et
al., 1986; Verhulst et al., 2011). In contrast, the growth rates for the LL portion alone are

much more compressive (as low as 0.1 dB/dB; Goodman et al., 2011) and the growth
rates for the SL portion alone are much less compressive (up to 1 dB/dB). Analysis of the
TEOAE across a broad time window also accounts for the oft-cited level-dependency of
TEOAE latency: latency decreases as stimulus level increases (Neely et al., 1988;
Tognola et al., 1997; Sisto & Moleti, 2007; Rasetshwane et al., 2013). Recall that the
latencies of the different TEOAE portions do not change with level (Goodman et al.,
2009; 2011; Moleti et al., 2012a); however, each portion’s magnitude is strongly level
dependent. At low stimulus levels, the magnitude of the LL portion is larger than that of
the SL portion; however, as stimulus level increases, the magnitude of the SL portion

 
  118  

quickly exceeds that of the LL portion, due to the former’s less compressive growth.
When overall latency is calculated as the energy-weighted mean of the time-window (as
is often the case), the greatest weight is attributed to the moment in time corresponding to
the largest magnitude peak in the overall TEOAE, resulting in a latency shift with
increasing stimulus level.
5.1.2 Proposed Mechanisms of Generation
There is general consensus that at low levels, the TEOAE is generated through
linear coherent reflection (LCR), each frequency component of the TEOAE arising
primarily from its tonotopic place on the basilar membrane (BM; Tavartkiladze, et al.,
1994; Shera & Guinan, 1999; Kalluri & Shera, 2007; Moleti et al., 2013; Zettner &
Folsom, 2003;). Since the LL portion dominates the low-level TEOAE, the generation of
the LL portion may be attributed to LCR. The highly compressive growth of the LL
portion agrees well with mechanical measurements of BM compression (Ruggero et al.,
1997; Rhode, 2007) as well as psychophysical estimates (Oxenham and Plack, 1997;
Epstein and Florentine, 2005) at the tonotopic location: Growth rates across all of these
measures are between 0.1 – 0.3 dB/dB. Accordingly, the LL portion presumably mirrors
the active mechanics of the basilar membrane at the tontopic place. Not surprisingly,
sensorineural hearing loss (SNHL) results in either a loss or decrease in amplitude of the

LL portion. For this reason, the LL portion of the TEOAE has been useful in detecting
hearing loss.
The generation of the TEOAE evoked by higher-level stimuli and its relationship
to the underlying mechanics of the active cochlea are more poorly understood. Since the
TEOAE evoked by higher-level stimuli includes both LL and SL portions, the uncertainty
may be attributed mostly to poor understanding of the origin of the SL portion. The
different latencies and growth rates of the LL and SL portions suggest that they are
generated through different mechanisms and/or at different locations along the BM. Some
researchers have attributed the SL portion to 2 f1 ! f2 intermodulation distortion (Withnell

 
  119  

and McKinley, 2005; Withnell et al., 2008; Goodman et al., 2009; Moleti et al., 2012a).
Under this hypothesis, the broad spectral content of the evoking stimulus results in
frequency interactions within the nonlinear cochlea, inducing intermodulation distortion
(Yates and Withnell, 1999). In mammals, the most robust intermodulation distortion is
the 2 f1 ! f2 distortion product and, therefore, has thus been speculated to be the source of
the SL portion of the TEOAE. Both the short latency and less compressive growth rate of
the SL portion are seemingly consistent with a 2 f1 ! f2 distortion-source OAE (Moleti et
al., 2012a).
Several recent studies have cast doubt on the role of intermodulation distortion in
the generation of the SL portion of the TEOAE. First, Goodman and colleagues (2011)
reported minimal suppression of the SL portion in the presence of a high-pass noise
masker whose bandwidth overlapped with the generation region of a 2 f1 ! f2 distortion-
source OAE. Second, SL components have been measured in stimulus-frequency (SF)
OAEs (Siegel et al., 2005; Sisto et al., 2013) with comparable latencies and growth rates
to the TEOAE SL portions (Moleti et al., 2013). In contrast to the stimuli used to evoke
TEOAEs, the bandwidths of SFOAE evoking stimuli are narrow and do not include
energy at the f1 and f2 primaries required to generate a 2 f1 ! f2 distortion-source OAE.
An alternative mechanism to intermodulation distortion that has been

hypothesized to account for SL TEOAE (and SFOAE) generation is basal reflection


(Goodman et al., 2011; Moleti et al., 2013). Per this theory, the SL portion is generated
through linear coherent reflection across BM regions basal to the tonotopic place. Zweig
and Shera (1995) originally hypothesized that as the level of the OAE evoking stimulus
increases, the region of coherent reflection would extend basally as the BM mechanics
become increasingly passive. Such a shift would result in OAEs with shorter latency and
less compressive growth, consistent with fewer phase rotations of the forward traveling
cochlear wave and linearization of BM growth basal to the tonotopic place (Ruggero et
al., 1997; Rhode, 2007). Recently, nonlinear cochlear model simulations have

 
  120  

demonstrated that linear coherent reflection basal to the tonotopic place is capable of
generating SL SFOAEs (Choi et al., 2008) and TEOAE (Moleti et al., 2013) portions,
thereby lending support to the basal reflection hypothesis.
The location along the BM where the SL portion of the TEOAE is generated
likely depends on its generation mechanism. Generation through 2 f1 ! f2 intermodulation
distortion would be expected to correspond to a generation region around the tonotopic
place of the f2 primary. However, if generated through some alternative mechanism,
such as basal reflection, the contributing region is not entirely obvious. For instance,
using a model simulation of TEOAE generation, Moleti and colleagues (2013) recently
concluded that generation of the SL TEOAE occurs around approximatley 1/4-octave
basal to the tonotopic place. In contrast, the BM regions at the tail portion of the cochlear
traveling wave have also been implicated in SL TEOAE/SFOAE generation (Guinan,
1990; Siegel et al., 2004; Siegel et al., 2005; Choi et al., 2008).
5.2 Experimental Work to Determine the
Origin of the SL TEOAE
5.2.1 Experiment 1
5.2.1.1. Design and Analysis
Experiment 1 was designed to test the hypothesis that the SL portion of the

TEOAE is generated through 2 f1 ! f2 intermodulation distortion (Withnell et al., 2008;


Goodman et al., 2009; Moleti et al., 2012a). TEOAEs were evoked by 2 kHz tone-bursts
of progressively longer durations: 3-, 6-, 12-, and 24-cycles. With each doubling in tone-
burst duration, the stimulus energy at the requisite frequencies for a 2 kHz DPOAE (

f1 = 2.5 kHz, f2 = 3.1 kHz) decreased by approximately 19 dB. Accordingly, upon each

doubling in stimulus duration, the input to any cubic intermodulation distortion generator
would decrease by 19 dB and the resulting SL TEOAE, if generated by intermodulation
distortion, would be expected to rapidly diminish in magnitude. In contrast to the
stimulus energy at 2.5 kHz and 3.1 kHz, energy at 2 kHz (tone-burst center frequency)

 
  121  

was constant across the different duration tone-bursts. If the SL TEOAE component is
dependent on the tone-burst center frequency, as opposed to off-center frequencies ( f1
and f2 ), then it should persist as tone-burst duration increases.
To evaluate the effect of tone-burst duration on the SL portion of the TEOAE,
three analysis strategies were utilized. First, the TEOAE waveform envelopes evoked by
the different duration tone-bursts were compared to each other and examined for the
presence of SL energy. Second, the latency-intensity functions of the TEOAEs were
compared across the different duration tone-bursts for evidence of similar dependency of
TEOAE latency on tone-burst level. As discussed above, when a broad time window is
used to analyze the bandpass filtered TEOAE, latency decreases with increasing stimulus
level due to the increasing magnitude of the SL portion (Sisto and Moleti, 2007;
Goodman et al., 2011; Moleti et al., 2012a; Rasetshwane et al., 2013). In the absence of
the SL portion, latency would not be expected to decrease since only the LL portion,
whose latency does not change with stimulus level, would be present. Accordingly, if
generation of the SL portion is dependent on stimulus energy at f1 and f2 , the latency of
the TEOAEs evoked by the longer-duration tone-bursts should be minimally dependent
on stimulus level. Third, input-output functions of TEOAE magnitudes evoked by the
different tone-burst durations were compared for evidence of similar compressive growth

rates. If the SL portion were generated through intermodulation distortion, its


contribution to the total TEOAE would systematically decrease as tone-burst duration
increased. Accordingly, the TEOAEs evoked by the longer-duration tone-bursts would
exhibit more compressive growth than the TEOAEs evoked by the shorter-duration tone-
bursts. Similarity between the TEOAE envelopes, latency-intensity functions, and input-
output functions across the different tone-burst durations would implicate a mechanism
other than intermodulation distortion, such as basal reflection, in the generation of the SL
portion.

 
  122  

5.2.1.2. Results
TEOAE envelopes included identifiable SL and LL portions at all tone-burst
durations. The SL portions grew less compressively than the LL portions. Within each
subject, the latencies of the SL and LL portions were both level- and bandwidth-
invariant, meaning that they did not change with either stimulus level or stimulus
bandwidth (duration). Within each subject, the TEOAE latency-intensity functions were
similar across the different tone-burst durations. In certain cases, the latency of the
longest duration tone-burst was slightly longer than the other shorter duration tone-bursts,
resulting in a statistically significant effect of tone-burst duration on TEOAE latency.
However, the dependency of latency on stimulus level was not significant across the
different tone-burst durations. Latency decreased at a rate of ~0.11 ms per dB increase in
stimulus level. The different tone-burst duration input-output functions for a given
subject were nearly identical. Neither tone-burst duration nor the interaction between
tone-burst duration and stimulus level significantly influenced TEOAE magnitude.
TEOAE magnitude grew at a rate of approximately 0.45 dB per dB increase in stimulus
level.
5.2.1.3. Discussion
Results demonstrate that intermodulation distortion is not the primary generation

mechanism of the SL portion of the TEOAE in humans. As such, the current findings are
in disagreement with previous studies implicating 2 f1 ! f2 distortion in the generation of
the SL portion in humans (Withnell et al., 2008; Goodman et al., 2009; Moleti et al.,
2012a). None of these studies directly tested the hypothesis that the SL portion is a

2 f1 ! f2 distortion-source OAE, however. Rather, the investigators assumed a role of

cubic intermodulation distortion given the latency and growth rate of the TEOAE coupled
with the broad bandwidth of the evoking click stimulus. As discussed in earlier chapters,
short latency and less compressive growth may also occur as a result of basal linear
coherent reflection (Goodman et al., 2011; Moleti et al., 2013).

 
  123  

The similarity between latencies and magnitudes across the different duration
tone-burst OAEs implies a shared generation mechanism. Commonality between the LL
portion of the TEOAE and low-level SFOAEs has previously been established (Kalluri
and Shera, 2007). The current findings support such commonality but also generalize it to
apply to all components of the TEOAE and SFOAE across a wide range of stimulus
levels. In other words, despite their temporal dissimilarity, the TEOAE and SFOAE
appear to be generated in the same way, regardless of stimulus level.
5.2.2. Experiment 2
5.2.2.1. Design and Analysis
Experiment 2 was designed to estimate the generation locations of the different
latency portions of the TEOAE along the BM. TEOAEs were evoked by 3-cycle, 2 kHz
tone-bursts in the presence of various suppressor stimuli spanning a frequency range from
0.5 – 16 kHz. It was assumed that the suppressor frequency that caused the greatest
reduction of the TEOAE amplitude was closest to the region of generation. Previous
studies examining the generation region of the TEOAE using suppressors have focused
only on the LL portion (Tavartkiladze, et al., 1994; Zettner & Folsom, 2003).
To determine the approximate region of generation for each portion of the
TEOAE, time windows corresponding to the temporal locations of the SL and LL

portions were first identified from a control condition. For computational purposes, the
SL portion was further sub-divided into a shorter-latency (SL2) and longer-latency (SL1)
portion. This sub-division was performed individually for each subject based on the
location of magnitude peaks in the TEOAE waveform envelopes. The magnitude of the
suppressed TEOAE within each time-window (SL2, SL1, and LL) was then calculated
for each of the different suppressor frequencies. A larger magnitude corresponded to a
greater two-tone suppressive effect caused by the suppressor. The generation region was
assumed to correspond to the suppressor frequency which caused the greatest decrease in
TEOAE magnitude.

 
  124  

5.2.2.2. Results
The LL portion was most sensitive to suppression by a 2.07 kHz tone while the
SL1 and SL2 components were most sensitive to suppression by 2.55 and 3.14 kHz tones,
respectively. Suppressors above 4 kHz generally induced minimal suppression of the
TEOAE.
5.2.2.3. Discussion
The different latency portions of the TEOAE are generated at different locations
along the BM. Consistent with previous two-tone suppression studies (Tavartkiladze, et
al., 1994; Zettner & Folsom, 2003) and estimates based on LL TEOAE latency (Kalluri
and Shera, 2007), generation of the LL portion was found to occur near its tonotopic
location. Both of the SL portions (SL1 and SL2) were generated basal to the tonotopic
location: The SL1 portion was generated approximately 1/3-octave basal while the SL2
portion was generated approximately 3/5-octave basal. Contributions from the very base
of the cochlea, corresponding to the tail regions of the cochlear traveling wave (Guinan,
1990; Siegel et al., 2005), were minimal.
The current study’s empirical estimate of the generation place of the SL1 portion
is comparable to a recent estimate derived from a nonlinear cochlear model simulation of
TEOAE generation (Moleti et al., 2013). In their model, Moleti and colleagues

systematically “turned off” impedance discontinuities at basilar membrane regions basal


to the tonotopic place. They found the greatest contribution to the SL portion came from
the region approximately 1/4-octave basal to the tonotopic place, which is similar to the
empirical estimate obtained in the current study. However, the current study also
estimated the generation place of an even earlier portion of the TEOAE (SL2), finding it
to be even more basal.

 
  125  

5.3. Conclusions and Future Directions


Taken together, the experimental results from this thesis suggest that TEOAEs are
generated by a linear coherent reflection mechanism, regardless of stimulus level. At low
stimulus levels, the TEOAE is dominated by reflections coming from the peak of the
traveling wave at the tonotopic place along the BM. As stimulus level increases, basal
shifts and broadening of the cochlear traveling waves result in the TEOAE becoming
dominated by coherent reflections coming from more basal regions.
While the generation mechanism and generation location of the SL portion of the
TEOAE is better understood as a result of the experiments described in this thesis,
opportunities for future research are apparent. One issue concerning the different latency
TEOAE components is their relationship to underlying basilar membrane mechanics. It
was hypothesized during the discussion of Experiment 2 that the different latency
portions of the TEOAE assay cochlear mechanics, namely, basilar membrane growth, at
each component’s generation region. Comparisons between the growth rates and relative
magnitudes of the different portions of the TEOAE to analogous mechanical
measurements of the BM response support this hypothesis (see Figs. 4.10 and 4.11). In
animals, this hypothesis might be more fully tested through comparing TEOAEs to
mechanical measurements of the BM response (see Rhode, 2007).

While it was touched on in Experiment 1, the origin of the temporal fine structure
of the TEOAE deserves further study. Temporal fine structure, as used in this paper,
referred to the presence of magnitude peaks and nulls in the TEOAE envelope. This fine
structure was used to identify the different-temporal portions of the TEOAE in
Experiment 2. Two hypotheses were presented to explain the differences in temporal fine
structure across subjects as well as across tone-burst durations (see Chapter 4:
Experiment 1). Recall that the TEOAE envelopes for several subjects were more
continuous in nature (reduced temporal fine structure) compared to other subjects where
TEOAE envelopes were punctuated by distinct peaks and nulls (increased temporal fine

 
  126  

structure). One hypothesis suggests that TEOAE morphology is a byproduct of temporal


overlap between distinct packets of similar frequency TEOAE energy. The resolution of
individual peaks and nulls in the envelope will depend on the latencies and phases of
these different components. Alternatively, TEOAE envelope morphology may be due to
wave-interference between different frequency TEOAE components. The morphology of
the envelope would thus depend on cochlear reflectivity or, the underlying roughness
pattern along the basilar membrane that causes back-scattering of the forward traveling
cochlear wave. Neither of these mechanisms is necessarily exclusive of the other and it is
likely that both contribute to the morphology of the bandpass filtered TEOAE envelope.
A final area of study deserving further attention and that may benefit from the
current work is the effect of sensorineural hearing loss (SNHL) on TEOAE magnitude
and latency. SNHL reduces the magnitude of the TEOAE at frequencies affected by the
hearing loss. However, a handful of studies have shown that basal trauma to the cochlea,
whether induced experimentally or through natural environmental noise exposure, may
also decrease the magnitude of lower-frequency OAEs where normal hearing is preserved
(Avan et al., 1995; 1997; Withnell et al., 2000; Lucertini et al., 2002). In light of the
current findings that implicate basal regions in the generation of the SL portion of the
TEOAE, such a finding is not necessarily surprising. Future work investigating the effect

of SNHL on TEOAE magnitude may benefit from examining the different temporal
portions of the TEOAE.
In general, the effect of SNHL on TEOAE latency is poorly understood. In as
much as latency is related to the bandwidth of the cochlear filters (Zweig, 1976), latency
would be expected to decrease in the presence of SNHL since the gain of the cochlear
amplifier is reduced. Several studies have measured a decrease (Konrad-Martin & Keefe,
2005; Keefe, 2012); however, other studies have reported either no change (Prieve et al.,
1996) or an increase in latency (Lucertini et al., 2002) as a result of SNHL. It is possible

 
  127  

that these seemingly contradictory findings may be explained through a differential effect
of the hearing loss on the different temporal portions of the TEOAE.

 
  128  

REFERENCES

Abdala, C. Maturation of the human cochlear amplifier: Distortion product otoacoustic


emission suppression tuning curves recorded at low and high primary levels. J Acoust
Soc Am, 2001, 110(3), 1465-1476.

Abdala, C.; Dhar, S.; Kalluri, R. Level dependence of distortion product otoacoustic
emission phase is attributed to component mixing. J Acoust Soc Am, 2011, 129(5),
3123-3133.

Abdala, C.; Sininger, Y. S.; Ekelid, M; Zeng, F-G. Distortion product otoacoustic
emission suppression tuning curves in human adults and neonates. Hear Res, 1996, 98,
38-53.

Avan, P.; Bonfils, P.; Loth, D.; et al. Quantitative assessment of human cochlear
function by evoked otoacoustic emissions. Hear Res, 1991, 52, 99-112.

Avan, P.; Bonfils, P.; Loth, D.; et al. Transient-evoked otoacoustic emissions and high-
frequency acoustic trauma in the guinea pig. J Acoust Soc Am, 1995, 97(5), 3012-3020.

Avan, P.; Bonfils, P.; Loth, D.; Wit, H. P. Temporal patterns of transient-evoked
otoacoustic emissions in normal and impaired cochleae. Hear Res, 1993, 70, 109-120.

Avan, P.; Elbez, M.; Bonfils, P. Click-evoked otoacoustic emissions and the influence
of high-frequency hearing losses in humans. J Acoust Soc Am, 1997, 101(5), 2771-
2777.

Békésy, G. von. Description of some mechanical properties of the organ of Corti. J


Acoust Soc Am, 1953, 25(4), 770-785.

Békésy, G. von. On the resonance curve and the decay period at various points on the
cochler partition. J Acoust Soc Am, 1949, 21(3), 245-254.

Békésy, G. von. The variation in phase along the basilar membrane with sinusoidal
vibrations. J Acoust Soc Am, 1947, 19(3), 452-460.

Bian, L; Chertoff, M. E.; Miller, E. Deriving a cochlear transducer function from low-
frequency modulation of distortion product otoacoustic emissions. J Acoust Soc Am,
2002, 112(1), 198-210.

Boege, P.; Janssen, T. Pure-tone threshold estimation from extrapolated distortion


product otoacoustic emission I/O-functions in normal and cochlear hearing loss ears. J
Acoust Soc Am, 2002, 111(4), 1810-1818.

 
  129  

Boer, E. de. Mechanics of the cochlea: Modeling efforts. In The Cochlea; Dallos, P.;
Popper, A. N.; Fay, R. R., Eds.; Springer: New York, 1996; Vol. 8; 258-317.

Boer, E. de. No sharpening? A challenge for cochlear mechanics. J Acoust Soc Am,
1983a, 73(2), 567-573.

Boer, E. de. On active and passive cochlear models - Toward a generalized analysis. J
Acoust Soc Am, 1983b, 73(2), 574-576.

Brass, D.; Kemp, D. T. Suppression of stimulus frequency otoacoustic emissions. J


Acoust Soc Am, 1993, 93(2), 920-939.

Brownell, W. E.; Bader, C. R.; Bertrand, D.; Ribaupierre, Y. de. Evoked mechanical
responses of isolated cochlear outer hair cells. Science, 1985, 227, 194-196.

Camalet, S.; Duke, T.; Jülicher, F.; Prost, J. Auditory sensitivity provided by self-tuned
critical oscillations of hair cells. Proc Natl Acad Sci, 2000, 97(7), 3183-3188.

Carvalho, S.; Büki, B.; Bonfils, P.; Avan, P. Effect of click intensity on click-evoked
otoacoustic emission waveforms: Implications for the origin of emissions. Hear Res,
2003, 175, 215-225.

Cheatham, M. A.; Huynh, K. H.; Gao, J; et al. Cochlear function in Prestin knockout
mice. J Physiol, 2004, 560(3), 821-830.

Choi, Y-S.; Lee, S-Y.; Parham, K.; Neely, S. T.; Kim, D. O. Stimulus-frequency
otoacoustic emission: Measurements in humans and simulations with an active cochlear
model. J Acoust Soc Am, 2008, 123(5), 2651-2669.

Collet, L.; Gartner, M.; Moulin, A.; et al. Evoked otoacoustic emissions and
sensorineural hearing loss. Arch Otolaryngol Head Neck Surg, 1989, 115, 1060-1062.

Collet, L.; Veuillet, E.; Chanal, J. M.; Morgon, A. Evoked otoacoustic emissions:
Correlates between spectrum analysis and audiogram. Int J Audiol, 1991, 30(3), 164-
172.

Cooper, N. P.; Rhode, W. S. Basilar mechanics in the hook region of cat and guinea pig
cochlea: Sharp tuning and nonlinearity in the absence of baseline position shifts. Hear
Res, 1992, 63, 163-190.

Dallos, P.; Wu, X.; Cheatham, M. A.; et al. Prestin-based outer hair cell motility is
necessary for mammalian cochlear amplification. Neuron, 2008, 58, 333-339.

Davis, H. An active process in cochlear mechanics. Hear Res, 1983, 9, 79-90.

Dhar, S.; Long, G. R.; Talmadge, C. L.; Tubis, A. The effect of stimulus-frequency

 
  130  

ratio on distortion product otoacoustic emission components. J Acoust Soc Am, 2005,
117(6), 3766-3776.

Don, M.; Eggermont, J. J. Analysis of the click-evoked brainstem potentials in man


using high-pass noise masking. J Acoust Soc Am, 1978, 63(4), 1084-1092.

Don, M.; Ponton, C. W.; Eggermont, J. J.; Kwong, B. The effects of sensory hearing
loss on cochlear filter times estimated from auditory brainstem response latencies. J
Acoust Soc Am, 1998, 104(4), 2280-2289.

Dorn, P. A.; Konrad-Martin, D.; Neely, S. T.; et al. Distortion product otoacoustic
emission input/output functions in normal-hearing and hearing-impaired human ears. J
Acoust Soc Am, 2001, 110(6), 3119-3131.

Dorn, P. A.; Piskorski, P.; Gorga, M. P.; et al. Predicting audiometric status from
distortion product otoacoustic emissions using multivariate analyses. Ear Hear, 1999,
20(2), 149-163.

Epstein, M.; Florentine, M. Inferring basilar-membrane motion from tone-burst


otoacoustic emissions and psychoacoustic measurements. J Acoust Soc Am, 2005,
117(1), 263-274.

Evans, B. N.; Dallos, P. Stereocilia displacement induced somatic motility of cochlear


outer hair cells. Proc Natl Acad Sci USA, 1993, 90, 8347-8351.

Evans, E. F. The frequency response and other properties of single fibres in the
abnormal cochlea. J Physiol, 1972, 226, 263-287.

Evans, E. F. The sharpening of cochlear frequency selectivity in the normal and


abnormal cochlea. Audiology, 1975, 14, 419-444.

Fitzgerald, T. S.; Prieve, B. A. Detection of hearing loss using 2f2-f1 and 2f1-f2
distortion-product otoacoustic emissions. J Sp Lang Hear Res, 2005, 48, 1165-1186.

Gaskill, S. A.; Brown, A. M. The behavior of the acoustic distortion product, 2f1-f2,
from the human ear and its relation to auditory sensitivity. J Acoust Soc Am, 1990,
88(2), 821-839.

Geisler, C. D. From Sound to Synapse: Physiology of the Mammalian Ear, 1st ed;
Oxford University Press: New York, 1998.

Geisler, C. D.; Nuttall, A. L. Two-tone suppression of basilar membrane vibrations in


the base of the guinea pig cochlea using "low-side" suppressors. J Acoust Soc Am,
1997, 102(1), 430-440.

Geisler, C. D.; Yates, G. K.; Patuzzi, R. B.; Johnstone, B. M. Saturation of outer hair

 
  131  

cell receptor currents causes two-tone suppression. Hear Res, 1990, 44, 241-256.

Gilman, S.; Dirks, D. D. Acoustics of ear canal measurement of eardrum SPL in


simulators. J Acoust Soc Am, 1986, 80(3), 783-793.

Gold, T. Hearing. II. The physical basis of the action of the cochlea. Proc R Soc
London B, 1948, 135, 492-498.

Gold, T.; Pumphrey, R. J. Hearing. I. The cochlea as a frequency analyzer. Proc R Soc
London B, 1948, 135, 462-491.

Goodman S. S.; Withnell, R. H.; Shera, C. A. The origin of SFOAE microstructure in


the guinea pig. Hear Res, 2003, 183, 7-17.

Goodman, S. S.; Fitzpatrick, D. F.; Ellison, J. C.; et al. High-frequency click-evoked


otoacoustic emissions and behavioral thresholds in humans. J Acoust Soc Am, 2009,
125(2), 1014-1032.

Goodman, S. S.; Mertes, I. B.; Scheperle, R. A. Delays and growth rates of multiple
TEOAE components. In What Fire is in Mine Ears: Progress in Auditory
Biomechanics; AIP Conf Proc, 2011, 1403, 279-285.

Gorga, M. P.; Neely, S. T.; Dierking, D. M.; Dorn, P. A.; Hoover, B. M.; Fitzpatrick,
D. F. Distortion product otoacoustic emission suppression tuning curves in normal-
hearing and hearing-impaired human ears. J Acoust Soc Am, 2003, 114(1), 263-278.

Gorga, M. P.; Neely, S. T.; Dierking, D. M.; et al. Distortion product otoacoustic
emission suppression tuning curves in normal-hearing and hearing-impaired human
ears. J Acoust Soc Am, 2003, 114(1), 263-278.

Gorga, M. P.; Neely, S. T.; Dorn, P. A.; Hoover, B. M. Further efforts to predict pure-
tone thresholds from distortion product otoacoustic emission input/output functions. J
Acoust Soc Am, 2003, 113(6), 3275-3284.

Gorga, M. P.; Neely, S. T.; Kopun, J; Tan, H. Distortion-product otoacoustic emission


suppression tuning curves in humans. J Acoust Soc Am, 2011, 129(2), 817-827.

Grandori, F. Nonlinear phenomena in click- and tone-burst-evoked otoacoustic


emissions from human ears. Audiology, 1985, 24, 71-80.

Greenwood, D. D. A cochlear frequency-position function for several species - 29 years


later. J Acoust Soc Am, 1990, 87(6), 2592-2605.

Greenwood, D. D. Critical bandwidth and frequency coordinates of the basilar


membrane. J Acoust Soc Am, 1961, 33(10), 1344-1356.

 
  132  

Guinan, J. J. Changes in stimulus frequency otoacoustic emissions produced by two-


tone suppression and efferent stimulation in cat. In The Mechanics and Biophysics of
Hearing; Dallos, P.; Geisler, C. D.; Matthews, J.; Ruggero, M. A.; Steele, C., Eds.;
Springer-Verlag: Madison, 1990; 170-177.

Hall, J. L. Two-tone distortion products in a nonlinear model of the basilar membrane.


J Acoust Soc Am, 1974, 56(6), 1818-1828.

Harte, J. M.; Pigasse, G.; Dau, T. Comparison of cochlear delay estimates using
otoacoustic emissions and auditory brainstem responses. J Acoust Soc Am, 2009,
126(3), 1291-1301.

Heitmann, J.; Waldmann, B.; Schnitlzer, H-U.; et al. Suppression of distortion product
otoacoustic emissions (DPOAE) near 2f1-f2 removes DP-gram fine structure -
Evidence for a secondary generator. J Acoust Soc Am, 1998, 103(3), 1527-1531.

Hoaglin, D.; Mosteller, F.; Tukey, J. W. Understanding Robust and Exploratory Data
Analysis. Wiley: New York, 1983.

Hubbard, A. A traveling-wave amplifier model of the cochlea. Science, 1993, 259, 68-
71.

Hubbard, A. E.; Mountain, D. C. Alternating current delivered into the scala media
alters sound pressure at the eardrum. Science, 1983, 222, 510-512.

Hussain, D. M.; Gorga, M. P.; Neely, S. T.; et al. Transient evoked otoacoustic
emissions in patients with normal hearing and in patients with hearing loss. Ear Hear,
1998, 19(6), 434-449.

Iwasa, K. H.; Chadwick, R. S. Elasticity and active force generation of cochlear outer
hair cells. J Acoust Soc Am, 1992, 92(6), 3169-3173.

Jedrzejczak, W. W.; Blinowska, K. J.; Konopka, W. Time-frequency analysis of


transiently evoked otoacoustic emissions of subjects exposed to noise. Hear Res, 2005,
205, 249-255.

Johnson, T. A.; Neely, S. T.; Kopun, J. G.; et al. Distortion product otoacoustic
emissions: Cochlear-source contributions and clinical test performance. J Acoust Soc
Am, 2007, 122(6), 3539-3553.

Johnstone, B. M.; Patuzzi, R.; Yates, G. K. Basilar membrane measurements and the
travelling wave. Hear Res, 1986, 22, 147-153.

Kalluri, R.; Shera, C. A. Distortion-product source unmixing: A test of the two-


mechanism model for DPOAE generation. J Acoust Soc Am, 2001, 109(2), 622-637.

 
  133  

Kalluri, R.; Shera, C. A. Near equivalence of human click-evoked and stimulus-


frequency otoacoustic emissions. J Acoust Soc Am, 2007, 121(4), 2097-2110.

Kalluri, R.; Shera, C. A. Measuring stimulus-frequency otoacoustic emissions using


swept tones. J Acoust Soc Am, 2013, 134(1), 356-368.

Keefe, D. H. Double-evoked otoacoustic emissions. I. Measurement theory and


nonlinear coherence. J Acoust Soc Am, 1998, 103(6), 3489-3498.

Keefe, D. H. Moments of click-evoked otoacoustic emissions in human ears: Group


delay and spread, instantaneous frequency and bandwidth. J Acoust Soc Am, 2012,
132(5), 3319-3350.

Keefe, D. H.; Ellison, J. C.; Fitzpatrick, D. F.; Gorga, M. P. Two-tone suppression of


stimulus frequency otoacoustic emissions. J Acoust Soc Am, 2008, 123(3), 1479-1494.

Keefe, D. H.; Ling, R. Double-evoked otoacoustic emissions. II. Intermittent noise


rejection, calibration and ear-canal measurements. J Acoust Soc Am, 1998, 103(6),
3499-3508.

Keefe, D. H.; Schairer, K. S. Specification of absorbed-sound power in the ear canal:


Application to suppression of stimulus frequency otoacoustic emissions. J Acoust Soc
Am, 2011, 129(2), 779-791.

Kemp, D. T. Evidence of mechanical nonlinearity and frequency selective wave


amplification in the cochlea. Arch Oto-Rhino-Laryngol, 1979a, 224, 37-45.

Kemp, D. T. Otoacoustic emissions, travelling waves and cochlear mechanisms. Hear


Res, 1986, 22, 95-104.

Kemp, D. T. Stimulated acoustic emissions from the human auditory system. J Acoust
Soc Am, 1978, 64(5), 1386-1391.

Kemp, D. T. The evoked cochlear mechanical response and the auditory microstructure
- Evidence for a new element in cochlear mechanics. Scand Audiol Suppl, 1979b, 9, 35-
47.

Kemp, D. T.; Bray, P.; Alexander, L.; Brown, A. M. Acoustic emission cochleography
- practical aspects. Scand Audiol Suppl, 1986, 25, 71-95

Kemp, D. T.; Chum, R. A. Properties of the generator of stimululated otoacoustic


emissions. Hear Res, 1980, 2, 213-232.

Kemp, D. T.; Ryan, S.; Bray, P. A guide to the effective use of otoacoustic emissions.
Ear Hear, 1990, 11, 93-105.

 
  134  

Khanna, S. M.; Leonard, D. G. B. Relationship between basilar membrane tuning and


hair cell condition. Hear Res, 1986, 23, 55-70.

Killan, E. C.; Lutman, M. E.; Montelpare, W. J.; Thyer, N. J. A mechanism for


simultaneous suppression of tone burst-evoked otoacoustic emissions. Hear Res, 2012,
285, 58-64.

Knight, R. D.; Kemp, D. T. Indications of different distortion product otoacoustic


emission mechanisms from a detailed f1, f2 area study. J Acoust Soc Am, 2000, 107(1),
457-473.

Knight, R. D.; Kemp, D. T. Relationships between DPOAE and TEOAE amplitude and
phase characteristics. J Acoust Soc Am, 1999, 106(3), 1420-1435.

Knight, R. D.; Kemp, D. T. Wave and place fixed DPOAE maps of the human ear. J
Acoust Soc Am, 2001, 109(4), 1513-1525.

Konrad-Martin, D.; Keefe, D. H. Time-frequency analyses of transient-evoked


stimulus-frequency and distortion-product otoacoustic emissions: Testing cochlear
model predictions. J Acoust Soc Am, 2003, 114(4), 2021-2043.

Konrad-Martin, D.; Keefe, D. H. Transient-evoked stimulus-frequency and distortion-


product otoacoustic emissions in normal and impaired ears. J Acoust Soc Am, 2005,
117(6), 3799-3815.

Konrad-Martin, D.; Neely, S. T; Keefe, D. H.; et al. Sources of distortion product


otoacoustic emissions revealed by suppression experiments and inverse fast Fourier
transforms in normal ears. J Acoust Soc Am, 2001, 109(6), 2862-2879.

Kros, C. T.; Rusch, A.; Richardson, G. P. Mechano-electrical transducer currents in


hair cells of the cultured neonatal mouse cochlea. Proc R Soc London B, 1992, 249,
185-193.

Lewis, J. D.; Goodman, S. S. The generation mechanism and location of early TEOAE
components. Poster presented at the American Auditory Society Annual Meeting, 2013,
89 Phys07.

Liberman, M. C.; Gao, J.; He, D. Z. Z.; et al. Prestin is required for electromotility of
the outer hair cell and for the cochlear amplifier. Nature, 2002, 419(19), 300-304.

Lichtenhan, J. T. Effects of low-frequency biasing on otoacoustic and neural measures


suggest that stimulus-frequency otoacoustic emissions originate near the peak region of
the traveling wave. J Assoc Res Otolaryngol, 2012, 13(1), 17-28.

Long, G. R.; Talmadge, C. L.; Lee, J. Measuring distortion product otoacoustic


emissions using continuously sweeping primaries. J Acoust Soc Am, 2008, 124(3),

 
  135  

1613-1626.

Lucertini, M.; Moleti, A.; Sisto, R. On the detection of early cochlear damage by
otoacoustic emission analysis. J Acoust Soc Am, 2002, 111(2), 972-978.

Lukashkin, A. N.; Russell, I. J. A descriptive model of the receptor potential


nonlinearities generated by the hair cell mechanoelectrical transducer. J Acoust Soc Am,
1998, 103(2), 973-980.

Lukashkin, A. N.; Russell, I. J. Analysis of the f2-f1 and 2f1-f2 distortion components
generated by the hair cell mechanoelectrical transducer: Dependence on the amplitudes
of the primaries and feedback gain. J Acoust Soc Am, 1999, 106(5), 2661-2668.

Maat, B.; Wit, H. P.; Dijk, P van. Noise-evoked otoacoustic emissions in humans. J
Acoust Soc Am, 2000, 108(5), 2272-2280.

Martin, G. K.; Jassir, D.; Stagner, B. B.; et al. Locus of generation for the 2f1-f2 vs 2f2-
f1 distortion-product otoacoustic emissions in normal-hearing humans revealed by
suppression tuning, onset latencies, and amplitude correlations. J Acoust Soc Am, 1998,
103(4), 1957-1971.

Martin, G. K.; Lonsbury-Martin, B. L.; Probst, T.; et al. Acoustic distortion products in
rabbit ear canal. II. Sites of origin revealed by suppression contours and pure-tone
exposures. Hear Res, 1987, 28, 191-208.

Martin, G. K.; Probst, R.; Lonsbury-Martin, B. L. Otoacoustic emissions in human ears:


Normative findings. Ear Hear, 1990, 11(2), 106-120.

Martin, G. K.; Stagner, B. B.; Lonsbury-Martin, B. L. Time-domain demonstration of


distributed distortion-product otoacoustic emission components. J Acoust Soc Am,
2013, 134(1), 342-355.

Martin, P.; Hudspeth, A. J.; Jülicher, F. Comparison of a hair bundle's spontaneous


oscillations with its response to mechanical stimulation reveals the underlying active
process. Proc Natl Acad Sci, 2001, 98(25), 14380-14385.

Mauermann, M.; Kollmeier, B. Distortion product otoacoustic emission (DPOAE)


input/output functions and the influence of the second DPOAE source. J Acoust Soc
Am, 2004, 116(4), 2199-2212.

Mauermann, M.; Uppenkamp, S.; Hengel, P. W. J. van.; Kollmeier, B. Evidence for the
distortion product frequency place as a source of distortion product otoacoustic
emission (DPOAE) fine structure in humans. I. Fine structure and higher-order DPOAE
as a function of the frequency ratio f2/f1. J Acoust Soc Am, 1999, 106(6), 3473-3483.

Mertes, I. B.; Goodman, S. S. Short-latency transient-evoked otoacoustic emissions as

 
  136  

predictors of hearing status and thresholds. J Acoust Soc Am, 2013, 134(3), 2127-2135.

Moleti, A.; Al-Maamury, A. M.; Bertaccini, D.; et al. Generation place of the long- and
short-latency components of transient-evoked otoacoustic emissions in a nonlinear
cochlear model. J Acoust Soc Am, 2013, 133(6), 4098-4108.

Moleti, A.; Botti, T.; Sisto, R. Transient-evoked otoacoustic emission generators in a


nonlinear cochlea. J Acoust Soc Am, 2012, 131(4), 2891-2903.

Moleti, A.; Longo, F.; Sisto, R. Time-frequency domain filtering of evoked otoacoustic
emissions. J Acoust Soc Am, 2012, 132(4), 2455-2467.

Moleti, A.; Sisto, R. Objective estimates of cochlear tuning by otoacoustic emission


analysis. J Acoust Soc Am, 2003, 113(1), 423-429.

Neely, S. T.; Gorga, M. P. Comparison between intensity and pressure as measures of


sound level in the ear canal. J Acoust Soc Am, 1998, 104(5), 2925-2934.

Neely, S. T.; Kim, D. O. An active cochlear model showing sharp tuning and high
sensitivity. Hear Res, 1983, 9, 123-130

Neely, S. T.; Kim, D. O. Cochlear models incorporating active processes. In Active


Processes and Otoacoustic Emissions; Manley, G. A.; Fay, R.R.; Popper, A. N., Eds.;
Springer: New York; Vol. 30; 381-394.

Neely, S. T.; Norton, S. J.; Gorga, M. P.; Jestaedt, W. Latency of auditory brain-stem
responses and otoacoustic emissions using tone-burst stimuli. J Acoust Soc Am, 1988,
83(2), 652-656.

Neely, S. T.; Stenfelt, S.; Schairer, K. S. Alternative ear-canal measures related to


absorbance. Ear Hear, 2013, 34(S1), 72-77.

Neumann, J.; Uppenkamp, S.; Kollmeier, B. Chirp evoked otoacousitc emissions. Hear
Res, 1994, 79, 17-25.

Norton, S. J.; Neely, S. T. Tone-burst-evoked otoacoustic emissions from normal-


hearing subjects. J Acoust Soc Am, 1987, 81(6), 1860-1872.

Oxenham, A. J.; Bacon, S. P. Cochlear compression: Perceptual measures and


implications for normal and impaired hearing. Ear Hear, 2003, 24(5), 353-366.

Oxenham, A. J.; Plack, C. J. A behavioral measure of basilar-membrane nonlinearity in


listeners with normal and impaired hearing. J Acoust Soc Am, 1997, 101(6), 3666-3675.

Patuzzi, R. Cochlear mechanics and macromechanics. In The Cochlea; Dallos, P.;


Popper, A. N.; Fay, R. R., Eds.; Springer: New York, 1996; Vol. 8; 186-257.

 
  137  

Patuzzi, R.; Robertson, D. Tuning in the mamallian cochlea. Phyiol Rev, 1988, 68(4),
1009-1082.

Patuzzi, R.; Sellick, P. M.; Johnstone, B. M. The modulation of the sensitivity of the
mammalian cochlea by low frequency tones. III. Basilar membrane motion. Hear Res,
1984, 13, 19-28.

Penner, M. J.; Zhang, T. Prevalence of spontaneous otoacoustic emissions in adults


revisited. Hear Res, 1997, 103, 28-34.

Prieve, B. A.; Gorga, M. P.; Neely, S. T. Click- and tone-burst-evoked otoacoustic


emissions in normal-hearing and hearing-impaired ears. J Acoust Soc Am, 1996, 99(5),
3077-3086.

Probst, R.; Coats, A. C.; Martin, G. K.; Lonsbury-Martin, B. L. Spontaneous, click-,


and toneburst-evoked otoacoustic emissions from normal ears. Hear Res, 1986, 21,
261-275.

Probst, R.; Lonsbury-Martin, B. L.; Martin, G. K.; Coats, A. C. Otoacoustic emissions


in ears with hearing loss. Am J Otolaryngol, 1987, 8, 73-81.

Rasetshwane, D. M.; Argenyi, M.; Neely, S. T.; et al. Latency of tone-burst-evoked


auditory brain stem responses and otoacoustic emissions: Level, frequency, and rise-
time effects. J Acoust Soc Am, 2013, 133(5), 2803-2817.

Recio, A.; Rhode, W. S. Basilar membrane responses to broadband stimuli. J Acoust


Soc Am, 2000, 108(5), 2281-2298.

Recio, A.; Rich, N. C.; Narayan, S. S.; Ruggero, M. A. Basilar-membrane responses to


clicks at the base of the chinchilla cochlea. J Acoust Soc Am, 1998, 103(4), 1972-1989.

Ren, T. Longitudinal pattern of basilar membrane vibration in the sensitive cochlea. J


Neurosci, 2002, 99, 17101-17106.

Ren, T.; Nuttall, A. L. Basilar membrane vibration in the basal turn of the sensitive
gerbil cochlea. Hear Res, 2001, 151, 48-60.

Ren, T.; Nuttall, A. L. Extracochlear electrically evoked otoacoustic emissions: A


model for in vivo assessment of outer hair cell motility. Hear Res, 1996, 92, 178-183.

Rhode, W. S. Basilar membrane mechanics in the 6-9 kHz region of sensitive chinchilla
cochleae. J Acoust Soc Am, 2007, 121(5), 2792-2804.

Rhode, W. S. Distortion product otoacoustic emissions and basilar membrane vibration


in the 6-9 kHz region of sensitive chinchilla cochleae. J Acoust Soc Am, 2007, 122(5),

 
  138  

2725-2737.

Rhode, W. S. Observations of the vibration of the basilar membrane in squirrel


monkeys using the Mössbauer technique. J Acoust Soc Am, 1971, 49(4), 1218-1231.

Rhode, W. S. Some observations on cochlear mechanics. J Acoust Soc Am, 1978, 64(1),
158-176.

Rhode, W. S.; Recio, A. Study of mechanical motions in the basal region of the
chinchilla cochlea. J Acoust Soc Am, 2000, 107(6), 3317-3332.

Robles, L.; Ruggero, M. A. Mechanics of the mammalian cochlea. Physiol Rev, 2001,
81(3), 1305-1352.

Robles, L.; Ruggero, M. A.; Rich, N. C. Basilar membrane mechanics at the base of the
chinchilla cochlea. I. Input-output functions, tuning curves, and response phases. J
Acoust Soc Am, 1986, 80(5), 1364-1374.

Robles, L.; Ruggero, M. A.; Rich, N. C. Two-tone distortion in the basilar membrane of
the cochlea. Nature, 1991, 349, 413-414.

Ruggero, M. A. Responses to sound of the basilar membrane of the mammalian


cochlea. Curr Opin Neurobiol, 1992, 2, 449-456.

Ruggero, M. A.; Rich, N. C.; Recio, A.; et al. Basilar-membrane responses to tones at
the base of the chinchilla cochlea. J Acoust Soc Am, 1997, 101(4), 2151-2163.

Ruggero, M. A.; Robles, L.; Rich, N. C. Two-tone suppression in the basilar membrane
of the cochlea: Mechanical basis of auditory-nerve rate suppression. J Neurophysiol,
1992, 68(4), 1087-1099.

Sachs, R. M.; Burkhard, M. D. Insert earphone pressure response in real ears and
couplers. J Acoust Soc Am, 1972, 52(1A), 183.

Santos-Sacchi, J. On the frequency limit and phase of outer hair cell motility: Effects of
the membrane filter. J Neurosci, 1992, 12(5), 1906-1916.

Santos-Sacchi, J.; Song, L; Zheng, J; Nuttall, A. L. Control of mammalian cochlear


amplification chloride anions. J Neurosci, 2006, 26(15), 3992-3998.

Schairer, K. S.; Ellison, J. C.; Fitzpatrick, D.; Keefe, D. H. Use of stimulus-frequency


otoacoustic emission latency and level to investigate cochlear mechanics in human ears.
J Acoust Soc Am, 2006, 120(2), 901-914.

Schairer, K. S.; Fitzpatrick, D.; Keefe, D. H. Input-output functions for stimulus-


frequency otoacoustic emissions in normal-hearing adult ears. J Acoust Soc Am, 2003,

 
  139  

114(2), 944-966.

Sellick, P. M.; Patuzzi, R.; Johnstone, B. M. Measurement of basilar membrane motion


in the guinea pig using the Mössbauer technique. J Acoust Soc Am, 1982, 72(1), 131-
141.

Sellick, P. M.; Patuzzi, R.; Johnstone, B. M. Modulation of responses of spiral ganglion


cells in the guinea pig cochlea by low frequency sound. Hear Res, 1982, 7, 199-221.

Shera, C. A. Frequency glides in click responses of the basilar membrane and auditory
nerve: Their scaling behavior and origin in traveling-wave dispersion. J Acoust Soc Am,
2001, 109(5), 2023-2034.

Shera, C. A. Mammalian spontaneous otoacoustic emissions are amplitude-stabilized


cochlear standing wave. J Acoust Soc Am, 2003, 114(1), 244-262.

Shera, C. A.; Guinan, J. J. Evoked otoacoustic emissions arise by two fundamentally


different mechanisms: A taxonomy for mammalian OAEs. J Acoust Soc Am, 1999,
105(2), 782-798.

Shera, C. A.; Guinan, J. J. Mechanisms of mammalian otoacoustic emission. In Active


Processes and Otoacoustic Emissions; Manley, G. A.; Fay, R. R.; Popper, A. N., Eds;
Springer: New York; 2008; 30; 305-342.

Shera, C. A.; Guinan, J. J. Stimulus-frequency-emission group delay: A test of coherent


reflection filtering and a window on cochlear tuning. J Acoust Soc Am, 2003, 113(5),
2762-2772.

Shera, C. A.; Guinan, J. J.; Oxenham, A. J. Otoacoustic estimation of cochlear tuning:


Validation in the chinchilla. J Assoc Res Otolaryngol, 2010, 11, 343-365.

Shera, C. A.; Guinan, J. J.; Oxenham, A. J. Revised estimates of human cochlear tuning
from otoacoustic and behavioral measurements. Proc Natl Acad Sci, 2002, 99(5), 3318-
3323.

Shera, C. A.; Tubis, A.; Talmadge, C. L. Testing coherent reflection in chinchilla:


Auditory-nerve responses predict stimulus-frequency emissions. J Acoust Soc Am,
2008, 124(1), 381-395.

Shera, C. A.; Tubis, A.; Talmadge, C. L.; Guinan, J. J. The dual effect of "suppressor"
tones on stimulus-frequency otoacoustic emissions. Assoc Res Otolaryngol Mid-Winter
Meeting Abstract, 2004, 27, 776.

Shera, C. A.; Bergevin, C. Obtaining reliable phase-gradient delays from otoacoustic


emission data. J Acoust Soc Am, 2012, 132(2), 927-943.

 
  140  

Shower, E. G.; Biddulph, R. Differential pitch sensitivity of the ear. J Acoust Soc Am,
1931, 3(2), 275-287.

Siegel, J. H. Ear-canal standing waves and high-frequency sound calibration using


otoacoustic emission probes. J Acoust Soc Am, 1994, 95(5), 2589-2597.

Siegel, J. H.; Cerka, A. J.; Recio-Spinoso, A.; et al. Delays of stimulus-frequency


otoacoustic emissions and cochlear vibrations contradict the theory of coherent
reflection filtering. J Acoust Soc Am, 2005, 118(4), 2434-2443.

Siegel, J. H.; Cerka, A. J.; Temchin, A. N.; Ruggero, M. Similar two-tone suppression
patterns in SFOAEs and the cochlear microphonics indicate comparable spatial
summation of underlying generators. Assoc Res Otolaryngol Mid-Winter Meeting
Abstract, 2004, 27, 365.

Siegel, J. Species differences in low-level otoacoustic emissions may be explained by


'hot regions' in the cochlea. J Acoust Soc Am, 2008, 123(5), 3852.

Sisto R.; Moleti, A. Transient evoked otoacoustic emission latency and cochlear tuning
at different stimulus levels. J Acoust Soc Am, 2007, 122(4), 2183-2190.

Sisto, R.; Moleti, A. Transient evoked otoacoustic emission input/output function and
cochlear reflectivity: Experiment and model. J Acoust Soc Am, 2008, 124(5), 2995-
3008.

Sisto, R.; Moleti, A.; Lucertini, M. Spontaneous otoacoustic emissions and relaxation
dynamics of long decay time OAEs in audiometrically normal and impaired subjects. J
Acoust Soc Am, 2001, 109(2), 638-647.

Sisto, R.; Moleti, A.; Shera, C. A. Cochlear reflectivity in transmission-line models and
otoacoustic emission characteristic time delays. J Acoust Soc Am, 2007, 122(6), 3554-
3561.

Sisto, R.; Sanjust, F.; Moleti, A. Input/output functions of different-latency components


of transient-evoked and stimulus-frequency otoacoustic emissions. J Acoust Soc Am,
2013, 133(4), 2240-2253.

Smurznyski, J.; Leonard, G.; Kim, D. O.; et al. Distortion product otoacoustic
emissions in normal and impaired adult ears. Arch Otolaryngol Head Neck Surg, 1990,
116, 1309-1316.

Stinson, M. R.; Lawton, B. W. Specification of the geometry of the human ear canal for
the prediction of sound-pressure level distribution. J Acoust Soc Am, 1989, 85(6), 2492-
2503.

Stinson, M. R.; Shaw, E. A. G.; Lawton, B. W. Estimation of acoustical energy

 
  141  

reflectance at the eardrum from measurements of pressure distribution in the human ear
canal. J Acoust Soc Am, 1982, 72(3), 766-773.

Talmadge, C. L.; Long, G. R.; Murphy, W. J.; Tubis, A. New off-line method for
detecting spontaneous otoacoustic emissions in human subjects. Hear Res, 1993, 71,
170-182.

Talmadge, C. L.; Long, G. R.; Tubis, A.; Dhar, S. Experimental confirmation of the
two-source interference model for the fine structure of distortion product otoacoustic
emissions. J Acoust Soc Am, 1999, 105(1), 275-292.

Talmadge, C. L.; Tubis, A. On modeling the connection between spontaneous and


evoked otoacoustic emissions. In Biophysics of Hair Cell Sensory Systems; Duifhuis,
H.; Horst, J. W.; Difjk, P. van; Netten, S. M. van, Eds.; World Scientific: Singapore,
1993; 25-32.

Talmadge, C. L.; Tubis, A.; Long, G. R.; Tong, C. Modeling the combined effects of
basilar membrane nonlinearity and roughness on stimulus frequency otoacoustic
emission fine structure. J Acoust Soc Am, 2000, 108(6), 2911-2932.

Tavartkiladze, G. A.; Frolenkov, G. I.; Kruglov, A. V.; Artamasov, S. V. Ipsilateral


suppression effects on transient evoked otoacoustic emission. Brit J Audiol, 1994, 28,
193-204.

Tognola, G.; Grandori, F.; Ravazzani, P. Time-frequency distributions of click-evoked


otoacoustic emissions. Hear Res, 1997, 106, 112-122.

Verhulst, S.; Dau, T.; Shera, C. A. Nonlinear time-domain cochlear model for transient
stimulation and human otoacoustic emission. J Acoust Soc Am, 2012, 132(6), 3842-
3848.

Verhulst, S.; Harte, J. M.; Dau, T. Temporal suppression of the click-evoked


otoacoustic emission level-curve. J Acoust Soc Am, 2011, 129(3), 1452-1463.

Vetešník, A.; Turcanu, D.; Dalhoff, E.; Gummer, A. W. Extraction of sources of


distortion product otoacoustic emissions by onset-decomposition. Hear Res, 2009, 256,
21-38.

Weiss, T. F.; Leong, R. A model for signal transmission in an ear having hair cells with
free-standing stereocilia. IV. Mechanoelectric transduction stage. Hear Res, 1985, 20,
175-195.

Whitehead, M. L.; McCoy, M. J.; Lonsbury-Martin, B. L.; Martin, G. K. Dependence


of distortion-product otoacoustic emissions on primary levels in normal and impaired
ears. I. Effects of decreasing L2 below L1. J Acoust Soc Am, 1995a, 97(4), 2346-2358.

 
  142  

Whitehead, M. L.; Stagner, B. B.; Martin, G. K.; Lonsbury-Martin, B. L. Visualization


of the onset of distortion-produce otoacoustic emissions, and measurement of their
latency. J Acoust Soc Am, 1996, 100(3), 1663-1679.

Whitehead, M. L.; Stagner, B. B.; McCoy, M. J.; et al. Dependence of distortion-


product otoacoustic emissions on primary levels in normal and impaired ears. II.
Asymmetry in L1, L2 space. J Acoust Soc Am, 1995b, 97(4), 2359-2377.

Wilson, J. P. Evidence for a cochlear origin for acoustic re-emissions, threshold fine-
structure and tonal tinnitus. Hear Res, 1980, 2, 233-252.

Withnell, R. H.; Hazlewood, C.; Knowlton, A. Reconciling the origin of the transient
evoked otoacoustic emission in humans. J Acoust Soc Am, 2008, 123(1), 212-221.

Withnell, R. H.; McKinley, S. Delay dependence for the origin of the nonlinear derived
transient evoked otoacoustic emission. J Acoust Soc Am, 2005, 117(1), 281-291.

Withnell, R. H.; Yates, G. K. Enhancement of the transient-evoked otoacoustic


emission produced by the addition of a pure tone in the guinea pig. J Acoust Soc Am,
1998, 104(1), 344-349.

Withnell, R. H.; Yates, G. K.; Kirk, D. L. Changes to low-frequency components of the


TEOAE following acoustic trauma to the base of the cochlea. Hear Res, 2000, 139, 1-
12.

Xu, L.; Probst, R.; Harris, F. P.; Roede, J. Peripheral analysis of frequency in human
ears revealed by tone burst evoked otoacoustic emissions. Hear Res, 1994, 74, 173-180.

Yates, G. K.; Winter, I. M.; Robertson, D. Basilar membrane nonlinearity determines


auditory nerve rate-intensity functions and cochlear dynamic range. Hear Res, 1990,
45, 203-220.

Yates, G. K.; Withnell, R. H. The role of intermodulation distortion in transient-evoked


otoacoustic emissions. Hear Res, 1999, 136, 49-64.

Zettner, E. M.; Folsom, R. C. Transient emission suppression tuning curve attributes in


relation to psychoacoustic threshold. J Acoust Soc Am, 2003, 113(4), 2031-2041.

Zhen, J.; Shen, W.; He, D. Z. Z.; et al. Prestin is the motor protein of cochlear outer
hair cells. Nature, 2000, 45, 149-155.

Zweig, G. Basilar membrane motion. In Cold Spring Harbor Symposia on Quantitative


Biology; Cold Spring Harbor Press; Vol. 40; 619-633.

You might also like